RULESETS=chrome/content/rules/default.rulesets

INDENT_CHAR=' '

TAG_DEFINITIONS='

SED_TRIM_CMD='

repeat_char() {

[ $2 -gt 0 ] || return

local i

for i in $(seq 1 $2); do printf "$1"; done

}

format_rulesets() {

local IFS tag idepth prebreaks postbreaks

local _indent _pre _post _sed_pre _sed_post _sed_oneshot

# Print pretty banner, very hardcoded

printf '\n'

printf '%15s | %s | %s | %s\n' 'tag name' 'indent' 'prebreak' 'postbreak'

printf '%15s-+-%s-+-%s-+-%s\n' \

'-------------' '------' '--------' '----------'

# Iterate through tags and add appropriate indentation and linebreaks

while read tag idepth prebreaks postbreaks; do

( [ "$tag" = '#' ] || [ ! $postbreaks ] ) && continue # Invalid; skip

unset _indent _pre _post _sed_pre _sed_post _sed_oneshot

printf "%15s | %6d | %8d | %9s\n" "$tag" $idepth $prebreaks $postbreaks

# Special characters (\n) need double escaping when saved to a variable

# since they are dereferenced and break everytime they're passed around.

# bash printf has a %q format character for this, but we're /bin/sh

# Should always be a prepended linebreak unless we opt out with -1

_pre="$(repeat_char '\\n' $((prebreaks+1)))"

_post="$(repeat_char '\\n' $postbreaks)"

_indent="$(repeat_char "$INDENT_CHAR" $idepth)"

# breaks before opening <tags> and <tags/>

_sed_pre="s:<${tag}[ />]:${_pre}${_indent}\0:g;"

# breaks after closing </tags>

_sed_post="s:</${tag}>:\n${_indent}\0${_post}:g;"

# breaks after oneshot <tags/>

_sed_oneshot="s:<${tag}\(/>\| [^>]\+/>\):\0${_post}:g;"

sed -ir "$_sed_pre $_sed_post $_sed_oneshot" $RULESETS

done <<- EOF

echo #padding for some distance after the tag table

}

rulesize() {

wc -c < $RULESETS

}

populate_rulesets() {

local xmlfile

# Under git bash, sed -i issues errors and sets the file "read only"

[ -f "$RULESETS" ] && chmod u+w $RULESETS

printf '<rulesetlibrary gitcommitid="%s">' \

"${GIT_COMMIT_ID:-unset}" > $RULESETS

# Include the filename.xml as the "f" attribute

for xmlfile in chrome/content/rules/*.xml; do

sed "s/<ruleset/\0 f=\"${xmlfile##*/}\"/g" "$xmlfile" >> $RULESETS

done

echo "</rulesetlibrary>" >> $RULESETS

}

flatten_file() {

# Strip *all* control chars; we'll re-add them soon as per tag definitions.

# tr cannot edit in-place so we need to temp, either in a file or a variable

echo "$(tr -d '[:cntrl:]' < $RULESETS | tr -s '[:space:]')" > $RULESETS

# Beware that this *assumes* the used shell accepts variable sizes of >2Mb.

}

cd src

echo "Creating ruleset library..."

populate_rulesets

echo "Removing control characters, whitespace and comments..."

PRECRUSH=$(rulesize)

flatten_file

echo "Formatting..."

format_rulesets

echo "Final touches..."

sed -ir "$SED_TRIM_CMD" $RULESETS

POSTCRUSH=$(rulesize)

printf "Crushed %d bytes of rulesets into %d (delta %d)\n" \

$PRECRUSH $POSTCRUSH $((POSTCRUSH-PRECRUSH))

touch -r chrome/content/rules $RULESETS

unset INDENT_CHAR TAG_DEFINITIONS SED_TRIM_CMD PRECRUSH POSTCRUSH

unset repeat_char format_rulesets rulesize populate_rulesets flatten_file

cd ..