Prepare unicode_names2_generator for initial release

It now needs to be published, as it is a build-dependency of
unicode_names2 itself.

build.rs

@@ -1,16 +1,18 @@
 use std::{env, path::PathBuf};
 use unicode_names2_generator as generator;
 
+const UNICODE_DATA: &str = include_str!("data/UnicodeData.txt");
+
 fn main() {
     let out_dir = PathBuf::from(env::var_os("OUT_DIR").unwrap());
     {
         let mut generated_path = out_dir.clone();
         generated_path.push("generated.rs");
-        generator::generate(Some(&generated_path), None);
+        generator::generate(UNICODE_DATA, Some(&generated_path), None);
     }
     {
         let mut generated_phf_path = out_dir;
         generated_phf_path.push("generated_phf.rs");
-        generator::generate_phf(Some(&generated_phf_path), None, 3, 2);
+        generator::generate_phf(UNICODE_DATA, Some(&generated_phf_path), None, 3, 2);
     }
 }

generator/Cargo.toml

@@ -2,8 +2,16 @@
 
 name = "unicode_names2_generator"
 edition = "2018"
-version = "0.0.1"
+rust-version = "1.63.0"
+version = "1.0.0"
 authors = ["Huon Wilson <dbau.pp@gmail.com>"]
+homepage = "https://github.com/progval/unicode_names2"
+repository = "https://github.com/progval/unicode_names2"
+documentation = "https://docs.rs/unicode_names2/"
+license = "MIT/Apache-2.0"
+description = """
+Generates the perfect-hash function used by `unicode_names2`.
+"""
 
 [features]
 unstable = []

generator/src/lib.rs

@@ -19,16 +19,14 @@ mod phf;
 mod trie;
 mod util;
 
-const UNICODE_DATA: &str = include_str!("../../data/UnicodeData.txt");
-
 const SPLITTERS: &[u8] = b"-";
 
 struct TableData {
     codepoint_names: Vec<(char, &'static str)>,
     cjk_ideograph_ranges: Vec<(char, char)>,
 }
 
-fn get_table_data() -> TableData {
+fn get_table_data(unicode_data: &'static str) -> TableData {
     fn extract(line: &'static str) -> Option<(char, &'static str)> {
         let splits: Vec<_> = line.splitn(15, ';').collect();
         assert_eq!(splits.len(), 15);
@@ -41,7 +39,7 @@ fn get_table_data() -> TableData {
         Some((c, name))
     }
 
-    let mut iter = UNICODE_DATA.split('\n');
+    let mut iter = unicode_data.split('\n');
 
     let mut codepoint_names = vec![];
     let mut cjk_ideograph_ranges = vec![];
@@ -362,20 +360,20 @@ fn make_context(path: Option<&Path>) -> Context {
 
 #[allow(clippy::type_complexity)]
 fn get_truncated_table_data(
-    truncate: Option<usize>,
+    unicode_data: &'static str, truncate: Option<usize>,
 ) -> (Vec<(char, &'static str)>, Vec<(char, char)>) {
     let TableData {
         mut codepoint_names,
         cjk_ideograph_ranges: cjk,
-    } = get_table_data();
+    } = get_table_data(unicode_data);
     if let Some(n) = truncate {
         codepoint_names.truncate(n)
     }
     (codepoint_names, cjk)
 }
 
-pub fn generate_phf(path: Option<&Path>, truncate: Option<usize>, lambda: usize, tries: usize) {
-    let (codepoint_names, _) = get_truncated_table_data(truncate);
+pub fn generate_phf(unicode_data: &'static str, path: Option<&Path>, truncate: Option<usize>, lambda: usize, tries: usize) {
+    let (codepoint_names, _) = get_truncated_table_data(unicode_data, truncate);
 
     let mut ctxt = make_context(path);
     let (n, disps, data) = phf::create_phf(&codepoint_names, lambda, tries);
@@ -390,8 +388,8 @@ pub fn generate_phf(path: Option<&Path>, truncate: Option<usize>, lambda: usize,
     }
 }
 
-pub fn generate(path: Option<&Path>, truncate: Option<usize>) {
-    let (codepoint_names, cjk) = get_truncated_table_data(truncate);
+pub fn generate(unicode_data: &'static str, path: Option<&Path>, truncate: Option<usize>) {
+    let (codepoint_names, cjk) = get_truncated_table_data(unicode_data, truncate);
     let mut ctxt = make_context(path);
 
     write_cjk_ideograph_ranges(&mut ctxt, &cjk);