Add initial unicodedata module.

RustPython · windelbouwman · Jul 1, 2019 · Jun 26, 2019 · Jun 30, 2019 · Jun 26, 2019
commit 4db83aca7f1e1fbc8b6f69cb230d4a8b1c0dfd7b
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/tests/snippets/unicode_fu.py b/tests/snippets/unicode_fu.py
@@ -11,3 +11,11 @@
 c = ᚴ*3
 
 assert c == '👋👋👋'
+
+import unicodedata
+assert unicodedata.category('a') == 'Ll'
+assert unicodedata.category('A') == 'Lu'
+assert unicodedata.name('a') == 'LATIN SMALL LETTER A'
+assert unicodedata.lookup('LATIN SMALL LETTER A') == 'a'
+assert unicodedata.bidirectional('a') == 'L'
+assert unicodedata.normalize('NFC', 'bla') == 'bla'
diff --git a/vm/Cargo.toml b/vm/Cargo.toml
@@ -32,6 +32,8 @@ indexmap = "1.0.2"
 crc = "^1.0.0"
 bincode = "1.1.4"
 unicode_categories = "0.1.1"
+unicode_names2 = "0.2.2"
+unic = "0.9.0"
 maplit = "1.0"
 proc-macro-hack = "0.5"
 bitflags = "1.1"

diff --git a/vm/src/stdlib/mod.rs b/vm/src/stdlib/mod.rs
@@ -16,6 +16,7 @@ mod string;
 mod thread;
 mod time_module;
 mod tokenize;
+mod unicodedata;
 mod warnings;
 mod weakref;
 use std::collections::HashMap;
@@ -54,6 +55,7 @@ pub fn get_module_inits() -> HashMap<String, StdlibInitFunc> {
         "tokenize".to_string() => Box::new(tokenize::make_module),
         "_weakref".to_string() => Box::new(weakref::make_module),
         "_imp".to_string() => Box::new(imp::make_module),
+        "unicodedata".to_string() => Box::new(unicodedata::make_module),
         "_warnings".to_string() => Box::new(warnings::make_module),
     };