revlog: make the rust test for node hex prefix resolution exercise the nodemap

rust/hg-core/src/revlog/mod.rs

@@ -1,825 +1,843 b''
 // Copyright 2018-2023 Georges Racinet <[e-mail]>
 //           and Mercurial contributors
 //
 // This software may be used and distributed according to the terms of the
 // GNU General Public License version 2 or any later version.
 //! Mercurial concepts for handling revision history
 
 pub mod node;
 pub mod nodemap;
 mod nodemap_docket;
 pub mod path_encode;
 pub use node::{FromHexError, Node, NodePrefix};
 pub mod changelog;
 pub mod filelog;
 pub mod index;
 pub mod manifest;
 pub mod patch;
 
 use std::borrow::Cow;
 use std::io::Read;
 use std::ops::Deref;
 use std::path::Path;
 
 use flate2::read::ZlibDecoder;
 use sha1::{Digest, Sha1};
 use std::cell::RefCell;
 use zstd;
 
 use self::node::{NODE_BYTES_LENGTH, NULL_NODE};
 use self::nodemap_docket::NodeMapDocket;
 use super::index::Index;
 use super::nodemap::{NodeMap, NodeMapError};
 use crate::errors::HgError;
 use crate::vfs::Vfs;
 
 /// Mercurial revision numbers
 ///
 /// As noted in revlog.c, revision numbers are actually encoded in
 /// 4 bytes, and are liberally converted to ints, whence the i32
 pub type Revision = i32;
 
 /// Marker expressing the absence of a parent
 ///
 /// Independently of the actual representation, `NULL_REVISION` is guaranteed
 /// to be smaller than all existing revisions.
 pub const NULL_REVISION: Revision = -1;
 
 /// Same as `mercurial.node.wdirrev`
 ///
 /// This is also equal to `i32::max_value()`, but it's better to spell
 /// it out explicitely, same as in `mercurial.node`
 #[allow(clippy::unreadable_literal)]
 pub const WORKING_DIRECTORY_REVISION: Revision = 0x7fffffff;
 
 pub const WORKING_DIRECTORY_HEX: &str =
     "ffffffffffffffffffffffffffffffffffffffff";
 
 /// The simplest expression of what we need of Mercurial DAGs.
 pub trait Graph {
     /// Return the two parents of the given `Revision`.
     ///
     /// Each of the parents can be independently `NULL_REVISION`
     fn parents(&self, rev: Revision) -> Result<[Revision; 2], GraphError>;
 }
 
 #[derive(Clone, Debug, PartialEq)]
 pub enum GraphError {
     ParentOutOfRange(Revision),
     WorkingDirectoryUnsupported,
 }
 
 /// The Mercurial Revlog Index
 ///
 /// This is currently limited to the minimal interface that is needed for
 /// the [`nodemap`](nodemap/index.html) module
 pub trait RevlogIndex {
     /// Total number of Revisions referenced in this index
     fn len(&self) -> usize;
 
     fn is_empty(&self) -> bool {
         self.len() == 0
     }
 
     /// Return a reference to the Node or `None` if rev is out of bounds
     ///
     /// `NULL_REVISION` is not considered to be out of bounds.
     fn node(&self, rev: Revision) -> Option<&Node>;
 }
 
 const REVISION_FLAG_CENSORED: u16 = 1 << 15;
 const REVISION_FLAG_ELLIPSIS: u16 = 1 << 14;
 const REVISION_FLAG_EXTSTORED: u16 = 1 << 13;
 const REVISION_FLAG_HASCOPIESINFO: u16 = 1 << 12;
 
 // Keep this in sync with REVIDX_KNOWN_FLAGS in
 // mercurial/revlogutils/flagutil.py
 const REVIDX_KNOWN_FLAGS: u16 = REVISION_FLAG_CENSORED
     | REVISION_FLAG_ELLIPSIS
     | REVISION_FLAG_EXTSTORED
     | REVISION_FLAG_HASCOPIESINFO;
 
 const NULL_REVLOG_ENTRY_FLAGS: u16 = 0;
 
 #[derive(Debug, derive_more::From)]
 pub enum RevlogError {
     InvalidRevision,
     /// Working directory is not supported
     WDirUnsupported,
     /// Found more than one entry whose ID match the requested prefix
     AmbiguousPrefix,
     #[from]
     Other(HgError),
 }
 
 impl From<NodeMapError> for RevlogError {
     fn from(error: NodeMapError) -> Self {
         match error {
             NodeMapError::MultipleResults => RevlogError::AmbiguousPrefix,
             NodeMapError::RevisionNotInIndex(rev) => RevlogError::corrupted(
                 format!("nodemap point to revision {} not in index", rev),
             ),
         }
     }
 }
 
 fn corrupted<S: AsRef<str>>(context: S) -> HgError {
     HgError::corrupted(format!("corrupted revlog, {}", context.as_ref()))
 }
 
 impl RevlogError {
     fn corrupted<S: AsRef<str>>(context: S) -> Self {
         RevlogError::Other(corrupted(context))
     }
 }
 
 /// Read only implementation of revlog.
 pub struct Revlog {
     /// When index and data are not interleaved: bytes of the revlog index.
     /// When index and data are interleaved: bytes of the revlog index and
     /// data.
     index: Index,
     /// When index and data are not interleaved: bytes of the revlog data
     data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>>,
     /// When present on disk: the persistent nodemap for this revlog
     nodemap: Option<nodemap::NodeTree>,
 }
 
 impl Revlog {
     /// Open a revlog index file.
     ///
     /// It will also open the associated data file if index and data are not
     /// interleaved.
     pub fn open(
         store_vfs: &Vfs,
         index_path: impl AsRef<Path>,
         data_path: Option<&Path>,
         use_nodemap: bool,
     ) -> Result<Self, HgError> {
+        Self::open_gen(store_vfs, index_path, data_path, use_nodemap, None)
+    }
+
+    fn open_gen(
+        store_vfs: &Vfs,
+        index_path: impl AsRef<Path>,
+        data_path: Option<&Path>,
+        use_nodemap: bool,
+        nodemap_for_test: Option<nodemap::NodeTree>,
+    ) -> Result<Self, HgError> {
         let index_path = index_path.as_ref();
         let index = {
             match store_vfs.mmap_open_opt(&index_path)? {
                 None => Index::new(Box::new(vec![])),
                 Some(index_mmap) => {
                     let index = Index::new(Box::new(index_mmap))?;
                     Ok(index)
                 }
             }
         }?;
 
         let default_data_path = index_path.with_extension("d");
 
         // type annotation required
         // won't recognize Mmap as Deref<Target = [u8]>
         let data_bytes: Option<Box<dyn Deref<Target = [u8]> + Send>> =
             if index.is_inline() {
                 None
             } else {
                 let data_path = data_path.unwrap_or(&default_data_path);
                 let data_mmap = store_vfs.mmap_open(data_path)?;
                 Some(Box::new(data_mmap))
             };
 
         let nodemap = if index.is_inline() || !use_nodemap {
             None
         } else {
             NodeMapDocket::read_from_file(store_vfs, index_path)?.map(
                 |(docket, data)| {
                     nodemap::NodeTree::load_bytes(
                         Box::new(data),
                         docket.data_length,
                     )
                 },
             )
         };
 
+        let nodemap = nodemap_for_test.or(nodemap);
+
         Ok(Revlog {
             index,
             data_bytes,
             nodemap,
         })
     }
 
     /// Return number of entries of the `Revlog`.
     pub fn len(&self) -> usize {
         self.index.len()
     }
 
     /// Returns `true` if the `Revlog` has zero `entries`.
     pub fn is_empty(&self) -> bool {
         self.index.is_empty()
     }
 
     /// Returns the node ID for the given revision number, if it exists in this
     /// revlog
     pub fn node_from_rev(&self, rev: Revision) -> Option<&Node> {
         if rev == NULL_REVISION {
             return Some(&NULL_NODE);
         }
         Some(self.index.get_entry(rev)?.hash())
     }
 
     /// Return the revision number for the given node ID, if it exists in this
     /// revlog
     pub fn rev_from_node(
         &self,
         node: NodePrefix,
     ) -> Result<Revision, RevlogError> {
         if let Some(nodemap) = &self.nodemap {
             nodemap
                 .find_bin(&self.index, node)?
                 .ok_or(RevlogError::InvalidRevision)
         } else {
             self.rev_from_node_no_persistent_nodemap(node)
         }
     }
 
     /// Same as `rev_from_node`, without using a persistent nodemap
     ///
     /// This is used as fallback when a persistent nodemap is not present.
     /// This happens when the persistent-nodemap experimental feature is not
     /// enabled, or for small revlogs.
     fn rev_from_node_no_persistent_nodemap(
         &self,
         node: NodePrefix,
     ) -> Result<Revision, RevlogError> {
         // Linear scan of the revlog
         // TODO: consider building a non-persistent nodemap in memory to
         // optimize these cases.
         let mut found_by_prefix = None;
         for rev in (-1..self.len() as Revision).rev() {
             let candidate_node = if rev == -1 {
                 NULL_NODE
             } else {
                 let index_entry =
                     self.index.get_entry(rev).ok_or_else(|| {
                         HgError::corrupted(
                             "revlog references a revision not in the index",
                         )
                     })?;
                 *index_entry.hash()
             };
             if node == candidate_node {
                 return Ok(rev);
             }
             if node.is_prefix_of(&candidate_node) {
                 if found_by_prefix.is_some() {
                     return Err(RevlogError::AmbiguousPrefix);
                 }
                 found_by_prefix = Some(rev)
             }
         }
         found_by_prefix.ok_or(RevlogError::InvalidRevision)
     }
 
     /// Returns whether the given revision exists in this revlog.
     pub fn has_rev(&self, rev: Revision) -> bool {
         self.index.get_entry(rev).is_some()
     }
 
     /// Return the full data associated to a revision.
     ///
     /// All entries required to build the final data out of deltas will be
     /// retrieved as needed, and the deltas will be applied to the inital
     /// snapshot to rebuild the final data.
     pub fn get_rev_data(
         &self,
         rev: Revision,
     ) -> Result<Cow<[u8]>, RevlogError> {
         if rev == NULL_REVISION {
             return Ok(Cow::Borrowed(&[]));
         };
         Ok(self.get_entry(rev)?.data()?)
     }
 
     /// Check the hash of some given data against the recorded hash.
     pub fn check_hash(
         &self,
         p1: Revision,
         p2: Revision,
         expected: &[u8],
         data: &[u8],
     ) -> bool {
         let e1 = self.index.get_entry(p1);
         let h1 = match e1 {
             Some(ref entry) => entry.hash(),
             None => &NULL_NODE,
         };
         let e2 = self.index.get_entry(p2);
         let h2 = match e2 {
             Some(ref entry) => entry.hash(),
             None => &NULL_NODE,
         };
 
         hash(data, h1.as_bytes(), h2.as_bytes()) == expected
     }
 
     /// Build the full data of a revision out its snapshot
     /// and its deltas.
     fn build_data_from_deltas(
         snapshot: RevlogEntry,
         deltas: &[RevlogEntry],
     ) -> Result<Vec<u8>, HgError> {
         let snapshot = snapshot.data_chunk()?;
         let deltas = deltas
             .iter()
             .rev()
             .map(RevlogEntry::data_chunk)
             .collect::<Result<Vec<_>, _>>()?;
         let patches: Vec<_> =
             deltas.iter().map(|d| patch::PatchList::new(d)).collect();
         let patch = patch::fold_patch_lists(&patches);
         Ok(patch.apply(&snapshot))
     }
 
     /// Return the revlog data.
     fn data(&self) -> &[u8] {
         match &self.data_bytes {
             Some(data_bytes) => data_bytes,
             None => panic!(
                 "forgot to load the data or trying to access inline data"
             ),
         }
     }
 
     pub fn make_null_entry(&self) -> RevlogEntry {
         RevlogEntry {
             revlog: self,
             rev: NULL_REVISION,
             bytes: b"",
             compressed_len: 0,
             uncompressed_len: 0,
             base_rev_or_base_of_delta_chain: None,
             p1: NULL_REVISION,
             p2: NULL_REVISION,
             flags: NULL_REVLOG_ENTRY_FLAGS,
             hash: NULL_NODE,
         }
     }
 
     /// Get an entry of the revlog.
     pub fn get_entry(
         &self,
         rev: Revision,
     ) -> Result<RevlogEntry, RevlogError> {
         if rev == NULL_REVISION {
             return Ok(self.make_null_entry());
         }
         let index_entry = self
             .index
             .get_entry(rev)
             .ok_or(RevlogError::InvalidRevision)?;
         let start = index_entry.offset();
         let end = start + index_entry.compressed_len() as usize;
         let data = if self.index.is_inline() {
             self.index.data(start, end)
         } else {
             &self.data()[start..end]
         };
         let entry = RevlogEntry {
             revlog: self,
             rev,
             bytes: data,
             compressed_len: index_entry.compressed_len(),
             uncompressed_len: index_entry.uncompressed_len(),
             base_rev_or_base_of_delta_chain: if index_entry
                 .base_revision_or_base_of_delta_chain()
                 == rev
             {
                 None
             } else {
                 Some(index_entry.base_revision_or_base_of_delta_chain())
             },
             p1: index_entry.p1(),
             p2: index_entry.p2(),
             flags: index_entry.flags(),
             hash: *index_entry.hash(),
         };
         Ok(entry)
     }
 
     /// when resolving internal references within revlog, any errors
     /// should be reported as corruption, instead of e.g. "invalid revision"
     fn get_entry_internal(
         &self,
         rev: Revision,
     ) -> Result<RevlogEntry, HgError> {
         self.get_entry(rev)
             .map_err(|_| corrupted(format!("revision {} out of range", rev)))
     }
 }
 
 /// The revlog entry's bytes and the necessary informations to extract
 /// the entry's data.
 #[derive(Clone)]
 pub struct RevlogEntry<'revlog> {
     revlog: &'revlog Revlog,
     rev: Revision,
     bytes: &'revlog [u8],
     compressed_len: u32,
     uncompressed_len: i32,
     base_rev_or_base_of_delta_chain: Option<Revision>,
     p1: Revision,
     p2: Revision,
     flags: u16,
     hash: Node,
 }
 
 thread_local! {
   // seems fine to [unwrap] here: this can only fail due to memory allocation
   // failing, and it's normal for that to cause panic.
   static ZSTD_DECODER : RefCell<zstd::bulk::Decompressor<'static>> =
       RefCell::new(zstd::bulk::Decompressor::new().ok().unwrap());
 }
 
 fn zstd_decompress_to_buffer(
     bytes: &[u8],
     buf: &mut Vec<u8>,
 ) -> Result<usize, std::io::Error> {
     ZSTD_DECODER
         .with(|decoder| decoder.borrow_mut().decompress_to_buffer(bytes, buf))
 }
 
 impl<'revlog> RevlogEntry<'revlog> {
     pub fn revision(&self) -> Revision {
         self.rev
     }
 
     pub fn node(&self) -> &Node {
         &self.hash
     }
 
     pub fn uncompressed_len(&self) -> Option<u32> {
         u32::try_from(self.uncompressed_len).ok()
     }
 
     pub fn has_p1(&self) -> bool {
         self.p1 != NULL_REVISION
     }
 
     pub fn p1_entry(
         &self,
     ) -> Result<Option<RevlogEntry<'revlog>>, RevlogError> {
         if self.p1 == NULL_REVISION {
             Ok(None)
         } else {
             Ok(Some(self.revlog.get_entry(self.p1)?))
         }
     }
 
     pub fn p2_entry(
         &self,
     ) -> Result<Option<RevlogEntry<'revlog>>, RevlogError> {
         if self.p2 == NULL_REVISION {
             Ok(None)
         } else {
             Ok(Some(self.revlog.get_entry(self.p2)?))
         }
     }
 
     pub fn p1(&self) -> Option<Revision> {
         if self.p1 == NULL_REVISION {
             None
         } else {
             Some(self.p1)
         }
     }
 
     pub fn p2(&self) -> Option<Revision> {
         if self.p2 == NULL_REVISION {
             None
         } else {
             Some(self.p2)
         }
     }
 
     pub fn is_censored(&self) -> bool {
         (self.flags & REVISION_FLAG_CENSORED) != 0
     }
 
     pub fn has_length_affecting_flag_processor(&self) -> bool {
         // Relevant Python code: revlog.size()
         // note: ELLIPSIS is known to not change the content
         (self.flags & (REVIDX_KNOWN_FLAGS ^ REVISION_FLAG_ELLIPSIS)) != 0
     }
 
     /// The data for this entry, after resolving deltas if any.
     pub fn rawdata(&self) -> Result<Cow<'revlog, [u8]>, HgError> {
         let mut entry = self.clone();
         let mut delta_chain = vec![];
 
         // The meaning of `base_rev_or_base_of_delta_chain` depends on
         // generaldelta. See the doc on `ENTRY_DELTA_BASE` in
         // `mercurial/revlogutils/constants.py` and the code in
         // [_chaininfo] and in [index_deltachain].
         let uses_generaldelta = self.revlog.index.uses_generaldelta();
         while let Some(base_rev) = entry.base_rev_or_base_of_delta_chain {
             let base_rev = if uses_generaldelta {
                 base_rev
             } else {
                 entry.rev - 1
             };
             delta_chain.push(entry);
             entry = self.revlog.get_entry_internal(base_rev)?;
         }
 
         let data = if delta_chain.is_empty() {
             entry.data_chunk()?
         } else {
             Revlog::build_data_from_deltas(entry, &delta_chain)?.into()
         };
 
         Ok(data)
     }
 
     fn check_data(
         &self,
         data: Cow<'revlog, [u8]>,
     ) -> Result<Cow<'revlog, [u8]>, HgError> {
         if self.revlog.check_hash(
             self.p1,
             self.p2,
             self.hash.as_bytes(),
             &data,
         ) {
             Ok(data)
         } else {
             if (self.flags & REVISION_FLAG_ELLIPSIS) != 0 {
                 return Err(HgError::unsupported(
                     "ellipsis revisions are not supported by rhg",
                 ));
             }
             Err(corrupted(format!(
                 "hash check failed for revision {}",
                 self.rev
             )))
         }
     }
 
     pub fn data(&self) -> Result<Cow<'revlog, [u8]>, HgError> {
         let data = self.rawdata()?;
         if self.rev == NULL_REVISION {
             return Ok(data);
         }
         if self.is_censored() {
             return Err(HgError::CensoredNodeError);
         }
         self.check_data(data)
     }
 
     /// Extract the data contained in the entry.
     /// This may be a delta. (See `is_delta`.)
     fn data_chunk(&self) -> Result<Cow<'revlog, [u8]>, HgError> {
         if self.bytes.is_empty() {
             return Ok(Cow::Borrowed(&[]));
         }
         match self.bytes[0] {
             // Revision data is the entirety of the entry, including this
             // header.
             b'\0' => Ok(Cow::Borrowed(self.bytes)),
             // Raw revision data follows.
             b'u' => Ok(Cow::Borrowed(&self.bytes[1..])),
             // zlib (RFC 1950) data.
             b'x' => Ok(Cow::Owned(self.uncompressed_zlib_data()?)),
             // zstd data.
             b'\x28' => Ok(Cow::Owned(self.uncompressed_zstd_data()?)),
             // A proper new format should have had a repo/store requirement.
             format_type => Err(corrupted(format!(
                 "unknown compression header '{}'",
                 format_type
             ))),
         }
     }
 
     fn uncompressed_zlib_data(&self) -> Result<Vec<u8>, HgError> {
         let mut decoder = ZlibDecoder::new(self.bytes);
         if self.is_delta() {
             let mut buf = Vec::with_capacity(self.compressed_len as usize);
             decoder
                 .read_to_end(&mut buf)
                 .map_err(|e| corrupted(e.to_string()))?;
             Ok(buf)
         } else {
             let cap = self.uncompressed_len.max(0) as usize;
             let mut buf = vec![0; cap];
             decoder
                 .read_exact(&mut buf)
                 .map_err(|e| corrupted(e.to_string()))?;
             Ok(buf)
         }
     }
 
     fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, HgError> {
         let cap = self.uncompressed_len.max(0) as usize;
         if self.is_delta() {
             // [cap] is usually an over-estimate of the space needed because
             // it's the length of delta-decoded data, but we're interested
             // in the size of the delta.
             // This means we have to [shrink_to_fit] to avoid holding on
             // to a large chunk of memory, but it also means we must have a
             // fallback branch, for the case when the delta is longer than
             // the original data (surprisingly, this does happen in practice)
             let mut buf = Vec::with_capacity(cap);
             match zstd_decompress_to_buffer(self.bytes, &mut buf) {
                 Ok(_) => buf.shrink_to_fit(),
                 Err(_) => {
                     buf.clear();
                     zstd::stream::copy_decode(self.bytes, &mut buf)
                         .map_err(|e| corrupted(e.to_string()))?;
                 }
             };
             Ok(buf)
         } else {
             let mut buf = Vec::with_capacity(cap);
             let len = zstd_decompress_to_buffer(self.bytes, &mut buf)
                 .map_err(|e| corrupted(e.to_string()))?;
             if len != self.uncompressed_len as usize {
                 Err(corrupted("uncompressed length does not match"))
             } else {
                 Ok(buf)
             }
         }
     }
 
     /// Tell if the entry is a snapshot or a delta
     /// (influences on decompression).
     fn is_delta(&self) -> bool {
         self.base_rev_or_base_of_delta_chain.is_some()
     }
 }
 
 /// Calculate the hash of a revision given its data and its parents.
 fn hash(
     data: &[u8],
     p1_hash: &[u8],
     p2_hash: &[u8],
 ) -> [u8; NODE_BYTES_LENGTH] {
     let mut hasher = Sha1::new();
     let (a, b) = (p1_hash, p2_hash);
     if a > b {
         hasher.update(b);
         hasher.update(a);
     } else {
         hasher.update(a);
         hasher.update(b);
     }
     hasher.update(data);
     *hasher.finalize().as_ref()
 }
 
 #[cfg(test)]
 mod tests {
     use super::*;
     use crate::index::{IndexEntryBuilder, INDEX_ENTRY_SIZE};
     use itertools::Itertools;
 
     #[test]
     fn test_empty() {
         let temp = tempfile::tempdir().unwrap();
         let vfs = Vfs { base: temp.path() };
         std::fs::write(temp.path().join("foo.i"), b"").unwrap();
         let revlog = Revlog::open(&vfs, "foo.i", None, false).unwrap();
         assert!(revlog.is_empty());
         assert_eq!(revlog.len(), 0);
         assert!(revlog.get_entry(0).is_err());
         assert!(!revlog.has_rev(0));
         assert_eq!(
             revlog.rev_from_node(NULL_NODE.into()).unwrap(),
             NULL_REVISION
         );
         let null_entry = revlog.get_entry(NULL_REVISION).ok().unwrap();
         assert_eq!(null_entry.revision(), NULL_REVISION);
         assert!(null_entry.data().unwrap().is_empty());
     }
 
     #[test]
     fn test_inline() {
         let temp = tempfile::tempdir().unwrap();
         let vfs = Vfs { base: temp.path() };
         let node0 = Node::from_hex("2ed2a3912a0b24502043eae84ee4b279c18b90dd")
             .unwrap();
         let node1 = Node::from_hex("b004912a8510032a0350a74daa2803dadfb00e12")
             .unwrap();
         let node2 = Node::from_hex("dd6ad206e907be60927b5a3117b97dffb2590582")
             .unwrap();
         let entry0_bytes = IndexEntryBuilder::new()
             .is_first(true)
             .with_version(1)
             .with_inline(true)
             .with_offset(INDEX_ENTRY_SIZE)
             .with_node(node0)
             .build();
         let entry1_bytes = IndexEntryBuilder::new()
             .with_offset(INDEX_ENTRY_SIZE)
             .with_node(node1)
             .build();
         let entry2_bytes = IndexEntryBuilder::new()
             .with_offset(INDEX_ENTRY_SIZE)
             .with_p1(0)
             .with_p2(1)
             .with_node(node2)
             .build();
         let contents = vec![entry0_bytes, entry1_bytes, entry2_bytes]
             .into_iter()
             .flatten()
             .collect_vec();
         std::fs::write(temp.path().join("foo.i"), contents).unwrap();
         let revlog = Revlog::open(&vfs, "foo.i", None, false).unwrap();
 
         let entry0 = revlog.get_entry(0).ok().unwrap();
         assert_eq!(entry0.revision(), 0);
         assert_eq!(*entry0.node(), node0);
         assert!(!entry0.has_p1());
         assert_eq!(entry0.p1(), None);
         assert_eq!(entry0.p2(), None);
         let p1_entry = entry0.p1_entry().unwrap();
         assert!(p1_entry.is_none());
         let p2_entry = entry0.p2_entry().unwrap();
         assert!(p2_entry.is_none());
 
         let entry1 = revlog.get_entry(1).ok().unwrap();
         assert_eq!(entry1.revision(), 1);
         assert_eq!(*entry1.node(), node1);
         assert!(!entry1.has_p1());
         assert_eq!(entry1.p1(), None);
         assert_eq!(entry1.p2(), None);
         let p1_entry = entry1.p1_entry().unwrap();
         assert!(p1_entry.is_none());
         let p2_entry = entry1.p2_entry().unwrap();
         assert!(p2_entry.is_none());
 
         let entry2 = revlog.get_entry(2).ok().unwrap();
         assert_eq!(entry2.revision(), 2);
         assert_eq!(*entry2.node(), node2);
         assert!(entry2.has_p1());
         assert_eq!(entry2.p1(), Some(0));
         assert_eq!(entry2.p2(), Some(1));
         let p1_entry = entry2.p1_entry().unwrap();
         assert!(p1_entry.is_some());
         assert_eq!(p1_entry.unwrap().revision(), 0);
         let p2_entry = entry2.p2_entry().unwrap();
         assert!(p2_entry.is_some());
         assert_eq!(p2_entry.unwrap().revision(), 1);
     }
 
     #[test]
     fn test_nodemap() {
         let temp = tempfile::tempdir().unwrap();
         let vfs = Vfs { base: temp.path() };
 
         // building a revlog with a forced Node starting with zeros
         // This is a corruption, but it does not preclude using the nodemap
         // if we don't try and access the data
         let node0 = Node::from_hex("00d2a3912a0b24502043eae84ee4b279c18b90dd")
             .unwrap();
         let node1 = Node::from_hex("b004912a8510032a0350a74daa2803dadfb00e12")
             .unwrap();
         let entry0_bytes = IndexEntryBuilder::new()
             .is_first(true)
             .with_version(1)
             .with_inline(true)
             .with_offset(INDEX_ENTRY_SIZE)
             .with_node(node0)
             .build();
         let entry1_bytes = IndexEntryBuilder::new()
             .with_offset(INDEX_ENTRY_SIZE)
             .with_node(node1)
             .build();
         let contents = vec![entry0_bytes, entry1_bytes]
             .into_iter()
             .flatten()
             .collect_vec();
         std::fs::write(temp.path().join("foo.i"), contents).unwrap();
-        let revlog = Revlog::open(&vfs, "foo.i", None, false).unwrap();
+
+        let mut idx = nodemap::tests::TestNtIndex::new();
+        idx.insert_node(0, node0).unwrap();
+        idx.insert_node(1, node1).unwrap();
+
+        let revlog =
+            Revlog::open_gen(&vfs, "foo.i", None, true, Some(idx.nt)).unwrap();
 
         // accessing the data shows the corruption
         revlog.get_entry(0).unwrap().data().unwrap_err();
 
         assert_eq!(revlog.rev_from_node(NULL_NODE.into()).unwrap(), -1);
         assert_eq!(revlog.rev_from_node(node0.into()).unwrap(), 0);
         assert_eq!(revlog.rev_from_node(node1.into()).unwrap(), 1);
         assert_eq!(
             revlog
                 .rev_from_node(NodePrefix::from_hex("000").unwrap())
                 .unwrap(),
             -1
         );
         assert_eq!(
             revlog
                 .rev_from_node(NodePrefix::from_hex("b00").unwrap())
                 .unwrap(),
             1
         );
         // RevlogError does not implement PartialEq
         // (ultimately because io::Error does not)
         match revlog
             .rev_from_node(NodePrefix::from_hex("00").unwrap())
             .expect_err("Expected to give AmbiguousPrefix error")
         {
             RevlogError::AmbiguousPrefix => (),
             e => {
                 panic!("Got another error than AmbiguousPrefix: {:?}", e);
             }
         };
     }
 }

rust/hg-core/src/revlog/nodemap.rs

@@ -1,1067 +1,1074 b''
 // Copyright 2018-2020 Georges Racinet <[e-mail]>
 //           and Mercurial contributors
 //
 // This software may be used and distributed according to the terms of the
 // GNU General Public License version 2 or any later version.
 //! Indexing facilities for fast retrieval of `Revision` from `Node`
 //!
 //! This provides a variation on the 16-ary radix tree that is
 //! provided as "nodetree" in revlog.c, ready for append-only persistence
 //! on disk.
 //!
 //! Following existing implicit conventions, the "nodemap" terminology
 //! is used in a more abstract context.
 
 use super::{
     node::NULL_NODE, Node, NodePrefix, Revision, RevlogIndex, NULL_REVISION,
 };
 
 use bytes_cast::{unaligned, BytesCast};
 use std::cmp::max;
 use std::fmt;
 use std::mem::{self, align_of, size_of};
 use std::ops::Deref;
 use std::ops::Index;
 
 #[derive(Debug, PartialEq)]
 pub enum NodeMapError {
     /// A `NodePrefix` matches several [`Revision`]s.
     ///
     /// This can be returned by methods meant for (at most) one match.
     MultipleResults,
     /// A `Revision` stored in the nodemap could not be found in the index
     RevisionNotInIndex(Revision),
 }
 
 /// Mapping system from Mercurial nodes to revision numbers.
 ///
 /// ## `RevlogIndex` and `NodeMap`
 ///
 /// One way to think about their relationship is that
 /// the `NodeMap` is a prefix-oriented reverse index of the [`Node`]
 /// information carried by a [`RevlogIndex`].
 ///
 /// Many of the methods in this trait take a `RevlogIndex` argument
 /// which is used for validation of their results. This index must naturally
 /// be the one the `NodeMap` is about, and it must be consistent.
 ///
 /// Notably, the `NodeMap` must not store
 /// information about more `Revision` values than there are in the index.
 /// In these methods, an encountered `Revision` is not in the index, a
 /// [RevisionNotInIndex](NodeMapError) error is returned.
 ///
 /// In insert operations, the rule is thus that the `NodeMap` must always
 /// be updated after the `RevlogIndex` it is about.
 pub trait NodeMap {
     /// Find the unique `Revision` having the given `Node`
     ///
     /// If no Revision matches the given `Node`, `Ok(None)` is returned.
     fn find_node(
         &self,
         index: &impl RevlogIndex,
         node: &Node,
     ) -> Result<Option<Revision>, NodeMapError> {
         self.find_bin(index, node.into())
     }
 
     /// Find the unique Revision whose `Node` starts with a given binary prefix
     ///
     /// If no Revision matches the given prefix, `Ok(None)` is returned.
     ///
     /// If several Revisions match the given prefix, a
     /// [MultipleResults](NodeMapError)  error is returned.
     fn find_bin(
         &self,
         idx: &impl RevlogIndex,
         prefix: NodePrefix,
     ) -> Result<Option<Revision>, NodeMapError>;
 
     /// Give the size of the shortest node prefix that determines
     /// the revision uniquely.
     ///
     /// From a binary node prefix, if it is matched in the node map, this
     /// returns the number of hexadecimal digits that would had sufficed
     /// to find the revision uniquely.
     ///
     /// Returns `None` if no [`Revision`] could be found for the prefix.
     ///
     /// If several Revisions match the given prefix, a
     /// [MultipleResults](NodeMapError)  error is returned.
     fn unique_prefix_len_bin(
         &self,
         idx: &impl RevlogIndex,
         node_prefix: NodePrefix,
     ) -> Result<Option<usize>, NodeMapError>;
 
     /// Same as [unique_prefix_len_bin](Self::unique_prefix_len_bin), with
     /// a full [`Node`] as input
     fn unique_prefix_len_node(
         &self,
         idx: &impl RevlogIndex,
         node: &Node,
     ) -> Result<Option<usize>, NodeMapError> {
         self.unique_prefix_len_bin(idx, node.into())
     }
 }
 
 pub trait MutableNodeMap: NodeMap {
     fn insert<I: RevlogIndex>(
         &mut self,
         index: &I,
         node: &Node,
         rev: Revision,
     ) -> Result<(), NodeMapError>;
 }
 
 /// Low level NodeTree [`Block`] elements
 ///
 /// These are exactly as for instance on persistent storage.
 type RawElement = unaligned::I32Be;
 
 /// High level representation of values in NodeTree
 /// [`Blocks`](struct.Block.html)
 ///
 /// This is the high level representation that most algorithms should
 /// use.
 #[derive(Clone, Debug, Eq, PartialEq)]
 enum Element {
     Rev(Revision),
     Block(usize),
     None,
 }
 
 impl From<RawElement> for Element {
     /// Conversion from low level representation, after endianness conversion.
     ///
     /// See [`Block`](struct.Block.html) for explanation about the encoding.
     fn from(raw: RawElement) -> Element {
         let int = raw.get();
         if int >= 0 {
             Element::Block(int as usize)
         } else if int == -1 {
             Element::None
         } else {
             Element::Rev(-int - 2)
         }
     }
 }
 
 impl From<Element> for RawElement {
     fn from(element: Element) -> RawElement {
         RawElement::from(match element {
             Element::None => 0,
             Element::Block(i) => i as i32,
             Element::Rev(rev) => -rev - 2,
         })
     }
 }
 
 const ELEMENTS_PER_BLOCK: usize = 16; // number of different values in a nybble
 
 /// A logical block of the [`NodeTree`], packed with a fixed size.
 ///
 /// These are always used in container types implementing `Index<Block>`,
 /// such as `&Block`
 ///
 /// As an array of integers, its ith element encodes that the
 /// ith potential edge from the block, representing the ith hexadecimal digit
 /// (nybble) `i` is either:
 ///
 /// - absent (value -1)
 /// - another `Block` in the same indexable container (value ≥ 0)
 ///  - a [`Revision`] leaf (value ≤ -2)
 ///
 /// Endianness has to be fixed for consistency on shared storage across
 /// different architectures.
 ///
 /// A key difference with the C `nodetree` is that we need to be
 /// able to represent the [`Block`] at index 0, hence -1 is the empty marker
 /// rather than 0 and the [`Revision`] range upper limit of -2 instead of -1.
 ///
 /// Another related difference is that `NULL_REVISION` (-1) is not
 /// represented at all, because we want an immutable empty nodetree
 /// to be valid.
 #[derive(Copy, Clone, BytesCast, PartialEq)]
 #[repr(transparent)]
 pub struct Block([RawElement; ELEMENTS_PER_BLOCK]);
 
 impl Block {
     fn new() -> Self {
         let absent_node = RawElement::from(-1);
         Block([absent_node; ELEMENTS_PER_BLOCK])
     }
 
     fn get(&self, nybble: u8) -> Element {
         self.0[nybble as usize].into()
     }
 
     fn set(&mut self, nybble: u8, element: Element) {
         self.0[nybble as usize] = element.into()
     }
 }
 
 impl fmt::Debug for Block {
     /// sparse representation for testing and debugging purposes
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         f.debug_map()
             .entries((0..16).filter_map(|i| match self.get(i) {
                 Element::None => None,
                 element => Some((i, element)),
             }))
             .finish()
     }
 }
 
 /// A mutable 16-radix tree with the root block logically at the end
 ///
 /// Because of the append only nature of our node trees, we need to
 /// keep the original untouched and store new blocks separately.
 ///
 /// The mutable root [`Block`] is kept apart so that we don't have to rebump
 /// it on each insertion.
 pub struct NodeTree {
     readonly: Box<dyn Deref<Target = [Block]> + Send>,
     growable: Vec<Block>,
     root: Block,
     masked_inner_blocks: usize,
 }
 
 impl Index<usize> for NodeTree {
     type Output = Block;
 
     fn index(&self, i: usize) -> &Block {
         let ro_len = self.readonly.len();
         if i < ro_len {
             &self.readonly[i]
         } else if i == ro_len + self.growable.len() {
             &self.root
         } else {
             &self.growable[i - ro_len]
         }
     }
 }
 
 /// Return `None` unless the [`Node`] for `rev` has given prefix in `idx`.
 fn has_prefix_or_none(
     idx: &impl RevlogIndex,
     prefix: NodePrefix,
     rev: Revision,
 ) -> Result<Option<Revision>, NodeMapError> {
     idx.node(rev)
         .ok_or(NodeMapError::RevisionNotInIndex(rev))
         .map(|node| {
             if prefix.is_prefix_of(node) {
                 Some(rev)
             } else {
                 None
             }
         })
 }
 
 /// validate that the candidate's node starts indeed with given prefix,
 /// and treat ambiguities related to [`NULL_REVISION`].
 ///
 /// From the data in the NodeTree, one can only conclude that some
 /// revision is the only one for a *subprefix* of the one being looked up.
 fn validate_candidate(
     idx: &impl RevlogIndex,
     prefix: NodePrefix,
     candidate: (Option<Revision>, usize),
 ) -> Result<(Option<Revision>, usize), NodeMapError> {
     let (rev, steps) = candidate;
     if let Some(nz_nybble) = prefix.first_different_nybble(&NULL_NODE) {
         rev.map_or(Ok((None, steps)), |r| {
             has_prefix_or_none(idx, prefix, r)
                 .map(|opt| (opt, max(steps, nz_nybble + 1)))
         })
     } else {
         // the prefix is only made of zeros; NULL_REVISION always matches it
         // and any other *valid* result is an ambiguity
         match rev {
             None => Ok((Some(NULL_REVISION), steps + 1)),
             Some(r) => match has_prefix_or_none(idx, prefix, r)? {
                 None => Ok((Some(NULL_REVISION), steps + 1)),
                 _ => Err(NodeMapError::MultipleResults),
             },
         }
     }
 }
 
 impl NodeTree {
     /// Initiate a NodeTree from an immutable slice-like of `Block`
     ///
     /// We keep `readonly` and clone its root block if it isn't empty.
     fn new(readonly: Box<dyn Deref<Target = [Block]> + Send>) -> Self {
         let root = readonly.last().cloned().unwrap_or_else(Block::new);
         NodeTree {
             readonly,
             growable: Vec::new(),
             root,
             masked_inner_blocks: 0,
         }
     }
 
     /// Create from an opaque bunch of bytes
     ///
     /// The created [`NodeTreeBytes`] from `bytes`,
     /// of which exactly `amount` bytes are used.
     ///
     /// - `buffer` could be derived from `PyBuffer` and `Mmap` objects.
     /// - `amount` is expressed in bytes, and is not automatically derived from
     ///   `bytes`, so that a caller that manages them atomically can perform
     ///   temporary disk serializations and still rollback easily if needed.
     ///   First use-case for this would be to support Mercurial shell hooks.
     ///
     /// panics if `buffer` is smaller than `amount`
     pub fn load_bytes(
         bytes: Box<dyn Deref<Target = [u8]> + Send>,
         amount: usize,
     ) -> Self {
         NodeTree::new(Box::new(NodeTreeBytes::new(bytes, amount)))
     }
 
     /// Retrieve added [`Block`]s and the original immutable data
     pub fn into_readonly_and_added(
         self,
     ) -> (Box<dyn Deref<Target = [Block]> + Send>, Vec<Block>) {
         let mut vec = self.growable;
         let readonly = self.readonly;
         if readonly.last() != Some(&self.root) {
             vec.push(self.root);
         }
         (readonly, vec)
     }
 
     /// Retrieve added [`Block]s as bytes, ready to be written to persistent
     /// storage
     pub fn into_readonly_and_added_bytes(
         self,
     ) -> (Box<dyn Deref<Target = [Block]> + Send>, Vec<u8>) {
         let (readonly, vec) = self.into_readonly_and_added();
         // Prevent running `v`'s destructor so we are in complete control
         // of the allocation.
         let vec = mem::ManuallyDrop::new(vec);
 
         // Transmute the `Vec<Block>` to a `Vec<u8>`. Blocks are contiguous
         // bytes, so this is perfectly safe.
         let bytes = unsafe {
             // Check for compatible allocation layout.
             // (Optimized away by constant-folding + dead code elimination.)
             assert_eq!(size_of::<Block>(), 64);
             assert_eq!(align_of::<Block>(), 1);
 
             // /!\ Any use of `vec` after this is use-after-free.
             // TODO: use `into_raw_parts` once stabilized
             Vec::from_raw_parts(
                 vec.as_ptr() as *mut u8,
                 vec.len() * size_of::<Block>(),
                 vec.capacity() * size_of::<Block>(),
             )
         };
         (readonly, bytes)
     }
 
     /// Total number of blocks
     fn len(&self) -> usize {
         self.readonly.len() + self.growable.len() + 1
     }
 
     /// Implemented for completeness
     ///
     /// A `NodeTree` always has at least the mutable root block.
     #[allow(dead_code)]
     fn is_empty(&self) -> bool {
         false
     }
 
     /// Main working method for `NodeTree` searches
     ///
     /// The first returned value is the result of analysing `NodeTree` data
     /// *alone*: whereas `None` guarantees that the given prefix is absent
     /// from the [`NodeTree`] data (but still could match [`NULL_NODE`]), with
     /// `Some(rev)`, it is to be understood that `rev` is the unique
     /// [`Revision`] that could match the prefix. Actually, all that can
     /// be inferred from
     /// the `NodeTree` data is that `rev` is the revision with the longest
     /// common node prefix with the given prefix.
     ///
     /// The second returned value is the size of the smallest subprefix
     /// of `prefix` that would give the same result, i.e. not the
     /// [MultipleResults](NodeMapError) error variant (again, using only the
     /// data of the [`NodeTree`]).
     fn lookup(
         &self,
         prefix: NodePrefix,
     ) -> Result<(Option<Revision>, usize), NodeMapError> {
         for (i, visit_item) in self.visit(prefix).enumerate() {
             if let Some(opt) = visit_item.final_revision() {
                 return Ok((opt, i + 1));
             }
         }
         Err(NodeMapError::MultipleResults)
     }
 
     fn visit(&self, prefix: NodePrefix) -> NodeTreeVisitor {
         NodeTreeVisitor {
             nt: self,
             prefix,
             visit: self.len() - 1,
             nybble_idx: 0,
             done: false,
         }
     }
     /// Return a mutable reference for `Block` at index `idx`.
     ///
     /// If `idx` lies in the immutable area, then the reference is to
     /// a newly appended copy.
     ///
     /// Returns (new_idx, glen, mut_ref) where
     ///
     /// - `new_idx` is the index of the mutable `Block`
     /// - `mut_ref` is a mutable reference to the mutable Block.
     /// - `glen` is the new length of `self.growable`
     ///
     /// Note: the caller wouldn't be allowed to query `self.growable.len()`
     /// itself because of the mutable borrow taken with the returned `Block`
     fn mutable_block(&mut self, idx: usize) -> (usize, &mut Block, usize) {
         let ro_blocks = &self.readonly;
         let ro_len = ro_blocks.len();
         let glen = self.growable.len();
         if idx < ro_len {
             self.masked_inner_blocks += 1;
             self.growable.push(ro_blocks[idx]);
             (glen + ro_len, &mut self.growable[glen], glen + 1)
         } else if glen + ro_len == idx {
             (idx, &mut self.root, glen)
         } else {
             (idx, &mut self.growable[idx - ro_len], glen)
         }
     }
 
     /// Main insertion method
     ///
     /// This will dive in the node tree to find the deepest `Block` for
     /// `node`, split it as much as needed and record `node` in there.
     /// The method then backtracks, updating references in all the visited
     /// blocks from the root.
     ///
     /// All the mutated `Block` are copied first to the growable part if
     /// needed. That happens for those in the immutable part except the root.
     pub fn insert<I: RevlogIndex>(
         &mut self,
         index: &I,
         node: &Node,
         rev: Revision,
     ) -> Result<(), NodeMapError> {
         let ro_len = &self.readonly.len();
 
         let mut visit_steps: Vec<_> = self.visit(node.into()).collect();
         let read_nybbles = visit_steps.len();
         // visit_steps cannot be empty, since we always visit the root block
         let deepest = visit_steps.pop().unwrap();
 
         let (mut block_idx, mut block, mut glen) =
             self.mutable_block(deepest.block_idx);
 
         if let Element::Rev(old_rev) = deepest.element {
             let old_node = index
                 .node(old_rev)
                 .ok_or(NodeMapError::RevisionNotInIndex(old_rev))?;
             if old_node == node {
                 return Ok(()); // avoid creating lots of useless blocks
             }
 
             // Looping over the tail of nybbles in both nodes, creating
             // new blocks until we find the difference
             let mut new_block_idx = ro_len + glen;
             let mut nybble = deepest.nybble;
             for nybble_pos in read_nybbles..node.nybbles_len() {
                 block.set(nybble, Element::Block(new_block_idx));
 
                 let new_nybble = node.get_nybble(nybble_pos);
                 let old_nybble = old_node.get_nybble(nybble_pos);
 
                 if old_nybble == new_nybble {
                     self.growable.push(Block::new());
                     block = &mut self.growable[glen];
                     glen += 1;
                     new_block_idx += 1;
                     nybble = new_nybble;
                 } else {
                     let mut new_block = Block::new();
                     new_block.set(old_nybble, Element::Rev(old_rev));
                     new_block.set(new_nybble, Element::Rev(rev));
                     self.growable.push(new_block);
                     break;
                 }
             }
         } else {
             // Free slot in the deepest block: no splitting has to be done
             block.set(deepest.nybble, Element::Rev(rev));
         }
 
         // Backtrack over visit steps to update references
         while let Some(visited) = visit_steps.pop() {
             let to_write = Element::Block(block_idx);
             if visit_steps.is_empty() {
                 self.root.set(visited.nybble, to_write);
                 break;
             }
             let (new_idx, block, _) = self.mutable_block(visited.block_idx);
             if block.get(visited.nybble) == to_write {
                 break;
             }
             block.set(visited.nybble, to_write);
             block_idx = new_idx;
         }
         Ok(())
     }
 
     /// Make the whole `NodeTree` logically empty, without touching the
     /// immutable part.
     pub fn invalidate_all(&mut self) {
         self.root = Block::new();
         self.growable = Vec::new();
         self.masked_inner_blocks = self.readonly.len();
     }
 
     /// Return the number of blocks in the readonly part that are currently
     /// masked in the mutable part.
     ///
     /// The `NodeTree` structure has no efficient way to know how many blocks
     /// are already unreachable in the readonly part.
     ///
     /// After a call to `invalidate_all()`, the returned number can be actually
     /// bigger than the whole readonly part, a conventional way to mean that
     /// all the readonly blocks have been masked. This is what is really
     /// useful to the caller and does not require to know how many were
     /// actually unreachable to begin with.
     pub fn masked_readonly_blocks(&self) -> usize {
         if let Some(readonly_root) = self.readonly.last() {
             if readonly_root == &self.root {
                 return 0;
             }
         } else {
             return 0;
         }
         self.masked_inner_blocks + 1
     }
 }
 
 pub struct NodeTreeBytes {
     buffer: Box<dyn Deref<Target = [u8]> + Send>,
     len_in_blocks: usize,
 }
 
 impl NodeTreeBytes {
     fn new(
         buffer: Box<dyn Deref<Target = [u8]> + Send>,
         amount: usize,
     ) -> Self {
         assert!(buffer.len() >= amount);
         let len_in_blocks = amount / size_of::<Block>();
         NodeTreeBytes {
             buffer,
             len_in_blocks,
         }
     }
 }
 
 impl Deref for NodeTreeBytes {
     type Target = [Block];
 
     fn deref(&self) -> &[Block] {
         Block::slice_from_bytes(&self.buffer, self.len_in_blocks)
             // `NodeTreeBytes::new` already asserted that `self.buffer` is
             // large enough.
             .unwrap()
             .0
     }
 }
 
 struct NodeTreeVisitor<'n> {
     nt: &'n NodeTree,
     prefix: NodePrefix,
     visit: usize,
     nybble_idx: usize,
     done: bool,
 }
 
 #[derive(Debug, PartialEq, Clone)]
 struct NodeTreeVisitItem {
     block_idx: usize,
     nybble: u8,
     element: Element,
 }
 
 impl<'n> Iterator for NodeTreeVisitor<'n> {
     type Item = NodeTreeVisitItem;
 
     fn next(&mut self) -> Option<Self::Item> {
         if self.done || self.nybble_idx >= self.prefix.nybbles_len() {
             return None;
         }
 
         let nybble = self.prefix.get_nybble(self.nybble_idx);
         self.nybble_idx += 1;
 
         let visit = self.visit;
         let element = self.nt[visit].get(nybble);
         if let Element::Block(idx) = element {
             self.visit = idx;
         } else {
             self.done = true;
         }
 
         Some(NodeTreeVisitItem {
             block_idx: visit,
             nybble,
             element,
         })
     }
 }
 
 impl NodeTreeVisitItem {
     // Return `Some(opt)` if this item is final, with `opt` being the
     // `Revision` that it may represent.
     //
     // If the item is not terminal, return `None`
     fn final_revision(&self) -> Option<Option<Revision>> {
         match self.element {
             Element::Block(_) => None,
             Element::Rev(r) => Some(Some(r)),
             Element::None => Some(None),
         }
     }
 }
 
 impl From<Vec<Block>> for NodeTree {
     fn from(vec: Vec<Block>) -> Self {
         Self::new(Box::new(vec))
     }
 }
 
 impl fmt::Debug for NodeTree {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         let readonly: &[Block] = &*self.readonly;
         write!(
             f,
             "readonly: {:?}, growable: {:?}, root: {:?}",
             readonly, self.growable, self.root
         )
     }
 }
 
 impl Default for NodeTree {
     /// Create a fully mutable empty NodeTree
     fn default() -> Self {
         NodeTree::new(Box::new(Vec::new()))
     }
 }
 
 impl NodeMap for NodeTree {
     fn find_bin<'a>(
         &self,
         idx: &impl RevlogIndex,
         prefix: NodePrefix,
     ) -> Result<Option<Revision>, NodeMapError> {
         validate_candidate(idx, prefix, self.lookup(prefix)?)
             .map(|(opt, _shortest)| opt)
     }
 
     fn unique_prefix_len_bin<'a>(
         &self,
         idx: &impl RevlogIndex,
         prefix: NodePrefix,
     ) -> Result<Option<usize>, NodeMapError> {
         validate_candidate(idx, prefix, self.lookup(prefix)?)
             .map(|(opt, shortest)| opt.map(|_rev| shortest))
     }
 }
 
 #[cfg(test)]
-mod tests {
+pub mod tests {
     use super::NodeMapError::*;
     use super::*;
     use crate::revlog::node::{hex_pad_right, Node};
     use std::collections::HashMap;
 
     /// Creates a `Block` using a syntax close to the `Debug` output
     macro_rules! block {
         {$($nybble:tt : $variant:ident($val:tt)),*} => (
             {
                 let mut block = Block::new();
                 $(block.set($nybble, Element::$variant($val)));*;
                 block
             }
         )
     }
 
     #[test]
     fn test_block_debug() {
         let mut block = Block::new();
         block.set(1, Element::Rev(3));
         block.set(10, Element::Block(0));
         assert_eq!(format!("{:?}", block), "{1: Rev(3), 10: Block(0)}");
     }
 
     #[test]
     fn test_block_macro() {
         let block = block! {5: Block(2)};
         assert_eq!(format!("{:?}", block), "{5: Block(2)}");
 
         let block = block! {13: Rev(15), 5: Block(2)};
         assert_eq!(format!("{:?}", block), "{5: Block(2), 13: Rev(15)}");
     }
 
     #[test]
     fn test_raw_block() {
         let mut raw = [255u8; 64];
 
         let mut counter = 0;
         for val in [0_i32, 15, -2, -1, -3].iter() {
             for byte in val.to_be_bytes().iter() {
                 raw[counter] = *byte;
                 counter += 1;
             }
         }
         let (block, _) = Block::from_bytes(&raw).unwrap();
         assert_eq!(block.get(0), Element::Block(0));
         assert_eq!(block.get(1), Element::Block(15));
         assert_eq!(block.get(3), Element::None);
         assert_eq!(block.get(2), Element::Rev(0));
         assert_eq!(block.get(4), Element::Rev(1));
     }
 
     type TestIndex = HashMap<Revision, Node>;
 
     impl RevlogIndex for TestIndex {
         fn node(&self, rev: Revision) -> Option<&Node> {
             self.get(&rev)
         }
 
         fn len(&self) -> usize {
             self.len()
         }
     }
 
     /// Pad hexadecimal Node prefix with zeros on the right
     ///
     /// This avoids having to repeatedly write very long hexadecimal
     /// strings for test data, and brings actual hash size independency.
     #[cfg(test)]
     fn pad_node(hex: &str) -> Node {
         Node::from_hex(&hex_pad_right(hex)).unwrap()
     }
 
     /// Pad hexadecimal Node prefix with zeros on the right, then insert
     fn pad_insert(idx: &mut TestIndex, rev: Revision, hex: &str) {
         idx.insert(rev, pad_node(hex));
     }
 
     fn sample_nodetree() -> NodeTree {
         NodeTree::from(vec![
             block![0: Rev(9)],
             block![0: Rev(0), 1: Rev(9)],
             block![0: Block(1), 1:Rev(1)],
         ])
     }
 
     fn hex(s: &str) -> NodePrefix {
         NodePrefix::from_hex(s).unwrap()
     }
 
     #[test]
     fn test_nt_debug() {
         let nt = sample_nodetree();
         assert_eq!(
             format!("{:?}", nt),
             "readonly: \
              [{0: Rev(9)}, {0: Rev(0), 1: Rev(9)}, {0: Block(1), 1: Rev(1)}], \
              growable: [], \
              root: {0: Block(1), 1: Rev(1)}",
         );
     }
 
     #[test]
     fn test_immutable_find_simplest() -> Result<(), NodeMapError> {
         let mut idx: TestIndex = HashMap::new();
         pad_insert(&mut idx, 1, "1234deadcafe");
 
         let nt = NodeTree::from(vec![block! {1: Rev(1)}]);
         assert_eq!(nt.find_bin(&idx, hex("1"))?, Some(1));
         assert_eq!(nt.find_bin(&idx, hex("12"))?, Some(1));
         assert_eq!(nt.find_bin(&idx, hex("1234de"))?, Some(1));
         assert_eq!(nt.find_bin(&idx, hex("1a"))?, None);
         assert_eq!(nt.find_bin(&idx, hex("ab"))?, None);
 
         // and with full binary Nodes
         assert_eq!(nt.find_node(&idx, idx.get(&1).unwrap())?, Some(1));
         let unknown = Node::from_hex(&hex_pad_right("3d")).unwrap();
         assert_eq!(nt.find_node(&idx, &unknown)?, None);
         Ok(())
     }
 
     #[test]
     fn test_immutable_find_one_jump() {
         let mut idx = TestIndex::new();
         pad_insert(&mut idx, 9, "012");
         pad_insert(&mut idx, 0, "00a");
 
         let nt = sample_nodetree();
 
         assert_eq!(nt.find_bin(&idx, hex("0")), Err(MultipleResults));
         assert_eq!(nt.find_bin(&idx, hex("01")), Ok(Some(9)));
         assert_eq!(nt.find_bin(&idx, hex("00")), Err(MultipleResults));
         assert_eq!(nt.find_bin(&idx, hex("00a")), Ok(Some(0)));
         assert_eq!(nt.unique_prefix_len_bin(&idx, hex("00a")), Ok(Some(3)));
         assert_eq!(nt.find_bin(&idx, hex("000")), Ok(Some(NULL_REVISION)));
     }
 
     #[test]
     fn test_mutated_find() -> Result<(), NodeMapError> {
         let mut idx = TestIndex::new();
         pad_insert(&mut idx, 9, "012");
         pad_insert(&mut idx, 0, "00a");
         pad_insert(&mut idx, 2, "cafe");
         pad_insert(&mut idx, 3, "15");
         pad_insert(&mut idx, 1, "10");
 
         let nt = NodeTree {
             readonly: sample_nodetree().readonly,
             growable: vec![block![0: Rev(1), 5: Rev(3)]],
             root: block![0: Block(1), 1:Block(3), 12: Rev(2)],
             masked_inner_blocks: 1,
         };
         assert_eq!(nt.find_bin(&idx, hex("10"))?, Some(1));
         assert_eq!(nt.find_bin(&idx, hex("c"))?, Some(2));
         assert_eq!(nt.unique_prefix_len_bin(&idx, hex("c"))?, Some(1));
         assert_eq!(nt.find_bin(&idx, hex("00")), Err(MultipleResults));
         assert_eq!(nt.find_bin(&idx, hex("000"))?, Some(NULL_REVISION));
         assert_eq!(nt.unique_prefix_len_bin(&idx, hex("000"))?, Some(3));
         assert_eq!(nt.find_bin(&idx, hex("01"))?, Some(9));
         assert_eq!(nt.masked_readonly_blocks(), 2);
         Ok(())
     }
 
-    struct TestNtIndex {
-        index: TestIndex,
-        nt: NodeTree,
+    pub struct TestNtIndex {
+        pub index: TestIndex,
+        pub nt: NodeTree,
     }
 
     impl TestNtIndex {
-        fn new() -> Self {
+        pub fn new() -> Self {
             TestNtIndex {
                 index: HashMap::new(),
                 nt: NodeTree::default(),
             }
         }
 
-        fn insert(
+        pub fn insert_node(
+            &mut self,
+            rev: Revision,
+            node: Node,
+        ) -> Result<(), NodeMapError> {
+            self.index.insert(rev, node);
+            self.nt.insert(&self.index, &node, rev)?;
+            Ok(())
+        }
+
+        pub fn insert(
             &mut self,
             rev: Revision,
             hex: &str,
         ) -> Result<(), NodeMapError> {
-            let node = pad_node(hex);
-            self.index.insert(rev, node);
-            self.nt.insert(&self.index, &node, rev)?;
-            Ok(())
+            return self.insert_node(rev, pad_node(hex));
         }
 
         fn find_hex(
             &self,
             prefix: &str,
         ) -> Result<Option<Revision>, NodeMapError> {
             self.nt.find_bin(&self.index, hex(prefix))
         }
 
         fn unique_prefix_len_hex(
             &self,
             prefix: &str,
         ) -> Result<Option<usize>, NodeMapError> {
             self.nt.unique_prefix_len_bin(&self.index, hex(prefix))
         }
 
         /// Drain `added` and restart a new one
         fn commit(self) -> Self {
             let mut as_vec: Vec<Block> =
                 self.nt.readonly.iter().copied().collect();
             as_vec.extend(self.nt.growable);
             as_vec.push(self.nt.root);
 
             Self {
                 index: self.index,
                 nt: NodeTree::from(as_vec),
             }
         }
     }
 
     #[test]
     fn test_insert_full_mutable() -> Result<(), NodeMapError> {
         let mut idx = TestNtIndex::new();
         idx.insert(0, "1234")?;
         assert_eq!(idx.find_hex("1")?, Some(0));
         assert_eq!(idx.find_hex("12")?, Some(0));
 
         // let's trigger a simple split
         idx.insert(1, "1a34")?;
         assert_eq!(idx.nt.growable.len(), 1);
         assert_eq!(idx.find_hex("12")?, Some(0));
         assert_eq!(idx.find_hex("1a")?, Some(1));
 
         // reinserting is a no_op
         idx.insert(1, "1a34")?;
         assert_eq!(idx.nt.growable.len(), 1);
         assert_eq!(idx.find_hex("12")?, Some(0));
         assert_eq!(idx.find_hex("1a")?, Some(1));
 
         idx.insert(2, "1a01")?;
         assert_eq!(idx.nt.growable.len(), 2);
         assert_eq!(idx.find_hex("1a"), Err(NodeMapError::MultipleResults));
         assert_eq!(idx.find_hex("12")?, Some(0));
         assert_eq!(idx.find_hex("1a3")?, Some(1));
         assert_eq!(idx.find_hex("1a0")?, Some(2));
         assert_eq!(idx.find_hex("1a12")?, None);
 
         // now let's make it split and create more than one additional block
         idx.insert(3, "1a345")?;
         assert_eq!(idx.nt.growable.len(), 4);
         assert_eq!(idx.find_hex("1a340")?, Some(1));
         assert_eq!(idx.find_hex("1a345")?, Some(3));
         assert_eq!(idx.find_hex("1a341")?, None);
 
         // there's no readonly block to mask
         assert_eq!(idx.nt.masked_readonly_blocks(), 0);
         Ok(())
     }
 
     #[test]
     fn test_unique_prefix_len_zero_prefix() {
         let mut idx = TestNtIndex::new();
         idx.insert(0, "00000abcd").unwrap();
 
         assert_eq!(idx.find_hex("000"), Err(NodeMapError::MultipleResults));
         // in the nodetree proper, this will be found at the first nybble
         // yet the correct answer for unique_prefix_len is not 1, nor 1+1,
         // but the first difference with `NULL_NODE`
         assert_eq!(idx.unique_prefix_len_hex("00000a"), Ok(Some(6)));
         assert_eq!(idx.unique_prefix_len_hex("00000ab"), Ok(Some(6)));
 
         // same with odd result
         idx.insert(1, "00123").unwrap();
         assert_eq!(idx.unique_prefix_len_hex("001"), Ok(Some(3)));
         assert_eq!(idx.unique_prefix_len_hex("0012"), Ok(Some(3)));
 
         // these are unchanged of course
         assert_eq!(idx.unique_prefix_len_hex("00000a"), Ok(Some(6)));
         assert_eq!(idx.unique_prefix_len_hex("00000ab"), Ok(Some(6)));
     }
 
     #[test]
     fn test_insert_extreme_splitting() -> Result<(), NodeMapError> {
         // check that the splitting loop is long enough
         let mut nt_idx = TestNtIndex::new();
         let nt = &mut nt_idx.nt;
         let idx = &mut nt_idx.index;
 
         let node0_hex = hex_pad_right("444444");
         let mut node1_hex = hex_pad_right("444444");
         node1_hex.pop();
         node1_hex.push('5');
         let node0 = Node::from_hex(&node0_hex).unwrap();
         let node1 = Node::from_hex(&node1_hex).unwrap();
 
         idx.insert(0, node0);
         nt.insert(idx, &node0, 0)?;
         idx.insert(1, node1);
         nt.insert(idx, &node1, 1)?;
 
         assert_eq!(nt.find_bin(idx, (&node0).into())?, Some(0));
         assert_eq!(nt.find_bin(idx, (&node1).into())?, Some(1));
         Ok(())
     }
 
     #[test]
     fn test_insert_partly_immutable() -> Result<(), NodeMapError> {
         let mut idx = TestNtIndex::new();
         idx.insert(0, "1234")?;
         idx.insert(1, "1235")?;
         idx.insert(2, "131")?;
         idx.insert(3, "cafe")?;
         let mut idx = idx.commit();
         assert_eq!(idx.find_hex("1234")?, Some(0));
         assert_eq!(idx.find_hex("1235")?, Some(1));
         assert_eq!(idx.find_hex("131")?, Some(2));
         assert_eq!(idx.find_hex("cafe")?, Some(3));
         // we did not add anything since init from readonly
         assert_eq!(idx.nt.masked_readonly_blocks(), 0);
 
         idx.insert(4, "123A")?;
         assert_eq!(idx.find_hex("1234")?, Some(0));
         assert_eq!(idx.find_hex("1235")?, Some(1));
         assert_eq!(idx.find_hex("131")?, Some(2));
         assert_eq!(idx.find_hex("cafe")?, Some(3));
         assert_eq!(idx.find_hex("123A")?, Some(4));
         // we masked blocks for all prefixes of "123", including the root
         assert_eq!(idx.nt.masked_readonly_blocks(), 4);
 
         eprintln!("{:?}", idx.nt);
         idx.insert(5, "c0")?;
         assert_eq!(idx.find_hex("cafe")?, Some(3));
         assert_eq!(idx.find_hex("c0")?, Some(5));
         assert_eq!(idx.find_hex("c1")?, None);
         assert_eq!(idx.find_hex("1234")?, Some(0));
         // inserting "c0" is just splitting the 'c' slot of the mutable root,
         // it doesn't mask anything
         assert_eq!(idx.nt.masked_readonly_blocks(), 4);
 
         Ok(())
     }
 
     #[test]
     fn test_invalidate_all() -> Result<(), NodeMapError> {
         let mut idx = TestNtIndex::new();
         idx.insert(0, "1234")?;
         idx.insert(1, "1235")?;
         idx.insert(2, "131")?;
         idx.insert(3, "cafe")?;
         let mut idx = idx.commit();
 
         idx.nt.invalidate_all();
 
         assert_eq!(idx.find_hex("1234")?, None);
         assert_eq!(idx.find_hex("1235")?, None);
         assert_eq!(idx.find_hex("131")?, None);
         assert_eq!(idx.find_hex("cafe")?, None);
         // all the readonly blocks have been masked, this is the
         // conventional expected response
         assert_eq!(idx.nt.masked_readonly_blocks(), idx.nt.readonly.len() + 1);
         Ok(())
     }
 
     #[test]
     fn test_into_added_empty() {
         assert!(sample_nodetree().into_readonly_and_added().1.is_empty());
         assert!(sample_nodetree()
             .into_readonly_and_added_bytes()
             .1
             .is_empty());
     }
 
     #[test]
     fn test_into_added_bytes() -> Result<(), NodeMapError> {
         let mut idx = TestNtIndex::new();
         idx.insert(0, "1234")?;
         let mut idx = idx.commit();
         idx.insert(4, "cafe")?;
         let (_, bytes) = idx.nt.into_readonly_and_added_bytes();
 
         // only the root block has been changed
         assert_eq!(bytes.len(), size_of::<Block>());
         // big endian for -2
         assert_eq!(&bytes[4..2 * 4], [255, 255, 255, 254]);
         // big endian for -6
         assert_eq!(&bytes[12 * 4..13 * 4], [255, 255, 255, 250]);
         Ok(())
     }
 }