Fix word-boundary regexs. (#228)

toitware · Aug 28, 2023 · 3df5553 · 3df5553
1 parent fc33eb6
commit 3df5553
Show file tree

Hide file tree

Showing 2 changed files with 59 additions and 57 deletions.
diff --git a/vscode/syntaxes/toit.tmLanguage.json b/vscode/syntaxes/toit.tmLanguage.json
@@ -23,7 +23,7 @@
   "repository": {
     "import-section": {
       "name": "meta.import.toit",
-      "begin": "^import\\b",
+      "begin": "^import(?!-)\\b",
       "end": "^(?!\\s)",
       "beginCaptures": {
         "0": {
@@ -36,11 +36,11 @@
         },
         {
           "name": "constant.language.import-export-all.show.toit",
-          "match": "\\bshow\\b"
+          "match": "\\b(?<!-)show(?!-)\\b"
         },
         {
           "name": "constant.language.import-export-all.as.toit",
-          "match": "\\bas\\b"
+          "match": "\\b(?<!)as(?!-)\\b"
         },
         {
           "name": "constant.language.import-export-all.import_all.toit",
@@ -50,7 +50,7 @@
     },
     "export-section": {
       "name": "meta.export.toit",
-      "begin": "^export\\b",
+      "begin": "^export(?!-)\\b",
       "end": "^(?!\\s)",
       "beginCaptures": {
         "0": {
@@ -69,7 +69,7 @@
     },
     "class-section": {
       "name": "meta.class.toit",
-      "begin": "^(?:(abstract)[ ]+)?(class|monitor|interface)\\b",
+      "begin": "^(?:(abstract)[ ]+)?(class|monitor|interface)(?!-)\\b",
       "end": "^(?=[^\\s/]|/[^/*])",
       "beginCaptures": {
         "1": {
@@ -98,7 +98,7 @@
         },
         {
           "name": "meta.class.signature",
-          "begin": "(\\b\\w+\\b)",
+          "begin": "(\\b(?<!-)[\\w-]+(?!-)\\b)",
           "end": ":",
           "beginCaptures": {
             "1": {
@@ -111,7 +111,7 @@
             },
             {
               "name": "meta.extends.clause.toit",
-              "match": "\\b(extends)\\s+([\\p{L}_][\\w+-]*)\\b",
+              "match": "\\b(?<!-)(extends)\\s+([\\p{L}_][\\w-]*)(?!-)\\b",
               "captures": {
                 "1": {
                   "name": "keyword.control.extends.toit"
@@ -123,11 +123,11 @@
             },
             {
               "name": "keyword.control.extends.toit",
-              "match": "\\bextends\\b"
+              "match": "\\b(?<!-)extends(?!-)\\b"
             },
             {
               "name": "keyword.control.implements.toit",
-              "match": "\\bimplements\\b"
+              "match": "\\b(?<!-)implements(?!-)\\b"
             },
             {
               "include": "#type-name"
@@ -185,11 +185,11 @@
         },
         {
           "name": "keyword.control.toit",
-          "match": "(\\bstatic\\b|\\babstract\\b|\\boperator\\b)"
+          "match": "\\b(?<!-)(static|abstract|operator)(?!-)\\b"
         },
         {
           "name": "meta.member.signature",
-          "begin": "(\\bconstructor\\b)(\\.[\\p{L}_][\\w-]*)?|([\\p{L}_][\\w-]*[=]?|==|<<|>>>|>>|<=|>=|<|>|\\+|-|\\*|/|%|\\^|&|\\||\\[\\]\\=|\\[\\]|\\[\\.\\.\\])",
+          "begin": "(?<!-)\\b(constructor)(?!-)\\b(\\.[\\p{L}_][\\w-]*)?|([\\p{L}_][\\w-]*[=]?|==|<<|>>>|>>|<=|>=|<|>|\\+|-|\\*|/|%|\\^|&|\\||\\[\\]\\=|\\[\\]|\\[\\.\\.\\])",
           "end": "(?=\\:(?![\\p{L}_])|^\\s{0,4}(?:[^\\s/]|/[^/*]))",
           "beginCaptures": {
             "1": {
@@ -333,31 +333,31 @@
       "patterns": [
         {
           "name": "keyword.control.toit",
-          "match": "\\b(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)\\b"
+          "match": "\\b(?<!-)(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)(?!-)\\b"
         },
         {
           "name": "keyword.control.pseudo.toit",
-          "match": "\\bunreachable\\b"
+          "match": "\\b(?<!-)unreachable(?!-)\\b"
         },
         {
           "name": "keyword.control.type_check.toit",
-          "match": "\\b(is|as)\\b"
+          "match": "\\b(?<!-)(is|as)(?!-)\\b"
         }
       ]
     },
     "invalid_non_expression": {
       "patterns": [
         {
           "name": "invalid.illegal.non_expression.toit",
-          "match": "\\b(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)\\b"
+          "match": "\\b(?<!-)(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)(?!-)\\b"
         },
         {
           "name": "invalid.illegal.non_expression.toit",
-          "match": "\\b(is|as)\\b"
+          "match": "\\b(?<!-)(is|as)(?!-)\\b"
         },
         {
           "name": "invalid.illegal.non_expression.toit",
-          "match": "\\b(super|this)\\b"
+          "match": "\\b(?<!-)(super|this)(?!-)\\b"
         }
       ]
     },
@@ -544,11 +544,11 @@
       "patterns": [
         {
           "name": "constant.language.toit",
-          "match": "\\b(null|true|false)\\b"
+          "match": "\\b(?<!-)(null|true|false)(?!-)\\b"
         },
         {
           "name": "constant.numeric.capitalized_user_constants.toit",
-          "match": "\\b_*[A-Z](-?[A-Z0-9_])+\\b"
+          "match": "\\b(?<!-)_*[A-Z](-?[A-Z0-9_])+(?!-)\\b"
         }
       ]
     },
@@ -595,15 +595,15 @@
       "patterns": [
         {
           "name": "constant.numeric.dec.toit",
-          "match": "(?<!\\w)-?([0-9](_(?=[0-9]))?)+\\b"
+          "match": "(?<!\\w)-?([0-9](_(?=[0-9]))?)+(?!-)\\b"
         },
         {
           "name": "constant.numeric.dec.hex.toit",
-          "match": "(?<!\\w)-?0[xX]([0-9a-fA-F](_(?=[0-9a-fA-F]))?)+\\b"
+          "match": "(?<!\\w)-?0[xX]([0-9a-fA-F](_(?=[0-9a-fA-F]))?)+(?!-)\\b"
         },
         {
           "name": "constant.numeric.dec.bin.toit",
-          "match": "(?<!\\w)-?0[bB]([01](_(?=[01]))?)+\\b"
+          "match": "(?<!\\w)-?0[bB]([01](_(?=[01]))?)+(?!-)\\b"
         }
       ]
     },
@@ -717,7 +717,7 @@
       "patterns": [
         {
           "name": "variable.language.special.toit",
-          "match": "\\b(this|it|super)\\b"
+          "match": "\\b(?<!-)(this|it|super)(?!-)\\b"
         }
       ]
     },
@@ -737,35 +737,35 @@
       "patterns": [
         {
           "name": "entity.name.type.toit",
-          "match": "\\b_?[A-Z][0-9]*\\b[?]?"
+          "match": "\\b(?<!-)_?[A-Z][0-9]*(?!-)\\b[?]?"
         },
         {
           "name": "entity.name.type.toit",
-          "match": "\\b_?[A-Z][A-Z_-]*[a-z][\\w-]*\\b[?]?"
+          "match": "\\b(?<!-)_?[A-Z][A-Z_-]*[a-z][\\w-]*(?!-)\\b[?]?"
         },
         {
           "name": "entity.name.type.shorts.toit",
-          "match": "\\b(int|bool|float|string)\\b[?]?"
+          "match": "\\b(?<!-)(int|bool|float|string)(?!-)\\b[?]?"
         },
         {
           "name": "entity.name.type.any_none.toit",
-          "match": "\\b(any|none)\\b"
+          "match": "\\b(?<!-)(any|none)(?!-)\\b"
         }
       ]
     },
     "variable": {
       "patterns": [
         {
           "name": "entity.name.function.call.toit",
-          "match": "\\b[\\p{L}_][\\w-]*\\b"
+          "match": "\\b(?<!-)[\\p{L}_][\\w-]*(?!-)\\b"
         }
       ]
     },
     "primitive": {
       "patterns": [
         {
           "name": "support.function.builtin.toit",
-          "match": "\\#primitive\\b"
+          "match": "\\#primitive(?!-)\\b"
         }
       ]
     },

diff --git a/vscode/syntaxes/toit.tmLanguage.yaml b/vscode/syntaxes/toit.tmLanguage.yaml
@@ -17,6 +17,8 @@ comment: >
 # followed by any number of letters, digits, underscores or dashes. Toit actually doesn't
 # allow multiple dashes in a row (or the dash to be the last character), but we don't
 # enforce that here. The syntax highlighter only needs to render valid code correctly.
+# For most `\b` (word-boundary) we use a negative-look-ahead or negative-look-behind
+# to ensure that the boundary is not a dash: `(?!-)\b` or `(?<!-)\b`.
 
 name: Toit
 
@@ -32,23 +34,23 @@ patterns:
 repository:
   import-section:
     name: meta.import.toit
-    begin: ^import\b    # Section starts with an `import` (no indentation).
+    begin: ^import(?!-)\b    # Section starts with an `import` (no indentation).
     end: ^(?!\s)        # Ends a non-indented different section. (Note the negative look-ahead to avoid consuming the next token).
     beginCaptures:
       0:
         name: constant.language.import-export-all.import.toit
     patterns:
     - include: "#comment"
     - name: constant.language.import-export-all.show.toit
-      match: \bshow\b
+      match: \b(?<!-)show(?!-)\b
     - name: constant.language.import-export-all.as.toit
-      match: \bas\b
+      match: \b(?<!)as(?!-)\b
     - name: constant.language.import-export-all.import_all.toit
       match: \*
 
   export-section:
     name: meta.export.toit
-    begin: ^export\b    # Section starts with an `export` (no indentation).
+    begin: ^export(?!-)\b    # Section starts with an `export` (no indentation).
     end: ^(?!\s)        # Ends a non-indented different section. (Note the negative look-ahead to avoid consuming the next token).
     beginCaptures:
       0:
@@ -60,7 +62,7 @@ repository:
 
   class-section:
     name: meta.class.toit
-    begin: ^(?:(abstract)[ ]+)?(class|monitor|interface)\b
+    begin: ^(?:(abstract)[ ]+)?(class|monitor|interface)(?!-)\b
     end: ^(?=[^\s/]|/[^/*])        # Ends with a non-indented different section. (Note the negative look-ahead to avoid consuming the next token).
     beginCaptures:
       1:
@@ -80,24 +82,24 @@ repository:
     # "inherited-class". Otherwise we just mark "extends" and "implements" as
     # keywords and the rest as types.
     - name: meta.class.signature
-      begin: (\b\w+\b)
+      begin: (\b(?<!-)[\w-]+(?!-)\b)
       end: ":"
       beginCaptures:
         1:
           name: storage.type.class.toit
       patterns:
       - include: "#comment"
       - name: meta.extends.clause.toit
-        match: \b(extends)\s+([\p{L}_][\w+-]*)\b
+        match: \b(?<!-)(extends)\s+([\p{L}_][\w-]*)(?!-)\b
         captures:
           1:
             name: keyword.control.extends.toit
           2:
             name: entity.other.inherited-class
       - name: keyword.control.extends.toit
-        match: \bextends\b
+        match: \b(?<!-)extends(?!-)\b
       - name: keyword.control.implements.toit
-        match: \bimplements\b
+        match: \b(?<!-)implements(?!-)\b
       - include: "#type-name"
 
   toplevel-section:
@@ -133,11 +135,11 @@ repository:
     patterns:
     - include: "#comment"
     - name: keyword.control.toit
-      match: (\bstatic\b|\babstract\b|\boperator\b)
+      match: \b(?<!-)(static|abstract|operator)(?!-)\b
     - name: meta.member.signature
       # A constructor (potentially named), or a function, setter, or operator.
       # We don't handle operators specially, and just require the symbols to be after some whitespace.
-      begin: (\bconstructor\b)(\.[\p{L}_][\w-]*)?|([\p{L}_][\w-]*[=]?|==|<<|>>>|>>|<=|>=|<|>|\+|-|\*|/|%|\^|&|\||\[\]\=|\[\]|\[\.\.\])
+      begin: (?<!-)\b(constructor)(?!-)\b(\.[\p{L}_][\w-]*)?|([\p{L}_][\w-]*[=]?|==|<<|>>>|>>|<=|>=|<|>|\+|-|\*|/|%|\^|&|\||\[\]\=|\[\]|\[\.\.\])
       # Ends (and thus starts the body) with a ":" (unless that's for a block),
       #   or something that isn't intended by 4.
       end: (?=\:(?![\p{L}_])|^\s{0,4}(?:[^\s/]|/[^/*]))
@@ -220,20 +222,20 @@ repository:
   keyword:
     patterns:
     - name: keyword.control.toit
-      match: &keywords \b(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)\b
+      match: &keywords \b(?<!-)(assert|and|or|not|if|for|else|try|finally|call|while|break|continue|throw|static|abstract|return)(?!-)\b
     - name: keyword.control.pseudo.toit
-      match: \bunreachable\b
+      match: \b(?<!-)unreachable(?!-)\b
     - name: keyword.control.type_check.toit
-      match: \b(is|as)\b
+      match: \b(?<!-)(is|as)(?!-)\b
 
   invalid_non_expression: # Some expressions that are not valid in the signature. These will be highlighted as red.
     patterns:
     - name: invalid.illegal.non_expression.toit
       match: *keywords
     - name: invalid.illegal.non_expression.toit
-      match: \b(is|as)\b
+      match: \b(?<!-)(is|as)(?!-)\b
     - name: invalid.illegal.non_expression.toit
-      match: \b(super|this)\b
+      match: \b(?<!-)(super|this)(?!-)\b
 
   expressions:
     patterns:
@@ -333,9 +335,9 @@ repository:
   constant:
     patterns:
     - name: constant.language.toit
-      match: \b(null|true|false)\b
+      match: \b(?<!-)(null|true|false)(?!-)\b
     - name: constant.numeric.capitalized_user_constants.toit
-      match: \b_*[A-Z](-?[A-Z0-9_])+\b
+      match: \b(?<!-)_*[A-Z](-?[A-Z0-9_])+(?!-)\b
 
   number:
     name: constant.numeric.toit
@@ -361,11 +363,11 @@ repository:
   number-decimal:
     patterns:
     - name: constant.numeric.dec.toit
-      match: (?<!\w)-?([0-9](_(?=[0-9]))?)+\b
+      match: (?<!\w)-?([0-9](_(?=[0-9]))?)+(?!-)\b
     - name: constant.numeric.dec.hex.toit
-      match: (?<!\w)-?0[xX]([0-9a-fA-F](_(?=[0-9a-fA-F]))?)+\b
+      match: (?<!\w)-?0[xX]([0-9a-fA-F](_(?=[0-9a-fA-F]))?)+(?!-)\b
     - name: constant.numeric.dec.bin.toit
-      match: (?<!\w)-?0[bB]([01](_(?=[01]))?)+\b
+      match: (?<!\w)-?0[bB]([01](_(?=[01]))?)+(?!-)\b
 
   character:
     patterns:
@@ -438,7 +440,7 @@ repository:
   special-variable:
     patterns:
     - name: variable.language.special.toit
-      match: \b(this|it|super)\b
+      match: \b(?<!-)(this|it|super)(?!-)\b
 
   named-arg:
     patterns:
@@ -456,23 +458,23 @@ repository:
     #  - single capitalized character followed by a number -> also a type: `class C1`.
     #  - capitalized identifier with at least on lower-case character: a type.
     - name: entity.name.type.toit
-      match: \b_?[A-Z][0-9]*\b[?]?
+      match: \b(?<!-)_?[A-Z][0-9]*(?!-)\b[?]?
     - name: entity.name.type.toit
-      match: \b_?[A-Z][A-Z_-]*[a-z][\w-]*\b[?]?
+      match: \b(?<!-)_?[A-Z][A-Z_-]*[a-z][\w-]*(?!-)\b[?]?
     - name: entity.name.type.shorts.toit
-      match: \b(int|bool|float|string)\b[?]?
+      match: \b(?<!-)(int|bool|float|string)(?!-)\b[?]?
     - name: entity.name.type.any_none.toit
-      match: \b(any|none)\b
+      match: \b(?<!-)(any|none)(?!-)\b
 
   variable:
     patterns:
       - name: entity.name.function.call.toit
-        match: \b[\p{L}_][\w-]*\b
+        match: \b(?<!-)[\p{L}_][\w-]*(?!-)\b
 
   primitive:
     patterns:
     - name: support.function.builtin.toit
-      match: \#primitive\b
+      match: \#primitive(?!-)\b
 
   comment:
     patterns: