Adding initial implementation of here-comments (block comments) Issue #368

lib/grammar.js

@@ -158,6 +158,8 @@
     Comment: [
       o("COMMENT", function() {
         return new CommentNode($1);
+      }), o("HERECOMMENT", function() {
+        return new CommentNode($1, 'herecomment');
       })
     ],
     // [The existential operator](http://jashkenas.github.com/coffee-script/#existence).

lib/lexer.js

@@ -187,12 +187,40 @@
         return false;
       }
       quote = match[1].substr(0, 1);
-      doc = this.sanitize_heredoc(match[2] || match[4], quote);
+      doc = this.sanitize_heredoc(match[2] || match[4], {
+        quote: quote
+      });
       this.interpolate_string(("" + quote + doc + quote));
       this.line += count(match[1], "\n");
       this.i += match[1].length;
       return true;
     };
+    // Matches and conumes comments. We pass through comments into JavaScript,
+    // so they're treated as real tokens, like any other part of the language.
+    Lexer.prototype.comment_token = function comment_token() {
+      var comment, i, lines, match;
+      if (!(match = this.chunk.match(COMMENT))) {
+        return false;
+      }
+      if (match[3]) {
+        comment = this.sanitize_heredoc(match[3], {
+          herecomment: true
+        });
+        this.token('HERECOMMENT', compact(comment.split(MULTILINER)));
+      } else {
+        lines = compact(match[1].replace(COMMENT_CLEANER, '').split(MULTILINER));
+        i = this.tokens.length - 1;
+        if (this.unfinished()) {
+          while (this.tokens[i] && !include(LINE_BREAK, this.tokens[i][0])) {
+            i -= 1;
+          }
+        }
+        this.tokens.splice(i + 1, 0, ['COMMENT', lines, this.line], ['TERMINATOR', '\n', this.line]);
+      }
+      this.line += count(match[1], "\n");
+      this.i += match[1].length;
+      return true;
+    };
     // Matches JavaScript interpolated directly into the source via backticks.
     Lexer.prototype.js_token = function js_token() {
       var script;
@@ -244,25 +272,6 @@
       delimited = __slice.call(arguments, 0, _d - 0);
       return balanced_string(this.chunk, delimited);
     };
-    // Matches and conumes comments. We pass through comments into JavaScript,
-    // so they're treated as real tokens, like any other part of the language.
-    Lexer.prototype.comment_token = function comment_token() {
-      var comment, i, lines;
-      if (!(comment = this.match(COMMENT, 1))) {
-        return false;
-      }
-      this.line += (comment.match(MULTILINER) || []).length;
-      lines = compact(comment.replace(COMMENT_CLEANER, '').split(MULTILINER));
-      i = this.tokens.length - 1;
-      if (this.unfinished()) {
-        while (this.tokens[i] && !include(LINE_BREAK, this.tokens[i][0])) {
-          i -= 1;
-        }
-      }
-      this.tokens.splice(i + 1, 0, ['COMMENT', lines, this.line], ['TERMINATOR', '\n', this.line]);
-      this.i += comment.length;
-      return true;
-    };
     // Matches newlines, indents, and outdents, and determines which is which.
     // If we can detect that the current line is continued onto the the next line,
     // then the newline is suppressed:
@@ -276,7 +285,7 @@
       if (!(indent = this.match(MULTI_DENT, 1))) {
         return false;
       }
-      this.line += indent.match(MULTILINER).length;
+      this.line += count(indent, "\n");
       this.i += indent.length;
       prev = this.prev(2);
       size = indent.match(LAST_DENTS).reverse()[0].match(LAST_DENT)[1].length;
@@ -405,12 +414,16 @@
         }
       }
     };
-    // Sanitize a heredoc by escaping internal double quotes and erasing all
-    // external indentation on the left-hand side.
-    Lexer.prototype.sanitize_heredoc = function sanitize_heredoc(doc, quote) {
+    // Sanitize a heredoc or herecomment by escaping internal double quotes and
+    // erasing all external indentation on the left-hand side.
+    Lexer.prototype.sanitize_heredoc = function sanitize_heredoc(doc, options) {
       var indent;
       indent = (doc.match(HEREDOC_INDENT) || ['']).sort()[0];
-      return doc.replace(new RegExp("^" + indent, 'gm'), '').replace(MULTILINER, "\\n").replace(new RegExp(quote, 'g'), '\\"');
+      doc = doc.replace(new RegExp("^" + indent, 'gm'), '');
+      if (options.herecomment) {
+        return doc;
+      }
+      return doc.replace(MULTILINER, "\\n").replace(new RegExp(options.quote, 'g'), '\\"');
     };
     // Tag a half assignment.
     Lexer.prototype.tag_half_assignment = function tag_half_assignment(tag) {
@@ -636,7 +649,7 @@
   INTERPOLATION = /^\$([a-zA-Z_@]\w*(\.\w+)*)/;
   OPERATOR = /^([+\*&|\/\-%=<>:!?]+)([ \t]*)/;
   WHITESPACE = /^([ \t]+)/;
-  COMMENT = /^(((\n?[ \t]*)?#[^\n]*)+)/;
+  COMMENT = /^((\n?[ \t]*)?#{3}(?!#)\n*([\s\S]*?)\n*([ \t]*)#{3}|((\n?[ \t]*)?#[^\n]*)+)/;
   CODE = /^((-|=)>)/;
   MULTI_DENT = /^((\n([ \t]*))+)(\.)?/;
   LAST_DENTS = /\n([ \t]*)/g;

lib/nodes.js

@@ -510,8 +510,9 @@
   // CoffeeScript passes through comments as JavaScript comments at the
   // same position.
   exports.CommentNode = (function() {
-    CommentNode = function CommentNode(lines) {
+    CommentNode = function CommentNode(lines, type) {
       this.lines = lines;
+      this.type = type;
       this;
       return this;
     };
@@ -520,7 +521,13 @@
       return this;
     };
     CommentNode.prototype.compile_node = function compile_node(o) {
-      return ("" + this.tab + "//") + this.lines.join(("\n" + this.tab + "//"));
+      var sep;
+      if (this.type === 'herecomment') {
+        sep = '\n' + this.idt(1);
+        return "" + this.tab + "/*" + sep + (this.lines.join(sep)) + "\n" + this.tab + "*/";
+      } else {
+        return ("" + this.tab + "//") + this.lines.join(("\n" + this.tab + "//"));
+      }
     };
     return CommentNode;
   })();

src/grammar.coffee

@@ -165,6 +165,7 @@ grammar: {
   # positions in which they can occur in the grammar.
   Comment: [
     o "COMMENT",                                -> new CommentNode $1
+    o "HERECOMMENT",                            -> new CommentNode $1, 'herecomment'
   ]
 
   # [The existential operator](http://jashkenas.github.com/coffee-script/#existence).

src/lexer.coffee

@@ -126,12 +126,29 @@ exports.Lexer: class Lexer
   heredoc_token: ->
     return false unless match: @chunk.match(HEREDOC)
     quote: match[1].substr 0, 1
-    doc: @sanitize_heredoc match[2] or match[4], quote
+    doc: @sanitize_heredoc match[2] or match[4], {quote}
     @interpolate_string "$quote$doc$quote"
     @line: + count match[1], "\n"
     @i: + match[1].length
     true
 
+  # Matches and conumes comments. We pass through comments into JavaScript,
+  # so they're treated as real tokens, like any other part of the language.
+  comment_token: ->
+    return false unless match: @chunk.match(COMMENT)
+    if match[3]
+      comment: @sanitize_heredoc match[3], {herecomment: true}
+      @token 'HERECOMMENT', compact comment.split MULTILINER
+    else
+      lines: compact match[1].replace(COMMENT_CLEANER, '').split MULTILINER
+      i: @tokens.length - 1
+      if @unfinished()
+        i: - 1 while @tokens[i] and not include LINE_BREAK, @tokens[i][0]
+      @tokens.splice(i + 1, 0, ['COMMENT', lines, @line], ['TERMINATOR', '\n', @line])
+    @line: + count match[1], "\n"
+    @i: + match[1].length
+    true
+
   # Matches JavaScript interpolated directly into the source via backticks.
   js_token: ->
     return false unless starts @chunk, '`'
@@ -165,19 +182,6 @@ exports.Lexer: class Lexer
   balanced_token: (delimited...) ->
     balanced_string @chunk, delimited
 
-  # Matches and conumes comments. We pass through comments into JavaScript,
-  # so they're treated as real tokens, like any other part of the language.
-  comment_token: ->
-    return false unless comment: @match COMMENT, 1
-    @line: + (comment.match(MULTILINER) or []).length
-    lines: compact comment.replace(COMMENT_CLEANER, '').split MULTILINER
-    i: @tokens.length - 1
-    if @unfinished()
-      i: - 1 while @tokens[i] and not include LINE_BREAK, @tokens[i][0]
-    @tokens.splice(i + 1, 0, ['COMMENT', lines, @line], ['TERMINATOR', '\n', @line])
-    @i: + comment.length
-    true
-
   # Matches newlines, indents, and outdents, and determines which is which.
   # If we can detect that the current line is continued onto the the next line,
   # then the newline is suppressed:
@@ -190,7 +194,7 @@ exports.Lexer: class Lexer
   # can close multiple indents, so we need to know how far in we happen to be.
   line_token: ->
     return false unless indent: @match MULTI_DENT, 1
-    @line: + indent.match(MULTILINER).length
+    @line: + count indent, "\n"
     @i   : + indent.length
     prev: @prev(2)
     size: indent.match(LAST_DENTS).reverse()[0].match(LAST_DENT)[1].length
@@ -286,13 +290,14 @@ exports.Lexer: class Lexer
       else
         @tag 1, 'PROPERTY_ACCESS'
 
-  # Sanitize a heredoc by escaping internal double quotes and erasing all
-  # external indentation on the left-hand side.
-  sanitize_heredoc: (doc, quote) ->
+  # Sanitize a heredoc or herecomment by escaping internal double quotes and
+  # erasing all external indentation on the left-hand side.
+  sanitize_heredoc: (doc, options) ->
     indent: (doc.match(HEREDOC_INDENT) or ['']).sort()[0]
-    doc.replace(new RegExp("^" +indent, 'gm'), '')
-       .replace(MULTILINER, "\\n")
-       .replace(new RegExp(quote, 'g'), '\\"')
+    doc: doc.replace(new RegExp("^" +indent, 'gm'), '')
+    return doc if options.herecomment
+    doc.replace(MULTILINER, "\\n")
+       .replace(new RegExp(options.quote, 'g'), '\\"')
 
   # Tag a half assignment.
   tag_half_assignment: (tag) ->
@@ -477,7 +482,7 @@ HEREDOC       : /^("{6}|'{6}|"{3}\n?([\s\S]*?)\n?([ \t]*)"{3}|'{3}\n?([\s\S]*?)\
 INTERPOLATION : /^\$([a-zA-Z_@]\w*(\.\w+)*)/
 OPERATOR      : /^([+\*&|\/\-%=<>:!?]+)([ \t]*)/
 WHITESPACE    : /^([ \t]+)/
-COMMENT       : /^(((\n?[ \t]*)?#[^\n]*)+)/
+COMMENT       : /^((\n?[ \t]*)?#{3}(?!#)\n*([\s\S]*?)\n*([ \t]*)#{3}|((\n?[ \t]*)?#[^\n]*)+)/
 CODE          : /^((-|=)>)/
 MULTI_DENT    : /^((\n([ \t]*))+)(\.)?/
 LAST_DENTS    : /\n([ \t]*)/g

src/nodes.coffee

@@ -358,15 +358,20 @@ children ValueNode, 'base', 'properties'
 # same position.
 exports.CommentNode: class CommentNode extends BaseNode
 
-  constructor: (lines) ->
+  constructor: (lines, type) ->
     @lines: lines
+    @type: type
     this
 
   make_return: ->
     this
 
   compile_node: (o) ->
-    "$@tab//" + @lines.join("\n$@tab//")
+    if @type is 'herecomment'
+      sep: '\n' + @idt(1)
+      "$@tab/*$sep${ @lines.join(sep) }\n$@tab*/"
+    else
+      "$@tab//" + @lines.join("\n$@tab//")
 
 statement CommentNode
 

test/test_comments.coffee

@@ -54,3 +54,8 @@ test:
   'test'
 
 ok test is 'test test test'
+
+###
+  This is a here-comment.
+  Kind of like a heredoc.
+###