Support for other languages

.verb.md

@@ -78,6 +78,42 @@ Results in:
        original: 'var foo = "bar";' } ] }
 ```
 
+**Different comment styles and other languages**
+
+The `options` parameter can be used to specify a file-type and load comment-patterns for
+different languages. With the following Handlebars-file:
+
+```hbs
+{{!
+  This is a handlebars-file
+}}
+{{title}}
+```
+
+the following snippet
+
+```js
+var extract = require('extract-comments');
+
+// pass a string of javascript, CSS, LESS etc
+extract(string, { filename: 'abc.hbs'});
+```
+
+results in
+
+```js
+{
+      '1': {
+        begin: 1,
+        code: '{{title}}',
+        codeStart: 5,
+        content: 'This is a handlebars-file\n',
+        end: 3
+      }
+}
+```
+
+
 ## Extracting from files
 
 Prior to v0.7.0, there was a method to extract code comments from files. Here is the equivalent code to accomplish the same thing:

README.md

@@ -49,7 +49,6 @@ Results in:
 
 _(The reason the key is the starting line number is that it's easy to use this format with templates)_
 
-
 **Customize output**
 
 ```js
@@ -105,14 +104,17 @@ function extractComments(patterns, opts) {
 ```
 
 ## Related
- * [parse-comments](https://github.com/jonschlinkert/parse-comments): Parse code comments from JavaScript or any language that uses the same format.
- * [code-context](https://github.com/jonschlinkert/code-context): Parse a string of javascript to determine the context for functions, variables and comments based on the code that follows.
- * [esprima-extract-comments](https://github.com/jonschlinkert/esprima-extract-comments): Extract code comments from string or from a glob of files using esprima.
+
+* [code-context](https://github.com/jonschlinkert/code-context): Parse a string of javascript to determine the context for… [more](https://github.com/jonschlinkert/code-context)
+* [esprima-extract-comments](https://github.com/jonschlinkert/esprima-extract-comments): Extract code comments from string or from a glob of… [more](https://github.com/jonschlinkert/esprima-extract-comments)
+* [parse-comments](https://github.com/jonschlinkert/parse-comments): Parse code comments from JavaScript or any language that uses… [more](https://github.com/jonschlinkert/parse-comments)
 
 ## Contributing
+
 Pull requests and stars are always welcome. For bugs and feature requests, [please create an issue](https://github.com/jonschlinkert/extract-comments/issues)
 
 ## Run tests
+
 Install dev dependencies:
 
 ```bash
@@ -123,15 +125,22 @@ npm i -d && npm test
 
 **Jon Schlinkert**
 
-+ [github/jonschlinkert](https://github.com/jonschlinkert)
-+ [twitter/jonschlinkert](http://twitter.com/jonschlinkert) 
+[github/jonschlinkert](https://github.com/jonschlinkert)
+[twitter/jonschlinkert](http://twitter.com/jonschlinkert)
 
 ## License
-Copyright (c) 2014-2015 Jon Schlinkert  
-Released under the MIT license
+
+Copyright (c) 2014-2015 Jon Schlinkert
+Released under the MIT license.
 
 ***
 
-_This file was generated by [verb-cli](https://github.com/assemble/verb-cli) on April 02, 2015._
+_This file was generated by [verb-cli](https://github.com/assemble/verb-cli) on May 02, 2015._
 
 [map-files]: https://github.com/jonschlinkert/map-files
+
+<!-- reflinks generated by verb-reflinks plugin -->
+
+[verb]: https://github.com/assemble/verb
+[template]: https://github.com/jonschlinkert/template
+[assemble]: http://assemble.io

fixtures/abc.hbs

@@ -0,0 +1,4 @@
+{{!
+  This is a handlebars-file
+}}
+{{title}}

index.js

@@ -8,6 +8,8 @@
 'use strict';
 
 var isWhitespace = require('is-whitespace');
+var commentPattern = require('comment-patterns');
+var qm = require('quotemeta');
 
 /**
  * expose `extract`
@@ -31,47 +33,73 @@ module.exports = extract;
  * });
  * ```
  *
- * @param  {String} `string`
+ * @param  {String} `str` the source code to extract comments from
+ * @param  {Function} [`fn`] the transformer-callback
+ * @param  {Object} `options` an object with optional paramters
+ * @param  {String} [`options.filename`] a file name to determine the comment-patterns
  * @return {Object} Object of code comments.
+ *
  * @api public
  */
 
-function extract(str, fn) {
-  var start = /^\/\*\*?/;
-  var middle = /^\*([^*][^\/])*/;
-  var end = /^\*\//;
+function extract(str, fn, options) {
+  if (typeof fn !== 'function' && typeof options === "undefined") {
+    options = fn;
+    fn = null;
+  }
+  // default filename is a javascript file (for backwards compatibility)
+  var filename = (options && options.filename) || "abc.js";
+  var patterns = buildRegexes(commentPattern(filename));
 
   var lines = str.split(/[\r\n]/);
-  var len = lines.length, i = 0, m;
+  var len = lines.length, i = 0;
   var comments = {};
-  var isComment = false, afterCount;
-  var from, to, b, o = {};
+  var afterCount;
+  var b, o = {};
+
+  // See the `checkStart`-function for possible values.
+  var currentComment = null;
 
   while (i < len) {
     var line = lines[i++].trim();
-
-    if (!isComment && start.test(line)) {
-      afterCount = 0;
-      isComment = true;
-      o = {begin: null, end: null};
-      o.begin = b = i;
-      o.code = '';
-      o.content = '';
-    }
-
-    if (isComment && end.test(line)) {
-      o.end = i;
-      comments[b] = o;
-      isComment = false;
-    }
-
-    if (isComment && i > b) {
-      if (isMiddle(line)) {
-        o.content += stripStars(line) + '\n';
+    if (currentComment == null) {
+      // Check if any of the comment-patterns matches
+      currentComment = checkStart(line,patterns);
+      if (currentComment != null) {
+        // We are now inside a comment
+        afterCount = 0;
+        o = {begin: null, end: null};
+        o.begin = b = i;
+        o.code = '';
+        o.content = '';
+      }
+    } else {
+      // Inside a comment...
+      if (currentComment.multi) {
+        // Inside a multiline comment
+        if (currentComment.pattern.end.test(line)) {
+          // End of comment reached
+          o.end = i;
+          comments[b] = o;
+          currentComment = null;
+        } else if (currentComment.pattern.middle.test(line)) {
+          o.content += line.replace(currentComment.pattern.middle,"") + '\n';
+        }
+      } else {
+        // Last line was a single-line comment
+        if (currentComment.pattern.test(line)) {
+          o.content += line.replace(currentComment.pattern,"") + '\n';
+        } else {
+          // End of comment was reached one line before
+          o.end = i-1;
+          comments[b] = o;
+          currentComment = null;
+        }
       }
     }
 
-    if (!isComment && o.end && i > o.end && afterCount < 2) {
+
+    if (currentComment == null && o.end && i > o.end && afterCount < 2) {
       if (!isWhitespace(line)) {
         o.codeStart = i;
       }
@@ -89,34 +117,70 @@ function extract(str, fn) {
     }
   }
   return comments;
-};
+}
 
 /**
- * Strip the leading `*` from a line, ensuring
- * not to eat too many whitespaces after the delimiter.
+ * Check, whether the current line starts a comment.
+ * Return the pattern-set marking the current comment:
+ * This can be either `null`, if the current line is no comment,
+ * or `{multi:true, pattern:{start:,middle:,end:}}` for multiline-comments
+ * or `{multi:false, pattern: 'string'}` for single-line-commments
+ * @param {String} `line` the current line of code
+ * @param {Object} `patterns` the comment-patterns for the current language
  */
-
-function stripStars(str) {
-  str = str.replace(/^\s*/, '');
-  if (str.charAt(0) === '/') {
-    str = str.slice(1);
-  }
-  if (str.charAt(0) === '*') {
-    str = str.slice(1);
+function checkStart(line, patterns) {
+
+  var i;
+  // Check for multiline-comment. `pattern.multiLine` is never `null` due to the `buildRegexes`-function
+  for (i = 0; i < patterns.multiLine.length; i++) {
+    if (patterns.multiLine[i].start.test(line)) {
+      return {
+        multi: true,
+        pattern: patterns.multiLine[i]
+      }
+    }
   }
-  if (str.charAt(0) === ' ') {
-    str = str.slice(1);
+  // Check for single-line-comment. `pattern.singleLine` is never `null` due to the `buildRegexes`-function
+  for (i = 0; i < patterns.singleLine.length; i++) {
+    if (patterns.singleLine[i].test(line)) {
+      return {
+        multi: false,
+        pattern: patterns.singleLine[i]
+      }
+    }
   }
-  return str;
+  // No comment start detected
+  return null;
 }
 
 /**
- * Detect if the given line is in the middle
- * of a comment.
+ * Convert the string-based comment-delimiters from the `comment-patterns` result
+ * into regex-based comment-patterns.
+ * @param {Object} `patterns` the `comment-patterns`
+ * @returns {{multiLine: Array, singleLine: Array}} regexes for multi-line and single-line comments
  */
-
-function isMiddle(str) {
-  return typeof str === 'string'
-   && str.charAt(0) === '*'
-   && str.charAt(1) !== '/';
+function buildRegexes(patterns) {
+  var i = 0;
+  var result = {
+    multiLine: [],
+    singleLine: []
+  };
+  if (patterns.multiLineComment) {
+    for (i = 0; i < patterns.multiLineComment.length; i++) {
+
+      var ml = patterns.multiLineComment[i];
+      result.multiLine.push({
+        start: new RegExp("^" + qm(ml.start)),
+        middle: new RegExp("^" + qm(ml.middle)+"\\s?"),
+        end: new RegExp("^" + qm(ml.end))
+      });
+    }
+  }
+  if (patterns.singleLineComment) {
+    for (i = 0; i < patterns.singleLineComment.length; i++) {
+      var sl = patterns.singleLineComment[i];
+      result.singleLine.push(new RegExp("^" + qm(sl)));
+    }
+  }
+  return result;
 }

package.json

@@ -29,7 +29,9 @@
     "test": "mocha"
   },
   "dependencies": {
-    "is-whitespace": "^0.3.0"
+    "comment-patterns": "0.0.7",
+    "is-whitespace": "^0.3.0",
+    "quotemeta": "0.0.0"
   },
   "devDependencies": {
     "code-context": "^0.5.0",

test.js

@@ -73,4 +73,16 @@ describe('extract comments', function () {
     var comments = extract(str);
     comments['1274'].codeStart.should.equal(1291);
   });
+
+  it('should extract Handlebars-style comments when the filename is set to abc.hbs', function() {
+    extract(read("abc.hbs"),{ filename: 'abc.hbs' }).should.eql( {
+      '1': {
+        begin: 1,
+        code: '{{title}}',
+        codeStart: 5,
+        content: 'This is a handlebars-file\n',
+        end: 3
+      }
+    });
+  })
 });