Bug 1423487 - Support multiple authors in RSS feeds.

Tag #1273

4 files changed, 122 insertions, 53 deletions

diff --git a/mailnews/extensions/newsblog/content/FeedItem.js b/mailnews/extensions/newsblog/content/FeedItem.js
index 09e4eb861..5863c420a 100644
--- a/mailnews/extensions/newsblog/content/FeedItem.js
+++ b/mailnews/extensions/newsblog/content/FeedItem.js
@@ -22,7 +22,10 @@ FeedItem.prototype =
   content: null,
   enclosures: [],
   title: null,
-  author: "anonymous",
+  // Author must be angle bracket enclosed to function as an addr-spec, in the
+  // absence of an addr-spec portion of an RFC5322 email address, as other
+  // functionality (gloda search) depends on this.
+  author: "<anonymous>",
   inReplyTo: "",
   keywords: [],
   mURL: null,
@@ -245,12 +248,10 @@ FeedItem.prototype =
   {
     FeedUtils.log.trace("FeedItem.writeToFolder: " + this.identity +
                         " writing to message folder " + this.feed.name);
-    // Convert the title to UTF-16 before performing our HTML entity
-    // replacement reg expressions.
-    let title = this.title;
 
     // The subject may contain HTML entities.  Convert these to their unencoded
     // state. i.e. &amp; becomes '&'.
+    let title = this.title;
     title = this.mParserUtils.convertToPlainText(
         title,
         Ci.nsIDocumentEncoder.OutputSelectionOnly |
@@ -274,32 +275,11 @@ FeedItem.prototype =
       ("References: " + this.inReplyTo + "\n" +
        "In-Reply-To: " + this.inReplyTo + "\n") : "";
 
-    // If there are keywords (categories), create the headers. In the case of
-    // a longer than RFC5322 recommended line length, create multiple folded
-    // lines (easier to parse than multiple Keywords headers).
-    let keywordsStr = "";
-    if (this.keywords.length)
-    {
-      let HEADER = "Keywords: ";
-      let MAXLEN = 78;
-      keywordsStr = HEADER;
-      let keyword;
-      let keywords = [].concat(this.keywords);
-      let lines = [];
-      while (keywords.length)
-      {
-        keyword = keywords.shift();
-        if (keywordsStr.length + keyword.length > MAXLEN)
-        {
-          lines.push(keywordsStr)
-          keywordsStr = " ".repeat(HEADER.length);
-        }
-        keywordsStr += keyword + ",";
-      }
-      keywordsStr = keywordsStr.replace(/,$/,"\n");
-      lines.push(keywordsStr)
-      keywordsStr = lines.join("\n");
-    }
+    // Support multiple authors in From.
+    let fromStr = this.createHeaderStrFromArray("From: ", this.author);
+
+    // If there are keywords (categories), create the headers.
+    let keywordsStr = this.createHeaderStrFromArray("Keywords: ", this.keywords);
 
     // Escape occurrences of "From " at the beginning of lines of
     // content per the mbox standard, since "From " denotes a new
@@ -323,7 +303,7 @@ FeedItem.prototype =
       'Received: by localhost; ' + FeedUtils.getValidRFC5322Date() + '\n' +
       'Date: ' + this.mDate + '\n' +
       'Message-Id: ' + this.normalizeMessageID(this.id) + '\n' +
-      'From: ' + this.author + '\n' +
+      fromStr +
       'MIME-Version: 1.0\n' +
       'Subject: ' + this.title + '\n' +
       inreplytoHdrsStr +
@@ -369,6 +349,49 @@ FeedItem.prototype =
   },
 
 /**
+ * Create a header string from an array. Intended for comma separated headers
+ * like From or Keywords. In the case of a longer than RFC5322 recommended
+ * line length, create multiple folded lines (easier to parse than multiple
+ * headers).
+ *
+ * @param  {String} headerName          - Name of the header.
+ * @param  {String[]} headerItemsArray  - An Array of strings to concatenate.
+ *
+ * @returns {String}                    - The header string.
+ */
+  createHeaderStrFromArray(headerName, headerItemsArray) {
+    let headerStr = "";
+    if (!headerItemsArray || headerItemsArray.length == 0) {
+      return headerStr;
+    }
+
+    const HEADER = headerName;
+    const LINELENGTH = 78;
+    const MAXLINELENGTH = 990;
+    let items = [].concat(headerItemsArray);
+    let lines = [];
+    headerStr = HEADER;
+    while (items.length) {
+      let item = items.shift();
+      if (headerStr.length + item.length > LINELENGTH &&
+          headerStr.length > HEADER.length) {
+        lines.push(headerStr);
+        headerStr = " ".repeat(HEADER.length);
+      }
+
+      headerStr += headerStr.length + item.length > MAXLINELENGTH ?
+                     item.substr(0, MAXLINELENGTH - headerStr.length) + "…, " :
+                     item + ", ";
+    }
+
+    headerStr = headerStr.replace(/,\s$/, "\n");
+    lines.push(headerStr);
+    headerStr = lines.join("\n");
+
+    return headerStr;
+  },
+
+/**
  * Autotag messages.
  *
  * @param  nsIMsgDBHdr aMsgDBHdr - message to tag
diff --git a/mailnews/extensions/newsblog/content/FeedUtils.jsm b/mailnews/extensions/newsblog/content/FeedUtils.jsm
index 6d5e64dd2..f8368ac25 100644
--- a/mailnews/extensions/newsblog/content/FeedUtils.jsm
+++ b/mailnews/extensions/newsblog/content/FeedUtils.jsm
@@ -42,6 +42,7 @@ var FeedUtils = {
   },
 
   DC_NS: "http://purl.org/dc/elements/1.1/",
+  get DC_PUBLISHER()    { return this.rdf.GetResource(this.DC_NS + "publisher"); },
   get DC_CREATOR()      { return this.rdf.GetResource(this.DC_NS + "creator") },
   get DC_SUBJECT()      { return this.rdf.GetResource(this.DC_NS + "subject") },
   get DC_DATE()         { return this.rdf.GetResource(this.DC_NS + "date") },
diff --git a/mailnews/extensions/newsblog/content/feed-parser.js b/mailnews/extensions/newsblog/content/feed-parser.js
index 660333422..03fc72b22 100644
--- a/mailnews/extensions/newsblog/content/feed-parser.js
+++ b/mailnews/extensions/newsblog/content/feed-parser.js
@@ -222,9 +222,10 @@ FeedParser.prototype =
       tags = this.childrenByTagNameNS(itemNode, nsURI, "author");
       if (!tags)
         tags = this.childrenByTagNameNS(itemNode, FeedUtils.DC_NS, "creator");
-      item.author = this.getNodeValue(tags ? tags[0] : null) ||
-                    aFeed.title ||
-                    item.author;
+      let author = this.getNodeValue(tags ? tags[0] : null) ||
+                   aFeed.title;
+      author = this.cleanAuthorName(author);
+      item.author = author ? ["<" + author + ">"] : item.author;
 
       tags = this.childrenByTagNameNS(itemNode, nsURI, "pubDate");
       if (!tags || !this.getNodeValue(tags[0]))
@@ -386,10 +387,12 @@ FeedParser.prototype =
       item.id = item.url;
       item.url = this.validLink(item.url);
 
-      item.author = this.getRDFTargetValue(ds, itemResource, FeedUtils.DC_CREATOR) ||
-                    this.getRDFTargetValue(ds, channel, FeedUtils.DC_CREATOR) ||
-                    aFeed.title ||
-                    item.author;
+      let author = this.getRDFTargetValue(ds, itemResource, FeedUtils.DC_CREATOR) ||
+                   this.getRDFTargetValue(ds, channel, FeedUtils.DC_CREATOR) ||
+                   aFeed.title;
+      author = this.cleanAuthorName(author);
+      item.author = author ? ["<" + author + ">"] : item.author;
+
       item.date = this.getRDFTargetValue(ds, itemResource, FeedUtils.DC_DATE) ||
                   item.date;
       item.content = this.getRDFTargetValueFormatted(ds, itemResource,
@@ -608,7 +611,7 @@ FeedParser.prototype =
         continue;
       }
 
-      // XXX Support multiple authors.
+      // Support multiple authors.
       tags = this.childrenByTagNameNS(itemNode, FeedUtils.ATOM_IETF_NS, "source");
       let source = tags ? tags[0] : null;
 
@@ -618,22 +621,42 @@ FeedParser.prototype =
       if (!tags)
         tags = this.childrenByTagNameNS(channel, FeedUtils.ATOM_IETF_NS, "author");
 
-      let authorEl = tags ? tags[0] : null;
-
-      let author = "";
-      if (authorEl)
-      {
-        tags = this.childrenByTagNameNS(authorEl, FeedUtils.ATOM_IETF_NS, "name");
+      let authorTags = tags || [];
+      let authors = [];
+      for (let authorTag of authorTags) {
+        let author = "";
+        tags = this.childrenByTagNameNS(authorTag, FeedUtils.ATOM_IETF_NS, "name");
         let name = this.getNodeValue(tags ? tags[0] : null);
-        tags = this.childrenByTagNameNS(authorEl, FeedUtils.ATOM_IETF_NS, "email");
+        tags = this.childrenByTagNameNS(authorTag, FeedUtils.ATOM_IETF_NS, "email");
         let email = this.getNodeValue(tags ? tags[0] : null);
-        if (name)
-          author = name + (email ? " <" + email + ">" : "");
-        else if (email)
+        if (name) {
+          name = this.cleanAuthorName(name);
+          if (email) {
+            if (!email.match(/^<.*>$/)) {
+              email = " <" + email + ">";
+            }
+            author = name + email;
+          } else {
+            author = "<" + name + ">";
+          }
+        } else if (email) {
           author = email;
+        }
+        if (author) {
+          authors.push(author);
+        }
       }
 
-      item.author = author || item.author || aFeed.title;
+      if (authors.length == 0) {
+        tags = this.childrenByTagNameNS(channel, FeedUtils.DC_NS, "publisher");
+        let author = this.getNodeValue(tags ? tags[0] : null) ||
+                     aFeed.title;
+        author = this.cleanAuthorName(author);
+        item.author = author ? ["<" + author + ">"] : item.author;
+      } else {
+        item.author = authors;
+      }
+      FeedUtils.log.trace("FeedParser.parseAsAtomIETF: author(s) - " + item.author);
 
       tags = this.childrenByTagNameNS(itemNode, FeedUtils.ATOM_IETF_NS, "updated");
       if (!tags || !this.getNodeValue(tags[0]))
@@ -801,6 +824,29 @@ FeedParser.prototype =
     return content ? content : null;
   },
 
+  /**
+   * Return a cleaned up author name value.
+   *
+   * @param {String} authorString  - A string.
+   * @returns {String}             - A clean string value.
+   */
+  cleanAuthorName(authorString) {
+    if (!authorString) {
+      return "";
+    }
+    FeedUtils.log.trace("FeedParser.cleanAuthor: author1 - " + authorString);
+    let author = authorString.replace(/[\n\r\t]+/g, " ")
+                             .replace(/"/g, '\\"')
+                             .trim();
+    // If the name contains special chars, quote it.
+    if (author.match(/[<>@,"]/)) {
+      author = '"' + author + '"';
+    }
+    FeedUtils.log.trace("FeedParser.cleanAuthor: author2 - " + author);
+
+    return author;
+  },
+
   getRDFTargetValue: function(ds, source, property)
   {
     let nodeValue = this.getRDFTargetValueRaw(ds, source, property);
diff --git a/mailnews/extensions/newsblog/content/feed-subscriptions.xul b/mailnews/extensions/newsblog/content/feed-subscriptions.xul
index d6f4ea18f..d8dff29ee 100644
--- a/mailnews/extensions/newsblog/content/feed-subscriptions.xul
+++ b/mailnews/extensions/newsblog/content/feed-subscriptions.xul
@@ -17,12 +17,11 @@
 ]>
 
 <window id="subscriptionsDialog"
+        xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul"
+        xmlns:html="http://www.w3.org/1999/xhtml"
         flex="1"
         title="&feedSubscriptions.label;"
         windowtype="Mail:News-BlogSubscriptions"
-        xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul"
-        xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
-        xmlns:nc="http://home.netscape.com/NC-rdf#"
         persist="width height screenX screenY sizemode"
         onload="FeedSubscriptions.onLoad();"
         onclose="return FeedSubscriptions.onClose();"