rust: don't swallow valuable error information

This helps when diagnosing corruption, and is in general good practice. The
information is here, valuable and can be used easily.

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

@@ -1,631 +1,644 b''
 use std::borrow::Cow;
 use std::convert::TryFrom;
 use std::io::Read;
 use std::ops::Deref;
 use std::path::Path;
 
 use flate2::read::ZlibDecoder;
 use micro_timer::timed;
 use sha1::{Digest, Sha1};
 use zstd;
 
 use super::index::Index;
 use super::node::{NodePrefix, NODE_BYTES_LENGTH, NULL_NODE};
 use super::nodemap;
 use super::nodemap::{NodeMap, NodeMapError};
 use super::nodemap_docket::NodeMapDocket;
 use super::patch;
 use crate::errors::HgError;
 use crate::revlog::Revision;
 use crate::vfs::Vfs;
 use crate::{Node, NULL_REVISION};
 
 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(_) => RevlogError::corrupted(),
+            NodeMapError::RevisionNotInIndex(rev) => RevlogError::corrupted(
+                format!("nodemap point to revision {} not in index", rev),
+            ),
         }
     }
 }
 
-fn corrupted() -> HgError {
-    HgError::corrupted("corrupted revlog")
+fn corrupted<S: AsRef<str>>(context: S) -> HgError {
+    HgError::corrupted(format!("corrupted revlog, {}", context.as_ref()))
 }
 
 impl RevlogError {
-    fn corrupted() -> Self {
-        RevlogError::Other(corrupted())
+    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.
     #[timed]
     pub fn open(
         store_vfs: &Vfs,
         index_path: impl AsRef<Path>,
         data_path: Option<&Path>,
         use_nodemap: bool,
     ) -> 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() {
             None
         } else if !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,
                     )
                 },
             )
         };
 
         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
     #[timed]
     pub fn rev_from_node(
         &self,
         node: NodePrefix,
     ) -> Result<Revision, RevlogError> {
         if node.is_prefix_of(&NULL_NODE) {
             return Ok(NULL_REVISION);
         }
 
         if let Some(nodemap) = &self.nodemap {
             return nodemap
                 .find_bin(&self.index, node)?
                 .ok_or(RevlogError::InvalidRevision);
         }
 
         // Fallback to linear scan when a persistent nodemap is not present.
         // This happens when the persistent-nodemap experimental feature is not
         // enabled, or for small revlogs.
         //
         // TODO: consider building a non-persistent nodemap in memory to
         // optimize these cases.
         let mut found_by_prefix = None;
         for rev in (0..self.len() as Revision).rev() {
             let index_entry =
                 self.index.get_entry(rev).ok_or(HgError::corrupted(
                     "revlog references a revision not in the index",
                 ))?;
             if node == *index_entry.hash() {
                 return Ok(rev);
             }
             if node.is_prefix_of(index_entry.hash()) {
                 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.
     #[timed]
     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.
     #[timed]
     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(ref 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> {
-        return self.get_entry(rev).map_err(|_| corrupted());
+        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<'a> {
     revlog: &'a Revlog,
     rev: Revision,
     bytes: &'a [u8],
     compressed_len: u32,
     uncompressed_len: i32,
     base_rev_or_base_of_delta_chain: Option<Revision>,
     p1: Revision,
     p2: Revision,
     flags: u16,
     hash: Node,
 }
 
 impl<'a> RevlogEntry<'a> {
     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>, 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>, 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<'a, [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<'a, [u8]>,
     ) -> Result<Cow<'a, [u8]>, HgError> {
         if self.revlog.check_hash(
             self.p1,
             self.p2,
             self.hash.as_bytes(),
             &data,
         ) {
             Ok(data)
         } else {
-            Err(corrupted())
+            Err(corrupted(format!(
+                "hash check failed for revision {}",
+                self.rev
+            )))
         }
     }
 
     pub fn data(&self) -> Result<Cow<'a, [u8]>, HgError> {
         let data = self.rawdata()?;
         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<'a, [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_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(|_| corrupted())?;
+            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(|_| corrupted())?;
+            decoder
+                .read_exact(&mut buf)
+                .map_err(|e| corrupted(e.to_string()))?;
             Ok(buf)
         }
     }
 
     fn uncompressed_zstd_data(&self) -> Result<Vec<u8>, HgError> {
         if self.is_delta() {
             let mut buf = Vec::with_capacity(self.compressed_len as usize);
             zstd::stream::copy_decode(self.bytes, &mut buf)
-                .map_err(|_| corrupted())?;
+                .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];
             let len = zstd::block::decompress_to_buffer(self.bytes, &mut buf)
-                .map_err(|_| corrupted())?;
+                .map_err(|e| corrupted(e.to_string()))?;
             if len != self.uncompressed_len as usize {
-                Err(corrupted())
+                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));
     }
 
     #[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);
     }
 }

rust/hg-cpython/src/revlog.rs

@@ -1,512 +1,515 b''
 // revlog.rs
 //
 // Copyright 2019-2020 Georges Racinet <[e-mail]>
 //
 // This software may be used and distributed according to the terms of the
 // GNU General Public License version 2 or any later version.
 
 use crate::{
     cindex,
     utils::{node_from_py_bytes, node_from_py_object},
 };
 use cpython::{
     buffer::{Element, PyBuffer},
     exc::{IndexError, ValueError},
     ObjectProtocol, PyBytes, PyClone, PyDict, PyErr, PyInt, PyModule,
     PyObject, PyResult, PyString, PyTuple, Python, PythonObject, ToPyObject,
 };
 use hg::{
     nodemap::{Block, NodeMapError, NodeTree},
     revlog::{nodemap::NodeMap, NodePrefix, RevlogIndex},
     Revision,
 };
 use std::cell::RefCell;
 
 /// Return a Struct implementing the Graph trait
 pub(crate) fn pyindex_to_graph(
     py: Python,
     index: PyObject,
 ) -> PyResult<cindex::Index> {
     match index.extract::<MixedIndex>(py) {
         Ok(midx) => Ok(midx.clone_cindex(py)),
         Err(_) => cindex::Index::new(py, index),
     }
 }
 
 py_class!(pub class MixedIndex |py| {
     data cindex: RefCell<cindex::Index>;
     data nt: RefCell<Option<NodeTree>>;
     data docket: RefCell<Option<PyObject>>;
     // Holds a reference to the mmap'ed persistent nodemap data
     data mmap: RefCell<Option<PyBuffer>>;
 
     def __new__(_cls, cindex: PyObject) -> PyResult<MixedIndex> {
         Self::new(py, cindex)
     }
 
     /// Compatibility layer used for Python consumers needing access to the C index
     ///
     /// Only use case so far is `scmutil.shortesthexnodeidprefix`,
     /// that may need to build a custom `nodetree`, based on a specified revset.
     /// With a Rust implementation of the nodemap, we will be able to get rid of
     /// this, by exposing our own standalone nodemap class,
     /// ready to accept `MixedIndex`.
     def get_cindex(&self) -> PyResult<PyObject> {
         Ok(self.cindex(py).borrow().inner().clone_ref(py))
     }
 
     // Index API involving nodemap, as defined in mercurial/pure/parsers.py
 
     /// Return Revision if found, raises a bare `error.RevlogError`
     /// in case of ambiguity, same as C version does
     def get_rev(&self, node: PyBytes) -> PyResult<Option<Revision>> {
         let opt = self.get_nodetree(py)?.borrow();
         let nt = opt.as_ref().unwrap();
         let idx = &*self.cindex(py).borrow();
         let node = node_from_py_bytes(py, &node)?;
         nt.find_bin(idx, node.into()).map_err(|e| nodemap_error(py, e))
     }
 
     /// same as `get_rev()` but raises a bare `error.RevlogError` if node
     /// is not found.
     ///
     /// No need to repeat `node` in the exception, `mercurial/revlog.py`
     /// will catch and rewrap with it
     def rev(&self, node: PyBytes) -> PyResult<Revision> {
         self.get_rev(py, node)?.ok_or_else(|| revlog_error(py))
     }
 
     /// return True if the node exist in the index
     def has_node(&self, node: PyBytes) -> PyResult<bool> {
         self.get_rev(py, node).map(|opt| opt.is_some())
     }
 
     /// find length of shortest hex nodeid of a binary ID
     def shortest(&self, node: PyBytes) -> PyResult<usize> {
         let opt = self.get_nodetree(py)?.borrow();
         let nt = opt.as_ref().unwrap();
         let idx = &*self.cindex(py).borrow();
         match nt.unique_prefix_len_node(idx, &node_from_py_bytes(py, &node)?)
         {
             Ok(Some(l)) => Ok(l),
             Ok(None) => Err(revlog_error(py)),
             Err(e) => Err(nodemap_error(py, e)),
         }
     }
 
     def partialmatch(&self, node: PyObject) -> PyResult<Option<PyBytes>> {
         let opt = self.get_nodetree(py)?.borrow();
         let nt = opt.as_ref().unwrap();
         let idx = &*self.cindex(py).borrow();
 
         let node_as_string = if cfg!(feature = "python3-sys") {
             node.cast_as::<PyString>(py)?.to_string(py)?.to_string()
         }
         else {
             let node = node.extract::<PyBytes>(py)?;
             String::from_utf8_lossy(node.data(py)).to_string()
         };
 
-        let prefix = NodePrefix::from_hex(&node_as_string).map_err(|_| PyErr::new::<ValueError, _>(py, "Invalid node or prefix"))?;
+        let prefix = NodePrefix::from_hex(&node_as_string)
+            .map_err(|_| PyErr::new::<ValueError, _>(
+                py, format!("Invalid node or prefix '{}'", node_as_string))
+            )?;
 
         nt.find_bin(idx, prefix)
             // TODO make an inner API returning the node directly
             .map(|opt| opt.map(
                 |rev| PyBytes::new(py, idx.node(rev).unwrap().as_bytes())))
             .map_err(|e| nodemap_error(py, e))
 
     }
 
     /// append an index entry
     def append(&self, tup: PyTuple) -> PyResult<PyObject> {
         if tup.len(py) < 8 {
             // this is better than the panic promised by tup.get_item()
             return Err(
                 PyErr::new::<IndexError, _>(py, "tuple index out of range"))
         }
         let node_bytes = tup.get_item(py, 7).extract(py)?;
         let node = node_from_py_object(py, &node_bytes)?;
 
         let mut idx = self.cindex(py).borrow_mut();
         let rev = idx.len() as Revision;
 
         idx.append(py, tup)?;
         self.get_nodetree(py)?.borrow_mut().as_mut().unwrap()
             .insert(&*idx, &node, rev)
             .map_err(|e| nodemap_error(py, e))?;
         Ok(py.None())
     }
 
     def __delitem__(&self, key: PyObject) -> PyResult<()> {
         // __delitem__ is both for `del idx[r]` and `del idx[r1:r2]`
         self.cindex(py).borrow().inner().del_item(py, key)?;
         let mut opt = self.get_nodetree(py)?.borrow_mut();
         let mut nt = opt.as_mut().unwrap();
         nt.invalidate_all();
         self.fill_nodemap(py, &mut nt)?;
         Ok(())
     }
 
     //
     // Reforwarded C index API
     //
 
     // index_methods (tp_methods). Same ordering as in revlog.c
 
     /// return the gca set of the given revs
     def ancestors(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "ancestors", args, kw)
     }
 
     /// return the heads of the common ancestors of the given revs
     def commonancestorsheads(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "commonancestorsheads", args, kw)
     }
 
     /// Clear the index caches and inner py_class data.
     /// It is Python's responsibility to call `update_nodemap_data` again.
     def clearcaches(&self, *args, **kw) -> PyResult<PyObject> {
         self.nt(py).borrow_mut().take();
         self.docket(py).borrow_mut().take();
         self.mmap(py).borrow_mut().take();
         self.call_cindex(py, "clearcaches", args, kw)
     }
 
     /// return the raw binary string representing a revision
     def entry_binary(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "entry_binary", args, kw)
     }
 
     /// return a binary packed version of the header
     def pack_header(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "pack_header", args, kw)
     }
 
     /// get an index entry
     def get(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "get", args, kw)
     }
 
     /// compute phases
     def computephasesmapsets(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "computephasesmapsets", args, kw)
     }
 
     /// reachableroots
     def reachableroots2(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "reachableroots2", args, kw)
     }
 
     /// get head revisions
     def headrevs(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "headrevs", args, kw)
     }
 
     /// get filtered head revisions
     def headrevsfiltered(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "headrevsfiltered", args, kw)
     }
 
     /// True if the object is a snapshot
     def issnapshot(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "issnapshot", args, kw)
     }
 
     /// Gather snapshot data in a cache dict
     def findsnapshots(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "findsnapshots", args, kw)
     }
 
     /// determine revisions with deltas to reconstruct fulltext
     def deltachain(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "deltachain", args, kw)
     }
 
     /// slice planned chunk read to reach a density threshold
     def slicechunktodensity(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "slicechunktodensity", args, kw)
     }
 
     /// stats for the index
     def stats(&self, *args, **kw) -> PyResult<PyObject> {
         self.call_cindex(py, "stats", args, kw)
     }
 
     // index_sequence_methods and index_mapping_methods.
     //
     // Since we call back through the high level Python API,
     // there's no point making a distinction between index_get
     // and index_getitem.
 
     def __len__(&self) -> PyResult<usize> {
         self.cindex(py).borrow().inner().len(py)
     }
 
     def __getitem__(&self, key: PyObject) -> PyResult<PyObject> {
         // this conversion seems needless, but that's actually because
         // `index_getitem` does not handle conversion from PyLong,
         // which expressions such as [e for e in index] internally use.
         // Note that we don't seem to have a direct way to call
         // PySequence_GetItem (does the job), which would possibly be better
         // for performance
         let key = match key.extract::<Revision>(py) {
             Ok(rev) => rev.to_py_object(py).into_object(),
             Err(_) => key,
         };
         self.cindex(py).borrow().inner().get_item(py, key)
     }
 
     def __setitem__(&self, key: PyObject, value: PyObject) -> PyResult<()> {
         self.cindex(py).borrow().inner().set_item(py, key, value)
     }
 
     def __contains__(&self, item: PyObject) -> PyResult<bool> {
         // ObjectProtocol does not seem to provide contains(), so
         // this is an equivalent implementation of the index_contains()
         // defined in revlog.c
         let cindex = self.cindex(py).borrow();
         match item.extract::<Revision>(py) {
             Ok(rev) => {
                 Ok(rev >= -1 && rev < cindex.inner().len(py)? as Revision)
             }
             Err(_) => {
                 cindex.inner().call_method(
                     py,
                     "has_node",
                     PyTuple::new(py, &[item]),
                     None)?
                 .extract(py)
             }
         }
     }
 
     def nodemap_data_all(&self) -> PyResult<PyBytes> {
         self.inner_nodemap_data_all(py)
     }
 
     def nodemap_data_incremental(&self) -> PyResult<PyObject> {
         self.inner_nodemap_data_incremental(py)
     }
     def update_nodemap_data(
         &self,
         docket: PyObject,
         nm_data: PyObject
     ) -> PyResult<PyObject> {
         self.inner_update_nodemap_data(py, docket, nm_data)
     }
 
     @property
     def entry_size(&self) -> PyResult<PyInt> {
         self.cindex(py).borrow().inner().getattr(py, "entry_size")?.extract::<PyInt>(py)
     }
 
     @property
     def rust_ext_compat(&self) -> PyResult<PyInt> {
         self.cindex(py).borrow().inner().getattr(py, "rust_ext_compat")?.extract::<PyInt>(py)
     }
 
 });
 
 impl MixedIndex {
     fn new(py: Python, cindex: PyObject) -> PyResult<MixedIndex> {
         Self::create_instance(
             py,
             RefCell::new(cindex::Index::new(py, cindex)?),
             RefCell::new(None),
             RefCell::new(None),
             RefCell::new(None),
         )
     }
 
     /// This is scaffolding at this point, but it could also become
     /// a way to start a persistent nodemap or perform a
     /// vacuum / repack operation
     fn fill_nodemap(
         &self,
         py: Python,
         nt: &mut NodeTree,
     ) -> PyResult<PyObject> {
         let index = self.cindex(py).borrow();
         for r in 0..index.len() {
             let rev = r as Revision;
             // in this case node() won't ever return None
             nt.insert(&*index, index.node(rev).unwrap(), rev)
                 .map_err(|e| nodemap_error(py, e))?
         }
         Ok(py.None())
     }
 
     fn get_nodetree<'a>(
         &'a self,
         py: Python<'a>,
     ) -> PyResult<&'a RefCell<Option<NodeTree>>> {
         if self.nt(py).borrow().is_none() {
             let readonly = Box::new(Vec::new());
             let mut nt = NodeTree::load_bytes(readonly, 0);
             self.fill_nodemap(py, &mut nt)?;
             self.nt(py).borrow_mut().replace(nt);
         }
         Ok(self.nt(py))
     }
 
     /// forward a method call to the underlying C index
     fn call_cindex(
         &self,
         py: Python,
         name: &str,
         args: &PyTuple,
         kwargs: Option<&PyDict>,
     ) -> PyResult<PyObject> {
         self.cindex(py)
             .borrow()
             .inner()
             .call_method(py, name, args, kwargs)
     }
 
     pub fn clone_cindex(&self, py: Python) -> cindex::Index {
         self.cindex(py).borrow().clone_ref(py)
     }
 
     /// Returns the full nodemap bytes to be written as-is to disk
     fn inner_nodemap_data_all(&self, py: Python) -> PyResult<PyBytes> {
         let nodemap = self.get_nodetree(py)?.borrow_mut().take().unwrap();
         let (readonly, bytes) = nodemap.into_readonly_and_added_bytes();
 
         // If there's anything readonly, we need to build the data again from
         // scratch
         let bytes = if readonly.len() > 0 {
             let mut nt = NodeTree::load_bytes(Box::new(vec![]), 0);
             self.fill_nodemap(py, &mut nt)?;
 
             let (readonly, bytes) = nt.into_readonly_and_added_bytes();
             assert_eq!(readonly.len(), 0);
 
             bytes
         } else {
             bytes
         };
 
         let bytes = PyBytes::new(py, &bytes);
         Ok(bytes)
     }
 
     /// Returns the last saved docket along with the size of any changed data
     /// (in number of blocks), and said data as bytes.
     fn inner_nodemap_data_incremental(
         &self,
         py: Python,
     ) -> PyResult<PyObject> {
         let docket = self.docket(py).borrow();
         let docket = match docket.as_ref() {
             Some(d) => d,
             None => return Ok(py.None()),
         };
 
         let node_tree = self.get_nodetree(py)?.borrow_mut().take().unwrap();
         let masked_blocks = node_tree.masked_readonly_blocks();
         let (_, data) = node_tree.into_readonly_and_added_bytes();
         let changed = masked_blocks * std::mem::size_of::<Block>();
 
         Ok((docket, changed, PyBytes::new(py, &data))
             .to_py_object(py)
             .into_object())
     }
 
     /// Update the nodemap from the new (mmaped) data.
     /// The docket is kept as a reference for later incremental calls.
     fn inner_update_nodemap_data(
         &self,
         py: Python,
         docket: PyObject,
         nm_data: PyObject,
     ) -> PyResult<PyObject> {
         let buf = PyBuffer::get(py, &nm_data)?;
         let len = buf.item_count();
 
         // Build a slice from the mmap'ed buffer data
         let cbuf = buf.buf_ptr();
         let bytes = if std::mem::size_of::<u8>() == buf.item_size()
             && buf.is_c_contiguous()
             && u8::is_compatible_format(buf.format())
         {
             unsafe { std::slice::from_raw_parts(cbuf as *const u8, len) }
         } else {
             return Err(PyErr::new::<ValueError, _>(
                 py,
                 "Nodemap data buffer has an invalid memory representation"
                     .to_string(),
             ));
         };
 
         // Keep a reference to the mmap'ed buffer, otherwise we get a dangling
         // pointer.
         self.mmap(py).borrow_mut().replace(buf);
 
         let mut nt = NodeTree::load_bytes(Box::new(bytes), len);
 
         let data_tip =
             docket.getattr(py, "tip_rev")?.extract::<Revision>(py)?;
         self.docket(py).borrow_mut().replace(docket.clone_ref(py));
         let idx = self.cindex(py).borrow();
         let current_tip = idx.len();
 
         for r in (data_tip + 1)..current_tip as Revision {
             let rev = r as Revision;
             // in this case node() won't ever return None
             nt.insert(&*idx, idx.node(rev).unwrap(), rev)
                 .map_err(|e| nodemap_error(py, e))?
         }
 
         *self.nt(py).borrow_mut() = Some(nt);
 
         Ok(py.None())
     }
 }
 
 fn revlog_error(py: Python) -> PyErr {
     match py
         .import("mercurial.error")
         .and_then(|m| m.get(py, "RevlogError"))
     {
         Err(e) => e,
         Ok(cls) => PyErr::from_instance(
             py,
             cls.call(py, (py.None(),), None).ok().into_py_object(py),
         ),
     }
 }
 
 fn rev_not_in_index(py: Python, rev: Revision) -> PyErr {
     PyErr::new::<ValueError, _>(
         py,
         format!(
             "Inconsistency: Revision {} found in nodemap \
              is not in revlog index",
             rev
         ),
     )
 }
 
 /// Standard treatment of NodeMapError
 fn nodemap_error(py: Python, err: NodeMapError) -> PyErr {
     match err {
         NodeMapError::MultipleResults => revlog_error(py),
         NodeMapError::RevisionNotInIndex(r) => rev_not_in_index(py, r),
     }
 }
 
 /// Create the module, with __package__ given from parent
 pub fn init_module(py: Python, package: &str) -> PyResult<PyModule> {
     let dotted_name = &format!("{}.revlog", package);
     let m = PyModule::new(py, dotted_name)?;
     m.add(py, "__package__", package)?;
     m.add(py, "__doc__", "RevLog - Rust implementations")?;
 
     m.add_class::<MixedIndex>(py)?;
 
     let sys = PyModule::import(py, "sys")?;
     let sys_modules: PyDict = sys.get(py, "modules")?.extract(py)?;
     sys_modules.set_item(py, dotted_name, &m)?;
 
     Ok(m)
 }

rust/rhg/src/commands/status.rs

@@ -1,544 +1,547 b''
 // status.rs
 //
 // Copyright 2020, Georges Racinet <[e-mail]>
 //
 // This software may be used and distributed according to the terms of the
 // GNU General Public License version 2 or any later version.
 
 use crate::error::CommandError;
 use crate::ui::Ui;
 use crate::utils::path_utils::RelativizePaths;
 use clap::{Arg, SubCommand};
 use format_bytes::format_bytes;
 use hg;
 use hg::config::Config;
 use hg::dirstate::has_exec_bit;
 use hg::dirstate::status::StatusPath;
 use hg::dirstate::TruncatedTimestamp;
 use hg::errors::{HgError, IoResultExt};
 use hg::lock::LockError;
 use hg::manifest::Manifest;
 use hg::matchers::AlwaysMatcher;
 use hg::repo::Repo;
 use hg::utils::files::get_bytes_from_os_string;
 use hg::utils::files::get_bytes_from_path;
 use hg::utils::files::get_path_from_bytes;
 use hg::utils::hg_path::{hg_path_to_path_buf, HgPath};
 use hg::DirstateStatus;
 use hg::PatternFileWarning;
 use hg::StatusError;
 use hg::StatusOptions;
 use log::info;
 use std::io;
 use std::path::PathBuf;
 
 pub const HELP_TEXT: &str = "
 Show changed files in the working directory
 
 This is a pure Rust version of `hg status`.
 
 Some options might be missing, check the list below.
 ";
 
 pub fn args() -> clap::App<'static, 'static> {
     SubCommand::with_name("status")
         .alias("st")
         .about(HELP_TEXT)
         .arg(
             Arg::with_name("all")
                 .help("show status of all files")
                 .short("-A")
                 .long("--all"),
         )
         .arg(
             Arg::with_name("modified")
                 .help("show only modified files")
                 .short("-m")
                 .long("--modified"),
         )
         .arg(
             Arg::with_name("added")
                 .help("show only added files")
                 .short("-a")
                 .long("--added"),
         )
         .arg(
             Arg::with_name("removed")
                 .help("show only removed files")
                 .short("-r")
                 .long("--removed"),
         )
         .arg(
             Arg::with_name("clean")
                 .help("show only clean files")
                 .short("-c")
                 .long("--clean"),
         )
         .arg(
             Arg::with_name("deleted")
                 .help("show only deleted files")
                 .short("-d")
                 .long("--deleted"),
         )
         .arg(
             Arg::with_name("unknown")
                 .help("show only unknown (not tracked) files")
                 .short("-u")
                 .long("--unknown"),
         )
         .arg(
             Arg::with_name("ignored")
                 .help("show only ignored files")
                 .short("-i")
                 .long("--ignored"),
         )
         .arg(
             Arg::with_name("copies")
                 .help("show source of copied files (DEFAULT: ui.statuscopies)")
                 .short("-C")
                 .long("--copies"),
         )
         .arg(
             Arg::with_name("no-status")
                 .help("hide status prefix")
                 .short("-n")
                 .long("--no-status"),
         )
 }
 
 /// Pure data type allowing the caller to specify file states to display
 #[derive(Copy, Clone, Debug)]
 pub struct DisplayStates {
     pub modified: bool,
     pub added: bool,
     pub removed: bool,
     pub clean: bool,
     pub deleted: bool,
     pub unknown: bool,
     pub ignored: bool,
 }
 
 pub const DEFAULT_DISPLAY_STATES: DisplayStates = DisplayStates {
     modified: true,
     added: true,
     removed: true,
     clean: false,
     deleted: true,
     unknown: true,
     ignored: false,
 };
 
 pub const ALL_DISPLAY_STATES: DisplayStates = DisplayStates {
     modified: true,
     added: true,
     removed: true,
     clean: true,
     deleted: true,
     unknown: true,
     ignored: true,
 };
 
 impl DisplayStates {
     pub fn is_empty(&self) -> bool {
         !(self.modified
             || self.added
             || self.removed
             || self.clean
             || self.deleted
             || self.unknown
             || self.ignored)
     }
 }
 
 pub fn run(invocation: &crate::CliInvocation) -> Result<(), CommandError> {
     // TODO: lift these limitations
     if invocation.config.get_bool(b"ui", b"tweakdefaults")? {
         return Err(CommandError::unsupported(
             "ui.tweakdefaults is not yet supported with rhg status",
         ));
     }
     if invocation.config.get_bool(b"ui", b"statuscopies")? {
         return Err(CommandError::unsupported(
             "ui.statuscopies is not yet supported with rhg status",
         ));
     }
     if invocation
         .config
         .get(b"commands", b"status.terse")
         .is_some()
     {
         return Err(CommandError::unsupported(
             "status.terse is not yet supported with rhg status",
         ));
     }
 
     let ui = invocation.ui;
     let config = invocation.config;
     let args = invocation.subcommand_args;
 
     let verbose = !ui.plain(None)
         && !args.is_present("print0")
         && (config.get_bool(b"ui", b"verbose")?
             || config.get_bool(b"commands", b"status.verbose")?);
     if verbose {
         return Err(CommandError::unsupported(
             "verbose status is not supported yet",
         ));
     }
 
     let all = args.is_present("all");
     let display_states = if all {
         // TODO when implementing `--quiet`: it excludes clean files
         // from `--all`
         ALL_DISPLAY_STATES
     } else {
         let requested = DisplayStates {
             modified: args.is_present("modified"),
             added: args.is_present("added"),
             removed: args.is_present("removed"),
             clean: args.is_present("clean"),
             deleted: args.is_present("deleted"),
             unknown: args.is_present("unknown"),
             ignored: args.is_present("ignored"),
         };
         if requested.is_empty() {
             DEFAULT_DISPLAY_STATES
         } else {
             requested
         }
     };
     let no_status = args.is_present("no-status");
     let list_copies = all
         || args.is_present("copies")
         || config.get_bool(b"ui", b"statuscopies")?;
 
     let repo = invocation.repo?;
 
     if repo.has_sparse() || repo.has_narrow() {
         return Err(CommandError::unsupported(
             "rhg status is not supported for sparse checkouts or narrow clones yet"
         ));
     }
 
     let mut dmap = repo.dirstate_map_mut()?;
 
     let options = StatusOptions {
         // we're currently supporting file systems with exec flags only
         // anyway
         check_exec: true,
         list_clean: display_states.clean,
         list_unknown: display_states.unknown,
         list_ignored: display_states.ignored,
         list_copies,
         collect_traversed_dirs: false,
     };
 
     type StatusResult<'a> =
         Result<(DirstateStatus<'a>, Vec<PatternFileWarning>), StatusError>;
 
     let after_status = |res: StatusResult| -> Result<_, CommandError> {
         let (mut ds_status, pattern_warnings) = res?;
         for warning in pattern_warnings {
             match warning {
                 hg::PatternFileWarning::InvalidSyntax(path, syntax) => ui
                     .write_stderr(&format_bytes!(
                         b"{}: ignoring invalid syntax '{}'\n",
                         get_bytes_from_path(path),
                         &*syntax
                     ))?,
                 hg::PatternFileWarning::NoSuchFile(path) => {
                     let path = if let Ok(relative) =
                         path.strip_prefix(repo.working_directory_path())
                     {
                         relative
                     } else {
                         &*path
                     };
                     ui.write_stderr(&format_bytes!(
                         b"skipping unreadable pattern file '{}': \
                           No such file or directory\n",
                         get_bytes_from_path(path),
                     ))?
                 }
             }
         }
 
         for (path, error) in ds_status.bad {
             let error = match error {
                 hg::BadMatch::OsError(code) => {
                     std::io::Error::from_raw_os_error(code).to_string()
                 }
                 hg::BadMatch::BadType(ty) => {
                     format!("unsupported file type (type is {})", ty)
                 }
             };
             ui.write_stderr(&format_bytes!(
                 b"{}: {}\n",
                 path.as_bytes(),
                 error.as_bytes()
             ))?
         }
         if !ds_status.unsure.is_empty() {
             info!(
                 "Files to be rechecked by retrieval from filelog: {:?}",
                 ds_status.unsure.iter().map(|s| &s.path).collect::<Vec<_>>()
             );
         }
         let mut fixup = Vec::new();
         if !ds_status.unsure.is_empty()
             && (display_states.modified || display_states.clean)
         {
             let p1 = repo.dirstate_parents()?.p1;
             let manifest = repo.manifest_for_node(p1).map_err(|e| {
                 CommandError::from((e, &*format!("{:x}", p1.short())))
             })?;
             for to_check in ds_status.unsure {
                 if unsure_is_modified(repo, &manifest, &to_check.path)? {
                     if display_states.modified {
                         ds_status.modified.push(to_check);
                     }
                 } else {
                     if display_states.clean {
                         ds_status.clean.push(to_check.clone());
                     }
                     fixup.push(to_check.path.into_owned())
                 }
             }
         }
         let relative_paths = (!ui.plain(None))
             && config
                 .get_option(b"commands", b"status.relative")?
                 .unwrap_or(config.get_bool(b"ui", b"relative-paths")?);
         let output = DisplayStatusPaths {
             ui,
             no_status,
             relativize: if relative_paths {
                 Some(RelativizePaths::new(repo)?)
             } else {
                 None
             },
         };
         if display_states.modified {
             output.display(b"M ", "status.modified", ds_status.modified)?;
         }
         if display_states.added {
             output.display(b"A ", "status.added", ds_status.added)?;
         }
         if display_states.removed {
             output.display(b"R ", "status.removed", ds_status.removed)?;
         }
         if display_states.deleted {
             output.display(b"! ", "status.deleted", ds_status.deleted)?;
         }
         if display_states.unknown {
             output.display(b"? ", "status.unknown", ds_status.unknown)?;
         }
         if display_states.ignored {
             output.display(b"I ", "status.ignored", ds_status.ignored)?;
         }
         if display_states.clean {
             output.display(b"C ", "status.clean", ds_status.clean)?;
         }
 
         let dirstate_write_needed = ds_status.dirty;
         let filesystem_time_at_status_start =
             ds_status.filesystem_time_at_status_start;
 
         Ok((
             fixup,
             dirstate_write_needed,
             filesystem_time_at_status_start,
         ))
     };
     let (fixup, mut dirstate_write_needed, filesystem_time_at_status_start) =
         dmap.with_status(
             &AlwaysMatcher,
             repo.working_directory_path().to_owned(),
             ignore_files(repo, config),
             options,
             after_status,
         )?;
 
     if (fixup.is_empty() || filesystem_time_at_status_start.is_none())
         && !dirstate_write_needed
     {
         // Nothing to update
         return Ok(());
     }
 
     // Update the dirstate on disk if we can
     let with_lock_result =
         repo.try_with_wlock_no_wait(|| -> Result<(), CommandError> {
             if let Some(mtime_boundary) = filesystem_time_at_status_start {
                 for hg_path in fixup {
                     use std::os::unix::fs::MetadataExt;
                     let fs_path = hg_path_to_path_buf(&hg_path)
                         .expect("HgPath conversion");
                     // Specifically do not reuse `fs_metadata` from
                     // `unsure_is_clean` which was needed before reading
                     // contents. Here we access metadata again after reading
                     // content, in case it changed in the meantime.
                     let fs_metadata = repo
                         .working_directory_vfs()
                         .symlink_metadata(&fs_path)?;
                     if let Some(mtime) =
                         TruncatedTimestamp::for_reliable_mtime_of(
                             &fs_metadata,
                             &mtime_boundary,
                         )
                         .when_reading_file(&fs_path)?
                     {
                         let mode = fs_metadata.mode();
                         let size = fs_metadata.len();
                         dmap.set_clean(&hg_path, mode, size as u32, mtime)?;
                         dirstate_write_needed = true
                     }
                 }
             }
             drop(dmap); // Avoid "already mutably borrowed" RefCell panics
             if dirstate_write_needed {
                 repo.write_dirstate()?
             }
             Ok(())
         });
     match with_lock_result {
         Ok(closure_result) => closure_result?,
         Err(LockError::AlreadyHeld) => {
             // Not updating the dirstate is not ideal but not critical:
             // don’t keep our caller waiting until some other Mercurial
             // process releases the lock.
         }
         Err(LockError::Other(HgError::IoError { error, .. }))
             if error.kind() == io::ErrorKind::PermissionDenied =>
         {
             // `hg status` on a read-only repository is fine
         }
         Err(LockError::Other(error)) => {
             // Report other I/O errors
             Err(error)?
         }
     }
     Ok(())
 }
 
 fn ignore_files(repo: &Repo, config: &Config) -> Vec<PathBuf> {
     let mut ignore_files = Vec::new();
     let repo_ignore = repo.working_directory_vfs().join(".hgignore");
     if repo_ignore.exists() {
         ignore_files.push(repo_ignore)
     }
     for (key, value) in config.iter_section(b"ui") {
         if key == b"ignore" || key.starts_with(b"ignore.") {
             let path = get_path_from_bytes(value);
             // TODO: expand "~/" and environment variable here, like Python
             // does with `os.path.expanduser` and `os.path.expandvars`
 
             let joined = repo.working_directory_path().join(path);
             ignore_files.push(joined);
         }
     }
     ignore_files
 }
 
 struct DisplayStatusPaths<'a> {
     ui: &'a Ui,
     no_status: bool,
     relativize: Option<RelativizePaths>,
 }
 
 impl DisplayStatusPaths<'_> {
     // Probably more elegant to use a Deref or Borrow trait rather than
     // harcode HgPathBuf, but probably not really useful at this point
     fn display(
         &self,
         status_prefix: &[u8],
         label: &'static str,
         mut paths: Vec<StatusPath<'_>>,
     ) -> Result<(), CommandError> {
         paths.sort_unstable();
         // TODO: get the stdout lock once for the whole loop
         // instead of in each write
         for StatusPath { path, copy_source } in paths {
             let relative;
             let path = if let Some(relativize) = &self.relativize {
                 relative = relativize.relativize(&path);
                 &*relative
             } else {
                 path.as_bytes()
             };
             // TODO: Add a way to use `write_bytes!` instead of `format_bytes!`
             // in order to stream to stdout instead of allocating an
             // itermediate `Vec<u8>`.
             if !self.no_status {
                 self.ui.write_stdout_labelled(status_prefix, label)?
             }
             self.ui
                 .write_stdout_labelled(&format_bytes!(b"{}\n", path), label)?;
             if let Some(source) = copy_source {
                 let label = "status.copied";
                 self.ui.write_stdout_labelled(
                     &format_bytes!(b"  {}\n", source.as_bytes()),
                     label,
                 )?
             }
         }
         Ok(())
     }
 }
 
 /// Check if a file is modified by comparing actual repo store and file system.
 ///
 /// This meant to be used for those that the dirstate cannot resolve, due
 /// to time resolution limits.
 fn unsure_is_modified(
     repo: &Repo,
     manifest: &Manifest,
     hg_path: &HgPath,
 ) -> Result<bool, HgError> {
     let vfs = repo.working_directory_vfs();
     let fs_path = hg_path_to_path_buf(hg_path).expect("HgPath conversion");
     let fs_metadata = vfs.symlink_metadata(&fs_path)?;
     let is_symlink = fs_metadata.file_type().is_symlink();
     // TODO: Also account for `FALLBACK_SYMLINK` and `FALLBACK_EXEC` from the
     // dirstate
     let fs_flags = if is_symlink {
         Some(b'l')
     } else if has_exec_bit(&fs_metadata) {
         Some(b'x')
     } else {
         None
     };
 
     let entry = manifest
         .find_by_path(hg_path)?
         .expect("ambgious file not in p1");
     if entry.flags != fs_flags {
         return Ok(true);
     }
     let filelog = repo.filelog(hg_path)?;
     let fs_len = fs_metadata.len();
-    let filelog_entry =
-        filelog.entry_for_node(entry.node_id()?).map_err(|_| {
-            HgError::corrupted("filelog missing node from manifest")
+    let file_node = entry.node_id()?;
+    let filelog_entry = filelog.entry_for_node(file_node).map_err(|_| {
+        HgError::corrupted(format!(
+            "filelog missing node {:?} from manifest",
+            file_node
+        ))
         })?;
     if filelog_entry.file_data_len_not_equal_to(fs_len) {
         // No need to read file contents:
         // it cannot be equal if it has a different length.
         return Ok(true);
     }
 
     let p1_filelog_data = filelog_entry.data()?;
     let p1_contents = p1_filelog_data.file_data()?;
     if p1_contents.len() as u64 != fs_len {
         // No need to read file contents:
         // it cannot be equal if it has a different length.
         return Ok(true);
     }
 
     let fs_contents = if is_symlink {
         get_bytes_from_os_string(vfs.read_link(fs_path)?.into_os_string())
     } else {
         vfs.read(fs_path)?
     };
     Ok(p1_contents != &*fs_contents)
 }