#!/bin/bash

# This is a simple-minded script to clean up the html
# generated by the InfoLister extension for Firefox,
# to turn it into valid HTML 4.01.
# (It's not really InfoLister's fault that the html 
# isn't valid in the first place -- it has to do with
# unencoded characters like " and & in the extensions' 
# descriptions.)
# It also adds homepage URLs if they're missing and 
# changes some homepage URLS I wasn't happy with.

# Note: InfoLister's output is UTF-8 and this script's
# charset is UTF-8.  The output can be (and is)
# served as ISO-8859-1, but it's just US-ASCII with 
# html entities.  I don't know how bash/my system
# handles the charsets, but this is what works for
# me.

# the file to be cleaned up
INPUTFILE=~/web/remarqs.net/firefox/infolistergenerated.htm.txt

# the cleaned file
OUTPUTFILE=~/web/remarqs.net/firefox/extlist.htm

# sessionfix url not so good, fixing
# errorzilla mod page now bad, use amo instead
# long titles page now just points to amo, so link there
# quickrestart page is 404
# active stop button broken bad url
# firesizer lacks a URL
# objection links to old homepage
# quickrestart has " marks in its description
# console^2 needs the ^2 encoded
# faviconpicker II lacks a URL
# page info button has quotes in description
# throbber button lacks a URL
# bookmark sync & sort URL is a 404
# chromelist URL is lousy
# middleclicknocollapse hacked at by me
# us dept of homeland idiocy URL is lousy
# save complete lacks a URL
# permatabs homepage has nothing on it yet
# strip out disabled extensions

cat $INPUTFILE |
sed 's~\(http://www\.defector\.de/blog/categor\)ies/5-Firefox-Extensions~\1y/firefox-extensions/session-fix/~' |
sed 's~http://www\.jaybaldwin\.com/News\.aspx.*=51~https://addons.mozilla.org/firefox/addon/3336~' |
sed 's~".*\~robin0.*\"~"https://addons.mozilla.org/en-US/firefox/addon/1715"~' |
sed 's~http://quickaddons\.fastspace\.biz/~https://addons.mozilla.org/firefox/addon/3559~' |
sed 's~https://addons\.mozilla\.org/extensions/moreinfo\.php\?.*id=2025~https://addons.mozilla.org/firefox/addon/2025~' |
sed 's~href=\"\"\(.*Firesizer\)~href="https://addons.mozilla.org/firefox/addon/5792"\1~' |
sed 's~http://sourceforge\.net/projects/objection-ext/~http://objection.mozdev.org/~' |
sed 's~title=\"Adds a \"Restart Firefox\" item to the \"File\"~title="Adds a \"Restart Firefox\" item to the \"File\"~' |
sed 's/Console²/Console\²/' |
sed 's/href=\"\"\(.*Favicon[[:space:]]Picker\)/href=\"https:\/\/addons.mozilla.org\/addon\/3176\"\1/' |
sed 's/\"Page Info\"/\"Page Info\"/' |
sed 's/href=\"\"\(.*Throbber\)/href=\"https:\/\/addons.mozilla.org\/addon\/2821\"\1/' |
sed 's/href=\".*syncandsort.*html\"/href=\"https:\/\/addons.mozilla.org\/addon\/2367\"/' |
sed 's/http:\/\/www\.gijsk\.com\//http:\/\/www.gijsk.com\/mozilla\/chromelist\//' |
sed 's/\"http:\/\/visuallinkindicator\.mozdev\.org\/\"/\"bumped.htm\"/' |
sed 's/href=\"http:\/\/www\.taupehat\.com\"/href=\"https:\/\/addons.mozilla.org\/addon\/1501\"/' |
sed 's/href=\"\"\(.*Save[[:space:]]Complete\)/href=\"https:\/\/addons.mozilla.org\/addon\/4723\"\1/' |
sed 's~\"http://www\.donesmart\.com/\"~\"https://addons.mozilla.org/firefox/addon/2558\"~' |
sed 's/.*\[disabled\]<br>$//' \
> $OUTPUTFILE