rust: Remove hex parsing from the nodemap

Separating concerns simplifies error types.

Differential Revision: https://phab.mercurial-scm.org/D9864

rust/hg-core/examples/nodemap/main.rs

@@ -1,146 +1,146
 // 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 clap::*;
 use hg::revlog::node::*;
 use hg::revlog::nodemap::*;
 use hg::revlog::*;
 use memmap::MmapOptions;
 use rand::Rng;
 use std::fs::File;
 use std::io;
 use std::io::Write;
 use std::path::{Path, PathBuf};
 use std::str::FromStr;
 use std::time::Instant;
 
 mod index;
 use index::Index;
 
 fn mmap_index(repo_path: &Path) -> Index {
     let mut path = PathBuf::from(repo_path);
     path.extend([".hg", "store", "00changelog.i"].iter());
     Index::load_mmap(path)
 }
 
 fn mmap_nodemap(path: &Path) -> NodeTree {
     let file = File::open(path).unwrap();
     let mmap = unsafe { MmapOptions::new().map(&file).unwrap() };
     let len = mmap.len();
     NodeTree::load_bytes(Box::new(mmap), len)
 }
 
 /// Scan the whole index and create the corresponding nodemap file at `path`
 fn create(index: &Index, path: &Path) -> io::Result<()> {
     let mut file = File::create(path)?;
     let start = Instant::now();
     let mut nm = NodeTree::default();
     for rev in 0..index.len() {
         let rev = rev as Revision;
         nm.insert(index, index.node(rev).unwrap(), rev).unwrap();
     }
     eprintln!("Nodemap constructed in RAM in {:?}", start.elapsed());
     file.write(&nm.into_readonly_and_added_bytes().1)?;
     eprintln!("Nodemap written to disk");
     Ok(())
 }
 
 fn query(index: &Index, nm: &NodeTree, prefix: &str) {
     let start = Instant::now();
-    let res = nm.find_hex(index, prefix);
+    let res = NodePrefix::from_hex(prefix).map(|p| nm.find_bin(index, p));
     println!("Result found in {:?}: {:?}", start.elapsed(), res);
 }
 
 fn bench(index: &Index, nm: &NodeTree, queries: usize) {
     let len = index.len() as u32;
     let mut rng = rand::thread_rng();
     let nodes: Vec<Node> = (0..queries)
         .map(|_| {
             index
                 .node((rng.gen::<u32>() % len) as Revision)
                 .unwrap()
                 .clone()
         })
         .collect();
     if queries < 10 {
         let nodes_hex: Vec<String> =
             nodes.iter().map(|n| format!("{:x}", n)).collect();
         println!("Nodes: {:?}", nodes_hex);
     }
     let mut last: Option<Revision> = None;
     let start = Instant::now();
     for node in nodes.iter() {
         last = nm.find_bin(index, node.into()).unwrap();
     }
     let elapsed = start.elapsed();
     println!(
         "Did {} queries in {:?} (mean {:?}), last was {:x} with result {:?}",
         queries,
         elapsed,
         elapsed / (queries as u32),
         nodes.last().unwrap(),
         last
     );
 }
 
 fn main() {
     let matches = App::new("Nodemap pure Rust example")
         .arg(
             Arg::with_name("REPOSITORY")
                 .help("Path to the repository, always necessary for its index")
                 .required(true),
         )
         .arg(
             Arg::with_name("NODEMAP_FILE")
                 .help("Path to the nodemap file, independent of REPOSITORY")
                 .required(true),
         )
         .subcommand(
             SubCommand::with_name("create")
                 .about("Create NODEMAP_FILE by scanning repository index"),
         )
         .subcommand(
             SubCommand::with_name("query")
                 .about("Query NODEMAP_FILE for PREFIX")
                 .arg(Arg::with_name("PREFIX").required(true)),
         )
         .subcommand(
             SubCommand::with_name("bench")
                 .about(
                     "Perform #QUERIES random successful queries on NODEMAP_FILE")
                 .arg(Arg::with_name("QUERIES").required(true)),
         )
         .get_matches();
 
     let repo = matches.value_of("REPOSITORY").unwrap();
     let nm_path = matches.value_of("NODEMAP_FILE").unwrap();
 
     let index = mmap_index(&Path::new(repo));
 
     if let Some(_) = matches.subcommand_matches("create") {
         println!("Creating nodemap file {} for repository {}", nm_path, repo);
         create(&index, &Path::new(nm_path)).unwrap();
         return;
     }
 
     let nm = mmap_nodemap(&Path::new(nm_path));
     if let Some(matches) = matches.subcommand_matches("query") {
         let prefix = matches.value_of("PREFIX").unwrap();
         println!(
             "Querying {} in nodemap file {} of repository {}",
             prefix, nm_path, repo
         );
         query(&index, &nm, prefix);
     }
     if let Some(matches) = matches.subcommand_matches("bench") {
         let queries =
             usize::from_str(matches.value_of("QUERIES").unwrap()).unwrap();
         println!(
             "Doing {} random queries in nodemap file {} of repository {}",
             queries, nm_path, repo
         );
         bench(&index, &nm, queries);
     }
 }

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

@@ -1,1098 +1,1069
 // 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, FromHexError, Node, NodePrefix, Revision, RevlogIndex,
-    NULL_REVISION,
+    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 {
     MultipleResults,
-    InvalidNodePrefix,
     /// A `Revision` stored in the nodemap could not be found in the index
     RevisionNotInIndex(Revision),
 }
 
-impl From<FromHexError> for NodeMapError {
-    fn from(_: FromHexError) -> Self {
-        NodeMapError::InvalidNodePrefix
-    }
-}
-
 /// 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`] error is returned.
 ///
 /// In insert operations, the rule is thus that the `NodeMap` must always
 /// be updated after the `RevlogIndex`
 /// be updated first, and the `NodeMap` second.
 ///
 /// [`RevisionNotInIndex`]: enum.NodeMapError.html#variant.RevisionNotInIndex
 /// [`RevlogIndex`]: ../trait.RevlogIndex.html
 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`]
     /// error is returned.
     fn find_bin<'a>(
         &self,
         idx: &impl RevlogIndex,
         prefix: NodePrefix,
     ) -> Result<Option<Revision>, NodeMapError>;
 
-    /// Find the unique Revision whose `Node` hexadecimal string representation
-    /// starts with a given prefix
-    ///
-    /// If no Revision matches the given prefix, `Ok(None)` is returned.
-    ///
-    /// If several Revisions match the given prefix, a [`MultipleResults`]
-    /// error is returned.
-    fn find_hex(
-        &self,
-        idx: &impl RevlogIndex,
-        prefix: &str,
-    ) -> Result<Option<Revision>, NodeMapError> {
-        self.find_bin(idx, NodePrefix::from_hex(prefix)?)
-    }
-
     /// 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`]
     /// error is returned.
     fn unique_prefix_len_bin<'a>(
         &self,
         idx: &impl RevlogIndex,
         node_prefix: NodePrefix,
     ) -> Result<Option<usize>, NodeMapError>;
 
-    /// Same as `unique_prefix_len_bin`, with the hexadecimal representation
-    /// of the prefix as input.
-    fn unique_prefix_len_hex(
-        &self,
-        idx: &impl RevlogIndex,
-        prefix: &str,
-    ) -> Result<Option<usize>, NodeMapError> {
-        self.unique_prefix_len_bin(idx, NodePrefix::from_hex(prefix)?)
-    }
-
     /// Same as `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 [`Blocks`] 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,
         })
     }
 }
 
 /// 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.
 
 const ELEMENTS_PER_BLOCK: usize = 16; // number of different values in a nybble
 
 #[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 `index`.
 fn has_prefix_or_none(
     idx: &impl RevlogIndex,
     prefix: NodePrefix,
     rev: Revision,
 ) -> Result<Option<Revision>, NodeMapError> {
     idx.node(rev)
         .ok_or_else(|| 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 `buffer`,
     /// of which exactly `amount` bytes are used.
     ///
     /// - `buffer` could be derived from `PyBuffer` and `Mmap` objects.
     /// - `offset` allows for the final file format to include fixed data
     ///   (generation number, behavioural flags)
     /// - `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` 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 `Blocks` 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` 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<'n>(&'n self, prefix: NodePrefix) -> NodeTreeVisitor<'n> {
         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_else(|| 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 {
     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_hex(&idx, "1")?, Some(1));
-        assert_eq!(nt.find_hex(&idx, "12")?, Some(1));
-        assert_eq!(nt.find_hex(&idx, "1234de")?, Some(1));
-        assert_eq!(nt.find_hex(&idx, "1a")?, None);
-        assert_eq!(nt.find_hex(&idx, "ab")?, None);
+        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_hex(&idx, "0"), Err(MultipleResults));
-        assert_eq!(nt.find_hex(&idx, "01"), Ok(Some(9)));
-        assert_eq!(nt.find_hex(&idx, "00"), Err(MultipleResults));
-        assert_eq!(nt.find_hex(&idx, "00a"), Ok(Some(0)));
-        assert_eq!(nt.unique_prefix_len_hex(&idx, "00a"), Ok(Some(3)));
-        assert_eq!(nt.find_hex(&idx, "000"), Ok(Some(NULL_REVISION)));
+        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_hex(&idx, "10")?, Some(1));
-        assert_eq!(nt.find_hex(&idx, "c")?, Some(2));
-        assert_eq!(nt.unique_prefix_len_hex(&idx, "c")?, Some(1));
-        assert_eq!(nt.find_hex(&idx, "00"), Err(MultipleResults));
-        assert_eq!(nt.find_hex(&idx, "000")?, Some(NULL_REVISION));
-        assert_eq!(nt.unique_prefix_len_hex(&idx, "000")?, Some(3));
-        assert_eq!(nt.find_hex(&idx, "01")?, Some(9));
+        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,
     }
 
     impl TestNtIndex {
         fn new() -> Self {
             TestNtIndex {
                 index: HashMap::new(),
                 nt: NodeTree::default(),
             }
         }
 
         fn insert(
             &mut self,
             rev: Revision,
             hex: &str,
         ) -> Result<(), NodeMapError> {
             let node = pad_node(hex);
             self.index.insert(rev, node.clone());
             self.nt.insert(&self.index, &node, rev)?;
             Ok(())
         }
 
         fn find_hex(
             &self,
             prefix: &str,
         ) -> Result<Option<Revision>, NodeMapError> {
-            self.nt.find_hex(&self.index, prefix)
+            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_hex(&self.index, prefix)
+            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().map(|block| block.clone()).collect();
             as_vec.extend(self.nt.growable);
             as_vec.push(self.nt.root);
 
             Self {
                 index: self.index,
                 nt: NodeTree::from(as_vec).into(),
             }
         }
     }
 
     #[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").clone();
         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.clone());
         nt.insert(idx, &node0, 0)?;
         idx.insert(1, node1.clone());
         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(())
     }
 }

rust/hg-cpython/src/revlog.rs

@@ -1,491 +1,490
 // 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, PyModule, PyObject,
     PyResult, PyString, PyTuple, Python, PythonObject, ToPyObject,
 };
 use hg::{
     nodemap::{Block, NodeMapError, NodeTree},
-    revlog::{nodemap::NodeMap, RevlogIndex},
+    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()
         };
 
-        nt.find_hex(idx, &node_as_string)
+        let prefix = NodePrefix::from_hex(&node_as_string).map_err(|_| PyErr::new::<ValueError, _>(py, "Invalid node or prefix"))?;
+
+        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)
     }
 
     /// 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)
     }
 
 
 });
 
 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),
     }
 }
 
 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),
-        NodeMapError::InvalidNodePrefix => {
-            PyErr::new::<ValueError, _>(py, "Invalid node or prefix")
-        }
     }
 }
 
 /// 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)
 }