Roll DomDistillerJs (and make updater generate BUG=)

The updater now pulls BUG= lines out of the git log and 
processes/sorts/uniqifies them and prints them for even better 
commit messages.

Picked up changes:
6ef4764 Extend visibility checks to images that are extracted.
6e8c385 Updated Boilerpipe with some more modern tags.
cde81dc Add extract_text_only option
0a31800 filter out invisible elements
82b35ba Fix crash when content extraction fails.

BUG=354157,356336,362085,364356,367243

Review URL: https://codereview.chromium.org/297563002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@271582 0039d316-1c4b-4281-b951-d872f2087c98

third_party/dom_distiller_js/README.chromium

 Name: dom-distiller-js
 URL: https://code.google.com/p/dom-distiller
-Version: a15dc85486
+Version: 6ef4764253
 License: BSD
 Security Critical: yes
 

third_party/dom_distiller_js/package/proto/dom_distiller.proto

@@ -27,4 +27,6 @@ message DomDistillerResult {
 }
 
 message DomDistillerOptions {
+  // Whether to extract only the text (or to include the containing html).
+  optional bool extract_text_only = 1;
 }

third_party/dom_distiller_js/package/proto_gen/third_party/dom_distiller_js/dom_distiller_json_converter.h

@@ -167,6 +167,13 @@ namespace dom_distiller {
           dom_distiller::proto::DomDistillerOptions message;
           const base::DictionaryValue* dict;
           if (!json->GetAsDictionary(&dict)) goto error;
+          if (dict->HasKey("1")) {
+            bool field_value;
+            if (!dict->GetBoolean("1", &field_value)) {
+              goto error;
+            }
+            message.set_extract_text_only(field_value);
+          }
           return message;
 
         error:
@@ -175,6 +182,9 @@ namespace dom_distiller {
 
         static scoped_ptr<base::Value> WriteToValue(const dom_distiller::proto::DomDistillerOptions& message) {
           scoped_ptr<base::DictionaryValue> dict(new base::DictionaryValue());
+          if (message.has_extract_text_only()) {
+            dict->SetBoolean("1", message.extract_text_only());
+          }
           return dict.PassAs<base::Value>();
         }
       };

third_party/dom_distiller_js/update_domdistiller_js.sh

@@ -15,7 +15,8 @@
   dom_distiller_js_package=$dom_distiller_js_path/package
   readme_chromium=$dom_distiller_js_path/README.chromium
   tmpdir=/tmp/domdistiller-$$
-  changes=/tmp/domdistiller.changes
+  changes=$tmpdir/domdistiller.changes
+  bugs=$tmpdir/domdistiller.bugs
   curr_gitsha=$(grep 'Version:' $readme_chromium | awk '{print $2}')
 
   rm -rf $tmpdir
@@ -23,8 +24,27 @@
 
   pushd $tmpdir
   git clone https://code.google.com/p/dom-distiller/ .
+
   new_gitsha=$(git rev-parse --short=10 HEAD)
   git log --oneline ${curr_gitsha}.. > $changes
+
+  echo -n BUG= > $bugs
+
+  # This extracts BUG= lines from the log, extracts the numbers part, removes
+  # whitespace and deletes empty lines. Then, split on ',', sort, uniquify and
+  # rejoin. Finally, remove the trailing ',' and concat to $bugs.
+  git log \
+    | grep BUG= \
+    | sed -e 's/.*BUG=\(.*\)/\1/' -e 's/\s*//g' -e '/^$/d' \
+    | tr ',' '\n' \
+    | sort \
+    | uniq \
+    | tr '\n' ',' \
+    | head --bytes=-1 \
+    >> $bugs
+
+  echo >> $bugs  # add a newline
+
   ant package
   popd
 
@@ -34,13 +54,18 @@
   cp $tmpdir/LICENSE $dom_distiller_js_path/
   sed -i "s/Version: [0-9a-f]*/Version: $new_gitsha/" $readme_chromium
 
+  echo
+  echo
+  echo "---Generated commit message---"
+  echo
   echo "Picked up changes:"
   cat $changes
+  echo
+  cat $bugs
 
   # Run checklicenses.py on the pulled files, but only print the output on
   # failures.
   tools/checklicenses/checklicenses.py $dom_distiller_js_path > $tmpdir/checklicenses.out || cat $tmpdir/checklicenses.out
 
   rm -rf $tmpdir
-  rm $changes
 )