Show More
@@ -1,1173 +1,1343 b'' | |||||
1 | use std::collections::hash_map::RandomState; |
|
1 | use std::collections::hash_map::RandomState; | |
2 | use std::collections::HashSet; |
|
2 | use std::collections::HashSet; | |
3 | use std::fmt::Debug; |
|
3 | use std::fmt::Debug; | |
4 | use std::ops::Deref; |
|
4 | use std::ops::Deref; | |
5 | use std::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard}; |
|
5 | use std::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard}; | |
6 |
|
6 | |||
7 | use byteorder::{BigEndian, ByteOrder}; |
|
7 | use byteorder::{BigEndian, ByteOrder}; | |
8 | use bytes_cast::{unaligned, BytesCast}; |
|
8 | use bytes_cast::{unaligned, BytesCast}; | |
9 |
|
9 | |||
10 | use super::REVIDX_KNOWN_FLAGS; |
|
10 | use super::REVIDX_KNOWN_FLAGS; | |
11 | use crate::errors::HgError; |
|
11 | use crate::errors::HgError; | |
12 | use crate::node::{NODE_BYTES_LENGTH, NULL_NODE, STORED_NODE_ID_BYTES}; |
|
12 | use crate::node::{NODE_BYTES_LENGTH, NULL_NODE, STORED_NODE_ID_BYTES}; | |
13 | use crate::revlog::node::Node; |
|
13 | use crate::revlog::node::Node; | |
14 | use crate::revlog::{Revision, NULL_REVISION}; |
|
14 | use crate::revlog::{Revision, NULL_REVISION}; | |
15 | use crate::{ |
|
15 | use crate::{ | |
16 | dagops, BaseRevision, FastHashMap, Graph, GraphError, RevlogError, |
|
16 | dagops, BaseRevision, FastHashMap, Graph, GraphError, RevlogError, | |
17 | RevlogIndex, UncheckedRevision, |
|
17 | RevlogIndex, UncheckedRevision, | |
18 | }; |
|
18 | }; | |
19 |
|
19 | |||
20 | pub const INDEX_ENTRY_SIZE: usize = 64; |
|
20 | pub const INDEX_ENTRY_SIZE: usize = 64; | |
21 | pub const COMPRESSION_MODE_INLINE: u8 = 2; |
|
21 | pub const COMPRESSION_MODE_INLINE: u8 = 2; | |
22 |
|
22 | |||
|
23 | #[derive(Debug)] | |||
23 | pub struct IndexHeader { |
|
24 | pub struct IndexHeader { | |
24 | pub(super) header_bytes: [u8; 4], |
|
25 | pub(super) header_bytes: [u8; 4], | |
25 | } |
|
26 | } | |
26 |
|
27 | |||
27 | #[derive(Copy, Clone)] |
|
28 | #[derive(Copy, Clone)] | |
28 | pub struct IndexHeaderFlags { |
|
29 | pub struct IndexHeaderFlags { | |
29 | flags: u16, |
|
30 | flags: u16, | |
30 | } |
|
31 | } | |
31 |
|
32 | |||
32 | /// Corresponds to the high bits of `_format_flags` in python |
|
33 | /// Corresponds to the high bits of `_format_flags` in python | |
33 | impl IndexHeaderFlags { |
|
34 | impl IndexHeaderFlags { | |
34 | /// Corresponds to FLAG_INLINE_DATA in python |
|
35 | /// Corresponds to FLAG_INLINE_DATA in python | |
35 | pub fn is_inline(self) -> bool { |
|
36 | pub fn is_inline(self) -> bool { | |
36 | self.flags & 1 != 0 |
|
37 | self.flags & 1 != 0 | |
37 | } |
|
38 | } | |
38 | /// Corresponds to FLAG_GENERALDELTA in python |
|
39 | /// Corresponds to FLAG_GENERALDELTA in python | |
39 | pub fn uses_generaldelta(self) -> bool { |
|
40 | pub fn uses_generaldelta(self) -> bool { | |
40 | self.flags & 2 != 0 |
|
41 | self.flags & 2 != 0 | |
41 | } |
|
42 | } | |
42 | } |
|
43 | } | |
43 |
|
44 | |||
44 | /// Corresponds to the INDEX_HEADER structure, |
|
45 | /// Corresponds to the INDEX_HEADER structure, | |
45 | /// which is parsed as a `header` variable in `_loadindex` in `revlog.py` |
|
46 | /// which is parsed as a `header` variable in `_loadindex` in `revlog.py` | |
46 | impl IndexHeader { |
|
47 | impl IndexHeader { | |
47 | fn format_flags(&self) -> IndexHeaderFlags { |
|
48 | fn format_flags(&self) -> IndexHeaderFlags { | |
48 | // No "unknown flags" check here, unlike in python. Maybe there should |
|
49 | // No "unknown flags" check here, unlike in python. Maybe there should | |
49 | // be. |
|
50 | // be. | |
50 | IndexHeaderFlags { |
|
51 | IndexHeaderFlags { | |
51 | flags: BigEndian::read_u16(&self.header_bytes[0..2]), |
|
52 | flags: BigEndian::read_u16(&self.header_bytes[0..2]), | |
52 | } |
|
53 | } | |
53 | } |
|
54 | } | |
54 |
|
55 | |||
55 | /// The only revlog version currently supported by rhg. |
|
56 | /// The only revlog version currently supported by rhg. | |
56 | const REVLOGV1: u16 = 1; |
|
57 | const REVLOGV1: u16 = 1; | |
57 |
|
58 | |||
58 | /// Corresponds to `_format_version` in Python. |
|
59 | /// Corresponds to `_format_version` in Python. | |
59 | fn format_version(&self) -> u16 { |
|
60 | fn format_version(&self) -> u16 { | |
60 | BigEndian::read_u16(&self.header_bytes[2..4]) |
|
61 | BigEndian::read_u16(&self.header_bytes[2..4]) | |
61 | } |
|
62 | } | |
62 |
|
63 | |||
63 | pub fn parse(index_bytes: &[u8]) -> Result<Option<IndexHeader>, HgError> { |
|
64 | pub fn parse(index_bytes: &[u8]) -> Result<Option<IndexHeader>, HgError> { | |
64 | if index_bytes.is_empty() { |
|
65 | if index_bytes.is_empty() { | |
65 | return Ok(None); |
|
66 | return Ok(None); | |
66 | } |
|
67 | } | |
67 | if index_bytes.len() < 4 { |
|
68 | if index_bytes.len() < 4 { | |
68 | return Err(HgError::corrupted( |
|
69 | return Err(HgError::corrupted( | |
69 | "corrupted revlog: can't read the index format header", |
|
70 | "corrupted revlog: can't read the index format header", | |
70 | )); |
|
71 | )); | |
71 | } |
|
72 | } | |
72 | Ok(Some(IndexHeader { |
|
73 | Ok(Some(IndexHeader { | |
73 | header_bytes: { |
|
74 | header_bytes: { | |
74 | let bytes: [u8; 4] = |
|
75 | let bytes: [u8; 4] = | |
75 | index_bytes[0..4].try_into().expect("impossible"); |
|
76 | index_bytes[0..4].try_into().expect("impossible"); | |
76 | bytes |
|
77 | bytes | |
77 | }, |
|
78 | }, | |
78 | })) |
|
79 | })) | |
79 | } |
|
80 | } | |
80 | } |
|
81 | } | |
81 |
|
82 | |||
82 | /// Abstracts the access to the index bytes since they can be spread between |
|
83 | /// Abstracts the access to the index bytes since they can be spread between | |
83 | /// the immutable (bytes) part and the mutable (added) part if any appends |
|
84 | /// the immutable (bytes) part and the mutable (added) part if any appends | |
84 | /// happened. This makes it transparent for the callers. |
|
85 | /// happened. This makes it transparent for the callers. | |
85 | struct IndexData { |
|
86 | struct IndexData { | |
86 | /// Immutable bytes, most likely taken from disk |
|
87 | /// Immutable bytes, most likely taken from disk | |
87 | bytes: Box<dyn Deref<Target = [u8]> + Send>, |
|
88 | bytes: Box<dyn Deref<Target = [u8]> + Send>, | |
88 | /// Used when stripping index contents, keeps track of the start of the |
|
89 | /// Used when stripping index contents, keeps track of the start of the | |
89 | /// first stripped revision, which is used to give a slice of the |
|
90 | /// first stripped revision, which is used to give a slice of the | |
90 | /// `bytes` field. |
|
91 | /// `bytes` field. | |
91 | truncation: Option<usize>, |
|
92 | truncation: Option<usize>, | |
92 | /// Bytes that were added after reading the index |
|
93 | /// Bytes that were added after reading the index | |
93 | added: Vec<u8>, |
|
94 | added: Vec<u8>, | |
94 | } |
|
95 | } | |
95 |
|
96 | |||
96 | impl IndexData { |
|
97 | impl IndexData { | |
97 | pub fn new(bytes: Box<dyn Deref<Target = [u8]> + Send>) -> Self { |
|
98 | pub fn new(bytes: Box<dyn Deref<Target = [u8]> + Send>) -> Self { | |
98 | Self { |
|
99 | Self { | |
99 | bytes, |
|
100 | bytes, | |
100 | truncation: None, |
|
101 | truncation: None, | |
101 | added: vec![], |
|
102 | added: vec![], | |
102 | } |
|
103 | } | |
103 | } |
|
104 | } | |
104 |
|
105 | |||
105 | pub fn len(&self) -> usize { |
|
106 | pub fn len(&self) -> usize { | |
106 | match self.truncation { |
|
107 | match self.truncation { | |
107 | Some(truncation) => truncation + self.added.len(), |
|
108 | Some(truncation) => truncation + self.added.len(), | |
108 | None => self.bytes.len() + self.added.len(), |
|
109 | None => self.bytes.len() + self.added.len(), | |
109 | } |
|
110 | } | |
110 | } |
|
111 | } | |
111 |
|
112 | |||
112 | fn remove( |
|
113 | fn remove( | |
113 | &mut self, |
|
114 | &mut self, | |
114 | rev: Revision, |
|
115 | rev: Revision, | |
115 | offsets: Option<&[usize]>, |
|
116 | offsets: Option<&[usize]>, | |
116 | ) -> Result<(), RevlogError> { |
|
117 | ) -> Result<(), RevlogError> { | |
117 | let rev = rev.0 as usize; |
|
118 | let rev = rev.0 as usize; | |
118 | let truncation = if let Some(offsets) = offsets { |
|
119 | let truncation = if let Some(offsets) = offsets { | |
119 | offsets[rev] |
|
120 | offsets[rev] | |
120 | } else { |
|
121 | } else { | |
121 | rev * INDEX_ENTRY_SIZE |
|
122 | rev * INDEX_ENTRY_SIZE | |
122 | }; |
|
123 | }; | |
123 | if truncation < self.bytes.len() { |
|
124 | if truncation < self.bytes.len() { | |
124 | self.truncation = Some(truncation); |
|
125 | self.truncation = Some(truncation); | |
125 | self.added.clear(); |
|
126 | self.added.clear(); | |
126 | } else { |
|
127 | } else { | |
127 | self.added.truncate(truncation - self.bytes.len()); |
|
128 | self.added.truncate(truncation - self.bytes.len()); | |
128 | } |
|
129 | } | |
129 | Ok(()) |
|
130 | Ok(()) | |
130 | } |
|
131 | } | |
131 |
|
132 | |||
132 | fn is_new(&self) -> bool { |
|
133 | fn is_new(&self) -> bool { | |
133 | self.bytes.is_empty() |
|
134 | self.bytes.is_empty() | |
134 | } |
|
135 | } | |
135 | } |
|
136 | } | |
136 |
|
137 | |||
137 | impl std::ops::Index<std::ops::Range<usize>> for IndexData { |
|
138 | impl std::ops::Index<std::ops::Range<usize>> for IndexData { | |
138 | type Output = [u8]; |
|
139 | type Output = [u8]; | |
139 |
|
140 | |||
140 | fn index(&self, index: std::ops::Range<usize>) -> &Self::Output { |
|
141 | fn index(&self, index: std::ops::Range<usize>) -> &Self::Output { | |
141 | let start = index.start; |
|
142 | let start = index.start; | |
142 | let end = index.end; |
|
143 | let end = index.end; | |
143 | let immutable_len = match self.truncation { |
|
144 | let immutable_len = match self.truncation { | |
144 | Some(truncation) => truncation, |
|
145 | Some(truncation) => truncation, | |
145 | None => self.bytes.len(), |
|
146 | None => self.bytes.len(), | |
146 | }; |
|
147 | }; | |
147 | if start < immutable_len { |
|
148 | if start < immutable_len { | |
148 | if end > immutable_len { |
|
149 | if end > immutable_len { | |
149 | panic!("index data cannot span existing and added ranges"); |
|
150 | panic!("index data cannot span existing and added ranges"); | |
150 | } |
|
151 | } | |
151 | &self.bytes[index] |
|
152 | &self.bytes[index] | |
152 | } else { |
|
153 | } else { | |
153 | &self.added[start - immutable_len..end - immutable_len] |
|
154 | &self.added[start - immutable_len..end - immutable_len] | |
154 | } |
|
155 | } | |
155 | } |
|
156 | } | |
156 | } |
|
157 | } | |
157 |
|
158 | |||
158 | #[derive(Debug, PartialEq, Eq)] |
|
159 | #[derive(Debug, PartialEq, Eq)] | |
159 | pub struct RevisionDataParams { |
|
160 | pub struct RevisionDataParams { | |
160 | pub flags: u16, |
|
161 | pub flags: u16, | |
161 | pub data_offset: u64, |
|
162 | pub data_offset: u64, | |
162 | pub data_compressed_length: i32, |
|
163 | pub data_compressed_length: i32, | |
163 | pub data_uncompressed_length: i32, |
|
164 | pub data_uncompressed_length: i32, | |
164 | pub data_delta_base: i32, |
|
165 | pub data_delta_base: i32, | |
165 | pub link_rev: i32, |
|
166 | pub link_rev: i32, | |
166 | pub parent_rev_1: i32, |
|
167 | pub parent_rev_1: i32, | |
167 | pub parent_rev_2: i32, |
|
168 | pub parent_rev_2: i32, | |
168 | pub node_id: [u8; NODE_BYTES_LENGTH], |
|
169 | pub node_id: [u8; NODE_BYTES_LENGTH], | |
169 | pub _sidedata_offset: u64, |
|
170 | pub _sidedata_offset: u64, | |
170 | pub _sidedata_compressed_length: i32, |
|
171 | pub _sidedata_compressed_length: i32, | |
171 | pub data_compression_mode: u8, |
|
172 | pub data_compression_mode: u8, | |
172 | pub _sidedata_compression_mode: u8, |
|
173 | pub _sidedata_compression_mode: u8, | |
173 | pub _rank: i32, |
|
174 | pub _rank: i32, | |
174 | } |
|
175 | } | |
175 |
|
176 | |||
176 | impl Default for RevisionDataParams { |
|
177 | impl Default for RevisionDataParams { | |
177 | fn default() -> Self { |
|
178 | fn default() -> Self { | |
178 | Self { |
|
179 | Self { | |
179 | flags: 0, |
|
180 | flags: 0, | |
180 | data_offset: 0, |
|
181 | data_offset: 0, | |
181 | data_compressed_length: 0, |
|
182 | data_compressed_length: 0, | |
182 | data_uncompressed_length: 0, |
|
183 | data_uncompressed_length: 0, | |
183 | data_delta_base: -1, |
|
184 | data_delta_base: -1, | |
184 | link_rev: -1, |
|
185 | link_rev: -1, | |
185 | parent_rev_1: -1, |
|
186 | parent_rev_1: -1, | |
186 | parent_rev_2: -1, |
|
187 | parent_rev_2: -1, | |
187 | node_id: [0; NODE_BYTES_LENGTH], |
|
188 | node_id: [0; NODE_BYTES_LENGTH], | |
188 | _sidedata_offset: 0, |
|
189 | _sidedata_offset: 0, | |
189 | _sidedata_compressed_length: 0, |
|
190 | _sidedata_compressed_length: 0, | |
190 | data_compression_mode: COMPRESSION_MODE_INLINE, |
|
191 | data_compression_mode: COMPRESSION_MODE_INLINE, | |
191 | _sidedata_compression_mode: COMPRESSION_MODE_INLINE, |
|
192 | _sidedata_compression_mode: COMPRESSION_MODE_INLINE, | |
192 | _rank: -1, |
|
193 | _rank: -1, | |
193 | } |
|
194 | } | |
194 | } |
|
195 | } | |
195 | } |
|
196 | } | |
196 |
|
197 | |||
197 | #[derive(BytesCast)] |
|
198 | #[derive(BytesCast)] | |
198 | #[repr(C)] |
|
199 | #[repr(C)] | |
199 | pub struct RevisionDataV1 { |
|
200 | pub struct RevisionDataV1 { | |
200 | data_offset_or_flags: unaligned::U64Be, |
|
201 | data_offset_or_flags: unaligned::U64Be, | |
201 | data_compressed_length: unaligned::I32Be, |
|
202 | data_compressed_length: unaligned::I32Be, | |
202 | data_uncompressed_length: unaligned::I32Be, |
|
203 | data_uncompressed_length: unaligned::I32Be, | |
203 | data_delta_base: unaligned::I32Be, |
|
204 | data_delta_base: unaligned::I32Be, | |
204 | link_rev: unaligned::I32Be, |
|
205 | link_rev: unaligned::I32Be, | |
205 | parent_rev_1: unaligned::I32Be, |
|
206 | parent_rev_1: unaligned::I32Be, | |
206 | parent_rev_2: unaligned::I32Be, |
|
207 | parent_rev_2: unaligned::I32Be, | |
207 | node_id: [u8; STORED_NODE_ID_BYTES], |
|
208 | node_id: [u8; STORED_NODE_ID_BYTES], | |
208 | } |
|
209 | } | |
209 |
|
210 | |||
210 | fn _static_assert_size_of_revision_data_v1() { |
|
211 | fn _static_assert_size_of_revision_data_v1() { | |
211 | let _ = std::mem::transmute::<RevisionDataV1, [u8; 64]>; |
|
212 | let _ = std::mem::transmute::<RevisionDataV1, [u8; 64]>; | |
212 | } |
|
213 | } | |
213 |
|
214 | |||
214 | impl RevisionDataParams { |
|
215 | impl RevisionDataParams { | |
215 | pub fn validate(&self) -> Result<(), RevlogError> { |
|
216 | pub fn validate(&self) -> Result<(), RevlogError> { | |
216 | if self.flags & !REVIDX_KNOWN_FLAGS != 0 { |
|
217 | if self.flags & !REVIDX_KNOWN_FLAGS != 0 { | |
217 | return Err(RevlogError::corrupted(format!( |
|
218 | return Err(RevlogError::corrupted(format!( | |
218 | "unknown revlog index flags: {}", |
|
219 | "unknown revlog index flags: {}", | |
219 | self.flags |
|
220 | self.flags | |
220 | ))); |
|
221 | ))); | |
221 | } |
|
222 | } | |
222 | if self.data_compression_mode != COMPRESSION_MODE_INLINE { |
|
223 | if self.data_compression_mode != COMPRESSION_MODE_INLINE { | |
223 | return Err(RevlogError::corrupted(format!( |
|
224 | return Err(RevlogError::corrupted(format!( | |
224 | "invalid data compression mode: {}", |
|
225 | "invalid data compression mode: {}", | |
225 | self.data_compression_mode |
|
226 | self.data_compression_mode | |
226 | ))); |
|
227 | ))); | |
227 | } |
|
228 | } | |
228 | // FIXME isn't this only for v2 or changelog v2? |
|
229 | // FIXME isn't this only for v2 or changelog v2? | |
229 | if self._sidedata_compression_mode != COMPRESSION_MODE_INLINE { |
|
230 | if self._sidedata_compression_mode != COMPRESSION_MODE_INLINE { | |
230 | return Err(RevlogError::corrupted(format!( |
|
231 | return Err(RevlogError::corrupted(format!( | |
231 | "invalid sidedata compression mode: {}", |
|
232 | "invalid sidedata compression mode: {}", | |
232 | self._sidedata_compression_mode |
|
233 | self._sidedata_compression_mode | |
233 | ))); |
|
234 | ))); | |
234 | } |
|
235 | } | |
235 | Ok(()) |
|
236 | Ok(()) | |
236 | } |
|
237 | } | |
237 |
|
238 | |||
238 | pub fn into_v1(self) -> RevisionDataV1 { |
|
239 | pub fn into_v1(self) -> RevisionDataV1 { | |
239 | let data_offset_or_flags = self.data_offset << 16 | self.flags as u64; |
|
240 | let data_offset_or_flags = self.data_offset << 16 | self.flags as u64; | |
240 | let mut node_id = [0; STORED_NODE_ID_BYTES]; |
|
241 | let mut node_id = [0; STORED_NODE_ID_BYTES]; | |
241 | node_id[..NODE_BYTES_LENGTH].copy_from_slice(&self.node_id); |
|
242 | node_id[..NODE_BYTES_LENGTH].copy_from_slice(&self.node_id); | |
242 | RevisionDataV1 { |
|
243 | RevisionDataV1 { | |
243 | data_offset_or_flags: data_offset_or_flags.into(), |
|
244 | data_offset_or_flags: data_offset_or_flags.into(), | |
244 | data_compressed_length: self.data_compressed_length.into(), |
|
245 | data_compressed_length: self.data_compressed_length.into(), | |
245 | data_uncompressed_length: self.data_uncompressed_length.into(), |
|
246 | data_uncompressed_length: self.data_uncompressed_length.into(), | |
246 | data_delta_base: self.data_delta_base.into(), |
|
247 | data_delta_base: self.data_delta_base.into(), | |
247 | link_rev: self.link_rev.into(), |
|
248 | link_rev: self.link_rev.into(), | |
248 | parent_rev_1: self.parent_rev_1.into(), |
|
249 | parent_rev_1: self.parent_rev_1.into(), | |
249 | parent_rev_2: self.parent_rev_2.into(), |
|
250 | parent_rev_2: self.parent_rev_2.into(), | |
250 | node_id, |
|
251 | node_id, | |
251 | } |
|
252 | } | |
252 | } |
|
253 | } | |
253 | } |
|
254 | } | |
254 |
|
255 | |||
255 | /// A Revlog index |
|
256 | /// A Revlog index | |
256 | pub struct Index { |
|
257 | pub struct Index { | |
257 | bytes: IndexData, |
|
258 | bytes: IndexData, | |
258 | /// Offsets of starts of index blocks. |
|
259 | /// Offsets of starts of index blocks. | |
259 | /// Only needed when the index is interleaved with data. |
|
260 | /// Only needed when the index is interleaved with data. | |
260 | offsets: RwLock<Option<Vec<usize>>>, |
|
261 | offsets: RwLock<Option<Vec<usize>>>, | |
261 | uses_generaldelta: bool, |
|
262 | uses_generaldelta: bool, | |
262 | is_inline: bool, |
|
263 | is_inline: bool, | |
263 | /// Cache of the head revisions in this index, kept in sync. Should |
|
264 | /// Cache of the head revisions in this index, kept in sync. Should | |
264 | /// be accessed via the [`Self::head_revs`] method. |
|
265 | /// be accessed via the [`Self::head_revs`] method. | |
265 | head_revs: Vec<Revision>, |
|
266 | head_revs: Vec<Revision>, | |
266 | /// Cache of the last filtered revisions in this index, used to make sure |
|
267 | /// Cache of the last filtered revisions in this index, used to make sure | |
267 | /// we haven't changed filters when returning the cached `head_revs`. |
|
268 | /// we haven't changed filters when returning the cached `head_revs`. | |
268 | filtered_revs: HashSet<Revision>, |
|
269 | filtered_revs: HashSet<Revision>, | |
269 | } |
|
270 | } | |
270 |
|
271 | |||
271 | impl Debug for Index { |
|
272 | impl Debug for Index { | |
272 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
|
273 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { | |
273 | f.debug_struct("Index") |
|
274 | f.debug_struct("Index") | |
274 | .field("offsets", &self.offsets) |
|
275 | .field("offsets", &self.offsets) | |
275 | .field("uses_generaldelta", &self.uses_generaldelta) |
|
276 | .field("uses_generaldelta", &self.uses_generaldelta) | |
276 | .finish() |
|
277 | .finish() | |
277 | } |
|
278 | } | |
278 | } |
|
279 | } | |
279 |
|
280 | |||
280 | impl Graph for Index { |
|
281 | impl Graph for Index { | |
281 | fn parents(&self, rev: Revision) -> Result<[Revision; 2], GraphError> { |
|
282 | fn parents(&self, rev: Revision) -> Result<[Revision; 2], GraphError> { | |
282 | let err = || GraphError::ParentOutOfRange(rev); |
|
283 | let err = || GraphError::ParentOutOfRange(rev); | |
283 | match self.get_entry(rev) { |
|
284 | match self.get_entry(rev) { | |
284 | Some(entry) => { |
|
285 | Some(entry) => { | |
285 | // The C implementation checks that the parents are valid |
|
286 | // The C implementation checks that the parents are valid | |
286 | // before returning |
|
287 | // before returning | |
287 | Ok([ |
|
288 | Ok([ | |
288 | self.check_revision(entry.p1()).ok_or_else(err)?, |
|
289 | self.check_revision(entry.p1()).ok_or_else(err)?, | |
289 | self.check_revision(entry.p2()).ok_or_else(err)?, |
|
290 | self.check_revision(entry.p2()).ok_or_else(err)?, | |
290 | ]) |
|
291 | ]) | |
291 | } |
|
292 | } | |
292 | None => Ok([NULL_REVISION, NULL_REVISION]), |
|
293 | None => Ok([NULL_REVISION, NULL_REVISION]), | |
293 | } |
|
294 | } | |
294 | } |
|
295 | } | |
295 | } |
|
296 | } | |
296 |
|
297 | |||
297 | /// A cache suitable for find_snapshots |
|
298 | /// A cache suitable for find_snapshots | |
298 | /// |
|
299 | /// | |
299 | /// Logically equivalent to a mapping whose keys are [`BaseRevision`] and |
|
300 | /// Logically equivalent to a mapping whose keys are [`BaseRevision`] and | |
300 | /// values sets of [`BaseRevision`] |
|
301 | /// values sets of [`BaseRevision`] | |
301 | /// |
|
302 | /// | |
302 | /// TODO the dubious part is insisting that errors must be RevlogError |
|
303 | /// TODO the dubious part is insisting that errors must be RevlogError | |
303 | /// we would probably need to sprinkle some magic here, such as an associated |
|
304 | /// we would probably need to sprinkle some magic here, such as an associated | |
304 | /// type that would be Into<RevlogError> but even that would not be |
|
305 | /// type that would be Into<RevlogError> but even that would not be | |
305 | /// satisfactory, as errors potentially have nothing to do with the revlog. |
|
306 | /// satisfactory, as errors potentially have nothing to do with the revlog. | |
306 | pub trait SnapshotsCache { |
|
307 | pub trait SnapshotsCache { | |
307 | fn insert_for( |
|
308 | fn insert_for( | |
308 | &mut self, |
|
309 | &mut self, | |
309 | rev: BaseRevision, |
|
310 | rev: BaseRevision, | |
310 | value: BaseRevision, |
|
311 | value: BaseRevision, | |
311 | ) -> Result<(), RevlogError>; |
|
312 | ) -> Result<(), RevlogError>; | |
312 | } |
|
313 | } | |
313 |
|
314 | |||
314 | impl SnapshotsCache for FastHashMap<BaseRevision, HashSet<BaseRevision>> { |
|
315 | impl SnapshotsCache for FastHashMap<BaseRevision, HashSet<BaseRevision>> { | |
315 | fn insert_for( |
|
316 | fn insert_for( | |
316 | &mut self, |
|
317 | &mut self, | |
317 | rev: BaseRevision, |
|
318 | rev: BaseRevision, | |
318 | value: BaseRevision, |
|
319 | value: BaseRevision, | |
319 | ) -> Result<(), RevlogError> { |
|
320 | ) -> Result<(), RevlogError> { | |
320 | let all_values = self.entry(rev).or_insert_with(HashSet::new); |
|
321 | let all_values = self.entry(rev).or_insert_with(HashSet::new); | |
321 | all_values.insert(value); |
|
322 | all_values.insert(value); | |
322 | Ok(()) |
|
323 | Ok(()) | |
323 | } |
|
324 | } | |
324 | } |
|
325 | } | |
325 |
|
326 | |||
326 | impl Index { |
|
327 | impl Index { | |
327 | /// Create an index from bytes. |
|
328 | /// Create an index from bytes. | |
328 | /// Calculate the start of each entry when is_inline is true. |
|
329 | /// Calculate the start of each entry when is_inline is true. | |
329 | pub fn new( |
|
330 | pub fn new( | |
330 | bytes: Box<dyn Deref<Target = [u8]> + Send>, |
|
331 | bytes: Box<dyn Deref<Target = [u8]> + Send>, | |
331 | default_header: IndexHeader, |
|
332 | default_header: IndexHeader, | |
332 | ) -> Result<Self, HgError> { |
|
333 | ) -> Result<Self, HgError> { | |
333 | let header = |
|
334 | let header = | |
334 | IndexHeader::parse(bytes.as_ref())?.unwrap_or(default_header); |
|
335 | IndexHeader::parse(bytes.as_ref())?.unwrap_or(default_header); | |
335 |
|
336 | |||
336 | if header.format_version() != IndexHeader::REVLOGV1 { |
|
337 | if header.format_version() != IndexHeader::REVLOGV1 { | |
337 | // A proper new version should have had a repo/store |
|
338 | // A proper new version should have had a repo/store | |
338 | // requirement. |
|
339 | // requirement. | |
339 | return Err(HgError::corrupted("unsupported revlog version")); |
|
340 | return Err(HgError::corrupted("unsupported revlog version")); | |
340 | } |
|
341 | } | |
341 |
|
342 | |||
342 | // This is only correct because we know version is REVLOGV1. |
|
343 | // This is only correct because we know version is REVLOGV1. | |
343 | // In v2 we always use generaldelta, while in v0 we never use |
|
344 | // In v2 we always use generaldelta, while in v0 we never use | |
344 | // generaldelta. Similar for [is_inline] (it's only used in v1). |
|
345 | // generaldelta. Similar for [is_inline] (it's only used in v1). | |
345 | let uses_generaldelta = header.format_flags().uses_generaldelta(); |
|
346 | let uses_generaldelta = header.format_flags().uses_generaldelta(); | |
346 |
|
347 | |||
347 | if header.format_flags().is_inline() { |
|
348 | if header.format_flags().is_inline() { | |
348 | let mut offset: usize = 0; |
|
349 | let mut offset: usize = 0; | |
349 | let mut offsets = Vec::new(); |
|
350 | let mut offsets = Vec::new(); | |
350 |
|
351 | |||
351 | while offset + INDEX_ENTRY_SIZE <= bytes.len() { |
|
352 | while offset + INDEX_ENTRY_SIZE <= bytes.len() { | |
352 | offsets.push(offset); |
|
353 | offsets.push(offset); | |
353 | let end = offset + INDEX_ENTRY_SIZE; |
|
354 | let end = offset + INDEX_ENTRY_SIZE; | |
354 | let entry = IndexEntry { |
|
355 | let entry = IndexEntry { | |
355 | bytes: &bytes[offset..end], |
|
356 | bytes: &bytes[offset..end], | |
356 | offset_override: None, |
|
357 | offset_override: None, | |
357 | }; |
|
358 | }; | |
358 |
|
359 | |||
359 | offset += INDEX_ENTRY_SIZE + entry.compressed_len() as usize; |
|
360 | offset += INDEX_ENTRY_SIZE + entry.compressed_len() as usize; | |
360 | } |
|
361 | } | |
361 |
|
362 | |||
362 | if offset == bytes.len() { |
|
363 | if offset == bytes.len() { | |
363 | Ok(Self { |
|
364 | Ok(Self { | |
364 | bytes: IndexData::new(bytes), |
|
365 | bytes: IndexData::new(bytes), | |
365 | offsets: RwLock::new(Some(offsets)), |
|
366 | offsets: RwLock::new(Some(offsets)), | |
366 | uses_generaldelta, |
|
367 | uses_generaldelta, | |
367 | is_inline: true, |
|
368 | is_inline: true, | |
368 | head_revs: vec![], |
|
369 | head_revs: vec![], | |
369 | filtered_revs: HashSet::new(), |
|
370 | filtered_revs: HashSet::new(), | |
370 | }) |
|
371 | }) | |
371 | } else { |
|
372 | } else { | |
372 | Err(HgError::corrupted("unexpected inline revlog length")) |
|
373 | Err(HgError::corrupted("unexpected inline revlog length")) | |
373 | } |
|
374 | } | |
374 | } else { |
|
375 | } else { | |
375 | Ok(Self { |
|
376 | Ok(Self { | |
376 | bytes: IndexData::new(bytes), |
|
377 | bytes: IndexData::new(bytes), | |
377 | offsets: RwLock::new(None), |
|
378 | offsets: RwLock::new(None), | |
378 | uses_generaldelta, |
|
379 | uses_generaldelta, | |
379 | is_inline: false, |
|
380 | is_inline: false, | |
380 | head_revs: vec![], |
|
381 | head_revs: vec![], | |
381 | filtered_revs: HashSet::new(), |
|
382 | filtered_revs: HashSet::new(), | |
382 | }) |
|
383 | }) | |
383 | } |
|
384 | } | |
384 | } |
|
385 | } | |
385 |
|
386 | |||
386 | pub fn uses_generaldelta(&self) -> bool { |
|
387 | pub fn uses_generaldelta(&self) -> bool { | |
387 | self.uses_generaldelta |
|
388 | self.uses_generaldelta | |
388 | } |
|
389 | } | |
389 |
|
390 | |||
390 | /// Value of the inline flag. |
|
391 | /// Value of the inline flag. | |
391 | pub fn is_inline(&self) -> bool { |
|
392 | pub fn is_inline(&self) -> bool { | |
392 | self.is_inline |
|
393 | self.is_inline | |
393 | } |
|
394 | } | |
394 |
|
395 | |||
395 | /// Return a slice of bytes if `revlog` is inline. Panic if not. |
|
396 | /// Return a slice of bytes if `revlog` is inline. Panic if not. | |
396 | pub fn data(&self, start: usize, end: usize) -> &[u8] { |
|
397 | pub fn data(&self, start: usize, end: usize) -> &[u8] { | |
397 | if !self.is_inline() { |
|
398 | if !self.is_inline() { | |
398 | panic!("tried to access data in the index of a revlog that is not inline"); |
|
399 | panic!("tried to access data in the index of a revlog that is not inline"); | |
399 | } |
|
400 | } | |
400 | &self.bytes[start..end] |
|
401 | &self.bytes[start..end] | |
401 | } |
|
402 | } | |
402 |
|
403 | |||
403 | /// Return number of entries of the revlog index. |
|
404 | /// Return number of entries of the revlog index. | |
404 | pub fn len(&self) -> usize { |
|
405 | pub fn len(&self) -> usize { | |
405 | if let Some(offsets) = &*self.get_offsets() { |
|
406 | if let Some(offsets) = &*self.get_offsets() { | |
406 | offsets.len() |
|
407 | offsets.len() | |
407 | } else { |
|
408 | } else { | |
408 | self.bytes.len() / INDEX_ENTRY_SIZE |
|
409 | self.bytes.len() / INDEX_ENTRY_SIZE | |
409 | } |
|
410 | } | |
410 | } |
|
411 | } | |
411 |
|
412 | |||
412 | pub fn get_offsets(&self) -> RwLockReadGuard<Option<Vec<usize>>> { |
|
413 | pub fn get_offsets(&self) -> RwLockReadGuard<Option<Vec<usize>>> { | |
413 | if self.is_inline() { |
|
414 | if self.is_inline() { | |
414 | { |
|
415 | { | |
415 | // Wrap in a block to drop the read guard |
|
416 | // Wrap in a block to drop the read guard | |
416 | // TODO perf? |
|
417 | // TODO perf? | |
417 | let mut offsets = self.offsets.write().unwrap(); |
|
418 | let mut offsets = self.offsets.write().unwrap(); | |
418 | if offsets.is_none() { |
|
419 | if offsets.is_none() { | |
419 | offsets.replace(inline_scan(&self.bytes.bytes).1); |
|
420 | offsets.replace(inline_scan(&self.bytes.bytes).1); | |
420 | } |
|
421 | } | |
421 | } |
|
422 | } | |
422 | } |
|
423 | } | |
423 | self.offsets.read().unwrap() |
|
424 | self.offsets.read().unwrap() | |
424 | } |
|
425 | } | |
425 |
|
426 | |||
426 | pub fn get_offsets_mut(&mut self) -> RwLockWriteGuard<Option<Vec<usize>>> { |
|
427 | pub fn get_offsets_mut(&mut self) -> RwLockWriteGuard<Option<Vec<usize>>> { | |
427 | let mut offsets = self.offsets.write().unwrap(); |
|
428 | let mut offsets = self.offsets.write().unwrap(); | |
428 | if self.is_inline() && offsets.is_none() { |
|
429 | if self.is_inline() && offsets.is_none() { | |
429 | offsets.replace(inline_scan(&self.bytes.bytes).1); |
|
430 | offsets.replace(inline_scan(&self.bytes.bytes).1); | |
430 | } |
|
431 | } | |
431 | offsets |
|
432 | offsets | |
432 | } |
|
433 | } | |
433 |
|
434 | |||
434 | /// Returns `true` if the `Index` has zero `entries`. |
|
435 | /// Returns `true` if the `Index` has zero `entries`. | |
435 | pub fn is_empty(&self) -> bool { |
|
436 | pub fn is_empty(&self) -> bool { | |
436 | self.len() == 0 |
|
437 | self.len() == 0 | |
437 | } |
|
438 | } | |
438 |
|
439 | |||
439 | /// Return the index entry corresponding to the given revision or `None` |
|
440 | /// Return the index entry corresponding to the given revision or `None` | |
440 | /// for [`NULL_REVISION`] |
|
441 | /// for [`NULL_REVISION`] | |
441 | /// |
|
442 | /// | |
442 | /// The specified revision being of the checked type, it always exists |
|
443 | /// The specified revision being of the checked type, it always exists | |
443 | /// if it was validated by this index. |
|
444 | /// if it was validated by this index. | |
444 | pub fn get_entry(&self, rev: Revision) -> Option<IndexEntry> { |
|
445 | pub fn get_entry(&self, rev: Revision) -> Option<IndexEntry> { | |
445 | if rev == NULL_REVISION { |
|
446 | if rev == NULL_REVISION { | |
446 | return None; |
|
447 | return None; | |
447 | } |
|
448 | } | |
448 | Some(if let Some(offsets) = &*self.get_offsets() { |
|
449 | Some(if let Some(offsets) = &*self.get_offsets() { | |
449 | self.get_entry_inline(rev, offsets.as_ref()) |
|
450 | self.get_entry_inline(rev, offsets.as_ref()) | |
450 | } else { |
|
451 | } else { | |
451 | self.get_entry_separated(rev) |
|
452 | self.get_entry_separated(rev) | |
452 | }) |
|
453 | }) | |
453 | } |
|
454 | } | |
454 |
|
455 | |||
455 | /// Return the binary content of the index entry for the given revision |
|
456 | /// Return the binary content of the index entry for the given revision | |
456 | /// |
|
457 | /// | |
457 | /// See [get_entry()](`Self::get_entry()`) for cases when `None` is |
|
458 | /// See [get_entry()](`Self::get_entry()`) for cases when `None` is | |
458 | /// returned. |
|
459 | /// returned. | |
459 | pub fn entry_binary(&self, rev: Revision) -> Option<&[u8]> { |
|
460 | pub fn entry_binary(&self, rev: Revision) -> Option<&[u8]> { | |
460 | self.get_entry(rev).map(|e| { |
|
461 | self.get_entry(rev).map(|e| { | |
461 | let bytes = e.as_bytes(); |
|
462 | let bytes = e.as_bytes(); | |
462 | if rev.0 == 0 { |
|
463 | if rev.0 == 0 { | |
463 | &bytes[4..] |
|
464 | &bytes[4..] | |
464 | } else { |
|
465 | } else { | |
465 | bytes |
|
466 | bytes | |
466 | } |
|
467 | } | |
467 | }) |
|
468 | }) | |
468 | } |
|
469 | } | |
469 |
|
470 | |||
470 | pub fn entry_as_params( |
|
471 | pub fn entry_as_params( | |
471 | &self, |
|
472 | &self, | |
472 | rev: UncheckedRevision, |
|
473 | rev: UncheckedRevision, | |
473 | ) -> Option<RevisionDataParams> { |
|
474 | ) -> Option<RevisionDataParams> { | |
474 | let rev = self.check_revision(rev)?; |
|
475 | let rev = self.check_revision(rev)?; | |
475 | self.get_entry(rev).map(|e| RevisionDataParams { |
|
476 | self.get_entry(rev).map(|e| RevisionDataParams { | |
476 | flags: e.flags(), |
|
477 | flags: e.flags(), | |
477 | data_offset: if rev.0 == 0 && !self.bytes.is_new() { |
|
478 | data_offset: if rev.0 == 0 && !self.bytes.is_new() { | |
478 | e.flags() as u64 |
|
479 | e.flags() as u64 | |
479 | } else { |
|
480 | } else { | |
480 | e.raw_offset() |
|
481 | e.raw_offset() | |
481 | }, |
|
482 | }, | |
482 | data_compressed_length: e.compressed_len().try_into().unwrap(), |
|
483 | data_compressed_length: e.compressed_len().try_into().unwrap(), | |
483 | data_uncompressed_length: e.uncompressed_len(), |
|
484 | data_uncompressed_length: e.uncompressed_len(), | |
484 | data_delta_base: e.base_revision_or_base_of_delta_chain().0, |
|
485 | data_delta_base: e.base_revision_or_base_of_delta_chain().0, | |
485 | link_rev: e.link_revision().0, |
|
486 | link_rev: e.link_revision().0, | |
486 | parent_rev_1: e.p1().0, |
|
487 | parent_rev_1: e.p1().0, | |
487 | parent_rev_2: e.p2().0, |
|
488 | parent_rev_2: e.p2().0, | |
488 | node_id: e.hash().as_bytes().try_into().unwrap(), |
|
489 | node_id: e.hash().as_bytes().try_into().unwrap(), | |
489 | ..Default::default() |
|
490 | ..Default::default() | |
490 | }) |
|
491 | }) | |
491 | } |
|
492 | } | |
492 |
|
493 | |||
493 | fn get_entry_inline( |
|
494 | fn get_entry_inline( | |
494 | &self, |
|
495 | &self, | |
495 | rev: Revision, |
|
496 | rev: Revision, | |
496 | offsets: &[usize], |
|
497 | offsets: &[usize], | |
497 | ) -> IndexEntry { |
|
498 | ) -> IndexEntry { | |
498 | let start = offsets[rev.0 as usize]; |
|
499 | let start = offsets[rev.0 as usize]; | |
499 | let end = start + INDEX_ENTRY_SIZE; |
|
500 | let end = start + INDEX_ENTRY_SIZE; | |
500 | let bytes = &self.bytes[start..end]; |
|
501 | let bytes = &self.bytes[start..end]; | |
501 |
|
502 | |||
502 | // See IndexEntry for an explanation of this override. |
|
503 | // See IndexEntry for an explanation of this override. | |
503 | let offset_override = Some(end); |
|
504 | let offset_override = Some(end); | |
504 |
|
505 | |||
505 | IndexEntry { |
|
506 | IndexEntry { | |
506 | bytes, |
|
507 | bytes, | |
507 | offset_override, |
|
508 | offset_override, | |
508 | } |
|
509 | } | |
509 | } |
|
510 | } | |
510 |
|
511 | |||
511 | fn get_entry_separated(&self, rev: Revision) -> IndexEntry { |
|
512 | fn get_entry_separated(&self, rev: Revision) -> IndexEntry { | |
512 | let start = rev.0 as usize * INDEX_ENTRY_SIZE; |
|
513 | let start = rev.0 as usize * INDEX_ENTRY_SIZE; | |
513 | let end = start + INDEX_ENTRY_SIZE; |
|
514 | let end = start + INDEX_ENTRY_SIZE; | |
514 | let bytes = &self.bytes[start..end]; |
|
515 | let bytes = &self.bytes[start..end]; | |
515 |
|
516 | |||
516 | // Override the offset of the first revision as its bytes are used |
|
517 | // Override the offset of the first revision as its bytes are used | |
517 | // for the index's metadata (saving space because it is always 0) |
|
518 | // for the index's metadata (saving space because it is always 0) | |
518 | let offset_override = if rev == Revision(0) { Some(0) } else { None }; |
|
519 | let offset_override = if rev == Revision(0) { Some(0) } else { None }; | |
519 |
|
520 | |||
520 | IndexEntry { |
|
521 | IndexEntry { | |
521 | bytes, |
|
522 | bytes, | |
522 | offset_override, |
|
523 | offset_override, | |
523 | } |
|
524 | } | |
524 | } |
|
525 | } | |
525 |
|
526 | |||
|
527 | fn null_entry(&self) -> IndexEntry { | |||
|
528 | IndexEntry { | |||
|
529 | bytes: &[0; INDEX_ENTRY_SIZE], | |||
|
530 | offset_override: Some(0), | |||
|
531 | } | |||
|
532 | } | |||
|
533 | ||||
526 | /// Return the head revisions of this index |
|
534 | /// Return the head revisions of this index | |
527 | pub fn head_revs(&mut self) -> Result<Vec<Revision>, GraphError> { |
|
535 | pub fn head_revs(&mut self) -> Result<Vec<Revision>, GraphError> { | |
528 | self.head_revs_filtered(&HashSet::new()) |
|
536 | self.head_revs_filtered(&HashSet::new()) | |
529 | } |
|
537 | } | |
530 |
|
538 | |||
531 | /// Return the head revisions of this index |
|
539 | /// Return the head revisions of this index | |
532 | pub fn head_revs_filtered( |
|
540 | pub fn head_revs_filtered( | |
533 | &mut self, |
|
541 | &mut self, | |
534 | filtered_revs: &HashSet<Revision>, |
|
542 | filtered_revs: &HashSet<Revision>, | |
535 | ) -> Result<Vec<Revision>, GraphError> { |
|
543 | ) -> Result<Vec<Revision>, GraphError> { | |
536 | if !self.head_revs.is_empty() && filtered_revs == &self.filtered_revs { |
|
544 | if !self.head_revs.is_empty() && filtered_revs == &self.filtered_revs { | |
537 | return Ok(self.head_revs.to_owned()); |
|
545 | return Ok(self.head_revs.to_owned()); | |
538 | } |
|
546 | } | |
539 | let mut revs: HashSet<Revision, RandomState> = |
|
547 | let mut revs: HashSet<Revision, RandomState> = | |
540 | if filtered_revs.is_empty() { |
|
548 | if filtered_revs.is_empty() { | |
541 | (0..self.len()) |
|
549 | (0..self.len()) | |
542 | .into_iter() |
|
550 | .into_iter() | |
543 | .map(|i| Revision(i as BaseRevision)) |
|
551 | .map(|i| Revision(i as BaseRevision)) | |
544 | .collect() |
|
552 | .collect() | |
545 | } else { |
|
553 | } else { | |
546 | (0..self.len()) |
|
554 | (0..self.len()) | |
547 | .into_iter() |
|
555 | .into_iter() | |
548 | .filter_map(|i| { |
|
556 | .filter_map(|i| { | |
549 | let r = Revision(i as BaseRevision); |
|
557 | let r = Revision(i as BaseRevision); | |
550 | if filtered_revs.contains(&r) { |
|
558 | if filtered_revs.contains(&r) { | |
551 | None |
|
559 | None | |
552 | } else { |
|
560 | } else { | |
553 | Some(r) |
|
561 | Some(r) | |
554 | } |
|
562 | } | |
555 | }) |
|
563 | }) | |
556 | .collect() |
|
564 | .collect() | |
557 | }; |
|
565 | }; | |
558 | dagops::retain_heads(self, &mut revs)?; |
|
566 | dagops::retain_heads(self, &mut revs)?; | |
559 | if self.is_empty() { |
|
567 | if self.is_empty() { | |
560 | revs.insert(NULL_REVISION); |
|
568 | revs.insert(NULL_REVISION); | |
561 | } |
|
569 | } | |
562 | let mut as_vec: Vec<Revision> = |
|
570 | let mut as_vec: Vec<Revision> = | |
563 | revs.into_iter().map(Into::into).collect(); |
|
571 | revs.into_iter().map(Into::into).collect(); | |
564 | as_vec.sort_unstable(); |
|
572 | as_vec.sort_unstable(); | |
565 | self.head_revs = as_vec.to_owned(); |
|
573 | self.head_revs = as_vec.to_owned(); | |
566 | self.filtered_revs = filtered_revs.to_owned(); |
|
574 | self.filtered_revs = filtered_revs.to_owned(); | |
567 | Ok(as_vec) |
|
575 | Ok(as_vec) | |
568 | } |
|
576 | } | |
569 |
|
577 | |||
570 | /// Obtain the delta chain for a revision. |
|
578 | /// Obtain the delta chain for a revision. | |
571 | /// |
|
579 | /// | |
572 | /// `stop_rev` specifies a revision to stop at. If not specified, we |
|
580 | /// `stop_rev` specifies a revision to stop at. If not specified, we | |
573 | /// stop at the base of the chain. |
|
581 | /// stop at the base of the chain. | |
574 | /// |
|
582 | /// | |
575 | /// Returns a 2-tuple of (chain, stopped) where `chain` is a vec of |
|
583 | /// Returns a 2-tuple of (chain, stopped) where `chain` is a vec of | |
576 | /// revs in ascending order and `stopped` is a bool indicating whether |
|
584 | /// revs in ascending order and `stopped` is a bool indicating whether | |
577 | /// `stoprev` was hit. |
|
585 | /// `stoprev` was hit. | |
578 | pub fn delta_chain( |
|
586 | pub fn delta_chain( | |
579 | &self, |
|
587 | &self, | |
580 | rev: Revision, |
|
588 | rev: Revision, | |
581 | stop_rev: Option<Revision>, |
|
589 | stop_rev: Option<Revision>, | |
582 | ) -> Result<(Vec<Revision>, bool), HgError> { |
|
590 | ) -> Result<(Vec<Revision>, bool), HgError> { | |
583 | let mut current_rev = rev; |
|
591 | let mut current_rev = rev; | |
584 | let mut entry = self.get_entry(rev).unwrap(); |
|
592 | let mut entry = self.get_entry(rev).unwrap(); | |
585 | let mut chain = vec![]; |
|
593 | let mut chain = vec![]; | |
586 | while current_rev.0 != entry.base_revision_or_base_of_delta_chain().0 |
|
594 | while current_rev.0 != entry.base_revision_or_base_of_delta_chain().0 | |
587 | && stop_rev.map(|r| r != current_rev).unwrap_or(true) |
|
595 | && stop_rev.map(|r| r != current_rev).unwrap_or(true) | |
588 | { |
|
596 | { | |
589 | chain.push(current_rev); |
|
597 | chain.push(current_rev); | |
590 | let new_rev = if self.uses_generaldelta() { |
|
598 | let new_rev = if self.uses_generaldelta() { | |
591 | entry.base_revision_or_base_of_delta_chain() |
|
599 | entry.base_revision_or_base_of_delta_chain() | |
592 | } else { |
|
600 | } else { | |
593 | UncheckedRevision(current_rev.0 - 1) |
|
601 | UncheckedRevision(current_rev.0 - 1) | |
594 | }; |
|
602 | }; | |
595 | if new_rev.0 == NULL_REVISION.0 { |
|
|||
596 | break; |
|
|||
597 | } |
|
|||
598 | current_rev = self.check_revision(new_rev).ok_or_else(|| { |
|
603 | current_rev = self.check_revision(new_rev).ok_or_else(|| { | |
599 | HgError::corrupted(format!("Revision {new_rev} out of range")) |
|
604 | HgError::corrupted(format!("Revision {new_rev} out of range")) | |
600 | })?; |
|
605 | })?; | |
|
606 | if current_rev.0 == NULL_REVISION.0 { | |||
|
607 | break; | |||
|
608 | } | |||
601 | entry = self.get_entry(current_rev).unwrap() |
|
609 | entry = self.get_entry(current_rev).unwrap() | |
602 | } |
|
610 | } | |
603 |
|
611 | |||
604 | let stopped = if stop_rev.map(|r| current_rev == r).unwrap_or(false) { |
|
612 | let stopped = if stop_rev.map(|r| current_rev == r).unwrap_or(false) { | |
605 | true |
|
613 | true | |
606 | } else { |
|
614 | } else { | |
607 | chain.push(current_rev); |
|
615 | chain.push(current_rev); | |
608 | false |
|
616 | false | |
609 | }; |
|
617 | }; | |
610 | chain.reverse(); |
|
618 | chain.reverse(); | |
611 | Ok((chain, stopped)) |
|
619 | Ok((chain, stopped)) | |
612 | } |
|
620 | } | |
613 |
|
621 | |||
614 | pub fn find_snapshots( |
|
622 | pub fn find_snapshots( | |
615 | &self, |
|
623 | &self, | |
616 | start_rev: UncheckedRevision, |
|
624 | start_rev: UncheckedRevision, | |
617 | end_rev: UncheckedRevision, |
|
625 | end_rev: UncheckedRevision, | |
618 | cache: &mut impl SnapshotsCache, |
|
626 | cache: &mut impl SnapshotsCache, | |
619 | ) -> Result<(), RevlogError> { |
|
627 | ) -> Result<(), RevlogError> { | |
620 | let mut start_rev = start_rev.0; |
|
628 | let mut start_rev = start_rev.0; | |
621 | let mut end_rev = end_rev.0; |
|
629 | let mut end_rev = end_rev.0; | |
622 | end_rev += 1; |
|
630 | end_rev += 1; | |
623 | let len = self.len().try_into().unwrap(); |
|
631 | let len = self.len().try_into().unwrap(); | |
624 | if end_rev > len { |
|
632 | if end_rev > len { | |
625 | end_rev = len; |
|
633 | end_rev = len; | |
626 | } |
|
634 | } | |
627 | if start_rev < 0 { |
|
635 | if start_rev < 0 { | |
628 | start_rev = 0; |
|
636 | start_rev = 0; | |
629 | } |
|
637 | } | |
630 | for rev in start_rev..end_rev { |
|
638 | for rev in start_rev..end_rev { | |
631 | if !self.is_snapshot_unchecked(Revision(rev))? { |
|
639 | if !self.is_snapshot_unchecked(Revision(rev))? { | |
632 | continue; |
|
640 | continue; | |
633 | } |
|
641 | } | |
634 | let mut base = self |
|
642 | let mut base = self | |
635 | .get_entry(Revision(rev)) |
|
643 | .get_entry(Revision(rev)) | |
636 | .unwrap() |
|
644 | .unwrap() | |
637 | .base_revision_or_base_of_delta_chain(); |
|
645 | .base_revision_or_base_of_delta_chain(); | |
638 | if base.0 == rev { |
|
646 | if base.0 == rev { | |
639 | base = NULL_REVISION.into(); |
|
647 | base = NULL_REVISION.into(); | |
640 | } |
|
648 | } | |
641 | cache.insert_for(base.0, rev)?; |
|
649 | cache.insert_for(base.0, rev)?; | |
642 | } |
|
650 | } | |
643 | Ok(()) |
|
651 | Ok(()) | |
644 | } |
|
652 | } | |
645 |
|
653 | |||
646 | /// TODO move this to the trait probably, along with other things |
|
654 | /// TODO move this to the trait probably, along with other things | |
647 | pub fn append( |
|
655 | pub fn append( | |
648 | &mut self, |
|
656 | &mut self, | |
649 | revision_data: RevisionDataParams, |
|
657 | revision_data: RevisionDataParams, | |
650 | ) -> Result<(), RevlogError> { |
|
658 | ) -> Result<(), RevlogError> { | |
651 | revision_data.validate()?; |
|
659 | revision_data.validate()?; | |
652 | let new_offset = self.bytes.len(); |
|
660 | let new_offset = self.bytes.len(); | |
653 | if let Some(offsets) = &mut *self.get_offsets_mut() { |
|
661 | if let Some(offsets) = &mut *self.get_offsets_mut() { | |
654 | offsets.push(new_offset) |
|
662 | offsets.push(new_offset) | |
655 | } |
|
663 | } | |
656 | self.bytes.added.extend(revision_data.into_v1().as_bytes()); |
|
664 | self.bytes.added.extend(revision_data.into_v1().as_bytes()); | |
657 | self.head_revs.clear(); |
|
665 | self.head_revs.clear(); | |
658 | Ok(()) |
|
666 | Ok(()) | |
659 | } |
|
667 | } | |
660 |
|
668 | |||
661 | pub fn pack_header(&self, header: i32) -> [u8; 4] { |
|
669 | pub fn pack_header(&self, header: i32) -> [u8; 4] { | |
662 | header.to_be_bytes() |
|
670 | header.to_be_bytes() | |
663 | } |
|
671 | } | |
664 |
|
672 | |||
665 | pub fn remove(&mut self, rev: Revision) -> Result<(), RevlogError> { |
|
673 | pub fn remove(&mut self, rev: Revision) -> Result<(), RevlogError> { | |
666 | let offsets = self.get_offsets().clone(); |
|
674 | let offsets = self.get_offsets().clone(); | |
667 | self.bytes.remove(rev, offsets.as_deref())?; |
|
675 | self.bytes.remove(rev, offsets.as_deref())?; | |
668 | if let Some(offsets) = &mut *self.get_offsets_mut() { |
|
676 | if let Some(offsets) = &mut *self.get_offsets_mut() { | |
669 | offsets.truncate(rev.0 as usize) |
|
677 | offsets.truncate(rev.0 as usize) | |
670 | } |
|
678 | } | |
671 | self.head_revs.clear(); |
|
679 | self.head_revs.clear(); | |
672 | Ok(()) |
|
680 | Ok(()) | |
673 | } |
|
681 | } | |
674 |
|
682 | |||
675 | pub fn clear_caches(&mut self) { |
|
683 | pub fn clear_caches(&mut self) { | |
676 | // We need to get the 'inline' value from Python at init and use this |
|
684 | // We need to get the 'inline' value from Python at init and use this | |
677 | // instead of offsets to determine whether we're inline since we might |
|
685 | // instead of offsets to determine whether we're inline since we might | |
678 | // clear caches. This implies re-populating the offsets on-demand. |
|
686 | // clear caches. This implies re-populating the offsets on-demand. | |
679 | self.offsets = RwLock::new(None); |
|
687 | self.offsets = RwLock::new(None); | |
680 | self.head_revs.clear(); |
|
688 | self.head_revs.clear(); | |
681 | } |
|
689 | } | |
682 |
|
690 | |||
683 | /// Unchecked version of `is_snapshot`. |
|
691 | /// Unchecked version of `is_snapshot`. | |
684 | /// Assumes the caller checked that `rev` is within a valid revision range. |
|
692 | /// Assumes the caller checked that `rev` is within a valid revision range. | |
685 | pub fn is_snapshot_unchecked( |
|
693 | pub fn is_snapshot_unchecked( | |
686 | &self, |
|
694 | &self, | |
687 | mut rev: Revision, |
|
695 | mut rev: Revision, | |
688 | ) -> Result<bool, RevlogError> { |
|
696 | ) -> Result<bool, RevlogError> { | |
689 | while rev.0 >= 0 { |
|
697 | while rev.0 >= 0 { | |
690 | let entry = self.get_entry(rev).unwrap(); |
|
698 | let entry = self.get_entry(rev).unwrap(); | |
691 | let mut base = entry.base_revision_or_base_of_delta_chain().0; |
|
699 | let mut base = entry.base_revision_or_base_of_delta_chain().0; | |
692 | if base == rev.0 { |
|
700 | if base == rev.0 { | |
693 | base = NULL_REVISION.0; |
|
701 | base = NULL_REVISION.0; | |
694 | } |
|
702 | } | |
695 | if base == NULL_REVISION.0 { |
|
703 | if base == NULL_REVISION.0 { | |
696 | return Ok(true); |
|
704 | return Ok(true); | |
697 | } |
|
705 | } | |
698 | let [mut p1, mut p2] = self |
|
706 | let [mut p1, mut p2] = self | |
699 | .parents(rev) |
|
707 | .parents(rev) | |
700 | .map_err(|_| RevlogError::InvalidRevision)?; |
|
708 | .map_err(|_| RevlogError::InvalidRevision)?; | |
701 | while let Some(p1_entry) = self.get_entry(p1) { |
|
709 | while let Some(p1_entry) = self.get_entry(p1) { | |
702 | if p1_entry.compressed_len() != 0 || p1.0 == 0 { |
|
710 | if p1_entry.compressed_len() != 0 || p1.0 == 0 { | |
703 | break; |
|
711 | break; | |
704 | } |
|
712 | } | |
705 | let parent_base = |
|
713 | let parent_base = | |
706 | p1_entry.base_revision_or_base_of_delta_chain(); |
|
714 | p1_entry.base_revision_or_base_of_delta_chain(); | |
707 | if parent_base.0 == p1.0 { |
|
715 | if parent_base.0 == p1.0 { | |
708 | break; |
|
716 | break; | |
709 | } |
|
717 | } | |
710 | p1 = self |
|
718 | p1 = self | |
711 | .check_revision(parent_base) |
|
719 | .check_revision(parent_base) | |
712 | .ok_or(RevlogError::InvalidRevision)?; |
|
720 | .ok_or(RevlogError::InvalidRevision)?; | |
713 | } |
|
721 | } | |
714 | while let Some(p2_entry) = self.get_entry(p2) { |
|
722 | while let Some(p2_entry) = self.get_entry(p2) { | |
715 | if p2_entry.compressed_len() != 0 || p2.0 == 0 { |
|
723 | if p2_entry.compressed_len() != 0 || p2.0 == 0 { | |
716 | break; |
|
724 | break; | |
717 | } |
|
725 | } | |
718 | let parent_base = |
|
726 | let parent_base = | |
719 | p2_entry.base_revision_or_base_of_delta_chain(); |
|
727 | p2_entry.base_revision_or_base_of_delta_chain(); | |
720 | if parent_base.0 == p2.0 { |
|
728 | if parent_base.0 == p2.0 { | |
721 | break; |
|
729 | break; | |
722 | } |
|
730 | } | |
723 | p2 = self |
|
731 | p2 = self | |
724 | .check_revision(parent_base) |
|
732 | .check_revision(parent_base) | |
725 | .ok_or(RevlogError::InvalidRevision)?; |
|
733 | .ok_or(RevlogError::InvalidRevision)?; | |
726 | } |
|
734 | } | |
727 | if base == p1.0 || base == p2.0 { |
|
735 | if base == p1.0 || base == p2.0 { | |
728 | return Ok(false); |
|
736 | return Ok(false); | |
729 | } |
|
737 | } | |
730 | rev = self |
|
738 | rev = self | |
731 | .check_revision(base.into()) |
|
739 | .check_revision(base.into()) | |
732 | .ok_or(RevlogError::InvalidRevision)?; |
|
740 | .ok_or(RevlogError::InvalidRevision)?; | |
733 | } |
|
741 | } | |
734 | Ok(rev == NULL_REVISION) |
|
742 | Ok(rev == NULL_REVISION) | |
735 | } |
|
743 | } | |
736 |
|
744 | |||
737 | /// Return whether the given revision is a snapshot. Returns an error if |
|
745 | /// Return whether the given revision is a snapshot. Returns an error if | |
738 | /// `rev` is not within a valid revision range. |
|
746 | /// `rev` is not within a valid revision range. | |
739 | pub fn is_snapshot( |
|
747 | pub fn is_snapshot( | |
740 | &self, |
|
748 | &self, | |
741 | rev: UncheckedRevision, |
|
749 | rev: UncheckedRevision, | |
742 | ) -> Result<bool, RevlogError> { |
|
750 | ) -> Result<bool, RevlogError> { | |
743 | let rev = self |
|
751 | let rev = self | |
744 | .check_revision(rev) |
|
752 | .check_revision(rev) | |
745 | .ok_or_else(|| RevlogError::corrupted("test"))?; |
|
753 | .ok_or_else(|| RevlogError::corrupted("test"))?; | |
746 | self.is_snapshot_unchecked(rev) |
|
754 | self.is_snapshot_unchecked(rev) | |
747 | } |
|
755 | } | |
|
756 | ||||
|
757 | /// Slice revs to reduce the amount of unrelated data to be read from disk. | |||
|
758 | /// | |||
|
759 | /// The index is sliced into groups that should be read in one time. | |||
|
760 | /// | |||
|
761 | /// The initial chunk is sliced until the overall density | |||
|
762 | /// (payload/chunks-span ratio) is above `target_density`. | |||
|
763 | /// No gap smaller than `min_gap_size` is skipped. | |||
|
764 | pub fn slice_chunk_to_density( | |||
|
765 | &self, | |||
|
766 | revs: &[Revision], | |||
|
767 | target_density: f64, | |||
|
768 | min_gap_size: usize, | |||
|
769 | ) -> Vec<Vec<Revision>> { | |||
|
770 | if revs.is_empty() { | |||
|
771 | return vec![]; | |||
|
772 | } | |||
|
773 | if revs.len() == 1 { | |||
|
774 | return vec![revs.to_owned()]; | |||
|
775 | } | |||
|
776 | let delta_chain_span = self.segment_span(revs); | |||
|
777 | if delta_chain_span < min_gap_size { | |||
|
778 | return vec![revs.to_owned()]; | |||
|
779 | } | |||
|
780 | let entries: Vec<_> = revs | |||
|
781 | .iter() | |||
|
782 | .map(|r| { | |||
|
783 | (*r, self.get_entry(*r).unwrap_or_else(|| self.null_entry())) | |||
|
784 | }) | |||
|
785 | .collect(); | |||
|
786 | ||||
|
787 | let mut read_data = delta_chain_span; | |||
|
788 | let chain_payload: u32 = | |||
|
789 | entries.iter().map(|(_r, e)| e.compressed_len()).sum(); | |||
|
790 | let mut density = if delta_chain_span > 0 { | |||
|
791 | chain_payload as f64 / delta_chain_span as f64 | |||
|
792 | } else { | |||
|
793 | 1.0 | |||
|
794 | }; | |||
|
795 | ||||
|
796 | if density >= target_density { | |||
|
797 | return vec![revs.to_owned()]; | |||
|
798 | } | |||
|
799 | ||||
|
800 | // Store the gaps in a heap to have them sorted by decreasing size | |||
|
801 | let mut gaps = Vec::new(); | |||
|
802 | let mut previous_end = None; | |||
|
803 | ||||
|
804 | for (i, (_rev, entry)) in entries.iter().enumerate() { | |||
|
805 | let start = entry.c_start() as usize; | |||
|
806 | let length = entry.compressed_len(); | |||
|
807 | ||||
|
808 | // Skip empty revisions to form larger holes | |||
|
809 | if length == 0 { | |||
|
810 | continue; | |||
|
811 | } | |||
|
812 | ||||
|
813 | if let Some(end) = previous_end { | |||
|
814 | let gap_size = start - end; | |||
|
815 | // Only consider holes that are large enough | |||
|
816 | if gap_size > min_gap_size { | |||
|
817 | gaps.push((gap_size, i)); | |||
|
818 | } | |||
|
819 | } | |||
|
820 | previous_end = Some(start + length as usize); | |||
|
821 | } | |||
|
822 | if gaps.is_empty() { | |||
|
823 | return vec![revs.to_owned()]; | |||
|
824 | } | |||
|
825 | // sort the gaps to pop them from largest to small | |||
|
826 | gaps.sort_unstable(); | |||
|
827 | ||||
|
828 | // Collect the indices of the largest holes until | |||
|
829 | // the density is acceptable | |||
|
830 | let mut selected = vec![]; | |||
|
831 | while let Some((gap_size, gap_id)) = gaps.pop() { | |||
|
832 | if density >= target_density { | |||
|
833 | break; | |||
|
834 | } | |||
|
835 | selected.push(gap_id); | |||
|
836 | ||||
|
837 | // The gap sizes are stored as negatives to be sorted decreasingly | |||
|
838 | // by the heap | |||
|
839 | read_data -= gap_size; | |||
|
840 | density = if read_data > 0 { | |||
|
841 | chain_payload as f64 / read_data as f64 | |||
|
842 | } else { | |||
|
843 | 1.0 | |||
|
844 | }; | |||
|
845 | if density >= target_density { | |||
|
846 | break; | |||
|
847 | } | |||
|
848 | } | |||
|
849 | selected.sort_unstable(); | |||
|
850 | selected.push(revs.len()); | |||
|
851 | ||||
|
852 | // Cut the revs at collected indices | |||
|
853 | let mut previous_idx = 0; | |||
|
854 | let mut chunks = vec![]; | |||
|
855 | for idx in selected { | |||
|
856 | let chunk = self.trim_chunk(&entries, previous_idx, idx); | |||
|
857 | if !chunk.is_empty() { | |||
|
858 | chunks.push(chunk.iter().map(|(rev, _entry)| *rev).collect()); | |||
|
859 | } | |||
|
860 | previous_idx = idx; | |||
|
861 | } | |||
|
862 | let chunk = self.trim_chunk(&entries, previous_idx, entries.len()); | |||
|
863 | if !chunk.is_empty() { | |||
|
864 | chunks.push(chunk.iter().map(|(rev, _entry)| *rev).collect()); | |||
|
865 | } | |||
|
866 | ||||
|
867 | chunks | |||
|
868 | } | |||
|
869 | ||||
|
870 | /// Get the byte span of a segment of sorted revisions. | |||
|
871 | /// | |||
|
872 | /// Occurrences of [`NULL_REVISION`] are ignored at the beginning of | |||
|
873 | /// the `revs` segment. | |||
|
874 | /// | |||
|
875 | /// panics: | |||
|
876 | /// - if `revs` is empty or only made of `NULL_REVISION` | |||
|
877 | /// - if cannot retrieve entry for the last or first not null element of | |||
|
878 | /// `revs`. | |||
|
879 | fn segment_span(&self, revs: &[Revision]) -> usize { | |||
|
880 | if revs.is_empty() { | |||
|
881 | return 0; | |||
|
882 | } | |||
|
883 | let last_entry = &self.get_entry(revs[revs.len() - 1]).unwrap(); | |||
|
884 | let end = last_entry.c_start() + last_entry.compressed_len() as u64; | |||
|
885 | let first_rev = revs.iter().find(|r| r.0 != NULL_REVISION.0).unwrap(); | |||
|
886 | let start = if (*first_rev).0 == 0 { | |||
|
887 | 0 | |||
|
888 | } else { | |||
|
889 | self.get_entry(*first_rev).unwrap().c_start() | |||
|
890 | }; | |||
|
891 | (end - start) as usize | |||
|
892 | } | |||
|
893 | ||||
|
894 | /// Returns `&revs[startidx..endidx]` without empty trailing revs | |||
|
895 | fn trim_chunk<'a>( | |||
|
896 | &'a self, | |||
|
897 | revs: &'a [(Revision, IndexEntry)], | |||
|
898 | start: usize, | |||
|
899 | mut end: usize, | |||
|
900 | ) -> &'a [(Revision, IndexEntry)] { | |||
|
901 | // Trim empty revs at the end, except the very first rev of a chain | |||
|
902 | let last_rev = revs[end - 1].0; | |||
|
903 | if last_rev.0 < self.len() as BaseRevision { | |||
|
904 | while end > 1 | |||
|
905 | && end > start | |||
|
906 | && revs[end - 1].1.compressed_len() == 0 | |||
|
907 | { | |||
|
908 | end -= 1 | |||
|
909 | } | |||
|
910 | } | |||
|
911 | &revs[start..end] | |||
|
912 | } | |||
748 | } |
|
913 | } | |
749 | fn inline_scan(bytes: &[u8]) -> (usize, Vec<usize>) { |
|
914 | fn inline_scan(bytes: &[u8]) -> (usize, Vec<usize>) { | |
750 | let mut offset: usize = 0; |
|
915 | let mut offset: usize = 0; | |
751 | let mut offsets = Vec::new(); |
|
916 | let mut offsets = Vec::new(); | |
752 |
|
917 | |||
753 | while offset + INDEX_ENTRY_SIZE <= bytes.len() { |
|
918 | while offset + INDEX_ENTRY_SIZE <= bytes.len() { | |
754 | offsets.push(offset); |
|
919 | offsets.push(offset); | |
755 | let end = offset + INDEX_ENTRY_SIZE; |
|
920 | let end = offset + INDEX_ENTRY_SIZE; | |
756 | let entry = IndexEntry { |
|
921 | let entry = IndexEntry { | |
757 | bytes: &bytes[offset..end], |
|
922 | bytes: &bytes[offset..end], | |
758 | offset_override: None, |
|
923 | offset_override: None, | |
759 | }; |
|
924 | }; | |
760 |
|
925 | |||
761 | offset += INDEX_ENTRY_SIZE + entry.compressed_len() as usize; |
|
926 | offset += INDEX_ENTRY_SIZE + entry.compressed_len() as usize; | |
762 | } |
|
927 | } | |
763 | (offset, offsets) |
|
928 | (offset, offsets) | |
764 | } |
|
929 | } | |
765 |
|
930 | |||
766 | impl super::RevlogIndex for Index { |
|
931 | impl super::RevlogIndex for Index { | |
767 | fn len(&self) -> usize { |
|
932 | fn len(&self) -> usize { | |
768 | self.len() |
|
933 | self.len() | |
769 | } |
|
934 | } | |
770 |
|
935 | |||
771 | fn node(&self, rev: Revision) -> Option<&Node> { |
|
936 | fn node(&self, rev: Revision) -> Option<&Node> { | |
772 | if rev == NULL_REVISION { |
|
937 | if rev == NULL_REVISION { | |
773 | return Some(&NULL_NODE); |
|
938 | return Some(&NULL_NODE); | |
774 | } |
|
939 | } | |
775 | self.get_entry(rev).map(|entry| entry.hash()) |
|
940 | self.get_entry(rev).map(|entry| entry.hash()) | |
776 | } |
|
941 | } | |
777 | } |
|
942 | } | |
778 |
|
943 | |||
779 | #[derive(Debug)] |
|
944 | #[derive(Debug)] | |
780 | pub struct IndexEntry<'a> { |
|
945 | pub struct IndexEntry<'a> { | |
781 | bytes: &'a [u8], |
|
946 | bytes: &'a [u8], | |
782 | /// Allows to override the offset value of the entry. |
|
947 | /// Allows to override the offset value of the entry. | |
783 | /// |
|
948 | /// | |
784 | /// For interleaved index and data, the offset stored in the index |
|
949 | /// For interleaved index and data, the offset stored in the index | |
785 | /// corresponds to the separated data offset. |
|
950 | /// corresponds to the separated data offset. | |
786 | /// It has to be overridden with the actual offset in the interleaved |
|
951 | /// It has to be overridden with the actual offset in the interleaved | |
787 | /// index which is just after the index block. |
|
952 | /// index which is just after the index block. | |
788 | /// |
|
953 | /// | |
789 | /// For separated index and data, the offset stored in the first index |
|
954 | /// For separated index and data, the offset stored in the first index | |
790 | /// entry is mixed with the index headers. |
|
955 | /// entry is mixed with the index headers. | |
791 | /// It has to be overridden with 0. |
|
956 | /// It has to be overridden with 0. | |
792 | offset_override: Option<usize>, |
|
957 | offset_override: Option<usize>, | |
793 | } |
|
958 | } | |
794 |
|
959 | |||
795 | impl<'a> IndexEntry<'a> { |
|
960 | impl<'a> IndexEntry<'a> { | |
796 | /// Return the offset of the data. |
|
961 | /// Return the offset of the data. | |
797 | pub fn offset(&self) -> usize { |
|
962 | pub fn offset(&self) -> usize { | |
798 | if let Some(offset_override) = self.offset_override { |
|
963 | if let Some(offset_override) = self.offset_override { | |
799 | offset_override |
|
964 | offset_override | |
800 | } else { |
|
965 | } else { | |
801 | let mut bytes = [0; 8]; |
|
966 | let mut bytes = [0; 8]; | |
802 | bytes[2..8].copy_from_slice(&self.bytes[0..=5]); |
|
967 | bytes[2..8].copy_from_slice(&self.bytes[0..=5]); | |
803 | BigEndian::read_u64(&bytes[..]) as usize |
|
968 | BigEndian::read_u64(&bytes[..]) as usize | |
804 | } |
|
969 | } | |
805 | } |
|
970 | } | |
806 | pub fn raw_offset(&self) -> u64 { |
|
971 | pub fn raw_offset(&self) -> u64 { | |
807 | BigEndian::read_u64(&self.bytes[0..8]) |
|
972 | BigEndian::read_u64(&self.bytes[0..8]) | |
808 | } |
|
973 | } | |
809 |
|
974 | |||
|
975 | /// Same result (except potentially for rev 0) as C `index_get_start()` | |||
|
976 | fn c_start(&self) -> u64 { | |||
|
977 | self.raw_offset() >> 16 | |||
|
978 | } | |||
|
979 | ||||
810 | pub fn flags(&self) -> u16 { |
|
980 | pub fn flags(&self) -> u16 { | |
811 | BigEndian::read_u16(&self.bytes[6..=7]) |
|
981 | BigEndian::read_u16(&self.bytes[6..=7]) | |
812 | } |
|
982 | } | |
813 |
|
983 | |||
814 | /// Return the compressed length of the data. |
|
984 | /// Return the compressed length of the data. | |
815 | pub fn compressed_len(&self) -> u32 { |
|
985 | pub fn compressed_len(&self) -> u32 { | |
816 | BigEndian::read_u32(&self.bytes[8..=11]) |
|
986 | BigEndian::read_u32(&self.bytes[8..=11]) | |
817 | } |
|
987 | } | |
818 |
|
988 | |||
819 | /// Return the uncompressed length of the data. |
|
989 | /// Return the uncompressed length of the data. | |
820 | pub fn uncompressed_len(&self) -> i32 { |
|
990 | pub fn uncompressed_len(&self) -> i32 { | |
821 | BigEndian::read_i32(&self.bytes[12..=15]) |
|
991 | BigEndian::read_i32(&self.bytes[12..=15]) | |
822 | } |
|
992 | } | |
823 |
|
993 | |||
824 | /// Return the revision upon which the data has been derived. |
|
994 | /// Return the revision upon which the data has been derived. | |
825 | pub fn base_revision_or_base_of_delta_chain(&self) -> UncheckedRevision { |
|
995 | pub fn base_revision_or_base_of_delta_chain(&self) -> UncheckedRevision { | |
826 | // TODO Maybe return an Option when base_revision == rev? |
|
996 | // TODO Maybe return an Option when base_revision == rev? | |
827 | // Requires to add rev to IndexEntry |
|
997 | // Requires to add rev to IndexEntry | |
828 |
|
998 | |||
829 | BigEndian::read_i32(&self.bytes[16..]).into() |
|
999 | BigEndian::read_i32(&self.bytes[16..]).into() | |
830 | } |
|
1000 | } | |
831 |
|
1001 | |||
832 | pub fn link_revision(&self) -> UncheckedRevision { |
|
1002 | pub fn link_revision(&self) -> UncheckedRevision { | |
833 | BigEndian::read_i32(&self.bytes[20..]).into() |
|
1003 | BigEndian::read_i32(&self.bytes[20..]).into() | |
834 | } |
|
1004 | } | |
835 |
|
1005 | |||
836 | pub fn p1(&self) -> UncheckedRevision { |
|
1006 | pub fn p1(&self) -> UncheckedRevision { | |
837 | BigEndian::read_i32(&self.bytes[24..]).into() |
|
1007 | BigEndian::read_i32(&self.bytes[24..]).into() | |
838 | } |
|
1008 | } | |
839 |
|
1009 | |||
840 | pub fn p2(&self) -> UncheckedRevision { |
|
1010 | pub fn p2(&self) -> UncheckedRevision { | |
841 | BigEndian::read_i32(&self.bytes[28..]).into() |
|
1011 | BigEndian::read_i32(&self.bytes[28..]).into() | |
842 | } |
|
1012 | } | |
843 |
|
1013 | |||
844 | /// Return the hash of revision's full text. |
|
1014 | /// Return the hash of revision's full text. | |
845 | /// |
|
1015 | /// | |
846 | /// Currently, SHA-1 is used and only the first 20 bytes of this field |
|
1016 | /// Currently, SHA-1 is used and only the first 20 bytes of this field | |
847 | /// are used. |
|
1017 | /// are used. | |
848 | pub fn hash(&self) -> &'a Node { |
|
1018 | pub fn hash(&self) -> &'a Node { | |
849 | (&self.bytes[32..52]).try_into().unwrap() |
|
1019 | (&self.bytes[32..52]).try_into().unwrap() | |
850 | } |
|
1020 | } | |
851 |
|
1021 | |||
852 | pub fn as_bytes(&self) -> &'a [u8] { |
|
1022 | pub fn as_bytes(&self) -> &'a [u8] { | |
853 | self.bytes |
|
1023 | self.bytes | |
854 | } |
|
1024 | } | |
855 | } |
|
1025 | } | |
856 |
|
1026 | |||
857 | #[cfg(test)] |
|
1027 | #[cfg(test)] | |
858 | mod tests { |
|
1028 | mod tests { | |
859 | use super::*; |
|
1029 | use super::*; | |
860 | use crate::node::NULL_NODE; |
|
1030 | use crate::node::NULL_NODE; | |
861 |
|
1031 | |||
862 | #[cfg(test)] |
|
1032 | #[cfg(test)] | |
863 | #[derive(Debug, Copy, Clone)] |
|
1033 | #[derive(Debug, Copy, Clone)] | |
864 | pub struct IndexEntryBuilder { |
|
1034 | pub struct IndexEntryBuilder { | |
865 | is_first: bool, |
|
1035 | is_first: bool, | |
866 | is_inline: bool, |
|
1036 | is_inline: bool, | |
867 | is_general_delta: bool, |
|
1037 | is_general_delta: bool, | |
868 | version: u16, |
|
1038 | version: u16, | |
869 | offset: usize, |
|
1039 | offset: usize, | |
870 | compressed_len: usize, |
|
1040 | compressed_len: usize, | |
871 | uncompressed_len: usize, |
|
1041 | uncompressed_len: usize, | |
872 | base_revision_or_base_of_delta_chain: Revision, |
|
1042 | base_revision_or_base_of_delta_chain: Revision, | |
873 | link_revision: Revision, |
|
1043 | link_revision: Revision, | |
874 | p1: Revision, |
|
1044 | p1: Revision, | |
875 | p2: Revision, |
|
1045 | p2: Revision, | |
876 | node: Node, |
|
1046 | node: Node, | |
877 | } |
|
1047 | } | |
878 |
|
1048 | |||
879 | #[cfg(test)] |
|
1049 | #[cfg(test)] | |
880 | impl IndexEntryBuilder { |
|
1050 | impl IndexEntryBuilder { | |
881 | #[allow(clippy::new_without_default)] |
|
1051 | #[allow(clippy::new_without_default)] | |
882 | pub fn new() -> Self { |
|
1052 | pub fn new() -> Self { | |
883 | Self { |
|
1053 | Self { | |
884 | is_first: false, |
|
1054 | is_first: false, | |
885 | is_inline: false, |
|
1055 | is_inline: false, | |
886 | is_general_delta: true, |
|
1056 | is_general_delta: true, | |
887 | version: 1, |
|
1057 | version: 1, | |
888 | offset: 0, |
|
1058 | offset: 0, | |
889 | compressed_len: 0, |
|
1059 | compressed_len: 0, | |
890 | uncompressed_len: 0, |
|
1060 | uncompressed_len: 0, | |
891 | base_revision_or_base_of_delta_chain: Revision(0), |
|
1061 | base_revision_or_base_of_delta_chain: Revision(0), | |
892 | link_revision: Revision(0), |
|
1062 | link_revision: Revision(0), | |
893 | p1: NULL_REVISION, |
|
1063 | p1: NULL_REVISION, | |
894 | p2: NULL_REVISION, |
|
1064 | p2: NULL_REVISION, | |
895 | node: NULL_NODE, |
|
1065 | node: NULL_NODE, | |
896 | } |
|
1066 | } | |
897 | } |
|
1067 | } | |
898 |
|
1068 | |||
899 | pub fn is_first(&mut self, value: bool) -> &mut Self { |
|
1069 | pub fn is_first(&mut self, value: bool) -> &mut Self { | |
900 | self.is_first = value; |
|
1070 | self.is_first = value; | |
901 | self |
|
1071 | self | |
902 | } |
|
1072 | } | |
903 |
|
1073 | |||
904 | pub fn with_inline(&mut self, value: bool) -> &mut Self { |
|
1074 | pub fn with_inline(&mut self, value: bool) -> &mut Self { | |
905 | self.is_inline = value; |
|
1075 | self.is_inline = value; | |
906 | self |
|
1076 | self | |
907 | } |
|
1077 | } | |
908 |
|
1078 | |||
909 | pub fn with_general_delta(&mut self, value: bool) -> &mut Self { |
|
1079 | pub fn with_general_delta(&mut self, value: bool) -> &mut Self { | |
910 | self.is_general_delta = value; |
|
1080 | self.is_general_delta = value; | |
911 | self |
|
1081 | self | |
912 | } |
|
1082 | } | |
913 |
|
1083 | |||
914 | pub fn with_version(&mut self, value: u16) -> &mut Self { |
|
1084 | pub fn with_version(&mut self, value: u16) -> &mut Self { | |
915 | self.version = value; |
|
1085 | self.version = value; | |
916 | self |
|
1086 | self | |
917 | } |
|
1087 | } | |
918 |
|
1088 | |||
919 | pub fn with_offset(&mut self, value: usize) -> &mut Self { |
|
1089 | pub fn with_offset(&mut self, value: usize) -> &mut Self { | |
920 | self.offset = value; |
|
1090 | self.offset = value; | |
921 | self |
|
1091 | self | |
922 | } |
|
1092 | } | |
923 |
|
1093 | |||
924 | pub fn with_compressed_len(&mut self, value: usize) -> &mut Self { |
|
1094 | pub fn with_compressed_len(&mut self, value: usize) -> &mut Self { | |
925 | self.compressed_len = value; |
|
1095 | self.compressed_len = value; | |
926 | self |
|
1096 | self | |
927 | } |
|
1097 | } | |
928 |
|
1098 | |||
929 | pub fn with_uncompressed_len(&mut self, value: usize) -> &mut Self { |
|
1099 | pub fn with_uncompressed_len(&mut self, value: usize) -> &mut Self { | |
930 | self.uncompressed_len = value; |
|
1100 | self.uncompressed_len = value; | |
931 | self |
|
1101 | self | |
932 | } |
|
1102 | } | |
933 |
|
1103 | |||
934 | pub fn with_base_revision_or_base_of_delta_chain( |
|
1104 | pub fn with_base_revision_or_base_of_delta_chain( | |
935 | &mut self, |
|
1105 | &mut self, | |
936 | value: Revision, |
|
1106 | value: Revision, | |
937 | ) -> &mut Self { |
|
1107 | ) -> &mut Self { | |
938 | self.base_revision_or_base_of_delta_chain = value; |
|
1108 | self.base_revision_or_base_of_delta_chain = value; | |
939 | self |
|
1109 | self | |
940 | } |
|
1110 | } | |
941 |
|
1111 | |||
942 | pub fn with_link_revision(&mut self, value: Revision) -> &mut Self { |
|
1112 | pub fn with_link_revision(&mut self, value: Revision) -> &mut Self { | |
943 | self.link_revision = value; |
|
1113 | self.link_revision = value; | |
944 | self |
|
1114 | self | |
945 | } |
|
1115 | } | |
946 |
|
1116 | |||
947 | pub fn with_p1(&mut self, value: Revision) -> &mut Self { |
|
1117 | pub fn with_p1(&mut self, value: Revision) -> &mut Self { | |
948 | self.p1 = value; |
|
1118 | self.p1 = value; | |
949 | self |
|
1119 | self | |
950 | } |
|
1120 | } | |
951 |
|
1121 | |||
952 | pub fn with_p2(&mut self, value: Revision) -> &mut Self { |
|
1122 | pub fn with_p2(&mut self, value: Revision) -> &mut Self { | |
953 | self.p2 = value; |
|
1123 | self.p2 = value; | |
954 | self |
|
1124 | self | |
955 | } |
|
1125 | } | |
956 |
|
1126 | |||
957 | pub fn with_node(&mut self, value: Node) -> &mut Self { |
|
1127 | pub fn with_node(&mut self, value: Node) -> &mut Self { | |
958 | self.node = value; |
|
1128 | self.node = value; | |
959 | self |
|
1129 | self | |
960 | } |
|
1130 | } | |
961 |
|
1131 | |||
962 | pub fn build(&self) -> Vec<u8> { |
|
1132 | pub fn build(&self) -> Vec<u8> { | |
963 | let mut bytes = Vec::with_capacity(INDEX_ENTRY_SIZE); |
|
1133 | let mut bytes = Vec::with_capacity(INDEX_ENTRY_SIZE); | |
964 | if self.is_first { |
|
1134 | if self.is_first { | |
965 | bytes.extend(&match (self.is_general_delta, self.is_inline) { |
|
1135 | bytes.extend(&match (self.is_general_delta, self.is_inline) { | |
966 | (false, false) => [0u8, 0], |
|
1136 | (false, false) => [0u8, 0], | |
967 | (false, true) => [0u8, 1], |
|
1137 | (false, true) => [0u8, 1], | |
968 | (true, false) => [0u8, 2], |
|
1138 | (true, false) => [0u8, 2], | |
969 | (true, true) => [0u8, 3], |
|
1139 | (true, true) => [0u8, 3], | |
970 | }); |
|
1140 | }); | |
971 | bytes.extend(&self.version.to_be_bytes()); |
|
1141 | bytes.extend(&self.version.to_be_bytes()); | |
972 | // Remaining offset bytes. |
|
1142 | // Remaining offset bytes. | |
973 | bytes.extend(&[0u8; 2]); |
|
1143 | bytes.extend(&[0u8; 2]); | |
974 | } else { |
|
1144 | } else { | |
975 | // Offset stored on 48 bits (6 bytes) |
|
1145 | // Offset stored on 48 bits (6 bytes) | |
976 | bytes.extend(&(self.offset as u64).to_be_bytes()[2..]); |
|
1146 | bytes.extend(&(self.offset as u64).to_be_bytes()[2..]); | |
977 | } |
|
1147 | } | |
978 | bytes.extend(&[0u8; 2]); // Revision flags. |
|
1148 | bytes.extend(&[0u8; 2]); // Revision flags. | |
979 | bytes.extend(&(self.compressed_len as u32).to_be_bytes()); |
|
1149 | bytes.extend(&(self.compressed_len as u32).to_be_bytes()); | |
980 | bytes.extend(&(self.uncompressed_len as u32).to_be_bytes()); |
|
1150 | bytes.extend(&(self.uncompressed_len as u32).to_be_bytes()); | |
981 | bytes.extend( |
|
1151 | bytes.extend( | |
982 | &self.base_revision_or_base_of_delta_chain.0.to_be_bytes(), |
|
1152 | &self.base_revision_or_base_of_delta_chain.0.to_be_bytes(), | |
983 | ); |
|
1153 | ); | |
984 | bytes.extend(&self.link_revision.0.to_be_bytes()); |
|
1154 | bytes.extend(&self.link_revision.0.to_be_bytes()); | |
985 | bytes.extend(&self.p1.0.to_be_bytes()); |
|
1155 | bytes.extend(&self.p1.0.to_be_bytes()); | |
986 | bytes.extend(&self.p2.0.to_be_bytes()); |
|
1156 | bytes.extend(&self.p2.0.to_be_bytes()); | |
987 | bytes.extend(self.node.as_bytes()); |
|
1157 | bytes.extend(self.node.as_bytes()); | |
988 | bytes.extend(vec![0u8; 12]); |
|
1158 | bytes.extend(vec![0u8; 12]); | |
989 | bytes |
|
1159 | bytes | |
990 | } |
|
1160 | } | |
991 | } |
|
1161 | } | |
992 |
|
1162 | |||
993 | pub fn is_inline(index_bytes: &[u8]) -> bool { |
|
1163 | pub fn is_inline(index_bytes: &[u8]) -> bool { | |
994 | IndexHeader::parse(index_bytes) |
|
1164 | IndexHeader::parse(index_bytes) | |
995 | .expect("too short") |
|
1165 | .expect("too short") | |
996 | .unwrap() |
|
1166 | .unwrap() | |
997 | .format_flags() |
|
1167 | .format_flags() | |
998 | .is_inline() |
|
1168 | .is_inline() | |
999 | } |
|
1169 | } | |
1000 |
|
1170 | |||
1001 | pub fn uses_generaldelta(index_bytes: &[u8]) -> bool { |
|
1171 | pub fn uses_generaldelta(index_bytes: &[u8]) -> bool { | |
1002 | IndexHeader::parse(index_bytes) |
|
1172 | IndexHeader::parse(index_bytes) | |
1003 | .expect("too short") |
|
1173 | .expect("too short") | |
1004 | .unwrap() |
|
1174 | .unwrap() | |
1005 | .format_flags() |
|
1175 | .format_flags() | |
1006 | .uses_generaldelta() |
|
1176 | .uses_generaldelta() | |
1007 | } |
|
1177 | } | |
1008 |
|
1178 | |||
1009 | pub fn get_version(index_bytes: &[u8]) -> u16 { |
|
1179 | pub fn get_version(index_bytes: &[u8]) -> u16 { | |
1010 | IndexHeader::parse(index_bytes) |
|
1180 | IndexHeader::parse(index_bytes) | |
1011 | .expect("too short") |
|
1181 | .expect("too short") | |
1012 | .unwrap() |
|
1182 | .unwrap() | |
1013 | .format_version() |
|
1183 | .format_version() | |
1014 | } |
|
1184 | } | |
1015 |
|
1185 | |||
1016 | #[test] |
|
1186 | #[test] | |
1017 | fn flags_when_no_inline_flag_test() { |
|
1187 | fn flags_when_no_inline_flag_test() { | |
1018 | let bytes = IndexEntryBuilder::new() |
|
1188 | let bytes = IndexEntryBuilder::new() | |
1019 | .is_first(true) |
|
1189 | .is_first(true) | |
1020 | .with_general_delta(false) |
|
1190 | .with_general_delta(false) | |
1021 | .with_inline(false) |
|
1191 | .with_inline(false) | |
1022 | .build(); |
|
1192 | .build(); | |
1023 |
|
1193 | |||
1024 | assert!(!is_inline(&bytes)); |
|
1194 | assert!(!is_inline(&bytes)); | |
1025 | assert!(!uses_generaldelta(&bytes)); |
|
1195 | assert!(!uses_generaldelta(&bytes)); | |
1026 | } |
|
1196 | } | |
1027 |
|
1197 | |||
1028 | #[test] |
|
1198 | #[test] | |
1029 | fn flags_when_inline_flag_test() { |
|
1199 | fn flags_when_inline_flag_test() { | |
1030 | let bytes = IndexEntryBuilder::new() |
|
1200 | let bytes = IndexEntryBuilder::new() | |
1031 | .is_first(true) |
|
1201 | .is_first(true) | |
1032 | .with_general_delta(false) |
|
1202 | .with_general_delta(false) | |
1033 | .with_inline(true) |
|
1203 | .with_inline(true) | |
1034 | .build(); |
|
1204 | .build(); | |
1035 |
|
1205 | |||
1036 | assert!(is_inline(&bytes)); |
|
1206 | assert!(is_inline(&bytes)); | |
1037 | assert!(!uses_generaldelta(&bytes)); |
|
1207 | assert!(!uses_generaldelta(&bytes)); | |
1038 | } |
|
1208 | } | |
1039 |
|
1209 | |||
1040 | #[test] |
|
1210 | #[test] | |
1041 | fn flags_when_inline_and_generaldelta_flags_test() { |
|
1211 | fn flags_when_inline_and_generaldelta_flags_test() { | |
1042 | let bytes = IndexEntryBuilder::new() |
|
1212 | let bytes = IndexEntryBuilder::new() | |
1043 | .is_first(true) |
|
1213 | .is_first(true) | |
1044 | .with_general_delta(true) |
|
1214 | .with_general_delta(true) | |
1045 | .with_inline(true) |
|
1215 | .with_inline(true) | |
1046 | .build(); |
|
1216 | .build(); | |
1047 |
|
1217 | |||
1048 | assert!(is_inline(&bytes)); |
|
1218 | assert!(is_inline(&bytes)); | |
1049 | assert!(uses_generaldelta(&bytes)); |
|
1219 | assert!(uses_generaldelta(&bytes)); | |
1050 | } |
|
1220 | } | |
1051 |
|
1221 | |||
1052 | #[test] |
|
1222 | #[test] | |
1053 | fn test_offset() { |
|
1223 | fn test_offset() { | |
1054 | let bytes = IndexEntryBuilder::new().with_offset(1).build(); |
|
1224 | let bytes = IndexEntryBuilder::new().with_offset(1).build(); | |
1055 | let entry = IndexEntry { |
|
1225 | let entry = IndexEntry { | |
1056 | bytes: &bytes, |
|
1226 | bytes: &bytes, | |
1057 | offset_override: None, |
|
1227 | offset_override: None, | |
1058 | }; |
|
1228 | }; | |
1059 |
|
1229 | |||
1060 | assert_eq!(entry.offset(), 1) |
|
1230 | assert_eq!(entry.offset(), 1) | |
1061 | } |
|
1231 | } | |
1062 |
|
1232 | |||
1063 | #[test] |
|
1233 | #[test] | |
1064 | fn test_with_overridden_offset() { |
|
1234 | fn test_with_overridden_offset() { | |
1065 | let bytes = IndexEntryBuilder::new().with_offset(1).build(); |
|
1235 | let bytes = IndexEntryBuilder::new().with_offset(1).build(); | |
1066 | let entry = IndexEntry { |
|
1236 | let entry = IndexEntry { | |
1067 | bytes: &bytes, |
|
1237 | bytes: &bytes, | |
1068 | offset_override: Some(2), |
|
1238 | offset_override: Some(2), | |
1069 | }; |
|
1239 | }; | |
1070 |
|
1240 | |||
1071 | assert_eq!(entry.offset(), 2) |
|
1241 | assert_eq!(entry.offset(), 2) | |
1072 | } |
|
1242 | } | |
1073 |
|
1243 | |||
1074 | #[test] |
|
1244 | #[test] | |
1075 | fn test_compressed_len() { |
|
1245 | fn test_compressed_len() { | |
1076 | let bytes = IndexEntryBuilder::new().with_compressed_len(1).build(); |
|
1246 | let bytes = IndexEntryBuilder::new().with_compressed_len(1).build(); | |
1077 | let entry = IndexEntry { |
|
1247 | let entry = IndexEntry { | |
1078 | bytes: &bytes, |
|
1248 | bytes: &bytes, | |
1079 | offset_override: None, |
|
1249 | offset_override: None, | |
1080 | }; |
|
1250 | }; | |
1081 |
|
1251 | |||
1082 | assert_eq!(entry.compressed_len(), 1) |
|
1252 | assert_eq!(entry.compressed_len(), 1) | |
1083 | } |
|
1253 | } | |
1084 |
|
1254 | |||
1085 | #[test] |
|
1255 | #[test] | |
1086 | fn test_uncompressed_len() { |
|
1256 | fn test_uncompressed_len() { | |
1087 | let bytes = IndexEntryBuilder::new().with_uncompressed_len(1).build(); |
|
1257 | let bytes = IndexEntryBuilder::new().with_uncompressed_len(1).build(); | |
1088 | let entry = IndexEntry { |
|
1258 | let entry = IndexEntry { | |
1089 | bytes: &bytes, |
|
1259 | bytes: &bytes, | |
1090 | offset_override: None, |
|
1260 | offset_override: None, | |
1091 | }; |
|
1261 | }; | |
1092 |
|
1262 | |||
1093 | assert_eq!(entry.uncompressed_len(), 1) |
|
1263 | assert_eq!(entry.uncompressed_len(), 1) | |
1094 | } |
|
1264 | } | |
1095 |
|
1265 | |||
1096 | #[test] |
|
1266 | #[test] | |
1097 | fn test_base_revision_or_base_of_delta_chain() { |
|
1267 | fn test_base_revision_or_base_of_delta_chain() { | |
1098 | let bytes = IndexEntryBuilder::new() |
|
1268 | let bytes = IndexEntryBuilder::new() | |
1099 | .with_base_revision_or_base_of_delta_chain(Revision(1)) |
|
1269 | .with_base_revision_or_base_of_delta_chain(Revision(1)) | |
1100 | .build(); |
|
1270 | .build(); | |
1101 | let entry = IndexEntry { |
|
1271 | let entry = IndexEntry { | |
1102 | bytes: &bytes, |
|
1272 | bytes: &bytes, | |
1103 | offset_override: None, |
|
1273 | offset_override: None, | |
1104 | }; |
|
1274 | }; | |
1105 |
|
1275 | |||
1106 | assert_eq!(entry.base_revision_or_base_of_delta_chain(), 1.into()) |
|
1276 | assert_eq!(entry.base_revision_or_base_of_delta_chain(), 1.into()) | |
1107 | } |
|
1277 | } | |
1108 |
|
1278 | |||
1109 | #[test] |
|
1279 | #[test] | |
1110 | fn link_revision_test() { |
|
1280 | fn link_revision_test() { | |
1111 | let bytes = IndexEntryBuilder::new() |
|
1281 | let bytes = IndexEntryBuilder::new() | |
1112 | .with_link_revision(Revision(123)) |
|
1282 | .with_link_revision(Revision(123)) | |
1113 | .build(); |
|
1283 | .build(); | |
1114 |
|
1284 | |||
1115 | let entry = IndexEntry { |
|
1285 | let entry = IndexEntry { | |
1116 | bytes: &bytes, |
|
1286 | bytes: &bytes, | |
1117 | offset_override: None, |
|
1287 | offset_override: None, | |
1118 | }; |
|
1288 | }; | |
1119 |
|
1289 | |||
1120 | assert_eq!(entry.link_revision(), 123.into()); |
|
1290 | assert_eq!(entry.link_revision(), 123.into()); | |
1121 | } |
|
1291 | } | |
1122 |
|
1292 | |||
1123 | #[test] |
|
1293 | #[test] | |
1124 | fn p1_test() { |
|
1294 | fn p1_test() { | |
1125 | let bytes = IndexEntryBuilder::new().with_p1(Revision(123)).build(); |
|
1295 | let bytes = IndexEntryBuilder::new().with_p1(Revision(123)).build(); | |
1126 |
|
1296 | |||
1127 | let entry = IndexEntry { |
|
1297 | let entry = IndexEntry { | |
1128 | bytes: &bytes, |
|
1298 | bytes: &bytes, | |
1129 | offset_override: None, |
|
1299 | offset_override: None, | |
1130 | }; |
|
1300 | }; | |
1131 |
|
1301 | |||
1132 | assert_eq!(entry.p1(), 123.into()); |
|
1302 | assert_eq!(entry.p1(), 123.into()); | |
1133 | } |
|
1303 | } | |
1134 |
|
1304 | |||
1135 | #[test] |
|
1305 | #[test] | |
1136 | fn p2_test() { |
|
1306 | fn p2_test() { | |
1137 | let bytes = IndexEntryBuilder::new().with_p2(Revision(123)).build(); |
|
1307 | let bytes = IndexEntryBuilder::new().with_p2(Revision(123)).build(); | |
1138 |
|
1308 | |||
1139 | let entry = IndexEntry { |
|
1309 | let entry = IndexEntry { | |
1140 | bytes: &bytes, |
|
1310 | bytes: &bytes, | |
1141 | offset_override: None, |
|
1311 | offset_override: None, | |
1142 | }; |
|
1312 | }; | |
1143 |
|
1313 | |||
1144 | assert_eq!(entry.p2(), 123.into()); |
|
1314 | assert_eq!(entry.p2(), 123.into()); | |
1145 | } |
|
1315 | } | |
1146 |
|
1316 | |||
1147 | #[test] |
|
1317 | #[test] | |
1148 | fn node_test() { |
|
1318 | fn node_test() { | |
1149 | let node = Node::from_hex("0123456789012345678901234567890123456789") |
|
1319 | let node = Node::from_hex("0123456789012345678901234567890123456789") | |
1150 | .unwrap(); |
|
1320 | .unwrap(); | |
1151 | let bytes = IndexEntryBuilder::new().with_node(node).build(); |
|
1321 | let bytes = IndexEntryBuilder::new().with_node(node).build(); | |
1152 |
|
1322 | |||
1153 | let entry = IndexEntry { |
|
1323 | let entry = IndexEntry { | |
1154 | bytes: &bytes, |
|
1324 | bytes: &bytes, | |
1155 | offset_override: None, |
|
1325 | offset_override: None, | |
1156 | }; |
|
1326 | }; | |
1157 |
|
1327 | |||
1158 | assert_eq!(*entry.hash(), node); |
|
1328 | assert_eq!(*entry.hash(), node); | |
1159 | } |
|
1329 | } | |
1160 |
|
1330 | |||
1161 | #[test] |
|
1331 | #[test] | |
1162 | fn version_test() { |
|
1332 | fn version_test() { | |
1163 | let bytes = IndexEntryBuilder::new() |
|
1333 | let bytes = IndexEntryBuilder::new() | |
1164 | .is_first(true) |
|
1334 | .is_first(true) | |
1165 | .with_version(2) |
|
1335 | .with_version(2) | |
1166 | .build(); |
|
1336 | .build(); | |
1167 |
|
1337 | |||
1168 | assert_eq!(get_version(&bytes), 2) |
|
1338 | assert_eq!(get_version(&bytes), 2) | |
1169 | } |
|
1339 | } | |
1170 | } |
|
1340 | } | |
1171 |
|
1341 | |||
1172 | #[cfg(test)] |
|
1342 | #[cfg(test)] | |
1173 | pub use tests::IndexEntryBuilder; |
|
1343 | pub use tests::IndexEntryBuilder; |
@@ -1,928 +1,970 b'' | |||||
1 | // revlog.rs |
|
1 | // revlog.rs | |
2 | // |
|
2 | // | |
3 | // Copyright 2019-2020 Georges Racinet <georges.racinet@octobus.net> |
|
3 | // Copyright 2019-2020 Georges Racinet <georges.racinet@octobus.net> | |
4 | // |
|
4 | // | |
5 | // This software may be used and distributed according to the terms of the |
|
5 | // This software may be used and distributed according to the terms of the | |
6 | // GNU General Public License version 2 or any later version. |
|
6 | // GNU General Public License version 2 or any later version. | |
7 |
|
7 | |||
8 | use crate::{ |
|
8 | use crate::{ | |
9 | cindex, |
|
9 | cindex, | |
10 | conversion::rev_pyiter_collect, |
|
10 | conversion::rev_pyiter_collect, | |
11 | utils::{node_from_py_bytes, node_from_py_object}, |
|
11 | utils::{node_from_py_bytes, node_from_py_object}, | |
12 | PyRevision, |
|
12 | PyRevision, | |
13 | }; |
|
13 | }; | |
14 | use cpython::{ |
|
14 | use cpython::{ | |
15 | buffer::{Element, PyBuffer}, |
|
15 | buffer::{Element, PyBuffer}, | |
16 | exc::{IndexError, ValueError}, |
|
16 | exc::{IndexError, ValueError}, | |
17 | ObjectProtocol, PyBool, PyBytes, PyClone, PyDict, PyErr, PyInt, PyList, |
|
17 | ObjectProtocol, PyBool, PyBytes, PyClone, PyDict, PyErr, PyInt, PyList, | |
18 | PyModule, PyObject, PyResult, PySet, PyString, PyTuple, Python, |
|
18 | PyModule, PyObject, PyResult, PySet, PyString, PyTuple, Python, | |
19 | PythonObject, ToPyObject, |
|
19 | PythonObject, ToPyObject, | |
20 | }; |
|
20 | }; | |
21 | use hg::{ |
|
21 | use hg::{ | |
22 | errors::HgError, |
|
22 | errors::HgError, | |
23 | index::{IndexHeader, RevisionDataParams, SnapshotsCache}, |
|
23 | index::{IndexHeader, RevisionDataParams, SnapshotsCache}, | |
24 | nodemap::{Block, NodeMapError, NodeTree}, |
|
24 | nodemap::{Block, NodeMapError, NodeTree}, | |
25 | revlog::{nodemap::NodeMap, NodePrefix, RevlogError, RevlogIndex}, |
|
25 | revlog::{nodemap::NodeMap, NodePrefix, RevlogError, RevlogIndex}, | |
26 | BaseRevision, Revision, UncheckedRevision, NULL_REVISION, |
|
26 | BaseRevision, Revision, UncheckedRevision, NULL_REVISION, | |
27 | }; |
|
27 | }; | |
28 | use std::cell::RefCell; |
|
28 | use std::cell::RefCell; | |
29 |
|
29 | |||
30 | /// Return a Struct implementing the Graph trait |
|
30 | /// Return a Struct implementing the Graph trait | |
31 | pub(crate) fn pyindex_to_graph( |
|
31 | pub(crate) fn pyindex_to_graph( | |
32 | py: Python, |
|
32 | py: Python, | |
33 | index: PyObject, |
|
33 | index: PyObject, | |
34 | ) -> PyResult<cindex::Index> { |
|
34 | ) -> PyResult<cindex::Index> { | |
35 | match index.extract::<MixedIndex>(py) { |
|
35 | match index.extract::<MixedIndex>(py) { | |
36 | Ok(midx) => Ok(midx.clone_cindex(py)), |
|
36 | Ok(midx) => Ok(midx.clone_cindex(py)), | |
37 | Err(_) => cindex::Index::new(py, index), |
|
37 | Err(_) => cindex::Index::new(py, index), | |
38 | } |
|
38 | } | |
39 | } |
|
39 | } | |
40 |
|
40 | |||
41 | py_class!(pub class MixedIndex |py| { |
|
41 | py_class!(pub class MixedIndex |py| { | |
42 | data cindex: RefCell<cindex::Index>; |
|
42 | data cindex: RefCell<cindex::Index>; | |
43 | data index: RefCell<hg::index::Index>; |
|
43 | data index: RefCell<hg::index::Index>; | |
44 | data nt: RefCell<Option<NodeTree>>; |
|
44 | data nt: RefCell<Option<NodeTree>>; | |
45 | data docket: RefCell<Option<PyObject>>; |
|
45 | data docket: RefCell<Option<PyObject>>; | |
46 | // Holds a reference to the mmap'ed persistent nodemap data |
|
46 | // Holds a reference to the mmap'ed persistent nodemap data | |
47 | data nodemap_mmap: RefCell<Option<PyBuffer>>; |
|
47 | data nodemap_mmap: RefCell<Option<PyBuffer>>; | |
48 | // Holds a reference to the mmap'ed persistent index data |
|
48 | // Holds a reference to the mmap'ed persistent index data | |
49 | data index_mmap: RefCell<Option<PyBuffer>>; |
|
49 | data index_mmap: RefCell<Option<PyBuffer>>; | |
50 |
|
50 | |||
51 | def __new__( |
|
51 | def __new__( | |
52 | _cls, |
|
52 | _cls, | |
53 | cindex: PyObject, |
|
53 | cindex: PyObject, | |
54 | data: PyObject, |
|
54 | data: PyObject, | |
55 | default_header: u32, |
|
55 | default_header: u32, | |
56 | ) -> PyResult<MixedIndex> { |
|
56 | ) -> PyResult<MixedIndex> { | |
57 | Self::new(py, cindex, data, default_header) |
|
57 | Self::new(py, cindex, data, default_header) | |
58 | } |
|
58 | } | |
59 |
|
59 | |||
60 | /// Compatibility layer used for Python consumers needing access to the C index |
|
60 | /// Compatibility layer used for Python consumers needing access to the C index | |
61 | /// |
|
61 | /// | |
62 | /// Only use case so far is `scmutil.shortesthexnodeidprefix`, |
|
62 | /// Only use case so far is `scmutil.shortesthexnodeidprefix`, | |
63 | /// that may need to build a custom `nodetree`, based on a specified revset. |
|
63 | /// that may need to build a custom `nodetree`, based on a specified revset. | |
64 | /// With a Rust implementation of the nodemap, we will be able to get rid of |
|
64 | /// With a Rust implementation of the nodemap, we will be able to get rid of | |
65 | /// this, by exposing our own standalone nodemap class, |
|
65 | /// this, by exposing our own standalone nodemap class, | |
66 | /// ready to accept `MixedIndex`. |
|
66 | /// ready to accept `MixedIndex`. | |
67 | def get_cindex(&self) -> PyResult<PyObject> { |
|
67 | def get_cindex(&self) -> PyResult<PyObject> { | |
68 | Ok(self.cindex(py).borrow().inner().clone_ref(py)) |
|
68 | Ok(self.cindex(py).borrow().inner().clone_ref(py)) | |
69 | } |
|
69 | } | |
70 |
|
70 | |||
71 | // Index API involving nodemap, as defined in mercurial/pure/parsers.py |
|
71 | // Index API involving nodemap, as defined in mercurial/pure/parsers.py | |
72 |
|
72 | |||
73 | /// Return Revision if found, raises a bare `error.RevlogError` |
|
73 | /// Return Revision if found, raises a bare `error.RevlogError` | |
74 | /// in case of ambiguity, same as C version does |
|
74 | /// in case of ambiguity, same as C version does | |
75 | def get_rev(&self, node: PyBytes) -> PyResult<Option<PyRevision>> { |
|
75 | def get_rev(&self, node: PyBytes) -> PyResult<Option<PyRevision>> { | |
76 | let opt = self.get_nodetree(py)?.borrow(); |
|
76 | let opt = self.get_nodetree(py)?.borrow(); | |
77 | let nt = opt.as_ref().unwrap(); |
|
77 | let nt = opt.as_ref().unwrap(); | |
78 | let idx = &*self.cindex(py).borrow(); |
|
78 | let idx = &*self.cindex(py).borrow(); | |
79 | let ridx = &*self.index(py).borrow(); |
|
79 | let ridx = &*self.index(py).borrow(); | |
80 | let node = node_from_py_bytes(py, &node)?; |
|
80 | let node = node_from_py_bytes(py, &node)?; | |
81 | let rust_rev = |
|
81 | let rust_rev = | |
82 | nt.find_bin(ridx, node.into()).map_err(|e| nodemap_error(py, e))?; |
|
82 | nt.find_bin(ridx, node.into()).map_err(|e| nodemap_error(py, e))?; | |
83 | let c_rev = |
|
83 | let c_rev = | |
84 | nt.find_bin(idx, node.into()).map_err(|e| nodemap_error(py, e))?; |
|
84 | nt.find_bin(idx, node.into()).map_err(|e| nodemap_error(py, e))?; | |
85 | assert_eq!(rust_rev, c_rev); |
|
85 | assert_eq!(rust_rev, c_rev); | |
86 | Ok(rust_rev.map(Into::into)) |
|
86 | Ok(rust_rev.map(Into::into)) | |
87 |
|
87 | |||
88 | } |
|
88 | } | |
89 |
|
89 | |||
90 | /// same as `get_rev()` but raises a bare `error.RevlogError` if node |
|
90 | /// same as `get_rev()` but raises a bare `error.RevlogError` if node | |
91 | /// is not found. |
|
91 | /// is not found. | |
92 | /// |
|
92 | /// | |
93 | /// No need to repeat `node` in the exception, `mercurial/revlog.py` |
|
93 | /// No need to repeat `node` in the exception, `mercurial/revlog.py` | |
94 | /// will catch and rewrap with it |
|
94 | /// will catch and rewrap with it | |
95 | def rev(&self, node: PyBytes) -> PyResult<PyRevision> { |
|
95 | def rev(&self, node: PyBytes) -> PyResult<PyRevision> { | |
96 | self.get_rev(py, node)?.ok_or_else(|| revlog_error(py)) |
|
96 | self.get_rev(py, node)?.ok_or_else(|| revlog_error(py)) | |
97 | } |
|
97 | } | |
98 |
|
98 | |||
99 | /// return True if the node exist in the index |
|
99 | /// return True if the node exist in the index | |
100 | def has_node(&self, node: PyBytes) -> PyResult<bool> { |
|
100 | def has_node(&self, node: PyBytes) -> PyResult<bool> { | |
101 | // TODO OPTIM we could avoid a needless conversion here, |
|
101 | // TODO OPTIM we could avoid a needless conversion here, | |
102 | // to do when scaffolding for pure Rust switch is removed, |
|
102 | // to do when scaffolding for pure Rust switch is removed, | |
103 | // as `get_rev()` currently does the necessary assertions |
|
103 | // as `get_rev()` currently does the necessary assertions | |
104 | self.get_rev(py, node).map(|opt| opt.is_some()) |
|
104 | self.get_rev(py, node).map(|opt| opt.is_some()) | |
105 | } |
|
105 | } | |
106 |
|
106 | |||
107 | /// find length of shortest hex nodeid of a binary ID |
|
107 | /// find length of shortest hex nodeid of a binary ID | |
108 | def shortest(&self, node: PyBytes) -> PyResult<usize> { |
|
108 | def shortest(&self, node: PyBytes) -> PyResult<usize> { | |
109 | let opt = self.get_nodetree(py)?.borrow(); |
|
109 | let opt = self.get_nodetree(py)?.borrow(); | |
110 | let nt = opt.as_ref().unwrap(); |
|
110 | let nt = opt.as_ref().unwrap(); | |
111 | let idx = &*self.index(py).borrow(); |
|
111 | let idx = &*self.index(py).borrow(); | |
112 | match nt.unique_prefix_len_node(idx, &node_from_py_bytes(py, &node)?) |
|
112 | match nt.unique_prefix_len_node(idx, &node_from_py_bytes(py, &node)?) | |
113 | { |
|
113 | { | |
114 | Ok(Some(l)) => Ok(l), |
|
114 | Ok(Some(l)) => Ok(l), | |
115 | Ok(None) => Err(revlog_error(py)), |
|
115 | Ok(None) => Err(revlog_error(py)), | |
116 | Err(e) => Err(nodemap_error(py, e)), |
|
116 | Err(e) => Err(nodemap_error(py, e)), | |
117 | } |
|
117 | } | |
118 | } |
|
118 | } | |
119 |
|
119 | |||
120 | def partialmatch(&self, node: PyObject) -> PyResult<Option<PyBytes>> { |
|
120 | def partialmatch(&self, node: PyObject) -> PyResult<Option<PyBytes>> { | |
121 | let opt = self.get_nodetree(py)?.borrow(); |
|
121 | let opt = self.get_nodetree(py)?.borrow(); | |
122 | let nt = opt.as_ref().unwrap(); |
|
122 | let nt = opt.as_ref().unwrap(); | |
123 | let idx = &*self.index(py).borrow(); |
|
123 | let idx = &*self.index(py).borrow(); | |
124 |
|
124 | |||
125 | let node_as_string = if cfg!(feature = "python3-sys") { |
|
125 | let node_as_string = if cfg!(feature = "python3-sys") { | |
126 | node.cast_as::<PyString>(py)?.to_string(py)?.to_string() |
|
126 | node.cast_as::<PyString>(py)?.to_string(py)?.to_string() | |
127 | } |
|
127 | } | |
128 | else { |
|
128 | else { | |
129 | let node = node.extract::<PyBytes>(py)?; |
|
129 | let node = node.extract::<PyBytes>(py)?; | |
130 | String::from_utf8_lossy(node.data(py)).to_string() |
|
130 | String::from_utf8_lossy(node.data(py)).to_string() | |
131 | }; |
|
131 | }; | |
132 |
|
132 | |||
133 | let prefix = NodePrefix::from_hex(&node_as_string) |
|
133 | let prefix = NodePrefix::from_hex(&node_as_string) | |
134 | .map_err(|_| PyErr::new::<ValueError, _>( |
|
134 | .map_err(|_| PyErr::new::<ValueError, _>( | |
135 | py, format!("Invalid node or prefix '{}'", node_as_string)) |
|
135 | py, format!("Invalid node or prefix '{}'", node_as_string)) | |
136 | )?; |
|
136 | )?; | |
137 |
|
137 | |||
138 | nt.find_bin(idx, prefix) |
|
138 | nt.find_bin(idx, prefix) | |
139 | // TODO make an inner API returning the node directly |
|
139 | // TODO make an inner API returning the node directly | |
140 | .map(|opt| opt.map( |
|
140 | .map(|opt| opt.map( | |
141 | |rev| PyBytes::new(py, idx.node(rev).unwrap().as_bytes()))) |
|
141 | |rev| PyBytes::new(py, idx.node(rev).unwrap().as_bytes()))) | |
142 | .map_err(|e| nodemap_error(py, e)) |
|
142 | .map_err(|e| nodemap_error(py, e)) | |
143 |
|
143 | |||
144 | } |
|
144 | } | |
145 |
|
145 | |||
146 | /// append an index entry |
|
146 | /// append an index entry | |
147 | def append(&self, tup: PyTuple) -> PyResult<PyObject> { |
|
147 | def append(&self, tup: PyTuple) -> PyResult<PyObject> { | |
148 | if tup.len(py) < 8 { |
|
148 | if tup.len(py) < 8 { | |
149 | // this is better than the panic promised by tup.get_item() |
|
149 | // this is better than the panic promised by tup.get_item() | |
150 | return Err( |
|
150 | return Err( | |
151 | PyErr::new::<IndexError, _>(py, "tuple index out of range")) |
|
151 | PyErr::new::<IndexError, _>(py, "tuple index out of range")) | |
152 | } |
|
152 | } | |
153 | let node_bytes = tup.get_item(py, 7).extract(py)?; |
|
153 | let node_bytes = tup.get_item(py, 7).extract(py)?; | |
154 | let node = node_from_py_object(py, &node_bytes)?; |
|
154 | let node = node_from_py_object(py, &node_bytes)?; | |
155 |
|
155 | |||
156 | let rev = self.len(py)? as BaseRevision; |
|
156 | let rev = self.len(py)? as BaseRevision; | |
157 | let mut idx = self.cindex(py).borrow_mut(); |
|
157 | let mut idx = self.cindex(py).borrow_mut(); | |
158 |
|
158 | |||
159 | // This is ok since we will just add the revision to the index |
|
159 | // This is ok since we will just add the revision to the index | |
160 | let rev = Revision(rev); |
|
160 | let rev = Revision(rev); | |
161 | idx.append(py, tup.clone_ref(py))?; |
|
161 | idx.append(py, tup.clone_ref(py))?; | |
162 | self.index(py) |
|
162 | self.index(py) | |
163 | .borrow_mut() |
|
163 | .borrow_mut() | |
164 | .append(py_tuple_to_revision_data_params(py, tup)?) |
|
164 | .append(py_tuple_to_revision_data_params(py, tup)?) | |
165 | .unwrap(); |
|
165 | .unwrap(); | |
166 | self.get_nodetree(py)?.borrow_mut().as_mut().unwrap() |
|
166 | self.get_nodetree(py)?.borrow_mut().as_mut().unwrap() | |
167 | .insert(&*idx, &node, rev) |
|
167 | .insert(&*idx, &node, rev) | |
168 | .map_err(|e| nodemap_error(py, e))?; |
|
168 | .map_err(|e| nodemap_error(py, e))?; | |
169 | Ok(py.None()) |
|
169 | Ok(py.None()) | |
170 | } |
|
170 | } | |
171 |
|
171 | |||
172 | def __delitem__(&self, key: PyObject) -> PyResult<()> { |
|
172 | def __delitem__(&self, key: PyObject) -> PyResult<()> { | |
173 | // __delitem__ is both for `del idx[r]` and `del idx[r1:r2]` |
|
173 | // __delitem__ is both for `del idx[r]` and `del idx[r1:r2]` | |
174 | self.cindex(py).borrow().inner().del_item(py, &key)?; |
|
174 | self.cindex(py).borrow().inner().del_item(py, &key)?; | |
175 | let start = key.getattr(py, "start")?; |
|
175 | let start = key.getattr(py, "start")?; | |
176 | let start = UncheckedRevision(start.extract(py)?); |
|
176 | let start = UncheckedRevision(start.extract(py)?); | |
177 | let start = self.index(py) |
|
177 | let start = self.index(py) | |
178 | .borrow() |
|
178 | .borrow() | |
179 | .check_revision(start) |
|
179 | .check_revision(start) | |
180 | .ok_or_else(|| { |
|
180 | .ok_or_else(|| { | |
181 | nodemap_error(py, NodeMapError::RevisionNotInIndex(start)) |
|
181 | nodemap_error(py, NodeMapError::RevisionNotInIndex(start)) | |
182 | })?; |
|
182 | })?; | |
183 | self.index(py).borrow_mut().remove(start).unwrap(); |
|
183 | self.index(py).borrow_mut().remove(start).unwrap(); | |
184 | let mut opt = self.get_nodetree(py)?.borrow_mut(); |
|
184 | let mut opt = self.get_nodetree(py)?.borrow_mut(); | |
185 | let nt = opt.as_mut().unwrap(); |
|
185 | let nt = opt.as_mut().unwrap(); | |
186 | nt.invalidate_all(); |
|
186 | nt.invalidate_all(); | |
187 | self.fill_nodemap(py, nt)?; |
|
187 | self.fill_nodemap(py, nt)?; | |
188 | Ok(()) |
|
188 | Ok(()) | |
189 | } |
|
189 | } | |
190 |
|
190 | |||
191 | // |
|
191 | // | |
192 | // Reforwarded C index API |
|
192 | // Reforwarded C index API | |
193 | // |
|
193 | // | |
194 |
|
194 | |||
195 | // index_methods (tp_methods). Same ordering as in revlog.c |
|
195 | // index_methods (tp_methods). Same ordering as in revlog.c | |
196 |
|
196 | |||
197 | /// return the gca set of the given revs |
|
197 | /// return the gca set of the given revs | |
198 | def ancestors(&self, *args, **kw) -> PyResult<PyObject> { |
|
198 | def ancestors(&self, *args, **kw) -> PyResult<PyObject> { | |
199 | self.call_cindex(py, "ancestors", args, kw) |
|
199 | self.call_cindex(py, "ancestors", args, kw) | |
200 | } |
|
200 | } | |
201 |
|
201 | |||
202 | /// return the heads of the common ancestors of the given revs |
|
202 | /// return the heads of the common ancestors of the given revs | |
203 | def commonancestorsheads(&self, *args, **kw) -> PyResult<PyObject> { |
|
203 | def commonancestorsheads(&self, *args, **kw) -> PyResult<PyObject> { | |
204 | self.call_cindex(py, "commonancestorsheads", args, kw) |
|
204 | self.call_cindex(py, "commonancestorsheads", args, kw) | |
205 | } |
|
205 | } | |
206 |
|
206 | |||
207 | /// Clear the index caches and inner py_class data. |
|
207 | /// Clear the index caches and inner py_class data. | |
208 | /// It is Python's responsibility to call `update_nodemap_data` again. |
|
208 | /// It is Python's responsibility to call `update_nodemap_data` again. | |
209 | def clearcaches(&self, *args, **kw) -> PyResult<PyObject> { |
|
209 | def clearcaches(&self, *args, **kw) -> PyResult<PyObject> { | |
210 | self.nt(py).borrow_mut().take(); |
|
210 | self.nt(py).borrow_mut().take(); | |
211 | self.docket(py).borrow_mut().take(); |
|
211 | self.docket(py).borrow_mut().take(); | |
212 | self.nodemap_mmap(py).borrow_mut().take(); |
|
212 | self.nodemap_mmap(py).borrow_mut().take(); | |
213 | self.index(py).borrow_mut().clear_caches(); |
|
213 | self.index(py).borrow_mut().clear_caches(); | |
214 | self.call_cindex(py, "clearcaches", args, kw) |
|
214 | self.call_cindex(py, "clearcaches", args, kw) | |
215 | } |
|
215 | } | |
216 |
|
216 | |||
217 | /// return the raw binary string representing a revision |
|
217 | /// return the raw binary string representing a revision | |
218 | def entry_binary(&self, *args, **kw) -> PyResult<PyObject> { |
|
218 | def entry_binary(&self, *args, **kw) -> PyResult<PyObject> { | |
219 | let rindex = self.index(py).borrow(); |
|
219 | let rindex = self.index(py).borrow(); | |
220 | let rev = UncheckedRevision(args.get_item(py, 0).extract(py)?); |
|
220 | let rev = UncheckedRevision(args.get_item(py, 0).extract(py)?); | |
221 | let rust_bytes = rindex.check_revision(rev).and_then( |
|
221 | let rust_bytes = rindex.check_revision(rev).and_then( | |
222 | |r| rindex.entry_binary(r)) |
|
222 | |r| rindex.entry_binary(r)) | |
223 | .ok_or_else(|| rev_not_in_index(py, rev))?; |
|
223 | .ok_or_else(|| rev_not_in_index(py, rev))?; | |
224 | let rust_res = PyBytes::new(py, rust_bytes).into_object(); |
|
224 | let rust_res = PyBytes::new(py, rust_bytes).into_object(); | |
225 |
|
225 | |||
226 | let c_res = self.call_cindex(py, "entry_binary", args, kw)?; |
|
226 | let c_res = self.call_cindex(py, "entry_binary", args, kw)?; | |
227 | assert_py_eq(py, "entry_binary", &rust_res, &c_res)?; |
|
227 | assert_py_eq(py, "entry_binary", &rust_res, &c_res)?; | |
228 | Ok(rust_res) |
|
228 | Ok(rust_res) | |
229 | } |
|
229 | } | |
230 |
|
230 | |||
231 | /// return a binary packed version of the header |
|
231 | /// return a binary packed version of the header | |
232 | def pack_header(&self, *args, **kw) -> PyResult<PyObject> { |
|
232 | def pack_header(&self, *args, **kw) -> PyResult<PyObject> { | |
233 | let rindex = self.index(py).borrow(); |
|
233 | let rindex = self.index(py).borrow(); | |
234 | let packed = rindex.pack_header(args.get_item(py, 0).extract(py)?); |
|
234 | let packed = rindex.pack_header(args.get_item(py, 0).extract(py)?); | |
235 | let rust_res = PyBytes::new(py, &packed).into_object(); |
|
235 | let rust_res = PyBytes::new(py, &packed).into_object(); | |
236 |
|
236 | |||
237 | let c_res = self.call_cindex(py, "pack_header", args, kw)?; |
|
237 | let c_res = self.call_cindex(py, "pack_header", args, kw)?; | |
238 | assert_py_eq(py, "pack_header", &rust_res, &c_res)?; |
|
238 | assert_py_eq(py, "pack_header", &rust_res, &c_res)?; | |
239 | Ok(rust_res) |
|
239 | Ok(rust_res) | |
240 | } |
|
240 | } | |
241 |
|
241 | |||
242 | /// compute phases |
|
242 | /// compute phases | |
243 | def computephasesmapsets(&self, *args, **kw) -> PyResult<PyObject> { |
|
243 | def computephasesmapsets(&self, *args, **kw) -> PyResult<PyObject> { | |
244 | self.call_cindex(py, "computephasesmapsets", args, kw) |
|
244 | self.call_cindex(py, "computephasesmapsets", args, kw) | |
245 | } |
|
245 | } | |
246 |
|
246 | |||
247 | /// reachableroots |
|
247 | /// reachableroots | |
248 | def reachableroots2(&self, *args, **kw) -> PyResult<PyObject> { |
|
248 | def reachableroots2(&self, *args, **kw) -> PyResult<PyObject> { | |
249 | self.call_cindex(py, "reachableroots2", args, kw) |
|
249 | self.call_cindex(py, "reachableroots2", args, kw) | |
250 | } |
|
250 | } | |
251 |
|
251 | |||
252 | /// get head revisions |
|
252 | /// get head revisions | |
253 | def headrevs(&self, *args, **kw) -> PyResult<PyObject> { |
|
253 | def headrevs(&self, *args, **kw) -> PyResult<PyObject> { | |
254 | let rust_res = self.inner_headrevs(py)?; |
|
254 | let rust_res = self.inner_headrevs(py)?; | |
255 |
|
255 | |||
256 | let c_res = self.call_cindex(py, "headrevs", args, kw)?; |
|
256 | let c_res = self.call_cindex(py, "headrevs", args, kw)?; | |
257 | assert_py_eq(py, "headrevs", &rust_res, &c_res)?; |
|
257 | assert_py_eq(py, "headrevs", &rust_res, &c_res)?; | |
258 | Ok(rust_res) |
|
258 | Ok(rust_res) | |
259 | } |
|
259 | } | |
260 |
|
260 | |||
261 | /// get filtered head revisions |
|
261 | /// get filtered head revisions | |
262 | def headrevsfiltered(&self, *args, **kw) -> PyResult<PyObject> { |
|
262 | def headrevsfiltered(&self, *args, **kw) -> PyResult<PyObject> { | |
263 | let rust_res = self.inner_headrevsfiltered(py, &args.get_item(py, 0))?; |
|
263 | let rust_res = self.inner_headrevsfiltered(py, &args.get_item(py, 0))?; | |
264 | let c_res = self.call_cindex(py, "headrevsfiltered", args, kw)?; |
|
264 | let c_res = self.call_cindex(py, "headrevsfiltered", args, kw)?; | |
265 |
|
265 | |||
266 | assert_py_eq(py, "headrevsfiltered", &rust_res, &c_res)?; |
|
266 | assert_py_eq(py, "headrevsfiltered", &rust_res, &c_res)?; | |
267 | Ok(rust_res) |
|
267 | Ok(rust_res) | |
268 | } |
|
268 | } | |
269 |
|
269 | |||
270 | /// True if the object is a snapshot |
|
270 | /// True if the object is a snapshot | |
271 | def issnapshot(&self, *args, **kw) -> PyResult<bool> { |
|
271 | def issnapshot(&self, *args, **kw) -> PyResult<bool> { | |
272 | let index = self.index(py).borrow(); |
|
272 | let index = self.index(py).borrow(); | |
273 | let result = index |
|
273 | let result = index | |
274 | .is_snapshot(UncheckedRevision(args.get_item(py, 0).extract(py)?)) |
|
274 | .is_snapshot(UncheckedRevision(args.get_item(py, 0).extract(py)?)) | |
275 | .map_err(|e| { |
|
275 | .map_err(|e| { | |
276 | PyErr::new::<cpython::exc::ValueError, _>(py, e.to_string()) |
|
276 | PyErr::new::<cpython::exc::ValueError, _>(py, e.to_string()) | |
277 | })?; |
|
277 | })?; | |
278 | let cresult = self.call_cindex(py, "issnapshot", args, kw)?; |
|
278 | let cresult = self.call_cindex(py, "issnapshot", args, kw)?; | |
279 | assert_eq!(result, cresult.extract(py)?); |
|
279 | assert_eq!(result, cresult.extract(py)?); | |
280 | Ok(result) |
|
280 | Ok(result) | |
281 | } |
|
281 | } | |
282 |
|
282 | |||
283 | /// Gather snapshot data in a cache dict |
|
283 | /// Gather snapshot data in a cache dict | |
284 | def findsnapshots(&self, *args, **kw) -> PyResult<PyObject> { |
|
284 | def findsnapshots(&self, *args, **kw) -> PyResult<PyObject> { | |
285 | let index = self.index(py).borrow(); |
|
285 | let index = self.index(py).borrow(); | |
286 | let cache: PyDict = args.get_item(py, 0).extract(py)?; |
|
286 | let cache: PyDict = args.get_item(py, 0).extract(py)?; | |
287 | // this methods operates by setting new values in the cache, |
|
287 | // this methods operates by setting new values in the cache, | |
288 | // hence we will compare results by letting the C implementation |
|
288 | // hence we will compare results by letting the C implementation | |
289 | // operate over a deepcopy of the cache, and finally compare both |
|
289 | // operate over a deepcopy of the cache, and finally compare both | |
290 | // caches. |
|
290 | // caches. | |
291 | let c_cache = PyDict::new(py); |
|
291 | let c_cache = PyDict::new(py); | |
292 | for (k, v) in cache.items(py) { |
|
292 | for (k, v) in cache.items(py) { | |
293 | c_cache.set_item(py, k, PySet::new(py, v)?)?; |
|
293 | c_cache.set_item(py, k, PySet::new(py, v)?)?; | |
294 | } |
|
294 | } | |
295 |
|
295 | |||
296 | let start_rev = UncheckedRevision(args.get_item(py, 1).extract(py)?); |
|
296 | let start_rev = UncheckedRevision(args.get_item(py, 1).extract(py)?); | |
297 | let end_rev = UncheckedRevision(args.get_item(py, 2).extract(py)?); |
|
297 | let end_rev = UncheckedRevision(args.get_item(py, 2).extract(py)?); | |
298 | let mut cache_wrapper = PySnapshotsCache{ py, dict: cache }; |
|
298 | let mut cache_wrapper = PySnapshotsCache{ py, dict: cache }; | |
299 | index.find_snapshots( |
|
299 | index.find_snapshots( | |
300 | start_rev, |
|
300 | start_rev, | |
301 | end_rev, |
|
301 | end_rev, | |
302 | &mut cache_wrapper, |
|
302 | &mut cache_wrapper, | |
303 | ).map_err(|_| revlog_error(py))?; |
|
303 | ).map_err(|_| revlog_error(py))?; | |
304 |
|
304 | |||
305 | let c_args = PyTuple::new( |
|
305 | let c_args = PyTuple::new( | |
306 | py, |
|
306 | py, | |
307 | &[ |
|
307 | &[ | |
308 | c_cache.clone_ref(py).into_object(), |
|
308 | c_cache.clone_ref(py).into_object(), | |
309 | args.get_item(py, 1), |
|
309 | args.get_item(py, 1), | |
310 | args.get_item(py, 2) |
|
310 | args.get_item(py, 2) | |
311 | ] |
|
311 | ] | |
312 | ); |
|
312 | ); | |
313 | self.call_cindex(py, "findsnapshots", &c_args, kw)?; |
|
313 | self.call_cindex(py, "findsnapshots", &c_args, kw)?; | |
314 | assert_py_eq(py, "findsnapshots cache", |
|
314 | assert_py_eq(py, "findsnapshots cache", | |
315 | &cache_wrapper.into_object(), |
|
315 | &cache_wrapper.into_object(), | |
316 | &c_cache.into_object())?; |
|
316 | &c_cache.into_object())?; | |
317 | Ok(py.None()) |
|
317 | Ok(py.None()) | |
318 | } |
|
318 | } | |
319 |
|
319 | |||
320 | /// determine revisions with deltas to reconstruct fulltext |
|
320 | /// determine revisions with deltas to reconstruct fulltext | |
321 | def deltachain(&self, *args, **kw) -> PyResult<PyObject> { |
|
321 | def deltachain(&self, *args, **kw) -> PyResult<PyObject> { | |
322 | let index = self.index(py).borrow(); |
|
322 | let index = self.index(py).borrow(); | |
323 | let rev = args.get_item(py, 0).extract::<BaseRevision>(py)?.into(); |
|
323 | let rev = args.get_item(py, 0).extract::<BaseRevision>(py)?.into(); | |
324 | let stop_rev = |
|
324 | let stop_rev = | |
325 | args.get_item(py, 1).extract::<Option<BaseRevision>>(py)?; |
|
325 | args.get_item(py, 1).extract::<Option<BaseRevision>>(py)?; | |
326 | let rev = index.check_revision(rev).ok_or_else(|| { |
|
326 | let rev = index.check_revision(rev).ok_or_else(|| { | |
327 | nodemap_error(py, NodeMapError::RevisionNotInIndex(rev)) |
|
327 | nodemap_error(py, NodeMapError::RevisionNotInIndex(rev)) | |
328 | })?; |
|
328 | })?; | |
329 | let stop_rev = if let Some(stop_rev) = stop_rev { |
|
329 | let stop_rev = if let Some(stop_rev) = stop_rev { | |
330 | let stop_rev = UncheckedRevision(stop_rev); |
|
330 | let stop_rev = UncheckedRevision(stop_rev); | |
331 | Some(index.check_revision(stop_rev).ok_or_else(|| { |
|
331 | Some(index.check_revision(stop_rev).ok_or_else(|| { | |
332 | nodemap_error(py, NodeMapError::RevisionNotInIndex(stop_rev)) |
|
332 | nodemap_error(py, NodeMapError::RevisionNotInIndex(stop_rev)) | |
333 | })?) |
|
333 | })?) | |
334 | } else {None}; |
|
334 | } else {None}; | |
335 | let (chain, stopped) = index.delta_chain(rev, stop_rev).map_err(|e| { |
|
335 | let (chain, stopped) = index.delta_chain(rev, stop_rev).map_err(|e| { | |
336 | PyErr::new::<cpython::exc::ValueError, _>(py, e.to_string()) |
|
336 | PyErr::new::<cpython::exc::ValueError, _>(py, e.to_string()) | |
337 | })?; |
|
337 | })?; | |
338 |
|
338 | |||
339 | let cresult = self.call_cindex(py, "deltachain", args, kw)?; |
|
339 | let cresult = self.call_cindex(py, "deltachain", args, kw)?; | |
340 | let cchain: Vec<BaseRevision> = |
|
340 | let cchain: Vec<BaseRevision> = | |
341 | cresult.get_item(py, 0)?.extract::<Vec<BaseRevision>>(py)?; |
|
341 | cresult.get_item(py, 0)?.extract::<Vec<BaseRevision>>(py)?; | |
342 | let chain: Vec<_> = chain.into_iter().map(|r| r.0).collect(); |
|
342 | let chain: Vec<_> = chain.into_iter().map(|r| r.0).collect(); | |
343 | assert_eq!(chain, cchain); |
|
343 | assert_eq!(chain, cchain); | |
344 | assert_eq!(stopped, cresult.get_item(py, 1)?.extract(py)?); |
|
344 | assert_eq!(stopped, cresult.get_item(py, 1)?.extract(py)?); | |
345 |
|
345 | |||
346 | Ok( |
|
346 | Ok( | |
347 | PyTuple::new( |
|
347 | PyTuple::new( | |
348 | py, |
|
348 | py, | |
349 | &[ |
|
349 | &[ | |
350 | chain.into_py_object(py).into_object(), |
|
350 | chain.into_py_object(py).into_object(), | |
351 | stopped.into_py_object(py).into_object() |
|
351 | stopped.into_py_object(py).into_object() | |
352 | ] |
|
352 | ] | |
353 | ).into_object() |
|
353 | ).into_object() | |
354 | ) |
|
354 | ) | |
355 |
|
355 | |||
356 | } |
|
356 | } | |
357 |
|
357 | |||
358 | /// slice planned chunk read to reach a density threshold |
|
358 | /// slice planned chunk read to reach a density threshold | |
359 | def slicechunktodensity(&self, *args, **kw) -> PyResult<PyObject> { |
|
359 | def slicechunktodensity(&self, *args, **kw) -> PyResult<PyObject> { | |
360 |
self. |
|
360 | let rust_res = self.inner_slicechunktodensity( | |
|
361 | py, | |||
|
362 | args.get_item(py, 0), | |||
|
363 | args.get_item(py, 1).extract(py)?, | |||
|
364 | args.get_item(py, 2).extract(py)? | |||
|
365 | )?; | |||
|
366 | ||||
|
367 | let c_res = self.call_cindex(py, "slicechunktodensity", args, kw)?; | |||
|
368 | assert_eq!( | |||
|
369 | rust_res.len(), | |||
|
370 | c_res.len(py)?, | |||
|
371 | "chunks differ {:?} {}", | |||
|
372 | rust_res, c_res | |||
|
373 | ); | |||
|
374 | for (i, chunk) in rust_res.iter().enumerate() { | |||
|
375 | let c_chunk = c_res.get_item(py, i)?; | |||
|
376 | assert_eq!( | |||
|
377 | chunk.len(), | |||
|
378 | c_chunk.len(py)?, | |||
|
379 | "chunk {} length differ {:?} {}", | |||
|
380 | i, | |||
|
381 | chunk, | |||
|
382 | c_res | |||
|
383 | ); | |||
|
384 | for (j, rev) in chunk.iter().enumerate() { | |||
|
385 | let c_chunk: BaseRevision | |||
|
386 | = c_chunk.get_item(py, j)?.extract(py)?; | |||
|
387 | assert_eq!(c_chunk, rev.0); | |||
|
388 | } | |||
|
389 | } | |||
|
390 | Ok(c_res) | |||
361 | } |
|
391 | } | |
362 |
|
392 | |||
363 | /// stats for the index |
|
393 | /// stats for the index | |
364 | def stats(&self, *args, **kw) -> PyResult<PyObject> { |
|
394 | def stats(&self, *args, **kw) -> PyResult<PyObject> { | |
365 | self.call_cindex(py, "stats", args, kw) |
|
395 | self.call_cindex(py, "stats", args, kw) | |
366 | } |
|
396 | } | |
367 |
|
397 | |||
368 | // index_sequence_methods and index_mapping_methods. |
|
398 | // index_sequence_methods and index_mapping_methods. | |
369 | // |
|
399 | // | |
370 | // Since we call back through the high level Python API, |
|
400 | // Since we call back through the high level Python API, | |
371 | // there's no point making a distinction between index_get |
|
401 | // there's no point making a distinction between index_get | |
372 | // and index_getitem. |
|
402 | // and index_getitem. | |
373 | // gracinet 2023: this above is no longer true for the pure Rust impl |
|
403 | // gracinet 2023: this above is no longer true for the pure Rust impl | |
374 |
|
404 | |||
375 | def __len__(&self) -> PyResult<usize> { |
|
405 | def __len__(&self) -> PyResult<usize> { | |
376 | self.len(py) |
|
406 | self.len(py) | |
377 | } |
|
407 | } | |
378 |
|
408 | |||
379 | def __getitem__(&self, key: PyObject) -> PyResult<PyObject> { |
|
409 | def __getitem__(&self, key: PyObject) -> PyResult<PyObject> { | |
380 | let rust_res = self.inner_getitem(py, key.clone_ref(py))?; |
|
410 | let rust_res = self.inner_getitem(py, key.clone_ref(py))?; | |
381 |
|
411 | |||
382 | // this conversion seems needless, but that's actually because |
|
412 | // this conversion seems needless, but that's actually because | |
383 | // `index_getitem` does not handle conversion from PyLong, |
|
413 | // `index_getitem` does not handle conversion from PyLong, | |
384 | // which expressions such as [e for e in index] internally use. |
|
414 | // which expressions such as [e for e in index] internally use. | |
385 | // Note that we don't seem to have a direct way to call |
|
415 | // Note that we don't seem to have a direct way to call | |
386 | // PySequence_GetItem (does the job), which would possibly be better |
|
416 | // PySequence_GetItem (does the job), which would possibly be better | |
387 | // for performance |
|
417 | // for performance | |
388 | // gracinet 2023: the above comment can be removed when we use |
|
418 | // gracinet 2023: the above comment can be removed when we use | |
389 | // the pure Rust impl only. Note also that `key` can be a binary |
|
419 | // the pure Rust impl only. Note also that `key` can be a binary | |
390 | // node id. |
|
420 | // node id. | |
391 | let c_key = match key.extract::<BaseRevision>(py) { |
|
421 | let c_key = match key.extract::<BaseRevision>(py) { | |
392 | Ok(rev) => rev.to_py_object(py).into_object(), |
|
422 | Ok(rev) => rev.to_py_object(py).into_object(), | |
393 | Err(_) => key, |
|
423 | Err(_) => key, | |
394 | }; |
|
424 | }; | |
395 | let c_res = self.cindex(py).borrow().inner().get_item(py, c_key)?; |
|
425 | let c_res = self.cindex(py).borrow().inner().get_item(py, c_key)?; | |
396 |
|
426 | |||
397 | assert_py_eq(py, "__getitem__", &rust_res, &c_res)?; |
|
427 | assert_py_eq(py, "__getitem__", &rust_res, &c_res)?; | |
398 | Ok(rust_res) |
|
428 | Ok(rust_res) | |
399 | } |
|
429 | } | |
400 |
|
430 | |||
401 | def __contains__(&self, item: PyObject) -> PyResult<bool> { |
|
431 | def __contains__(&self, item: PyObject) -> PyResult<bool> { | |
402 | // ObjectProtocol does not seem to provide contains(), so |
|
432 | // ObjectProtocol does not seem to provide contains(), so | |
403 | // this is an equivalent implementation of the index_contains() |
|
433 | // this is an equivalent implementation of the index_contains() | |
404 | // defined in revlog.c |
|
434 | // defined in revlog.c | |
405 | let cindex = self.cindex(py).borrow(); |
|
435 | let cindex = self.cindex(py).borrow(); | |
406 | match item.extract::<i32>(py) { |
|
436 | match item.extract::<i32>(py) { | |
407 | Ok(rev) => { |
|
437 | Ok(rev) => { | |
408 | Ok(rev >= -1 && rev < self.len(py)? as BaseRevision) |
|
438 | Ok(rev >= -1 && rev < self.len(py)? as BaseRevision) | |
409 | } |
|
439 | } | |
410 | Err(_) => { |
|
440 | Err(_) => { | |
411 | let item_bytes: PyBytes = item.extract(py)?; |
|
441 | let item_bytes: PyBytes = item.extract(py)?; | |
412 | let rust_res = self.has_node(py, item_bytes)?; |
|
442 | let rust_res = self.has_node(py, item_bytes)?; | |
413 |
|
443 | |||
414 | let c_res = cindex.inner().call_method( |
|
444 | let c_res = cindex.inner().call_method( | |
415 | py, |
|
445 | py, | |
416 | "has_node", |
|
446 | "has_node", | |
417 | PyTuple::new(py, &[item.clone_ref(py)]), |
|
447 | PyTuple::new(py, &[item.clone_ref(py)]), | |
418 | None)? |
|
448 | None)? | |
419 | .extract(py)?; |
|
449 | .extract(py)?; | |
420 |
|
450 | |||
421 | assert_eq!(rust_res, c_res); |
|
451 | assert_eq!(rust_res, c_res); | |
422 | Ok(rust_res) |
|
452 | Ok(rust_res) | |
423 | } |
|
453 | } | |
424 | } |
|
454 | } | |
425 | } |
|
455 | } | |
426 |
|
456 | |||
427 | def nodemap_data_all(&self) -> PyResult<PyBytes> { |
|
457 | def nodemap_data_all(&self) -> PyResult<PyBytes> { | |
428 | self.inner_nodemap_data_all(py) |
|
458 | self.inner_nodemap_data_all(py) | |
429 | } |
|
459 | } | |
430 |
|
460 | |||
431 | def nodemap_data_incremental(&self) -> PyResult<PyObject> { |
|
461 | def nodemap_data_incremental(&self) -> PyResult<PyObject> { | |
432 | self.inner_nodemap_data_incremental(py) |
|
462 | self.inner_nodemap_data_incremental(py) | |
433 | } |
|
463 | } | |
434 | def update_nodemap_data( |
|
464 | def update_nodemap_data( | |
435 | &self, |
|
465 | &self, | |
436 | docket: PyObject, |
|
466 | docket: PyObject, | |
437 | nm_data: PyObject |
|
467 | nm_data: PyObject | |
438 | ) -> PyResult<PyObject> { |
|
468 | ) -> PyResult<PyObject> { | |
439 | self.inner_update_nodemap_data(py, docket, nm_data) |
|
469 | self.inner_update_nodemap_data(py, docket, nm_data) | |
440 | } |
|
470 | } | |
441 |
|
471 | |||
442 | @property |
|
472 | @property | |
443 | def entry_size(&self) -> PyResult<PyInt> { |
|
473 | def entry_size(&self) -> PyResult<PyInt> { | |
444 | self.cindex(py).borrow().inner().getattr(py, "entry_size")?.extract::<PyInt>(py) |
|
474 | self.cindex(py).borrow().inner().getattr(py, "entry_size")?.extract::<PyInt>(py) | |
445 | } |
|
475 | } | |
446 |
|
476 | |||
447 | @property |
|
477 | @property | |
448 | def rust_ext_compat(&self) -> PyResult<PyInt> { |
|
478 | def rust_ext_compat(&self) -> PyResult<PyInt> { | |
449 | self.cindex(py).borrow().inner().getattr(py, "rust_ext_compat")?.extract::<PyInt>(py) |
|
479 | self.cindex(py).borrow().inner().getattr(py, "rust_ext_compat")?.extract::<PyInt>(py) | |
450 | } |
|
480 | } | |
451 |
|
481 | |||
452 | }); |
|
482 | }); | |
453 |
|
483 | |||
454 | /// Take a (potentially) mmap'ed buffer, and return the underlying Python |
|
484 | /// Take a (potentially) mmap'ed buffer, and return the underlying Python | |
455 | /// buffer along with the Rust slice into said buffer. We need to keep the |
|
485 | /// buffer along with the Rust slice into said buffer. We need to keep the | |
456 | /// Python buffer around, otherwise we'd get a dangling pointer once the buffer |
|
486 | /// Python buffer around, otherwise we'd get a dangling pointer once the buffer | |
457 | /// is freed from Python's side. |
|
487 | /// is freed from Python's side. | |
458 | /// |
|
488 | /// | |
459 | /// # Safety |
|
489 | /// # Safety | |
460 | /// |
|
490 | /// | |
461 | /// The caller must make sure that the buffer is kept around for at least as |
|
491 | /// The caller must make sure that the buffer is kept around for at least as | |
462 | /// long as the slice. |
|
492 | /// long as the slice. | |
463 | #[deny(unsafe_op_in_unsafe_fn)] |
|
493 | #[deny(unsafe_op_in_unsafe_fn)] | |
464 | unsafe fn mmap_keeparound( |
|
494 | unsafe fn mmap_keeparound( | |
465 | py: Python, |
|
495 | py: Python, | |
466 | data: PyObject, |
|
496 | data: PyObject, | |
467 | ) -> PyResult<( |
|
497 | ) -> PyResult<( | |
468 | PyBuffer, |
|
498 | PyBuffer, | |
469 | Box<dyn std::ops::Deref<Target = [u8]> + Send + 'static>, |
|
499 | Box<dyn std::ops::Deref<Target = [u8]> + Send + 'static>, | |
470 | )> { |
|
500 | )> { | |
471 | let buf = PyBuffer::get(py, &data)?; |
|
501 | let buf = PyBuffer::get(py, &data)?; | |
472 | let len = buf.item_count(); |
|
502 | let len = buf.item_count(); | |
473 |
|
503 | |||
474 | // Build a slice from the mmap'ed buffer data |
|
504 | // Build a slice from the mmap'ed buffer data | |
475 | let cbuf = buf.buf_ptr(); |
|
505 | let cbuf = buf.buf_ptr(); | |
476 | let bytes = if std::mem::size_of::<u8>() == buf.item_size() |
|
506 | let bytes = if std::mem::size_of::<u8>() == buf.item_size() | |
477 | && buf.is_c_contiguous() |
|
507 | && buf.is_c_contiguous() | |
478 | && u8::is_compatible_format(buf.format()) |
|
508 | && u8::is_compatible_format(buf.format()) | |
479 | { |
|
509 | { | |
480 | unsafe { std::slice::from_raw_parts(cbuf as *const u8, len) } |
|
510 | unsafe { std::slice::from_raw_parts(cbuf as *const u8, len) } | |
481 | } else { |
|
511 | } else { | |
482 | return Err(PyErr::new::<ValueError, _>( |
|
512 | return Err(PyErr::new::<ValueError, _>( | |
483 | py, |
|
513 | py, | |
484 | "Nodemap data buffer has an invalid memory representation" |
|
514 | "Nodemap data buffer has an invalid memory representation" | |
485 | .to_string(), |
|
515 | .to_string(), | |
486 | )); |
|
516 | )); | |
487 | }; |
|
517 | }; | |
488 |
|
518 | |||
489 | Ok((buf, Box::new(bytes))) |
|
519 | Ok((buf, Box::new(bytes))) | |
490 | } |
|
520 | } | |
491 |
|
521 | |||
492 | fn py_tuple_to_revision_data_params( |
|
522 | fn py_tuple_to_revision_data_params( | |
493 | py: Python, |
|
523 | py: Python, | |
494 | tuple: PyTuple, |
|
524 | tuple: PyTuple, | |
495 | ) -> PyResult<RevisionDataParams> { |
|
525 | ) -> PyResult<RevisionDataParams> { | |
496 | if tuple.len(py) < 8 { |
|
526 | if tuple.len(py) < 8 { | |
497 | // this is better than the panic promised by tup.get_item() |
|
527 | // this is better than the panic promised by tup.get_item() | |
498 | return Err(PyErr::new::<IndexError, _>( |
|
528 | return Err(PyErr::new::<IndexError, _>( | |
499 | py, |
|
529 | py, | |
500 | "tuple index out of range", |
|
530 | "tuple index out of range", | |
501 | )); |
|
531 | )); | |
502 | } |
|
532 | } | |
503 | let offset_or_flags: u64 = tuple.get_item(py, 0).extract(py)?; |
|
533 | let offset_or_flags: u64 = tuple.get_item(py, 0).extract(py)?; | |
504 | let node_id = tuple |
|
534 | let node_id = tuple | |
505 | .get_item(py, 7) |
|
535 | .get_item(py, 7) | |
506 | .extract::<PyBytes>(py)? |
|
536 | .extract::<PyBytes>(py)? | |
507 | .data(py) |
|
537 | .data(py) | |
508 | .try_into() |
|
538 | .try_into() | |
509 | .unwrap(); |
|
539 | .unwrap(); | |
510 | let flags = (offset_or_flags & 0xFFFF) as u16; |
|
540 | let flags = (offset_or_flags & 0xFFFF) as u16; | |
511 | let data_offset = offset_or_flags >> 16; |
|
541 | let data_offset = offset_or_flags >> 16; | |
512 | Ok(RevisionDataParams { |
|
542 | Ok(RevisionDataParams { | |
513 | flags, |
|
543 | flags, | |
514 | data_offset, |
|
544 | data_offset, | |
515 | data_compressed_length: tuple.get_item(py, 1).extract(py)?, |
|
545 | data_compressed_length: tuple.get_item(py, 1).extract(py)?, | |
516 | data_uncompressed_length: tuple.get_item(py, 2).extract(py)?, |
|
546 | data_uncompressed_length: tuple.get_item(py, 2).extract(py)?, | |
517 | data_delta_base: tuple.get_item(py, 3).extract(py)?, |
|
547 | data_delta_base: tuple.get_item(py, 3).extract(py)?, | |
518 | link_rev: tuple.get_item(py, 4).extract(py)?, |
|
548 | link_rev: tuple.get_item(py, 4).extract(py)?, | |
519 | parent_rev_1: tuple.get_item(py, 5).extract(py)?, |
|
549 | parent_rev_1: tuple.get_item(py, 5).extract(py)?, | |
520 | parent_rev_2: tuple.get_item(py, 6).extract(py)?, |
|
550 | parent_rev_2: tuple.get_item(py, 6).extract(py)?, | |
521 | node_id, |
|
551 | node_id, | |
522 | ..Default::default() |
|
552 | ..Default::default() | |
523 | }) |
|
553 | }) | |
524 | } |
|
554 | } | |
525 | fn revision_data_params_to_py_tuple( |
|
555 | fn revision_data_params_to_py_tuple( | |
526 | py: Python, |
|
556 | py: Python, | |
527 | params: RevisionDataParams, |
|
557 | params: RevisionDataParams, | |
528 | ) -> PyTuple { |
|
558 | ) -> PyTuple { | |
529 | PyTuple::new( |
|
559 | PyTuple::new( | |
530 | py, |
|
560 | py, | |
531 | &[ |
|
561 | &[ | |
532 | params.data_offset.into_py_object(py).into_object(), |
|
562 | params.data_offset.into_py_object(py).into_object(), | |
533 | params |
|
563 | params | |
534 | .data_compressed_length |
|
564 | .data_compressed_length | |
535 | .into_py_object(py) |
|
565 | .into_py_object(py) | |
536 | .into_object(), |
|
566 | .into_object(), | |
537 | params |
|
567 | params | |
538 | .data_uncompressed_length |
|
568 | .data_uncompressed_length | |
539 | .into_py_object(py) |
|
569 | .into_py_object(py) | |
540 | .into_object(), |
|
570 | .into_object(), | |
541 | params.data_delta_base.into_py_object(py).into_object(), |
|
571 | params.data_delta_base.into_py_object(py).into_object(), | |
542 | params.link_rev.into_py_object(py).into_object(), |
|
572 | params.link_rev.into_py_object(py).into_object(), | |
543 | params.parent_rev_1.into_py_object(py).into_object(), |
|
573 | params.parent_rev_1.into_py_object(py).into_object(), | |
544 | params.parent_rev_2.into_py_object(py).into_object(), |
|
574 | params.parent_rev_2.into_py_object(py).into_object(), | |
545 | PyBytes::new(py, ¶ms.node_id) |
|
575 | PyBytes::new(py, ¶ms.node_id) | |
546 | .into_py_object(py) |
|
576 | .into_py_object(py) | |
547 | .into_object(), |
|
577 | .into_object(), | |
548 | params._sidedata_offset.into_py_object(py).into_object(), |
|
578 | params._sidedata_offset.into_py_object(py).into_object(), | |
549 | params |
|
579 | params | |
550 | ._sidedata_compressed_length |
|
580 | ._sidedata_compressed_length | |
551 | .into_py_object(py) |
|
581 | .into_py_object(py) | |
552 | .into_object(), |
|
582 | .into_object(), | |
553 | params |
|
583 | params | |
554 | .data_compression_mode |
|
584 | .data_compression_mode | |
555 | .into_py_object(py) |
|
585 | .into_py_object(py) | |
556 | .into_object(), |
|
586 | .into_object(), | |
557 | params |
|
587 | params | |
558 | ._sidedata_compression_mode |
|
588 | ._sidedata_compression_mode | |
559 | .into_py_object(py) |
|
589 | .into_py_object(py) | |
560 | .into_object(), |
|
590 | .into_object(), | |
561 | params._rank.into_py_object(py).into_object(), |
|
591 | params._rank.into_py_object(py).into_object(), | |
562 | ], |
|
592 | ], | |
563 | ) |
|
593 | ) | |
564 | } |
|
594 | } | |
565 |
|
595 | |||
566 | struct PySnapshotsCache<'p> { |
|
596 | struct PySnapshotsCache<'p> { | |
567 | py: Python<'p>, |
|
597 | py: Python<'p>, | |
568 | dict: PyDict, |
|
598 | dict: PyDict, | |
569 | } |
|
599 | } | |
570 |
|
600 | |||
571 | impl<'p> PySnapshotsCache<'p> { |
|
601 | impl<'p> PySnapshotsCache<'p> { | |
572 | fn into_object(self) -> PyObject { |
|
602 | fn into_object(self) -> PyObject { | |
573 | self.dict.into_object() |
|
603 | self.dict.into_object() | |
574 | } |
|
604 | } | |
575 | } |
|
605 | } | |
576 |
|
606 | |||
577 | impl<'p> SnapshotsCache for PySnapshotsCache<'p> { |
|
607 | impl<'p> SnapshotsCache for PySnapshotsCache<'p> { | |
578 | fn insert_for( |
|
608 | fn insert_for( | |
579 | &mut self, |
|
609 | &mut self, | |
580 | rev: BaseRevision, |
|
610 | rev: BaseRevision, | |
581 | value: BaseRevision, |
|
611 | value: BaseRevision, | |
582 | ) -> Result<(), RevlogError> { |
|
612 | ) -> Result<(), RevlogError> { | |
583 | let pyvalue = value.into_py_object(self.py).into_object(); |
|
613 | let pyvalue = value.into_py_object(self.py).into_object(); | |
584 | match self.dict.get_item(self.py, rev) { |
|
614 | match self.dict.get_item(self.py, rev) { | |
585 | Some(obj) => obj |
|
615 | Some(obj) => obj | |
586 | .extract::<PySet>(self.py) |
|
616 | .extract::<PySet>(self.py) | |
587 | .and_then(|set| set.add(self.py, pyvalue)), |
|
617 | .and_then(|set| set.add(self.py, pyvalue)), | |
588 | None => PySet::new(self.py, vec![pyvalue]) |
|
618 | None => PySet::new(self.py, vec![pyvalue]) | |
589 | .and_then(|set| self.dict.set_item(self.py, rev, set)), |
|
619 | .and_then(|set| self.dict.set_item(self.py, rev, set)), | |
590 | } |
|
620 | } | |
591 | .map_err(|_| { |
|
621 | .map_err(|_| { | |
592 | RevlogError::Other(HgError::unsupported( |
|
622 | RevlogError::Other(HgError::unsupported( | |
593 | "Error in Python caches handling", |
|
623 | "Error in Python caches handling", | |
594 | )) |
|
624 | )) | |
595 | }) |
|
625 | }) | |
596 | } |
|
626 | } | |
597 | } |
|
627 | } | |
598 |
|
628 | |||
599 | impl MixedIndex { |
|
629 | impl MixedIndex { | |
600 | fn new( |
|
630 | fn new( | |
601 | py: Python, |
|
631 | py: Python, | |
602 | cindex: PyObject, |
|
632 | cindex: PyObject, | |
603 | data: PyObject, |
|
633 | data: PyObject, | |
604 | header: u32, |
|
634 | header: u32, | |
605 | ) -> PyResult<MixedIndex> { |
|
635 | ) -> PyResult<MixedIndex> { | |
606 | // Safety: we keep the buffer around inside the class as `index_mmap` |
|
636 | // Safety: we keep the buffer around inside the class as `index_mmap` | |
607 | let (buf, bytes) = unsafe { mmap_keeparound(py, data)? }; |
|
637 | let (buf, bytes) = unsafe { mmap_keeparound(py, data)? }; | |
608 |
|
638 | |||
609 | Self::create_instance( |
|
639 | Self::create_instance( | |
610 | py, |
|
640 | py, | |
611 | RefCell::new(cindex::Index::new(py, cindex)?), |
|
641 | RefCell::new(cindex::Index::new(py, cindex)?), | |
612 | RefCell::new( |
|
642 | RefCell::new( | |
613 | hg::index::Index::new( |
|
643 | hg::index::Index::new( | |
614 | bytes, |
|
644 | bytes, | |
615 | IndexHeader::parse(&header.to_be_bytes()) |
|
645 | IndexHeader::parse(&header.to_be_bytes()) | |
616 | .expect("default header is broken") |
|
646 | .expect("default header is broken") | |
617 | .unwrap(), |
|
647 | .unwrap(), | |
618 | ) |
|
648 | ) | |
619 | .unwrap(), |
|
649 | .unwrap(), | |
620 | ), |
|
650 | ), | |
621 | RefCell::new(None), |
|
651 | RefCell::new(None), | |
622 | RefCell::new(None), |
|
652 | RefCell::new(None), | |
623 | RefCell::new(None), |
|
653 | RefCell::new(None), | |
624 | RefCell::new(Some(buf)), |
|
654 | RefCell::new(Some(buf)), | |
625 | ) |
|
655 | ) | |
626 | } |
|
656 | } | |
627 |
|
657 | |||
628 | fn len(&self, py: Python) -> PyResult<usize> { |
|
658 | fn len(&self, py: Python) -> PyResult<usize> { | |
629 | let rust_index_len = self.index(py).borrow().len(); |
|
659 | let rust_index_len = self.index(py).borrow().len(); | |
630 | let cindex_len = self.cindex(py).borrow().inner().len(py)?; |
|
660 | let cindex_len = self.cindex(py).borrow().inner().len(py)?; | |
631 | assert_eq!(rust_index_len, cindex_len); |
|
661 | assert_eq!(rust_index_len, cindex_len); | |
632 | Ok(cindex_len) |
|
662 | Ok(cindex_len) | |
633 | } |
|
663 | } | |
634 |
|
664 | |||
635 | /// This is scaffolding at this point, but it could also become |
|
665 | /// This is scaffolding at this point, but it could also become | |
636 | /// a way to start a persistent nodemap or perform a |
|
666 | /// a way to start a persistent nodemap or perform a | |
637 | /// vacuum / repack operation |
|
667 | /// vacuum / repack operation | |
638 | fn fill_nodemap( |
|
668 | fn fill_nodemap( | |
639 | &self, |
|
669 | &self, | |
640 | py: Python, |
|
670 | py: Python, | |
641 | nt: &mut NodeTree, |
|
671 | nt: &mut NodeTree, | |
642 | ) -> PyResult<PyObject> { |
|
672 | ) -> PyResult<PyObject> { | |
643 | let index = self.index(py).borrow(); |
|
673 | let index = self.index(py).borrow(); | |
644 | for r in 0..self.len(py)? { |
|
674 | for r in 0..self.len(py)? { | |
645 | let rev = Revision(r as BaseRevision); |
|
675 | let rev = Revision(r as BaseRevision); | |
646 | // in this case node() won't ever return None |
|
676 | // in this case node() won't ever return None | |
647 | nt.insert(&*index, index.node(rev).unwrap(), rev) |
|
677 | nt.insert(&*index, index.node(rev).unwrap(), rev) | |
648 | .map_err(|e| nodemap_error(py, e))? |
|
678 | .map_err(|e| nodemap_error(py, e))? | |
649 | } |
|
679 | } | |
650 | Ok(py.None()) |
|
680 | Ok(py.None()) | |
651 | } |
|
681 | } | |
652 |
|
682 | |||
653 | fn get_nodetree<'a>( |
|
683 | fn get_nodetree<'a>( | |
654 | &'a self, |
|
684 | &'a self, | |
655 | py: Python<'a>, |
|
685 | py: Python<'a>, | |
656 | ) -> PyResult<&'a RefCell<Option<NodeTree>>> { |
|
686 | ) -> PyResult<&'a RefCell<Option<NodeTree>>> { | |
657 | if self.nt(py).borrow().is_none() { |
|
687 | if self.nt(py).borrow().is_none() { | |
658 | let readonly = Box::<Vec<_>>::default(); |
|
688 | let readonly = Box::<Vec<_>>::default(); | |
659 | let mut nt = NodeTree::load_bytes(readonly, 0); |
|
689 | let mut nt = NodeTree::load_bytes(readonly, 0); | |
660 | self.fill_nodemap(py, &mut nt)?; |
|
690 | self.fill_nodemap(py, &mut nt)?; | |
661 | self.nt(py).borrow_mut().replace(nt); |
|
691 | self.nt(py).borrow_mut().replace(nt); | |
662 | } |
|
692 | } | |
663 | Ok(self.nt(py)) |
|
693 | Ok(self.nt(py)) | |
664 | } |
|
694 | } | |
665 |
|
695 | |||
666 | /// forward a method call to the underlying C index |
|
696 | /// forward a method call to the underlying C index | |
667 | fn call_cindex( |
|
697 | fn call_cindex( | |
668 | &self, |
|
698 | &self, | |
669 | py: Python, |
|
699 | py: Python, | |
670 | name: &str, |
|
700 | name: &str, | |
671 | args: &PyTuple, |
|
701 | args: &PyTuple, | |
672 | kwargs: Option<&PyDict>, |
|
702 | kwargs: Option<&PyDict>, | |
673 | ) -> PyResult<PyObject> { |
|
703 | ) -> PyResult<PyObject> { | |
674 | self.cindex(py) |
|
704 | self.cindex(py) | |
675 | .borrow() |
|
705 | .borrow() | |
676 | .inner() |
|
706 | .inner() | |
677 | .call_method(py, name, args, kwargs) |
|
707 | .call_method(py, name, args, kwargs) | |
678 | } |
|
708 | } | |
679 |
|
709 | |||
680 | pub fn clone_cindex(&self, py: Python) -> cindex::Index { |
|
710 | pub fn clone_cindex(&self, py: Python) -> cindex::Index { | |
681 | self.cindex(py).borrow().clone_ref(py) |
|
711 | self.cindex(py).borrow().clone_ref(py) | |
682 | } |
|
712 | } | |
683 |
|
713 | |||
684 | /// Returns the full nodemap bytes to be written as-is to disk |
|
714 | /// Returns the full nodemap bytes to be written as-is to disk | |
685 | fn inner_nodemap_data_all(&self, py: Python) -> PyResult<PyBytes> { |
|
715 | fn inner_nodemap_data_all(&self, py: Python) -> PyResult<PyBytes> { | |
686 | let nodemap = self.get_nodetree(py)?.borrow_mut().take().unwrap(); |
|
716 | let nodemap = self.get_nodetree(py)?.borrow_mut().take().unwrap(); | |
687 | let (readonly, bytes) = nodemap.into_readonly_and_added_bytes(); |
|
717 | let (readonly, bytes) = nodemap.into_readonly_and_added_bytes(); | |
688 |
|
718 | |||
689 | // If there's anything readonly, we need to build the data again from |
|
719 | // If there's anything readonly, we need to build the data again from | |
690 | // scratch |
|
720 | // scratch | |
691 | let bytes = if readonly.len() > 0 { |
|
721 | let bytes = if readonly.len() > 0 { | |
692 | let mut nt = NodeTree::load_bytes(Box::<Vec<_>>::default(), 0); |
|
722 | let mut nt = NodeTree::load_bytes(Box::<Vec<_>>::default(), 0); | |
693 | self.fill_nodemap(py, &mut nt)?; |
|
723 | self.fill_nodemap(py, &mut nt)?; | |
694 |
|
724 | |||
695 | let (readonly, bytes) = nt.into_readonly_and_added_bytes(); |
|
725 | let (readonly, bytes) = nt.into_readonly_and_added_bytes(); | |
696 | assert_eq!(readonly.len(), 0); |
|
726 | assert_eq!(readonly.len(), 0); | |
697 |
|
727 | |||
698 | bytes |
|
728 | bytes | |
699 | } else { |
|
729 | } else { | |
700 | bytes |
|
730 | bytes | |
701 | }; |
|
731 | }; | |
702 |
|
732 | |||
703 | let bytes = PyBytes::new(py, &bytes); |
|
733 | let bytes = PyBytes::new(py, &bytes); | |
704 | Ok(bytes) |
|
734 | Ok(bytes) | |
705 | } |
|
735 | } | |
706 |
|
736 | |||
707 | /// Returns the last saved docket along with the size of any changed data |
|
737 | /// Returns the last saved docket along with the size of any changed data | |
708 | /// (in number of blocks), and said data as bytes. |
|
738 | /// (in number of blocks), and said data as bytes. | |
709 | fn inner_nodemap_data_incremental( |
|
739 | fn inner_nodemap_data_incremental( | |
710 | &self, |
|
740 | &self, | |
711 | py: Python, |
|
741 | py: Python, | |
712 | ) -> PyResult<PyObject> { |
|
742 | ) -> PyResult<PyObject> { | |
713 | let docket = self.docket(py).borrow(); |
|
743 | let docket = self.docket(py).borrow(); | |
714 | let docket = match docket.as_ref() { |
|
744 | let docket = match docket.as_ref() { | |
715 | Some(d) => d, |
|
745 | Some(d) => d, | |
716 | None => return Ok(py.None()), |
|
746 | None => return Ok(py.None()), | |
717 | }; |
|
747 | }; | |
718 |
|
748 | |||
719 | let node_tree = self.get_nodetree(py)?.borrow_mut().take().unwrap(); |
|
749 | let node_tree = self.get_nodetree(py)?.borrow_mut().take().unwrap(); | |
720 | let masked_blocks = node_tree.masked_readonly_blocks(); |
|
750 | let masked_blocks = node_tree.masked_readonly_blocks(); | |
721 | let (_, data) = node_tree.into_readonly_and_added_bytes(); |
|
751 | let (_, data) = node_tree.into_readonly_and_added_bytes(); | |
722 | let changed = masked_blocks * std::mem::size_of::<Block>(); |
|
752 | let changed = masked_blocks * std::mem::size_of::<Block>(); | |
723 |
|
753 | |||
724 | Ok((docket, changed, PyBytes::new(py, &data)) |
|
754 | Ok((docket, changed, PyBytes::new(py, &data)) | |
725 | .to_py_object(py) |
|
755 | .to_py_object(py) | |
726 | .into_object()) |
|
756 | .into_object()) | |
727 | } |
|
757 | } | |
728 |
|
758 | |||
729 | /// Update the nodemap from the new (mmaped) data. |
|
759 | /// Update the nodemap from the new (mmaped) data. | |
730 | /// The docket is kept as a reference for later incremental calls. |
|
760 | /// The docket is kept as a reference for later incremental calls. | |
731 | fn inner_update_nodemap_data( |
|
761 | fn inner_update_nodemap_data( | |
732 | &self, |
|
762 | &self, | |
733 | py: Python, |
|
763 | py: Python, | |
734 | docket: PyObject, |
|
764 | docket: PyObject, | |
735 | nm_data: PyObject, |
|
765 | nm_data: PyObject, | |
736 | ) -> PyResult<PyObject> { |
|
766 | ) -> PyResult<PyObject> { | |
737 | // Safety: we keep the buffer around inside the class as `nodemap_mmap` |
|
767 | // Safety: we keep the buffer around inside the class as `nodemap_mmap` | |
738 | let (buf, bytes) = unsafe { mmap_keeparound(py, nm_data)? }; |
|
768 | let (buf, bytes) = unsafe { mmap_keeparound(py, nm_data)? }; | |
739 | let len = buf.item_count(); |
|
769 | let len = buf.item_count(); | |
740 | self.nodemap_mmap(py).borrow_mut().replace(buf); |
|
770 | self.nodemap_mmap(py).borrow_mut().replace(buf); | |
741 |
|
771 | |||
742 | let mut nt = NodeTree::load_bytes(bytes, len); |
|
772 | let mut nt = NodeTree::load_bytes(bytes, len); | |
743 |
|
773 | |||
744 | let data_tip = docket |
|
774 | let data_tip = docket | |
745 | .getattr(py, "tip_rev")? |
|
775 | .getattr(py, "tip_rev")? | |
746 | .extract::<BaseRevision>(py)? |
|
776 | .extract::<BaseRevision>(py)? | |
747 | .into(); |
|
777 | .into(); | |
748 | self.docket(py).borrow_mut().replace(docket.clone_ref(py)); |
|
778 | self.docket(py).borrow_mut().replace(docket.clone_ref(py)); | |
749 | let idx = self.index(py).borrow(); |
|
779 | let idx = self.index(py).borrow(); | |
750 | let data_tip = idx.check_revision(data_tip).ok_or_else(|| { |
|
780 | let data_tip = idx.check_revision(data_tip).ok_or_else(|| { | |
751 | nodemap_error(py, NodeMapError::RevisionNotInIndex(data_tip)) |
|
781 | nodemap_error(py, NodeMapError::RevisionNotInIndex(data_tip)) | |
752 | })?; |
|
782 | })?; | |
753 | let current_tip = idx.len(); |
|
783 | let current_tip = idx.len(); | |
754 |
|
784 | |||
755 | for r in (data_tip.0 + 1)..current_tip as BaseRevision { |
|
785 | for r in (data_tip.0 + 1)..current_tip as BaseRevision { | |
756 | let rev = Revision(r); |
|
786 | let rev = Revision(r); | |
757 | // in this case node() won't ever return None |
|
787 | // in this case node() won't ever return None | |
758 | nt.insert(&*idx, idx.node(rev).unwrap(), rev) |
|
788 | nt.insert(&*idx, idx.node(rev).unwrap(), rev) | |
759 | .map_err(|e| nodemap_error(py, e))? |
|
789 | .map_err(|e| nodemap_error(py, e))? | |
760 | } |
|
790 | } | |
761 |
|
791 | |||
762 | *self.nt(py).borrow_mut() = Some(nt); |
|
792 | *self.nt(py).borrow_mut() = Some(nt); | |
763 |
|
793 | |||
764 | Ok(py.None()) |
|
794 | Ok(py.None()) | |
765 | } |
|
795 | } | |
766 |
|
796 | |||
767 | fn inner_getitem(&self, py: Python, key: PyObject) -> PyResult<PyObject> { |
|
797 | fn inner_getitem(&self, py: Python, key: PyObject) -> PyResult<PyObject> { | |
768 | let idx = self.index(py).borrow(); |
|
798 | let idx = self.index(py).borrow(); | |
769 | Ok(match key.extract::<BaseRevision>(py) { |
|
799 | Ok(match key.extract::<BaseRevision>(py) { | |
770 | Ok(key_as_int) => { |
|
800 | Ok(key_as_int) => { | |
771 | let entry_params = if key_as_int == NULL_REVISION.0 { |
|
801 | let entry_params = if key_as_int == NULL_REVISION.0 { | |
772 | RevisionDataParams::default() |
|
802 | RevisionDataParams::default() | |
773 | } else { |
|
803 | } else { | |
774 | let rev = UncheckedRevision(key_as_int); |
|
804 | let rev = UncheckedRevision(key_as_int); | |
775 | match idx.entry_as_params(rev) { |
|
805 | match idx.entry_as_params(rev) { | |
776 | Some(e) => e, |
|
806 | Some(e) => e, | |
777 | None => { |
|
807 | None => { | |
778 | return Err(PyErr::new::<IndexError, _>( |
|
808 | return Err(PyErr::new::<IndexError, _>( | |
779 | py, |
|
809 | py, | |
780 | "revlog index out of range", |
|
810 | "revlog index out of range", | |
781 | )); |
|
811 | )); | |
782 | } |
|
812 | } | |
783 | } |
|
813 | } | |
784 | }; |
|
814 | }; | |
785 | revision_data_params_to_py_tuple(py, entry_params) |
|
815 | revision_data_params_to_py_tuple(py, entry_params) | |
786 | .into_object() |
|
816 | .into_object() | |
787 | } |
|
817 | } | |
788 | _ => self.get_rev(py, key.extract::<PyBytes>(py)?)?.map_or_else( |
|
818 | _ => self.get_rev(py, key.extract::<PyBytes>(py)?)?.map_or_else( | |
789 | || py.None(), |
|
819 | || py.None(), | |
790 | |py_rev| py_rev.into_py_object(py).into_object(), |
|
820 | |py_rev| py_rev.into_py_object(py).into_object(), | |
791 | ), |
|
821 | ), | |
792 | }) |
|
822 | }) | |
793 | } |
|
823 | } | |
794 |
|
824 | |||
795 | fn inner_headrevs(&self, py: Python) -> PyResult<PyObject> { |
|
825 | fn inner_headrevs(&self, py: Python) -> PyResult<PyObject> { | |
796 | let index = &mut *self.index(py).borrow_mut(); |
|
826 | let index = &mut *self.index(py).borrow_mut(); | |
797 | let as_vec: Vec<PyObject> = index |
|
827 | let as_vec: Vec<PyObject> = index | |
798 | .head_revs() |
|
828 | .head_revs() | |
799 | .map_err(|e| graph_error(py, e))? |
|
829 | .map_err(|e| graph_error(py, e))? | |
800 | .iter() |
|
830 | .iter() | |
801 | .map(|r| PyRevision::from(*r).into_py_object(py).into_object()) |
|
831 | .map(|r| PyRevision::from(*r).into_py_object(py).into_object()) | |
802 | .collect(); |
|
832 | .collect(); | |
803 | Ok(PyList::new(py, &as_vec).into_object()) |
|
833 | Ok(PyList::new(py, &as_vec).into_object()) | |
804 | } |
|
834 | } | |
805 |
|
835 | |||
806 | fn inner_headrevsfiltered( |
|
836 | fn inner_headrevsfiltered( | |
807 | &self, |
|
837 | &self, | |
808 | py: Python, |
|
838 | py: Python, | |
809 | filtered_revs: &PyObject, |
|
839 | filtered_revs: &PyObject, | |
810 | ) -> PyResult<PyObject> { |
|
840 | ) -> PyResult<PyObject> { | |
811 | let index = &mut *self.index(py).borrow_mut(); |
|
841 | let index = &mut *self.index(py).borrow_mut(); | |
812 | let filtered_revs = rev_pyiter_collect(py, filtered_revs, index)?; |
|
842 | let filtered_revs = rev_pyiter_collect(py, filtered_revs, index)?; | |
813 |
|
843 | |||
814 | let as_vec: Vec<PyObject> = index |
|
844 | let as_vec: Vec<PyObject> = index | |
815 | .head_revs_filtered(&filtered_revs) |
|
845 | .head_revs_filtered(&filtered_revs) | |
816 | .map_err(|e| graph_error(py, e))? |
|
846 | .map_err(|e| graph_error(py, e))? | |
817 | .iter() |
|
847 | .iter() | |
818 | .map(|r| PyRevision::from(*r).into_py_object(py).into_object()) |
|
848 | .map(|r| PyRevision::from(*r).into_py_object(py).into_object()) | |
819 | .collect(); |
|
849 | .collect(); | |
820 | Ok(PyList::new(py, &as_vec).into_object()) |
|
850 | Ok(PyList::new(py, &as_vec).into_object()) | |
821 | } |
|
851 | } | |
|
852 | ||||
|
853 | fn inner_slicechunktodensity( | |||
|
854 | &self, | |||
|
855 | py: Python, | |||
|
856 | revs: PyObject, | |||
|
857 | target_density: f64, | |||
|
858 | min_gap_size: usize, | |||
|
859 | ) -> PyResult<Vec<Vec<Revision>>> { | |||
|
860 | let index = &mut *self.index(py).borrow_mut(); | |||
|
861 | let revs: Vec<_> = rev_pyiter_collect(py, &revs, index)?; | |||
|
862 | Ok(index.slice_chunk_to_density(&revs, target_density, min_gap_size)) | |||
|
863 | } | |||
822 | } |
|
864 | } | |
823 |
|
865 | |||
824 | fn revlog_error(py: Python) -> PyErr { |
|
866 | fn revlog_error(py: Python) -> PyErr { | |
825 | match py |
|
867 | match py | |
826 | .import("mercurial.error") |
|
868 | .import("mercurial.error") | |
827 | .and_then(|m| m.get(py, "RevlogError")) |
|
869 | .and_then(|m| m.get(py, "RevlogError")) | |
828 | { |
|
870 | { | |
829 | Err(e) => e, |
|
871 | Err(e) => e, | |
830 | Ok(cls) => PyErr::from_instance( |
|
872 | Ok(cls) => PyErr::from_instance( | |
831 | py, |
|
873 | py, | |
832 | cls.call(py, (py.None(),), None).ok().into_py_object(py), |
|
874 | cls.call(py, (py.None(),), None).ok().into_py_object(py), | |
833 | ), |
|
875 | ), | |
834 | } |
|
876 | } | |
835 | } |
|
877 | } | |
836 |
|
878 | |||
837 | fn graph_error(py: Python, _err: hg::GraphError) -> PyErr { |
|
879 | fn graph_error(py: Python, _err: hg::GraphError) -> PyErr { | |
838 | // ParentOutOfRange is currently the only alternative |
|
880 | // ParentOutOfRange is currently the only alternative | |
839 | // in `hg::GraphError`. The C index always raises this simple ValueError. |
|
881 | // in `hg::GraphError`. The C index always raises this simple ValueError. | |
840 | PyErr::new::<ValueError, _>(py, "parent out of range") |
|
882 | PyErr::new::<ValueError, _>(py, "parent out of range") | |
841 | } |
|
883 | } | |
842 |
|
884 | |||
843 | fn nodemap_rev_not_in_index(py: Python, rev: UncheckedRevision) -> PyErr { |
|
885 | fn nodemap_rev_not_in_index(py: Python, rev: UncheckedRevision) -> PyErr { | |
844 | PyErr::new::<ValueError, _>( |
|
886 | PyErr::new::<ValueError, _>( | |
845 | py, |
|
887 | py, | |
846 | format!( |
|
888 | format!( | |
847 | "Inconsistency: Revision {} found in nodemap \ |
|
889 | "Inconsistency: Revision {} found in nodemap \ | |
848 | is not in revlog index", |
|
890 | is not in revlog index", | |
849 | rev |
|
891 | rev | |
850 | ), |
|
892 | ), | |
851 | ) |
|
893 | ) | |
852 | } |
|
894 | } | |
853 |
|
895 | |||
854 | fn rev_not_in_index(py: Python, rev: UncheckedRevision) -> PyErr { |
|
896 | fn rev_not_in_index(py: Python, rev: UncheckedRevision) -> PyErr { | |
855 | PyErr::new::<ValueError, _>( |
|
897 | PyErr::new::<ValueError, _>( | |
856 | py, |
|
898 | py, | |
857 | format!("revlog index out of range: {}", rev), |
|
899 | format!("revlog index out of range: {}", rev), | |
858 | ) |
|
900 | ) | |
859 | } |
|
901 | } | |
860 |
|
902 | |||
861 | /// Standard treatment of NodeMapError |
|
903 | /// Standard treatment of NodeMapError | |
862 | fn nodemap_error(py: Python, err: NodeMapError) -> PyErr { |
|
904 | fn nodemap_error(py: Python, err: NodeMapError) -> PyErr { | |
863 | match err { |
|
905 | match err { | |
864 | NodeMapError::MultipleResults => revlog_error(py), |
|
906 | NodeMapError::MultipleResults => revlog_error(py), | |
865 | NodeMapError::RevisionNotInIndex(r) => nodemap_rev_not_in_index(py, r), |
|
907 | NodeMapError::RevisionNotInIndex(r) => nodemap_rev_not_in_index(py, r), | |
866 | } |
|
908 | } | |
867 | } |
|
909 | } | |
868 |
|
910 | |||
869 | /// assert two Python objects to be equal from a Python point of view |
|
911 | /// assert two Python objects to be equal from a Python point of view | |
870 | /// |
|
912 | /// | |
871 | /// `method` is a label for the assertion error message, intended to be the |
|
913 | /// `method` is a label for the assertion error message, intended to be the | |
872 | /// name of the caller. |
|
914 | /// name of the caller. | |
873 | /// `normalizer` is a function that takes a Python variable name and returns |
|
915 | /// `normalizer` is a function that takes a Python variable name and returns | |
874 | /// an expression that the conparison will actually use. |
|
916 | /// an expression that the conparison will actually use. | |
875 | /// Foe example: `|v| format!("sorted({})", v)` |
|
917 | /// Foe example: `|v| format!("sorted({})", v)` | |
876 | fn assert_py_eq_normalized( |
|
918 | fn assert_py_eq_normalized( | |
877 | py: Python, |
|
919 | py: Python, | |
878 | method: &str, |
|
920 | method: &str, | |
879 | rust: &PyObject, |
|
921 | rust: &PyObject, | |
880 | c: &PyObject, |
|
922 | c: &PyObject, | |
881 | normalizer: impl FnOnce(&str) -> String + Copy, |
|
923 | normalizer: impl FnOnce(&str) -> String + Copy, | |
882 | ) -> PyResult<()> { |
|
924 | ) -> PyResult<()> { | |
883 | let locals = PyDict::new(py); |
|
925 | let locals = PyDict::new(py); | |
884 | locals.set_item(py, "rust".into_py_object(py).into_object(), rust)?; |
|
926 | locals.set_item(py, "rust".into_py_object(py).into_object(), rust)?; | |
885 | locals.set_item(py, "c".into_py_object(py).into_object(), c)?; |
|
927 | locals.set_item(py, "c".into_py_object(py).into_object(), c)?; | |
886 | // let lhs = format!(normalizer_fmt, "rust"); |
|
928 | // let lhs = format!(normalizer_fmt, "rust"); | |
887 | // let rhs = format!(normalizer_fmt, "c"); |
|
929 | // let rhs = format!(normalizer_fmt, "c"); | |
888 | let is_eq: PyBool = py |
|
930 | let is_eq: PyBool = py | |
889 | .eval( |
|
931 | .eval( | |
890 | &format!("{} == {}", &normalizer("rust"), &normalizer("c")), |
|
932 | &format!("{} == {}", &normalizer("rust"), &normalizer("c")), | |
891 | None, |
|
933 | None, | |
892 | Some(&locals), |
|
934 | Some(&locals), | |
893 | )? |
|
935 | )? | |
894 | .extract(py)?; |
|
936 | .extract(py)?; | |
895 | assert!( |
|
937 | assert!( | |
896 | is_eq.is_true(), |
|
938 | is_eq.is_true(), | |
897 | "{} results differ. Rust: {:?} C: {:?} (before any normalization)", |
|
939 | "{} results differ. Rust: {:?} C: {:?} (before any normalization)", | |
898 | method, |
|
940 | method, | |
899 | rust, |
|
941 | rust, | |
900 | c |
|
942 | c | |
901 | ); |
|
943 | ); | |
902 | Ok(()) |
|
944 | Ok(()) | |
903 | } |
|
945 | } | |
904 |
|
946 | |||
905 | fn assert_py_eq( |
|
947 | fn assert_py_eq( | |
906 | py: Python, |
|
948 | py: Python, | |
907 | method: &str, |
|
949 | method: &str, | |
908 | rust: &PyObject, |
|
950 | rust: &PyObject, | |
909 | c: &PyObject, |
|
951 | c: &PyObject, | |
910 | ) -> PyResult<()> { |
|
952 | ) -> PyResult<()> { | |
911 | assert_py_eq_normalized(py, method, rust, c, |v| v.to_owned()) |
|
953 | assert_py_eq_normalized(py, method, rust, c, |v| v.to_owned()) | |
912 | } |
|
954 | } | |
913 |
|
955 | |||
914 | /// Create the module, with __package__ given from parent |
|
956 | /// Create the module, with __package__ given from parent | |
915 | pub fn init_module(py: Python, package: &str) -> PyResult<PyModule> { |
|
957 | pub fn init_module(py: Python, package: &str) -> PyResult<PyModule> { | |
916 | let dotted_name = &format!("{}.revlog", package); |
|
958 | let dotted_name = &format!("{}.revlog", package); | |
917 | let m = PyModule::new(py, dotted_name)?; |
|
959 | let m = PyModule::new(py, dotted_name)?; | |
918 | m.add(py, "__package__", package)?; |
|
960 | m.add(py, "__package__", package)?; | |
919 | m.add(py, "__doc__", "RevLog - Rust implementations")?; |
|
961 | m.add(py, "__doc__", "RevLog - Rust implementations")?; | |
920 |
|
962 | |||
921 | m.add_class::<MixedIndex>(py)?; |
|
963 | m.add_class::<MixedIndex>(py)?; | |
922 |
|
964 | |||
923 | let sys = PyModule::import(py, "sys")?; |
|
965 | let sys = PyModule::import(py, "sys")?; | |
924 | let sys_modules: PyDict = sys.get(py, "modules")?.extract(py)?; |
|
966 | let sys_modules: PyDict = sys.get(py, "modules")?.extract(py)?; | |
925 | sys_modules.set_item(py, dotted_name, &m)?; |
|
967 | sys_modules.set_item(py, dotted_name, &m)?; | |
926 |
|
968 | |||
927 | Ok(m) |
|
969 | Ok(m) | |
928 | } |
|
970 | } |
General Comments 0
You need to be logged in to leave comments.
Login now