deps src: Automatic update

src/treesitter-stamp/download-treesitter.cmake

@@ -22,16 +22,16 @@ function(check_file_hash has_hash hash_is_good)
   set("${has_hash}" TRUE PARENT_SCOPE)
 
   message(STATUS "verifying file...
-       file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz'")
+       file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz'")
 
-  file("SHA256" "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz" actual_value)
+  file("SHA256" "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz" actual_value)
 
-  if(NOT "${actual_value}" STREQUAL "919b750da9af1260cd989498bc84c63391b72ee2aa2ec20fc84882544eb7a229")
+  if(NOT "${actual_value}" STREQUAL "6bc22ca7e0f81d77773462d922cf40b44bfd090d92abac75cb37dbae516c2417")
     set("${hash_is_good}" FALSE PARENT_SCOPE)
     message(STATUS "SHA256 hash of
-    /home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz
+    /home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz
   does not match expected value
-    expected: '919b750da9af1260cd989498bc84c63391b72ee2aa2ec20fc84882544eb7a229'
+    expected: '6bc22ca7e0f81d77773462d922cf40b44bfd090d92abac75cb37dbae516c2417'
       actual: '${actual_value}'")
   else()
     set("${hash_is_good}" TRUE PARENT_SCOPE)
@@ -71,32 +71,32 @@ function(sleep_before_download attempt)
   execute_process(COMMAND "${CMAKE_COMMAND}" -E sleep "${sleep_seconds}")
 endfunction()
 
-if(EXISTS "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz")
+if(EXISTS "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz")
   check_file_hash(has_hash hash_is_good)
   if(has_hash)
     if(hash_is_good)
       message(STATUS "File already exists and hash match (skip download):
-  file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz'
-  SHA256='919b750da9af1260cd989498bc84c63391b72ee2aa2ec20fc84882544eb7a229'"
+  file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz'
+  SHA256='6bc22ca7e0f81d77773462d922cf40b44bfd090d92abac75cb37dbae516c2417'"
       )
       return()
     else()
       message(STATUS "File already exists but hash mismatch. Removing...")
-      file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz")
+      file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz")
     endif()
   else()
     message(STATUS "File already exists but no hash specified (use URL_HASH):
-  file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz'
+  file='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz'
 Old file will be removed and new file downloaded from URL."
     )
-    file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz")
+    file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz")
   endif()
 endif()
 
 set(retry_number 5)
 
 message(STATUS "Downloading...
-   dst='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz'
+   dst='/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz'
    timeout='none'
    inactivity timeout='none'"
 )
@@ -107,7 +107,7 @@ foreach(i RANGE ${retry_number})
   if(status_code IN_LIST download_retry_codes)
     sleep_before_download(${i})
   endif()
-  foreach(url IN ITEMS [====[https://github.com/tree-sitter/tree-sitter/archive/v0.22.4.tar.gz]====])
+  foreach(url IN ITEMS [====[https://github.com/tree-sitter/tree-sitter/archive/v0.22.5.tar.gz]====])
     if(NOT url IN_LIST skip_url_list)
       message(STATUS "Using src='${url}'")
 
@@ -118,7 +118,7 @@ foreach(i RANGE ${retry_number})
 
       file(
         DOWNLOAD
-        "${url}" "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz"
+        "${url}" "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz"
 
         # no TIMEOUT
         # no INACTIVITY_TIMEOUT
@@ -135,7 +135,7 @@ foreach(i RANGE ${retry_number})
         check_file_hash(has_hash hash_is_good)
         if(has_hash AND NOT hash_is_good)
           message(STATUS "Hash mismatch, removing...")
-          file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz")
+          file(REMOVE "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz")
         else()
           message(STATUS "Downloading... done")
           return()

src/treesitter-stamp/extract-treesitter.cmake

@@ -5,7 +5,7 @@ cmake_minimum_required(VERSION 3.5)
 
 # Make file names absolute:
 #
-get_filename_component(filename "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.4.tar.gz" ABSOLUTE)
+get_filename_component(filename "/home/runner/work/deps/deps/neovim/deps/build/downloads/treesitter/v0.22.5.tar.gz" ABSOLUTE)
 get_filename_component(directory "/home/runner/work/deps/deps/neovim/deps/build/src/treesitter" ABSOLUTE)
 
 message(STATUS "extracting...

src/treesitter-stamp/treesitter-urlinfo.txt

@@ -6,7 +6,7 @@ method=url
 command=/usr/local/bin/cmake;-P;/home/runner/work/deps/deps/neovim/deps/build/src/treesitter-stamp/download-treesitter.cmake;COMMAND;/usr/local/bin/cmake;-P;/home/runner/work/deps/deps/neovim/deps/build/src/treesitter-stamp/verify-treesitter.cmake;COMMAND;/usr/local/bin/cmake;-P;/home/runner/work/deps/deps/neovim/deps/build/src/treesitter-stamp/extract-treesitter.cmake
 source_dir=/home/runner/work/deps/deps/neovim/deps/build/src/treesitter
 work_dir=/home/runner/work/deps/deps/neovim/deps/build/src
-url(s)=https://github.com/tree-sitter/tree-sitter/archive/v0.22.4.tar.gz
-hash=SHA256=919b750da9af1260cd989498bc84c63391b72ee2aa2ec20fc84882544eb7a229
+url(s)=https://github.com/tree-sitter/tree-sitter/archive/v0.22.5.tar.gz
+hash=SHA256=6bc22ca7e0f81d77773462d922cf40b44bfd090d92abac75cb37dbae516c2417
 no_extract=
 

src/treesitter/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## [0.22.3] - 2024-04-14
+
+### Performance
+
+- **cli**: Reduced the compile time of generated parsers by generating C code with fewer conditionals
+  (https://github.com/tree-sitter/tree-sitter/pull/3234)
+
 ## [0.22.1] — 2024-03-10
 
 ### Bug Fixes

src/treesitter/Cargo.toml

@@ -12,7 +12,7 @@ members = [
 resolver = "2"
 
 [workspace.package]
-version = "0.22.4"
+version = "0.22.5"
 authors = ["Max Brunsfeld <maxbrunsfeld@gmail.com>"]
 edition = "2021"
 rust-version = "1.74.1"

src/treesitter/Makefile

@@ -1,4 +1,4 @@
-VERSION := 0.22.4
+VERSION := 0.22.5
 
 # install directory layout
 PREFIX ?= /usr/local

src/treesitter/cli/npm/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tree-sitter-cli",
-  "version": "0.22.4",
+  "version": "0.22.5",
   "author": "Max Brunsfeld",
   "license": "MIT",
   "repository": {

src/treesitter/cli/src/generate/render.rs

@@ -64,7 +64,7 @@ struct Generator {
     main_lex_table: LexTable,
     keyword_lex_table: LexTable,
     large_character_sets: Vec<(Option<Symbol>, CharacterSet)>,
-    large_character_set_constant_names: Vec<String>,
+    large_character_set_info: Vec<LargeCharacterSetInfo>,
     large_state_count: usize,
     keyword_capture_token: Option<Symbol>,
     syntax_grammar: SyntaxGrammar,
@@ -81,6 +81,11 @@ struct Generator {
     abi_version: usize,
 }
 
+struct LargeCharacterSetInfo {
+    constant_name: String,
+    is_used: bool,
+}
+
 impl Generator {
     fn generate(mut self) -> String {
         self.init();
@@ -108,11 +113,7 @@ impl Generator {
             self.add_primary_state_id_list();
         }
 
-        // Generate a helper function for each large character set.
-        // let mut sorted_large_char_sets = self.large_character_sets.iter().collect::<Vec<_>>();
-        for ix in 0..self.large_character_sets.len() {
-            self.add_character_set(ix);
-        }
+        let buffer_offset_before_lex_functions = self.buffer.len();
 
         let mut main_lex_table = LexTable::default();
         swap(&mut main_lex_table, &mut self.main_lex_table);
@@ -124,6 +125,16 @@ impl Generator {
             self.add_lex_function("ts_lex_keywords", keyword_lex_table);
         }
 
+        // Once the lex functions are generated, and we've determined which large
+        // character sets are actually used, we can generate the large character set
+        // constants. Insert them into the output buffer before the lex functions.
+        let lex_functions = self.buffer[buffer_offset_before_lex_functions..].to_string();
+        self.buffer.truncate(buffer_offset_before_lex_functions);
+        for ix in 0..self.large_character_sets.len() {
+            self.add_character_set(ix);
+        }
+        self.buffer.push_str(&lex_functions);
+
         self.add_lex_modes_list();
         self.add_parse_table();
 
@@ -238,6 +249,23 @@ impl Generator {
             }
         }
 
+        for (ix, (symbol, _)) in self.large_character_sets.iter().enumerate() {
+            let count = self.large_character_sets[0..ix]
+                .iter()
+                .filter(|(sym, _)| sym == symbol)
+                .count()
+                + 1;
+            let constant_name = if let Some(symbol) = symbol {
+                format!("{}_character_set_{}", self.symbol_ids[&symbol], count)
+            } else {
+                format!("extras_character_set_{}", count)
+            };
+            self.large_character_set_info.push(LargeCharacterSetInfo {
+                constant_name,
+                is_used: false,
+            });
+        }
+
         // Determine which states should use the "small state" representation, and which should
         // use the normal array representation.
         let threshold = cmp::min(SMALL_STATE_THRESHOLD, self.parse_table.symbols.len() / 2);
@@ -829,10 +857,12 @@ impl Generator {
                     add!(self, "(!eof && ")
                 }
 
+                let char_set_info = &mut self.large_character_set_info[large_char_set_ix];
+                char_set_info.is_used = true;
                 add!(
                     self,
                     "set_contains({}, {}, lookahead)",
-                    &self.large_character_set_constant_names[large_char_set_ix],
+                    &char_set_info.constant_name,
                     large_set.range_count(),
                 );
                 if check_eof {
@@ -940,20 +970,17 @@ impl Generator {
     }
 
     fn add_character_set(&mut self, ix: usize) {
-        let (symbol, characters) = self.large_character_sets[ix].clone();
-        let count = self.large_character_sets[0..ix]
-            .iter()
-            .filter(|(sym, _)| *sym == symbol)
-            .count()
-            + 1;
+        let characters = self.large_character_sets[ix].1.clone();
+        let info = &self.large_character_set_info[ix];
+        if !info.is_used {
+            return;
+        }
 
-        let constant_name = if let Some(symbol) = symbol {
-            format!("{}_character_set_{}", self.symbol_ids[&symbol], count)
-        } else {
-            format!("extras_character_set_{}", count)
-        };
-        add_line!(self, "static TSCharacterRange {}[] = {{", constant_name);
-        self.large_character_set_constant_names.push(constant_name);
+        add_line!(
+            self,
+            "static TSCharacterRange {}[] = {{",
+            info.constant_name
+        );
 
         indent!(self);
         for (ix, range) in characters.ranges().enumerate() {
@@ -1694,7 +1721,7 @@ pub fn render_c_code(
         keyword_lex_table: tables.keyword_lex_table,
         keyword_capture_token: tables.word_token,
         large_character_sets: tables.large_character_sets,
-        large_character_set_constant_names: Vec::new(),
+        large_character_set_info: Vec::new(),
         syntax_grammar,
         lexical_grammar,
         default_aliases,

src/treesitter/cli/src/test.rs

@@ -525,11 +525,11 @@ fn parse_test_content(name: String, content: &str, file_path: Option<PathBuf>) -
         let mut test_name = String::new();
         let mut seen_marker = false;
 
-        for line in test_name_and_markers
-            .split(|&c| c == b'\n')
+        for line in str::from_utf8(test_name_and_markers)
+            .unwrap()
+            .lines()
             .filter(|s| !s.is_empty())
         {
-            let line = str::from_utf8(line).unwrap();
             match line.split('(').next().unwrap() {
                 ":skip" => (seen_marker, skip) = (true, true),
                 ":platform" => {

src/treesitter/lib/binding_rust/ffi.rs

@@ -8,7 +8,7 @@ include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
 #[cfg(not(feature = "bindgen"))]
 include!("./bindings.rs");
 
-#[cfg(unix)]
+#[cfg(not(windows))]
 extern "C" {
     pub(crate) fn _ts_dup(fd: std::os::raw::c_int) -> std::os::raw::c_int;
 }

src/treesitter/lib/binding_rust/lib.rs

@@ -3,8 +3,8 @@
 pub mod ffi;
 mod util;
 
-#[cfg(unix)]
-use std::os::unix::io::AsRawFd;
+#[cfg(not(windows))]
+use std::os::fd::AsRawFd;
 #[cfg(windows)]
 use std::os::windows::io::AsRawHandle;
 use std::{
@@ -942,10 +942,10 @@ impl Tree {
     #[doc(alias = "ts_tree_print_dot_graph")]
     pub fn print_dot_graph(
         &self,
-        #[cfg(unix)] file: &impl AsRawFd,
+        #[cfg(not(windows))] file: &impl AsRawFd,
         #[cfg(windows)] file: &impl AsRawHandle,
     ) {
-        #[cfg(unix)]
+        #[cfg(not(windows))]
         {
             let fd = file.as_raw_fd();
             unsafe { ffi::ts_tree_print_dot_graph(self.0.as_ptr(), fd) }

src/treesitter/lib/binding_web/package.json

@@ -1,6 +1,6 @@
 {
   "name": "web-tree-sitter",
-  "version": "0.22.4",
+  "version": "0.22.5",
   "description": "Tree-sitter bindings for the web",
   "main": "tree-sitter.js",
   "types": "tree-sitter-web.d.ts",