Simple Python script to let you (easily) apply favicons to bookmarklets.

Credit to http://www.nomachetejuggling.com/2012/03/30/how-to-clean-up-your-chrome-bookmark-bar/ for inspiration and explanation.

Export your Chrome bookmark HTML and do the following:

python show-missing-favicons.py userdata/bookmarks.unprocessed.html | python autodetect-favicons.py | python render-importable-html.py > userdata/import-me.html

This will create userdata/import-me.html which can be imported in to Chrome. Importing this file in to Chrome's bookmark manager will create a new folder called "Imported" (or "Imported (1)" if it already exists, etc.) which will contain copies of all bookmarklets for which favicon auto-detection succeeded.

Chrome (luckily) does not allow for two bookmarks with identical link data (in this case, javascript code) to have independent favicons; by importing these copies all of the original bookmarklets will have icons set with no extra work. It is safe to delete the "Imported" folder - note that Chrome Bookmark Manager will not necessarily show the icons until reloaded but the icons should be immediately visible on the bookmark bar.

• Improved auto-detection of favicons
• Potentially build a library of popular replacements
• Support meta tag declaration of favicon as well as favicon.ico
• More favicon sources
• Support for popular icon libraries such as Silk
• Support for importing local icons disk
• Better integration
• External graphical editor?
• Automating export/import/delete "Imported" folder
• Selenium?
• Doesn't seem to be designed for this
• AutoIt or Automator?
• Windows or Mac only
• Chrome extension
• http://developer.chrome.com/extensions/bookmarks.html has no support for icons
• http://code.google.com/p/chromium/issues/detail?id=59519 tracks this
• Direct editing of profile?
• Requires Chrome to be closed
• Requires re-calculating checksums
• Support for non-bookmarklet bookmarks
• Better documentation

• No threading of icon downloads.
• Minimal caching; will download favicons many times if they are used by many bookmarklets.
• Ignores hash and emits multiple (useless) copies of the same bookmarklet if it is present many times in the source.
• Manual rendering of json is hackish and inelegant.
• ujson dies if there is a trailing whitespace in input; currently using builtin json to circumvent this.

Each tool has a separate, simple, function.

Consumes a bookmark export file and produces a JSON dump of every bookmarklet without an icon. Does some crude matching to see if there's a source domain hiding in the bookmarklet. For example, here's the "popup with tags" bookmarklet from http://pinboard.in/howto/:

javascript:q%3Dlocation.href%3Bif(document.getSelection)%7Bd%3Ddocument.getSelection()%3B%7Delse%7Bd%3D%27%27%3B%7D%3Bp%3Ddocument.title%3Bvoid(open(%27https://pinboard.in/add%3Fshowtags%3Dyes%26url%3D%27%2BencodeURIComponent(q)%2B%27%26description%3D%27%2BencodeURIComponent(d)%2B%27%26title%3D%27%2BencodeURIComponent(p),%27Pinboard%27,%27toolbar%3Dno,scrollbars%3Dyes,width%3D750,height%3D700%27))%3B

This contains //pinboard.in/ - and .in is a valid TLD - so the autodetected domain is pinboard.in.

Bookmarklets are ignored if they already have an icon.

The output (JSON) contains for-future-use fields.

Consumes the JSON product of show-missing-favicons.py in order to download favicons.

The output is the same as show-missing-favicons.py but the "icon" field will be populated with the detected domain's favicon.

Currently lacks both threading and caching; may also download the same favicon several times as a result.

Consumes the JSON product of autodetect-favicons.py and produces an importable HTML bookmark file. Only bookmarklets with icons will be present in the output.