Merge pull request #20 from dtolnay/ucdparse

Replace ucd-generate dependency with ucd-parse

.gitattributes

@@ -1,5 +1,5 @@
-generate/src/ucd.rs linguist-generated
 src/tables.rs linguist-generated
 tests/fst/xid_continue.fst linguist-generated
 tests/fst/xid_start.fst linguist-generated
+tests/tables/tables.rs linguist-generated
 tests/trie/trie.rs linguist-generated

.github/workflows/ci.yml

@@ -20,15 +20,15 @@ jobs:
       - uses: actions/checkout@v3
       - uses: dtolnay/rust-toolchain@stable
       - id: ucd-generate
-        run: echo "version=$(grep 'ucd-generate [0-9]\+\.[0-9]\+\.[0-9]\+' generate/src/ucd.rs --only-matching)" >> $GITHUB_OUTPUT
+        run: echo "version=$(grep 'ucd-generate [0-9]\+\.[0-9]\+\.[0-9]\+' tests/tables/tables.rs --only-matching)" >> $GITHUB_OUTPUT
       - run: cargo install ucd-generate
       - run: curl -LO https://www.unicode.org/Public/zipped/latest/UCD.zip
       - run: unzip UCD.zip -d UCD
-      - run: ucd-generate property-bool UCD --include XID_Start,XID_Continue > generate/src/ucd.rs
+      - run: ucd-generate property-bool UCD --include XID_Start,XID_Continue > tests/tables/tables.rs
       - run: ucd-generate property-bool UCD --include XID_Start,XID_Continue --fst-dir tests/fst
       - run: ucd-generate property-bool UCD --include XID_Start,XID_Continue --trie-set > tests/trie/trie.rs
       - run: cargo run --manifest-path generate/Cargo.toml
-      - run: sed --in-place 's/ucd-generate [0-9]\+\.[0-9]\+\.[0-9]\+/${{steps.ucd-generate.outputs.version}}/' generate/src/ucd.rs tests/trie/trie.rs
+      - run: sed --in-place 's/ucd-generate [0-9]\+\.[0-9]\+\.[0-9]\+/${{steps.ucd-generate.outputs.version}}/' tests/tables/tables.rs tests/trie/trie.rs
       - run: git diff --exit-code
 
   test:

generate/Cargo.toml

@@ -4,3 +4,7 @@ version = "0.0.0"
 authors = ["David Tolnay <[email protected]>"]
 edition = "2018"
 publish = false
+
+[dependencies]
+anyhow = "1"
+ucd-parse = "0.1.10"

generate/src/main.rs

@@ -3,7 +3,7 @@
 // $ cargo install ucd-generate
 // $ curl -LO https://www.unicode.org/Public/zipped/15.0.0/UCD.zip
 // $ unzip UCD.zip -d UCD
-// $ ucd-generate property-bool UCD --include XID_Start,XID_Continue > generate/src/ucd.rs
+// $ ucd-generate property-bool UCD --include XID_Start,XID_Continue > tests/table/tables.rs
 // $ ucd-generate property-bool UCD --include XID_Start,XID_Continue --fst-dir tests/fst
 // $ ucd-generate property-bool UCD --include XID_Start,XID_Continue --trie-set > tests/trie/trie.rs
 // $ cargo run --manifest-path generate/Cargo.toml
@@ -12,50 +12,43 @@
     clippy::cast_lossless,
     clippy::cast_possible_truncation, // https://github.com/rust-lang/rust-clippy/issues/9613
     clippy::match_wild_err_arm,
+    clippy::module_name_repetitions,
     clippy::too_many_lines,
     clippy::uninlined_format_args
 )]
 
-#[rustfmt::skip]
-#[allow(dead_code, clippy::all, clippy::pedantic)]
-mod ucd;
-
 mod output;
+mod parse;
 mod write;
 
-use std::cmp::Ordering;
+use crate::parse::parse_xid_properties;
+use anyhow::Result;
 use std::collections::{BTreeMap as Map, VecDeque};
 use std::convert::TryFrom;
 use std::fs;
-use std::io;
+use std::io::{self, Write};
 use std::path::Path;
+use std::process;
 
 const CHUNK: usize = 64;
-const PATH: &str = "../src/tables.rs";
-
-fn is_xid_start(ch: char) -> bool {
-    search(ch, ucd::XID_START)
-}
-
-fn is_xid_continue(ch: char) -> bool {
-    search(ch, ucd::XID_CONTINUE)
-}
+const UCD: &str = "UCD";
+const TABLES: &str = "src/tables.rs";
+
+fn main() -> Result<()> {
+    let manifest_dir = Path::new(env!("CARGO_MANIFEST_DIR"));
+    let unicode_ident_dir = manifest_dir.parent().unwrap();
+    let ucd_dir = unicode_ident_dir.join(UCD);
+    if !ucd_dir.exists() {
+        writeln!(
+            io::stderr(),
+            "Not found: {}\nDownload from https://www.unicode.org/Public/zipped/l5.0.0/UCD.zip and unzip.",
+            ucd_dir.display(),
+        )?;
+        process::exit(1);
+    }
 
-fn search(ch: char, table: &[(u32, u32)]) -> bool {
-    table
-        .binary_search_by(|&(lo, hi)| {
-            if lo > ch as u32 {
-                Ordering::Greater
-            } else if hi < ch as u32 {
-                Ordering::Less
-            } else {
-                Ordering::Equal
-            }
-        })
-        .is_ok()
-}
+    let properties = parse_xid_properties(&ucd_dir)?;
 
-fn main() -> io::Result<()> {
     let mut chunkmap = Map::<[u8; CHUNK], u8>::new();
     let mut dense = Vec::<[u8; CHUNK]>::new();
     let mut new_chunk = |chunk| {
@@ -87,8 +80,8 @@ fn main() -> io::Result<()> {
                 let code = (i * CHUNK as u32 + j) * 8 + k;
                 if code >= 0x80 {
                     if let Some(ch) = char::from_u32(code) {
-                        *this_start |= (is_xid_start(ch) as u8) << k;
-                        *this_continue |= (is_xid_continue(ch) as u8) << k;
+                        *this_start |= (properties.is_xid_start(ch) as u8) << k;
+                        *this_continue |= (properties.is_xid_continue(ch) as u8) << k;
                     }
                 }
             }
@@ -163,7 +156,8 @@ fn main() -> io::Result<()> {
         *index = dense_to_halfdense[index];
     }
 
-    let out = write::output(&index_start, &index_continue, &halfdense);
-    let path = Path::new(env!("CARGO_MANIFEST_DIR")).join(PATH);
-    fs::write(path, out)
+    let out = write::output(&properties, &index_start, &index_continue, &halfdense);
+    let path = unicode_ident_dir.join(TABLES);
+    fs::write(path, out)?;
+    Ok(())
 }

generate/src/parse.rs

@@ -0,0 +1,40 @@
+use anyhow::Result;
+use std::collections::BTreeSet as Set;
+use std::path::Path;
+use ucd_parse::CoreProperty;
+
+pub struct Properties {
+    xid_start: Set<u32>,
+    xid_continue: Set<u32>,
+}
+
+impl Properties {
+    pub fn is_xid_start(&self, ch: char) -> bool {
+        self.xid_start.contains(&(ch as u32))
+    }
+
+    pub fn is_xid_continue(&self, ch: char) -> bool {
+        self.xid_continue.contains(&(ch as u32))
+    }
+}
+
+pub fn parse_xid_properties(ucd_dir: &Path) -> Result<Properties> {
+    let mut properties = Properties {
+        xid_start: Set::new(),
+        xid_continue: Set::new(),
+    };
+
+    let prop_list: Vec<CoreProperty> = ucd_parse::parse(ucd_dir)?;
+    for core in prop_list {
+        let set = match core.property.as_str() {
+            "XID_Start" => &mut properties.xid_start,
+            "XID_Continue" => &mut properties.xid_continue,
+            _ => continue,
+        };
+        for codepoint in core.codepoints {
+            set.insert(codepoint.value());
+        }
+    }
+
+    Ok(properties)
+}

generate/src/write.rs

@@ -1,5 +1,6 @@
 use crate::output::Output;
-use crate::{is_xid_continue, is_xid_start, CHUNK};
+use crate::parse::Properties;
+use crate::CHUNK;
 
 const HEAD: &str = "\
 // \x40generated by ../generate. To regenerate, run the following in the repo root:
@@ -17,7 +18,12 @@ pub(crate) struct Align8<T>(pub(crate) T);
 pub(crate) struct Align64<T>(pub(crate) T);
 ";
 
-pub fn output(index_start: &[u8], index_continue: &[u8], halfdense: &[u8]) -> Output {
+pub fn output(
+    properties: &Properties,
+    index_start: &[u8],
+    index_continue: &[u8],
+    halfdense: &[u8],
+) -> Output {
     let mut out = Output::new();
     writeln!(out, "{}", HEAD);
 
@@ -29,7 +35,8 @@ pub fn output(index_start: &[u8], index_continue: &[u8], halfdense: &[u8]) -> Ou
         write!(out, "   ");
         for j in 0..32 {
             let ch = (i * 32 + j) as char;
-            write!(out, " {},", if is_xid_start(ch) { 'T' } else { 'F' });
+            let is_xid_start = properties.is_xid_start(ch);
+            write!(out, " {},", if is_xid_start { 'T' } else { 'F' });
         }
         writeln!(out);
     }
@@ -44,7 +51,8 @@ pub fn output(index_start: &[u8], index_continue: &[u8], halfdense: &[u8]) -> Ou
         write!(out, "   ");
         for j in 0..32 {
             let ch = (i * 32 + j) as char;
-            write!(out, " {},", if is_xid_continue(ch) { 'T' } else { 'F' });
+            let is_xid_continue = properties.is_xid_continue(ch);
+            write!(out, " {},", if is_xid_continue { 'T' } else { 'F' });
         }
         writeln!(out);
     }

tests/static_size.rs

@@ -19,14 +19,13 @@ fn test_size() {
 #[test]
 fn test_xid_size() {
     #[deny(dead_code)]
-    #[allow(clippy::redundant_static_lifetimes)]
-    #[path = "../generate/src/ucd.rs"]
-    mod ucd;
+    #[path = "tables/mod.rs"]
+    mod tables;
 
-    let size = size_of_val(ucd::XID_START) + size_of_val(ucd::XID_CONTINUE);
+    let size = size_of_val(tables::XID_START) + size_of_val(tables::XID_CONTINUE);
     assert_eq!(11528, size);
 
-    let _ = ucd::BY_NAME;
+    let _ = tables::BY_NAME;
 }
 
 #[cfg(target_pointer_width = "64")]

tests/tables/mod.rs

@@ -0,0 +1,7 @@
+#![allow(clippy::module_inception)]
+
+#[allow(clippy::redundant_static_lifetimes)]
+#[rustfmt::skip]
+mod tables;
+
+pub(crate) use self::tables::*;