Merge pull request #11 from openimap/bodystructure

Resolves #10: Emit the bodystructure during parsing

Author: felixhammerl

README.md

@@ -68,6 +68,16 @@ parser.onbody = function(node, chunk){
 };
 ```
 
+#### Bodystructure
+
+Bodystructure is the original raw message stripped of bodies and multipart preambles. MIME stores like to store the bodystructure of MIME content in raw (loss-less) form, to later run through a MIME parser to answer IMAP or WebDAV type queries. Bodystructure is emitted in chunks of String. Define `onbodystructure` to catch these chunks:
+
+```javascript
+parser.onbodystructure = function(chunk){
+    console.log(chunk);
+};
+```
+
 #### Parse end
 
 When the parsing is finished, `onend` is called
@@ -80,7 +90,7 @@ parser.onend = function(){
 
 ## Quirks
 
-This seems like asynchronous but actually it is not. So always define `onheader`, `onbody` and `onend` before writing the first chunk of data to the parser.
+This seems like asynchronous but actually it is not. So always define `onheader`, `onbody`, `onbodystructure`, and `onend` before writing the first chunk of data to the parser.
 
 **message/rfc822** is automatically parsed if the mime part does not have a `Content-Disposition: attachment` header, otherwise it will be emitted as a regular attachment (as one long Uint8Array value).
 

package.json

@@ -1,6 +1,6 @@
 {
   "name": "emailjs-mime-parser",
-  "version": "1.0.0",
+  "version": "1.0.1-prerelease",
   "homepage": "https://github.com/emailjs/emailjs-mime-parser",
   "description": "Parse a mime tree, no magic included.",
   "author": "Andris Reinman <[email protected]>",

src/emailjs-mime-parser.js

@@ -323,6 +323,14 @@
      */
     MimeParser.prototype.onbody = function() {};
 
+    /**
+     * Override this function.
+     * Called when a bodystructure chunk is emitted. The bodystructure is the raw multipart message with empty bodies.
+     * @event
+     * @param {String} chunk Bodystructure chunk
+     */
+    MimeParser.prototype.onbodystructure = function() {};
+
     // NODE PROCESSING
 
     /**
@@ -421,7 +429,7 @@
     // Public methods
 
     /**
-     * Processes an enitre input line
+     * Processes an entire input line
      *
      * @param {String} line Entire input line as 'binary' string
      */
@@ -458,6 +466,7 @@
         if (!line) {
             this._parseHeaders();
             this._parser.onheader(this);
+            this._parser.onbodystructure(this.header.join('\n') + '\n\n');
             this._state = 'BODY';
             return;
         }
@@ -678,7 +687,7 @@
     };
 
     /**
-     * Parses Content-Trasnfer-Encoding value to see if the body needs to be converted
+     * Parses Content-Transfer-Encoding value to see if the body needs to be converted
      * before it can be emitted
      */
     MimeNode.prototype._processContentTransferEncoding = function() {
@@ -700,20 +709,22 @@
 
         if (this._isMultipart) {
             if (line === '--' + this._multipartBoundary) {
+                this._parser.onbodystructure(line + '\n');            
                 if (this._currentChild) {
                     this._currentChild.finalize();
                 }
                 this._currentChild = new MimeNode(this, this._parser);
                 this._childNodes.push(this._currentChild);
             } else if (line === '--' + this._multipartBoundary + '--') {
+                this._parser.onbodystructure(line + '\n');
                 if (this._currentChild) {
                     this._currentChild.finalize();
                 }
                 this._currentChild = false;
             } else if (this._currentChild) {
                 this._currentChild.writeLine(line);
             } else {
-                // Ignore body for multipart
+                // Ignore multipart preamble
             }
         } else if (this._isRfc822) {
             this._currentChild.writeLine(line);

test/mimeparser-unit.js

@@ -948,6 +948,59 @@
                 parser.write(fixture);
                 parser.end();
             });
+            
+            it('should emit bodystructure', function(done) {
+                var fixture =
+                    'MIME-Version: 1.0\n' +
+                    'Content-Type: multipart/mixed;\n' +
+                    ' boundary="------------304E429112E7D6AC36F087A8"\n' +
+                    '\n' +
+                    'This is a multi-part message in MIME format.\n' +
+                    '--------------304E429112E7D6AC36F087A8\n' +
+                    'Content-Type: text/html; charset=utf-8\n' +
+                    'Content-Transfer-Encoding: 7bit\n' +
+                    '\n' +
+                    '<html/>\n' +
+                    '--------------304E429112E7D6AC36F087A8\n' +
+                    'Content-Type: text/plain; charset=UTF-8; x-mac-type="0"; x-mac-creator="0";\n' +
+                    ' name="hello.mime"\n' +
+                    'Content-Transfer-Encoding: base64\n' +
+                    'Content-Disposition: attachment;\n' +
+                    ' filename="hello.mime"\n' +
+                    '\n' +
+                    'SGkgbW9tIQ==\n' +
+                    '--------------304E429112E7D6AC36F087A8--\n';
+
+                var expectedBodystructure =
+                    'MIME-Version: 1.0\n' +
+                    'Content-Type: multipart/mixed;\n' +
+                    ' boundary="------------304E429112E7D6AC36F087A8"\n' +
+                    '\n' +
+                    '--------------304E429112E7D6AC36F087A8\n' +
+                    'Content-Type: text/html; charset=utf-8\n' +
+                    'Content-Transfer-Encoding: 7bit\n' +
+                    '\n' +
+                    '--------------304E429112E7D6AC36F087A8\n' +
+                    'Content-Type: text/plain; charset=UTF-8; x-mac-type="0"; x-mac-creator="0";\n' +
+                    ' name="hello.mime"\n' +
+                    'Content-Transfer-Encoding: base64\n' +
+                    'Content-Disposition: attachment;\n' +
+                    ' filename="hello.mime"\n' +
+                    '\n' +
+                    '--------------304E429112E7D6AC36F087A8--\n';
+
+                var bodystructure = '';
+                parser.onbodystructure = function(chunk) {
+                    bodystructure += chunk;
+                };
+
+                parser.onend = function() {
+                    expect(bodystructure).to.equal(expectedBodystructure);
+                    done();
+                };
+                parser.write(fixture);
+                parser.end();
+            });
         });
     });
 }));