Read bookmarks from 'NETSCAPE-Bookmark-file's that don't have a beginning <DT> tag.

chrome/utility/importer/bookmark_html_reader.cc

Index: chrome/utility/importer/bookmark_html_reader.cc
diff --git a/chrome/utility/importer/bookmark_html_reader.cc b/chrome/utility/importer/bookmark_html_reader.cc
index a9edfebcc0e3d2c7c9ab59c5bb28ea0ecae9b9c6..703b7af20379dc139d1a7f438c6d7b3d8075c59a 100644
--- a/chrome/utility/importer/bookmark_html_reader.cc
+++ b/chrome/utility/importer/bookmark_html_reader.cc
@@ -91,6 +91,18 @@ void DataURLToFaviconUsage(const GURL& link_url,
 
 namespace bookmark_html_reader {
 
+static void stripDt(std::string& line) {
+  // Remove "<DT>" if the line starts with "<DT>".  This may not occur if
+  // "<DT>" was on the previous line.  Liberally accept entries that do not
+  // have an opening "<DT>" at all.
+  static const char kDtTag[] = "<DT>";
+  if (base::StartsWith(line, kDtTag,
+                       base::CompareCase::INSENSITIVE_ASCII)) {
+    line.erase(0, arraysize(kDtTag) - 1);
+    base::TrimString(line, " ", &line);
+  }
+}
+
 void ImportBookmarksFile(
     const base::Callback<bool(void)>& cancellation_callback,
     const base::Callback<bool(const GURL&)>& valid_url_callback,
@@ -298,16 +310,18 @@ bool ParseCharsetFromLine(const std::string& line, std::string* charset) {
   return false;
 }
 
-bool ParseFolderNameFromLine(const std::string& line,
+bool ParseFolderNameFromLine(std::string line,

[Ilya Sherman, 2016/04/14 02:22:53]

Even if you want to modify the variable in the method implementation, it's
preferred to pass the string by const-ref for the method signature, and then
make a copy inline, as needed.  This way, the interface does not change even if
the implementation changes to no longer need the copy; and moreover people
reading the interface don't spend any time wondering why one of the strings is
passed by const-ref, and the other by value.

[Tom (Use chromium acct), 2016/04/14 17:21:44]

Done.

                              const std::string& charset,
                              base::string16* folder_name,
                              bool* is_toolbar_folder,
                              base::Time* add_date) {
-  const char kFolderOpen[] = "<DT><H3";
+  const char kFolderOpen[] = "<H3";
   const char kFolderClose[] = "</H3>";
   const char kToolbarFolderAttribute[] = "PERSONAL_TOOLBAR_FOLDER";
   const char kAddDateAttribute[] = "ADD_DATE";
 
+  stripDt(line);

[Ilya Sherman, 2016/04/14 02:22:53]

Hmm, why do you need to also strip DT tags here, rather than just in the spot
where HRs are stripped?

[Tom (Use chromium acct), 2016/04/14 17:21:44]

I removed the code that strips the DT near where were strip the HR because some
unit tests were failing that expected the Parse*FromLine functions to work with
a leading <DT>.

+
   if (!base::StartsWith(line, kFolderOpen, base::CompareCase::SENSITIVE))
     return false;
 
@@ -343,7 +357,7 @@ bool ParseFolderNameFromLine(const std::string& line,
   return true;
 }
 
-bool ParseBookmarkFromLine(const std::string& line,
+bool ParseBookmarkFromLine(std::string line,
                            const std::string& charset,
                            base::string16* title,
                            GURL* url,
@@ -351,7 +365,7 @@ bool ParseBookmarkFromLine(const std::string& line,
                            base::string16* shortcut,
                            base::Time* add_date,
                            base::string16* post_data) {
-  const char kItemOpen[] = "<DT><A";
+  const char kItemOpen[] = "<A";
   const char kItemClose[] = "</A>";
   const char kFeedURLAttribute[] = "FEEDURL";
   const char kHrefAttribute[] = "HREF";
@@ -360,6 +374,7 @@ bool ParseBookmarkFromLine(const std::string& line,
   const char kAddDateAttribute[] = "ADD_DATE";
   const char kPostDataAttribute[] = "POST_DATA";
 
+  stripDt(line);
   title->clear();
   *url = GURL();
   *favicon = GURL();
@@ -430,15 +445,16 @@ bool ParseBookmarkFromLine(const std::string& line,
   return true;
 }
 
-bool ParseMinimumBookmarkFromLine(const std::string& line,
+bool ParseMinimumBookmarkFromLine(std::string line,
                                   const std::string& charset,
                                   base::string16* title,
                                   GURL* url) {
-  const char kItemOpen[] = "<DT><A";
+  const char kItemOpen[] = "<A";
   const char kItemClose[] = "</";
   const char kHrefAttributeUpper[] = "HREF";
   const char kHrefAttributeLower[] = "href";
 
+  stripDt(line);
   title->clear();
   *url = GURL();