rewrite comments

Fixes: #260
jeff-hykin · Jun 26, 2019 · b140eae · b140eae
1 parent bb966d7
commit b140eae
Show file tree

Hide file tree

Showing 4 changed files with 182 additions and 126 deletions.
diff --git a/language_tags/cpp.txt b/language_tags/cpp.txt
@@ -7,7 +7,7 @@ comment.block.preprocessor.else-branch.cpp
 comment.block.preprocessor.else-branch.in-block.cpp
 comment.block.preprocessor.if-branch.cpp
 comment.block.preprocessor.if-branch.in-block.cpp
-comment.line.banner.cpp
+comment.line.block.cpp
 comment.line.double-slash.cpp
 constant.character.escape.cpp
 constant.character.escape.line-continuation.cpp
@@ -397,6 +397,7 @@ meta.arguments.operator.noexcept.cpp
 meta.arguments.operator.sizeof.cpp
 meta.arguments.operator.typeid.cpp
 meta.asm.cpp
+meta.banner.character.cpp
 meta.block.class.cpp
 meta.block.cpp
 meta.block.enum.cpp
@@ -475,7 +476,7 @@ meta.tail.union.cpp
 meta.template.call.cpp
 meta.template.definition.cpp
 meta.toc-list.banner.block.cpp
-meta.toc-list.banner.line.cpp
+meta.toc-list.banner.double-slash.cpp
 meta.using-namespace.cpp
 punctuation.accessor.attribute.cpp
 punctuation.definition.begin.bracket.square.array.type.cpp

diff --git a/source/languages/cpp/generate.rb b/source/languages/cpp/generate.rb
@@ -2327,69 +2327,104 @@ def generateBlockFinder( name:"", tag_as:"", start_pattern:nil, needs_semicolon:
         ),
         includes: [:function_body_context]
     )
-    cpp_grammar[:comments_context] = {
-        patterns: [
-            {
-                captures: {
-                    "1" => {
-                        name: "meta.toc-list.banner.block"
-                    }
-                },
-                match: "^/\\* =(\\s*.*?)\\s*= \\*/$\\n?",
-                name: "comment.block"
-            },
-            {
-                begin: "/\\*",
-                beginCaptures: {
-                    "0" => {
-                        name: "punctuation.definition.comment.begin"
-                    }
-                },
-                end: "\\*/",
-                endCaptures: {
-                    "0" => {
-                        name: "punctuation.definition.comment.end"
-                    }
-                },
-                name: "comment.block"
-            },
-            {
-                captures: {
-                    "1" => {
-                        name: "meta.toc-list.banner.line"
-                    }
-                },
-                match: "^// =(\\s*.*?)\\s*=\\s*$\\n?",
-                name: "comment.line.banner"
-            },
-            {
-                begin: "(^[ \\t]+)?(?=//)",
-                beginCaptures: {
-                    "1" => {
-                        name: "punctuation.whitespace.comment.leading"
-                    }
-                },
-                end: "(?!\\G)",
-                patterns: [
-                    {
-                        begin: "//",
-                        beginCaptures: {
-                            "0" => {
-                                name: "punctuation.definition.comment"
-                            }
-                        },
-                        end: "(?=\\n)",
-                        name: "comment.line.double-slash",
-                        patterns: [
-                            {
-                                include: "#line_continuation_character"
-                            }
-                        ]
-                    }
-                ]
-            }
-        ]
-        }
+    cpp_grammar[:comments_context] = [
+        #
+        # file banner
+        # this matches emacs style file banners ex: /* = foo.c = */
+        # a file banner is a <comment start> <some spaces> <banner start> <some spaces>
+        # <comment contents> <banner end> <some spaces> <comment end>
+        # single line
+        newPattern(
+            should_fully_match: ["// ### test.c ###", "//=test.c - test util ="],
+            should_not_partial_match: ["// ### test.c #=#", "//=test.c - test util ~~~"],
+            match: /^/.maybe(@spaces).then(
+                match: newPattern(
+                    match: /\/\//,
+                    tag_as: "punctuation.definition.comment"
+                ).maybe(
+                    match: @spaces,
+                ).then(
+                    match: oneOrMoreOf(match: /[#;\/=*C~]+/, dont_back_track?: true).lookAheadToAvoid(/[#;\/=*C~]/),
+                    tag_as: "meta.banner.character",
+                    reference: "banner_part"
+                ).maybe(@spaces).then(/.+/).maybe(@spaces).backReference("banner_part")
+                .maybe(@spaces).then(/(?:\n|$)/),
+                tag_as: "comment.line.double-slash",
+            ),
+            # tag is a legacy name
+            tag_as: "meta.toc-list.banner.double-slash",
+        ),
+        # multi line comment
+        newPattern(
+            should_fully_match: ["/* ### test.c ###*/", "/*=test.c - test util =*/"],
+            should_not_partial_match: ["/* ### test.c #=# */", "/*=test.c - test util ~~~*/"],
+            match: /^/.maybe(@spaces).then(
+                match: newPattern(
+                    match: /\/\*/,
+                    tag_as: "punctuation.definition.comment"
+                ).maybe(
+                    match: @spaces,
+                    quantity_preference: :as_few_as_possible
+                ).then(
+                    match: oneOrMoreOf(match: /[#;\/=*C~]+/, dont_back_track?: true).lookAheadToAvoid(/[#;\/=*C~]/),
+                    tag_as: "meta.banner.character",
+                    reference: "banner_part"
+                ).maybe(@spaces).then(/.+/).maybe(@spaces).backReference("banner_part")
+                .maybe(@spaces).then(/\*\//),
+                tag_as: "comment.line.block",
+            ),
+            # tag is a legacy name
+            tag_as: "meta.toc-list.banner.block",
+        ),
+        #
+        # single line comment
+        #
+        PatternRange.new(
+            start_pattern: maybe(
+                # consumes extra space character
+                match: newPattern(
+                    match: @spaces,
+                    dont_back_track?: true
+                ),
+                tag_as: "punctuation.whitespace.comment.leading"
+            ).lookBehindToAvoid(/[\\]/).then(
+                match: /\/\//,
+                tag_as: "comment.line.double-slash punctuation.definition.comment"
+            ),
+            end_pattern: lookBehindToAvoid(/[\\]/).then(/\n/),
+            tag_content_as: "comment.line.double-slash",
+            includes: [
+               :line_continuation_characterm
+            ]
+        ),
+        #
+        # multi line comment
+        #
+        PatternRange.new(
+            start_pattern: newPattern(
+                should_fully_match: ["    /*", "/*"],
+                should_partial_match: ["/***/"],
+                match: maybe(
+                    # consumes extra space character
+                    match: newPattern(
+                        match: @spaces,
+                        dont_back_track?: true
+                    ),
+                    tag_as: "punctuation.whitespace.comment.leading"
+                ).lookBehindToAvoid(/[\\]/).then(
+                    match: /\/\*/,
+                    tag_as: "comment.block punctuation.definition.comment.begin"
+                )),
+            end_pattern: lookBehindToAvoid(/[\\]/).then(
+                match: /\*\//,
+                tag_as: "comment.block punctuation.definition.comment.end"
+            ),
+            tag_content_as: "comment.block",
+            includes: [
+               :line_continuation_character
+            ]
+        ),
+    ]
     cpp_grammar[:disabled] = {
         begin: "^\\s*#\\s*if(n?def)?\\b.*$",
         end: "^\\s*#\\s*endif\\b",

diff --git a/syntaxes/cpp.tmLanguage.json b/syntaxes/cpp.tmLanguage.json
@@ -9951,61 +9951,73 @@
     "comments_context": {
       "patterns": [
         {
+          "match": "^\\s*((\\/\\/)\\s*((?>[#;\\/=*C~]+)(?![#;\\/=*C~]))\\s*.+\\s*\\3\\s*(?:\\n|$))",
           "captures": {
             "1": {
-              "name": "meta.toc-list.banner.block.cpp"
+              "name": "comment.line.double-slash.cpp"
+            },
+            "2": {
+              "name": "punctuation.definition.comment.cpp"
+            },
+            "3": {
+              "name": "meta.banner.character.cpp"
             }
           },
-          "match": "^/\\* =(\\s*.*?)\\s*= \\*/$\\n?",
-          "name": "comment.block.cpp"
+          "name": "meta.toc-list.banner.double-slash.cpp"
         },
         {
-          "begin": "/\\*",
-          "beginCaptures": {
-            "0": {
-              "name": "punctuation.definition.comment.begin.cpp"
-            }
-          },
-          "end": "\\*/",
-          "endCaptures": {
-            "0": {
-              "name": "punctuation.definition.comment.end.cpp"
+          "match": "^\\s*((\\/\\*)\\s*?((?>[#;\\/=*C~]+)(?![#;\\/=*C~]))\\s*.+\\s*\\3\\s*\\*\\/)",
+          "captures": {
+            "1": {
+              "name": "comment.line.block.cpp"
+            },
+            "2": {
+              "name": "punctuation.definition.comment.cpp"
+            },
+            "3": {
+              "name": "meta.banner.character.cpp"
             }
           },
-          "name": "comment.block.cpp"
+          "name": "meta.toc-list.banner.block.cpp"
         },
         {
-          "captures": {
+          "contentName": "comment.line.double-slash.cpp",
+          "begin": "((?>\\s+)?)(?<![\\\\])(\\/\\/)",
+          "beginCaptures": {
             "1": {
-              "name": "meta.toc-list.banner.line.cpp"
+              "name": "punctuation.whitespace.comment.leading.cpp"
+            },
+            "2": {
+              "name": "comment.line.double-slash.cpp punctuation.definition.comment.cpp"
             }
           },
-          "match": "^// =(\\s*.*?)\\s*=\\s*$\\n?",
-          "name": "comment.line.banner.cpp"
+          "end": "(?<![\\\\])\\n",
+          "patterns": [
+            {
+              "include": "#line_continuation_characterm"
+            }
+          ]
         },
         {
-          "begin": "(^[ \\t]+)?(?=//)",
+          "contentName": "comment.block.cpp",
+          "begin": "(((?>\\s+)?)(?<![\\\\])(\\/\\*))",
           "beginCaptures": {
-            "1": {
+            "2": {
               "name": "punctuation.whitespace.comment.leading.cpp"
+            },
+            "3": {
+              "name": "comment.block.cpp punctuation.definition.comment.begin.cpp"
+            }
+          },
+          "end": "(?<![\\\\])(\\*\\/)",
+          "endCaptures": {
+            "1": {
+              "name": "comment.block.cpp punctuation.definition.comment.end.cpp"
             }
           },
-          "end": "(?!\\G)",
           "patterns": [
             {
-              "begin": "//",
-              "beginCaptures": {
-                "0": {
-                  "name": "punctuation.definition.comment.cpp"
-                }
-              },
-              "end": "(?=\\n)",
-              "name": "comment.line.double-slash.cpp",
-              "patterns": [
-                {
-                  "include": "#line_continuation_character"
-                }
-              ]
+              "include": "#line_continuation_character"
             }
           ]
         }

diff --git a/syntaxes/cpp.tmLanguage.yaml b/syntaxes/cpp.tmLanguage.yaml
@@ -5333,39 +5333,47 @@
     - include: "#function_body_context"
   comments_context:
     patterns:
-    - captures:
+    - match: "^\\s*((\\/\\/)\\s*((?>[#;\\/=*C~]+)(?![#;\\/=*C~]))\\s*.+\\s*\\3\\s*(?:\\n|$))"
+      captures:
         '1':
-          name: meta.toc-list.banner.block.cpp
-      match: "^/\\* =(\\s*.*?)\\s*= \\*/$\\n?"
-      name: comment.block.cpp
-    - begin: "/\\*"
-      beginCaptures:
-        '0':
-          name: punctuation.definition.comment.begin.cpp
-      end: "\\*/"
-      endCaptures:
-        '0':
-          name: punctuation.definition.comment.end.cpp
-      name: comment.block.cpp
-    - captures:
+          name: comment.line.double-slash.cpp
+        '2':
+          name: punctuation.definition.comment.cpp
+        '3':
+          name: meta.banner.character.cpp
+      name: meta.toc-list.banner.double-slash.cpp
+    - match: "^\\s*((\\/\\*)\\s*?((?>[#;\\/=*C~]+)(?![#;\\/=*C~]))\\s*.+\\s*\\3\\s*\\*\\/)"
+      captures:
         '1':
-          name: meta.toc-list.banner.line.cpp
-      match: "^// =(\\s*.*?)\\s*=\\s*$\\n?"
-      name: comment.line.banner.cpp
-    - begin: "(^[ \\t]+)?(?=//)"
+          name: comment.line.block.cpp
+        '2':
+          name: punctuation.definition.comment.cpp
+        '3':
+          name: meta.banner.character.cpp
+      name: meta.toc-list.banner.block.cpp
+    - contentName: comment.line.double-slash.cpp
+      begin: "((?>\\s+)?)(?<![\\\\])(\\/\\/)"
       beginCaptures:
         '1':
           name: punctuation.whitespace.comment.leading.cpp
-      end: "(?!\\G)"
+        '2':
+          name: comment.line.double-slash.cpp punctuation.definition.comment.cpp
+      end: "(?<![\\\\])\\n"
       patterns:
-      - begin: "//"
-        beginCaptures:
-          '0':
-            name: punctuation.definition.comment.cpp
-        end: "(?=\\n)"
-        name: comment.line.double-slash.cpp
-        patterns:
-        - include: "#line_continuation_character"
+      - include: "#line_continuation_characterm"
+    - contentName: comment.block.cpp
+      begin: "(((?>\\s+)?)(?<![\\\\])(\\/\\*))"
+      beginCaptures:
+        '2':
+          name: punctuation.whitespace.comment.leading.cpp
+        '3':
+          name: comment.block.cpp punctuation.definition.comment.begin.cpp
+      end: "(?<![\\\\])(\\*\\/)"
+      endCaptures:
+        '1':
+          name: comment.block.cpp punctuation.definition.comment.end.cpp
+      patterns:
+      - include: "#line_continuation_character"
   disabled:
     begin: "^\\s*#\\s*if(n?def)?\\b.*$"
     end: "^\\s*#\\s*endif\\b"