Merge pull request #73 from xandeer/master

fix: failed to parse headline tags like ":hello:你好:"

.changeset/quiet-walls-refuse.md

@@ -0,0 +1,5 @@
+---
+'uniorg-parse': patch
+---
+
+Fix parsing unicode characters in headline tags. The regex for parsing tags previously used `\w` (word) class, which does not behave correctly with unicode. Update it to use unicode's Letter and Number character properties instead.

packages/uniorg-parse/src/__snapshots__/parser.spec.ts.snap

@@ -2053,6 +2053,31 @@ children:
             value: "Something"
 `;
 
+exports[`org/parser headline statistics-cookie non-ascii characters in headline tags 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 21
+children:
+  - type: "section"
+    contentsBegin: 0
+    contentsEnd: 21
+    children:
+      - type: "headline"
+        level: 1
+        todoKeyword: null
+        priority: null
+        commented: false
+        rawValue: "headline"
+        tags:
+          - "hello"
+          - "你好"
+        contentsBegin: 2
+        contentsEnd: 10
+        children:
+          - type: "text"
+            value: "headline"
+`;
+
 exports[`org/parser headline statistics-cookie statistics cookie with long trailing space 1`] = `
 type: "org-data"
 contentsBegin: 0

packages/uniorg-parse/src/parser.spec.ts

@@ -124,6 +124,11 @@ describe('org/parser', () => {
         `* TODO [#A] COMMENT headline /italic/ title :some:tags: [1/3]`
       );
 
+      itParses(
+        'non-ascii characters in headline tags',
+        `* headline :hello:你好:`
+      );
+
       itParses('statistics cookie without trailing space', `* [/]hello`);
 
       itParses('statistics cookie with long trailing space', `* [/]    hello`);

packages/uniorg-parse/src/parser.ts

@@ -725,7 +725,9 @@ class Parser {
 
     const titleStart = this.r.offset();
 
-    const tagsM = this.r.lookingAt(/^(.*?)[ \t]+:([\w@#%:]+):[ \t]*$/m);
+    const tagsM = this.r.lookingAt(
+      /^(.*?)[ \t]+:([\p{L}\p{N}_@#%:]+):[ \t]*$/mu
+    );
     const tags = tagsM?.[2].split(':') ?? [];
     const titleEnd = tagsM
       ? titleStart + tagsM.index + tagsM[1].length