Merge pull request #42 from Marcondiro/master

Expose UNICODE_VERSION const

Author: dtolnay

generate/Cargo.toml

@@ -4,3 +4,6 @@ version = "0.0.0"
 authors = ["David Tolnay <[email protected]>"]
 edition = "2018"
 publish = false
+
+[dependencies]
+regex = "1.12.2"

generate/src/parse.rs

@@ -1,15 +1,21 @@
+use regex::Regex;
 use std::collections::BTreeSet as Set;
 use std::fs;
 use std::io::{self, Write};
 use std::path::Path;
 use std::process;
 
 pub struct Properties {
+    unicode_version: (u64, u64, u64),
     xid_start: Set<u32>,
     xid_continue: Set<u32>,
 }
 
 impl Properties {
+    pub const fn unicode_version(&self) -> (u64, u64, u64) {
+        self.unicode_version
+    }
+
     pub fn is_xid_start(&self, ch: char) -> bool {
         self.xid_start.contains(&(ch as u32))
     }
@@ -20,11 +26,6 @@ impl Properties {
 }
 
 pub fn parse_xid_properties(ucd_dir: &Path) -> Properties {
-    let mut properties = Properties {
-        xid_start: Set::new(),
-        xid_continue: Set::new(),
-    };
-
     let filename = "DerivedCoreProperties.txt";
     let path = ucd_dir.join(filename);
     let contents = fs::read_to_string(path).unwrap_or_else(|err| {
@@ -34,6 +35,12 @@ pub fn parse_xid_properties(ucd_dir: &Path) -> Properties {
         process::exit(1);
     });
 
+    let mut properties = Properties {
+        unicode_version: parse_unicode_version(filename, &contents),
+        xid_start: Set::new(),
+        xid_continue: Set::new(),
+    };
+
     for (i, line) in contents.lines().enumerate() {
         if line.starts_with('#') || line.trim().is_empty() {
             continue;
@@ -73,3 +80,24 @@ fn parse_line(line: &str) -> Option<(u32, u32, &str)> {
 fn parse_codepoint(s: &str) -> Option<u32> {
     u32::from_str_radix(s, 16).ok()
 }
+
+fn parse_unicode_version(filename: &str, contents: &str) -> (u64, u64, u64) {
+    let (name, extension) = filename
+        .rsplit_once('.')
+        .expect("Failed to split file name into name and extension");
+    let re = Regex::new(&format!(r"# {name}-(\d+).(\d+).(\d+).{extension}")).unwrap();
+    let caps = re
+        .captures(contents)
+        .expect("Failed to find unicode version in unicode data");
+    let v = caps
+        .iter()
+        .skip(1)
+        .map(|s| {
+            s.unwrap()
+                .as_str()
+                .parse()
+                .expect("Failed to parse unicode version")
+        })
+        .collect::<Vec<u64>>();
+    (v[0], v[1], v[2])
+}

generate/src/write.rs

@@ -24,6 +24,13 @@ pub fn output(
     let mut out = Output::new();
     writeln!(out, "{}", HEAD);
 
+    writeln!(
+        out,
+        "pub const UNICODE_VERSION: (u64, u64, u64) = {:?};",
+        properties.unicode_version()
+    );
+    writeln!(out);
+
     let ascii_start = (0u8..128)
         .map(|c| (properties.is_xid_start(c as char) as u128) << c)
         .sum::<u128>();

src/lib.rs

@@ -252,6 +252,7 @@
 #[rustfmt::skip]
 mod tables;
 
+pub use crate::tables::UNICODE_VERSION;
 use crate::tables::{ASCII_CONTINUE, ASCII_START, CHUNK, LEAF, TRIE_CONTINUE, TRIE_START};
 
 static ZERO: u8 = 0;