[209_7] 修复bash代码模式下的错误高亮 (#2672)

Co-authored-by: notfoundzzz <[email protected]>

Author: notfoundzzz

TeXmacs/plugins/bash/progs/code/bash-lang.scm

@@ -139,6 +139,12 @@
       "##" "#" "%%" "%"                   ;; ${var##pat} ${var%pat} ...
       )))
 
+;; Paths / urls
+(tm-define (parser-feature lan key)
+  (:require (and (== lan "bash") (== key "path")))
+  `(,(string->symbol key)
+    (enable)))
+
 ;; Numbers
 (tm-define (parser-feature lan key)
   (:require (and (== lan "bash") (== key "number")))
@@ -163,7 +169,8 @@
 (tm-define (parser-feature lan key)
   (:require (and (== lan "bash") (== key "comment")))
   `(,(string->symbol key)
-    (inline "#")))
+    (inline "#")
+    (inline_require_space)))
 
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;; Preferences for syntax highlighting

TeXmacs/tests/tmu/209_7.tmu

@@ -1,4 +1,4 @@
-<TMU|<tuple|1.1.0|2026.1.1>>
+<TMU|<tuple|1.1.0|2026.1.2>>
 
 <style|<tuple|generic|chinese|table-captions-above|number-europe|preview-ref|bash>>
 
@@ -24,6 +24,38 @@
 
     IFS=$'\\n\\t'
 
+    # ---------- path -------------
+
+    \;
+
+    tar xzvf nvim-linux-x86_64.tar.gz
+
+    sudo mv nvim-linux-x86_64 /opt/nvim
+
+    git clone [email protected]:XmacsLabs/mogan.git
+
+    cat chapter-3.1.tmu
+
+    3.2 4.5
+
+    \;
+
+    # ----------- commment ---------
+
+    echo $# # 行尾注释
+
+    \;
+
+    echo ${#PATH}#not_comment
+
+    echo "${#PATH}#still_not_comment"
+
+    \;
+
+    echo ${#PATH} # 这里才是注释
+
+    \;
+
     \;
 
     # ---------- globals ----------

devel/209_7.md

@@ -33,3 +33,13 @@
 
 关联 issue #2607
 
+## 2026/01/29
+### What
+- 修复 Bash 代码模式下 `#` 在 `$#`、`${#...}` 中被误识别为注释的问题
+- 增加路径/URL 识别规则，避免路径/文件名中的关键字与数字误高亮（如 `abc.git`）
+
+### How
+- 在 `prog_language.cpp` 增加可配置的 `path` 解析步骤，位于 `string_parser` 之后、`keyword/number` 之前；匹配到路径/URL token 则整体消费
+- `bash-lang.scm` 中启用 `path` 解析
+- 注释规则增加 `inline_require_space`：仅当 `#` 位于行首或空白后才作为注释起始
+

src/System/Language/impl_language.hpp

@@ -90,7 +90,11 @@ struct prog_language_rep : abstract_language_rep {
   void customize_string (tree config);
   void customize_preprocessor (tree config);
   void customize_comment (tree config);
+  void customize_path (tree config);
   tree get_parser_config (string lan, string key);
+
+  bool inline_comment_requires_space;
+  bool path_parser_enabled;
 };
 
 struct scheme_language_rep : language_rep {

src/System/Language/prog_language.cpp

@@ -25,6 +25,8 @@ prog_language_rep::prog_language_rep (string name)
     : abstract_language_rep (name) {
   if (DEBUG_PARSER)
     debug_packrat << "Building the " * name * " language parser" << LF;
+  inline_comment_requires_space= false;
+  path_parser_enabled          = false;
 
   string use_modules= "(use-modules (code " * name * "-lang))";
   eval (use_modules);
@@ -49,6 +51,9 @@ prog_language_rep::prog_language_rep (string name)
 
   tree preprocessor_config= get_parser_config (name, "preprocessor");
   customize_preprocessor (preprocessor_config);
+
+  tree path_config= get_parser_config (name, "path");
+  customize_path (path_config);
 }
 
 tree
@@ -207,6 +212,9 @@ prog_language_rep::customize_comment (tree config) {
       }
       inline_comment_parser.set_starts (inline_comment_starts);
     }
+    else if (label == "inline_require_space") {
+      inline_comment_requires_space= true;
+    }
   }
 }
 
@@ -227,6 +235,78 @@ prog_language_rep::customize_preprocessor (tree config) {
   if (DEBUG_PARSER) debug_packrat << preprocessor_parser.to_string ();
 }
 
+void
+prog_language_rep::customize_path (tree config) {
+  for (int i= 0; i < N (config); i++) {
+    tree   feature= config[i];
+    string label  = get_label (feature);
+    if (label == "enable") {
+      path_parser_enabled= true;
+    }
+  }
+}
+
+static bool
+is_path_token_delim (char c) {
+  return is_space (c) || c == '(' || c == ')' || c == '{' || c == '}' ||
+         c == '[' || c == ']' || c == ';' || c == '|' || c == '&' || c == '<' ||
+         c == '>';
+}
+
+static bool
+looks_like_path_token (string token) {
+  if (is_empty (token)) return false;
+  if (token == "." || token == "..") return false;
+
+  url parsed= url_path (token);
+  if (is_rooted (parsed) || is_rooted_web (parsed) || is_concat (parsed))
+    return true;
+
+  bool has_dot  = false;
+  bool has_alpha= false;
+  bool has_digit= false;
+  bool has_at   = false;
+  bool has_colon= false;
+  bool has_dash = false;
+  for (int i= 0; i < N (token); i++) {
+    char c= token[i];
+    if (c == '/' || c == '\\') return true;
+    if (c == '.') has_dot= true;
+    if (is_alpha (c)) has_alpha= true;
+    if (is_digit (c)) has_digit= true;
+    if (c == '-') has_dash= true;
+    if (c == '@') has_at= true;
+    if (c == ':') has_colon= true;
+    if (c == ':' && i + 2 < N (token) && token[i + 1] == '/' &&
+        token[i + 2] == '/')
+      return true;
+  }
+
+  if (N (token) >= 2 && token[0] == '.' && token[1] == '/') return true;
+  if (N (token) >= 3 && token[0] == '.' && token[1] == '.' && token[2] == '/')
+    return true;
+  if (N (token) >= 2 && token[0] == '~' && token[1] == '/') return true;
+  if (has_at && has_colon) return true;
+  if (has_dot && has_alpha) return true;
+  if (token[0] != '-' && has_dash && has_alpha && has_digit) return true;
+  return false;
+}
+
+static bool
+parse_path_token (string s, int& pos) {
+  if (pos >= N (s)) return false;
+  if (is_path_token_delim (s[pos])) return false;
+
+  int start= pos;
+  int end  = pos;
+  while (end < N (s) && !is_path_token_delim (s[end]))
+    end++;
+  string token= s (start, end);
+  if (!looks_like_path_token (token)) return false;
+  pos= end;
+  return true;
+}
+
 text_property
 prog_language_rep::advance (tree t, int& pos) {
   string s= t->label;
@@ -255,6 +335,10 @@ prog_language_rep::advance (tree t, int& pos) {
     current_parser= string_parser.get_parser_name ();
     return &tp_normal_rep;
   }
+  if (path_parser_enabled && parse_path_token (s, pos)) {
+    current_parser= "path_parser";
+    return &tp_normal_rep;
+  }
   if (keyword_parser.parse (s, pos)) {
     current_parser= keyword_parser.get_parser_name ();
     return &tp_normal_rep;
@@ -313,7 +397,8 @@ prog_language_rep::get_color (tree t, int start, int end) {
   int pos= 0;
   while (pos <= start) {
     if (inline_comment_parser.can_parse (s, pos)) {
-      return decode_color (lan_name, encode_color ("comment"));
+      if (!inline_comment_requires_space || (pos == 0 || is_space (s[pos - 1])))
+        return decode_color (lan_name, encode_color ("comment"));
     }
     pos++;
   }