A script to sort my Chrome bookmarks

process_bookmarks.sh

#!/bin/bash

set -e
set -u

BOOKMARKS_TXT="bookmarks.txt"

if [[ -z "$1" ]]; then
	echo "Usage: $0 $BOOKMARKS_TXT"
	exit -1
fi

rm *.txt
cat "$1" | grep 'HREF.*ADD' -o | cut -d= -f2 | awk '{print $1}' | sort -u > $BOOKMARKS_TXT

function sort_inplace_file() {
	cat $1 | sort -u > $1.sort
	mv $1.sort $1
}

function remove_lines_in_first_that_are_present_in_second() {
	#comm requires the input files to be sorted
	sort_inplace_file $1
	sort_inplace_file $2

	comm -23 $1 $2 > $1.new
	mv $1.new $1
}

function simple_tag() {
	echo "simple_tag $1"
	cat $BOOKMARKS_TXT | grep -i "$1" | tee $1.txt > /dev/null
	remove_lines_in_first_that_are_present_in_second $BOOKMARKS_TXT $1.txt
}

function append_tag_to_file() {
	echo "append_tag_to_file: $1 => $2"
	#XXX: for some reasone ">>" here crashed occasionally
	cat $BOOKMARKS_TXT | grep -i "$1" | tee -a $2.txt > /dev/null
	remove_lines_in_first_that_are_present_in_second $BOOKMARKS_TXT $2.txt
}

SIMPLE_TAGS=(
	'habrahabr'
	'wikipedia'
	'vk.com'
	'algo'
	'android'
	'blogspot'
	'blog'
	'book'
	'dsp'
	'erlang'
	'emacs'
	'fpga'
	'github'
	'gpu'
	'gsoc'
	'haskell'
	'livejournal'
	'linux'
	'math'
	'microsoft'
	'music'
	'ocaml'
	'opencv'
	'piano'
	'radio'
	'slideshare'
	'speakerdeck'
	'sport'
	'stackexchange'
	'wordpress'
)

for tag in ${SIMPLE_TAGS[@]}; do
	simple_tag "$tag"
done

append_tag_to_file "~" "personal_academia"

append_tag_to_file "qrz" "radio"

append_tag_to_file "stackoverflow" "stackexchange"

append_tag_to_file "fantlab" "fiction"

append_tag_to_file "opengl" "opengl"
append_tag_to_file "webgl" "opengl"
append_tag_to_file "glsl" "opengl"

append_tag_to_file "itep\.ru" "dsp"

append_tag_to_file "gamedev" "gamedev"
append_tag_to_file "gamerendering" "gamedev"
append_tag_to_file "gamasutra" "gamedev"
append_tag_to_file "geeks3d" "gamedev"
append_tag_to_file "render" "gamedev"

append_tag_to_file "electron" "electronics"
append_tag_to_file "ee" "electronics"
append_tag_to_file "eda" "electronics"

append_tag_to_file "asic" "fpga"
append_tag_to_file "verilog" "fpga"
append_tag_to_file "vhdl" "fpga"
append_tag_to_file "yosys" "fpga"
append_tag_to_file "uvm" "fpga"
append_tag_to_file "ovm" "fpga"

append_tag_to_file "courses" "courses"
append_tag_to_file "lectures" "courses"

append_tag_to_file "\.ac\.at" "edu"
append_tag_to_file "\.ac\.jp" "edu"
append_tag_to_file "\.ac\.kr" "edu"
append_tag_to_file "\.ac\.ru" "edu"
append_tag_to_file "\.ac\.uk" "edu"
append_tag_to_file "\.au\.dk" "edu"
append_tag_to_file "\.edu" "edu"
append_tag_to_file "\.ens\.fr" "edu"
append_tag_to_file "\.ethz\.ch" "edu"
append_tag_to_file "\.kuleven\.be" "edu"
append_tag_to_file "\.lmu\.de" "edu"
append_tag_to_file "\.thm\.de" "edu"
append_tag_to_file "\.dtu\.dk" "edu"
append_tag_to_file "\.ku\.dk" "edu"
append_tag_to_file "\.kth\.se" "edu"
append_tag_to_file "\.lth\.se" "edu"
append_tag_to_file "\.mipt\.ru" "edu"
append_tag_to_file "\.msu\.ru" "edu"
append_tag_to_file "\.ru\.nl" "edu"
append_tag_to_file "\.tue\.nl" "edu"
append_tag_to_file "\.ualberta\.ca" "edu"
append_tag_to_file "\.ubi\.pt" "edu"
append_tag_to_file "\.ufpe\.br" "edu"
append_tag_to_file "\.unibo\.it" "edu"
append_tag_to_file "\.uregina\.ca" "edu"
append_tag_to_file "\.uu\.nl" "edu"
append_tag_to_file "\.uu\.se" "edu"
append_tag_to_file "\.vu\.nl" "edu"