Mercurial > ec-dotfiles
changeset 159:75fe19903b74
Major cleanup
author | edhoprima@gmail.com <edhoprima@gmail.com> |
---|---|
date | Fri, 05 Jun 2009 15:20:36 +0000 |
parents | cba73f6a96bb |
children | 68227a30d0b3 |
files | moefetch.sh |
diffstat | 1 files changed, 199 insertions(+), 165 deletions(-) [+] |
line wrap: on
line diff
--- a/moefetch.sh Tue May 05 17:56:54 2009 +0000 +++ b/moefetch.sh Fri Jun 05 15:20:36 2009 +0000 @@ -13,39 +13,64 @@ # WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN # ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF # OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. - -# Version 0.1-alpha1. Lots of bugs here. Use with care +# +# Lots of bugs here. Use with care # USE WITH CARE - -# what it does: fetch every picture that has the specified tags. - +# +# what it does: fetch every picture that has the specified TAGS. # requirement: wget, libxslt, md5sum (or md5) -# configs -# program additional paths for: cut, sed, wc, md5(sum), wget, xsltproc, grep -extrapath= +# program additional paths for: cut, sed, wc, MD5(sum), wget, xsltproc, grep +ADDITIONAL_PATH= -# md5 calculation, expected output: <32digit md5><space(s)><filename> -# gnu: "md5sum", bsd: "md5 -r" -md5="md5 -r" +# custom md5 path with arguments, expected output: <32digit md5><space(s)><filename> +# Leave empty for "md5sum" (Linux, Solaris), "md5 -r" (*BSD) +MD5= -# server address. Danbooru only! I do not take responsibility of stupidity. -site="moe.imouto.org" +# default server address. Danbooru only! I do not take responsibility of stupidity. +DEFAULT_SITE="moe.imouto.org" # base directory. make sure it's writeable. I do not take responsibility if you don't own the folder and files as no check is done for this one. -# Structure is $basedir/<tags> -basedir="/home/ifail/test" +# Structure is ${BASE_DIR}/<TAGS> +# Absolute path only. +# Leave empty to use whatever folder you're running this at +BASE_DIR="" # not user modifiable from here -GENERATE() -{ +# useless welcome message. Also version +Msg_Welcome() { + MOEFETCHVERSION="0.1-beta" + cat <<EOF +moefetch ${MOEFETCHVERSION} +Copyright (c) 2009 edogawaconan <me@myconan.net> + +EOF +} + +# fatal error handler +Err_Fatal() { + echo "Fatal error: ${1}" + exit 1 +} + +# help message +Err_Help() { + cat <<EOF +Usage: moefetch (quick)fetch|status <TAGS> +EOF + exit 0 +} + +# generate link by transforming xml +Generate_Link() { + cd "${BASE_DIR}/temp" echo echo "Fetching xml file" - wget "http://$site/post/index.xml?tags=$tags&offset=0&limit=100000" -O "$outdir-xml" -e continue=off + wget "http://${SITE}/post/index.xml?tags=${TAGS}&offset=0&limit=100000" -O "${SITE_DIR}-${TARGET_DIR}-xml" -e continue=off echo "Processing XML file..." # xslt evilry - xsltproc - "$outdir-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([jp][pn]g\)/\1\2.\3/g' | grep ^http > "$outdir-list" + xsltproc - "${TARGET_DIR}-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([jp][pn]g\)/\1\2.\3/g' | grep ^http > "${TARGET_DIR}-list" <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0"> <xsl:output method="xml" indent="yes"/> <xsl:template match="post"> @@ -53,192 +78,201 @@ </xsl:template> </xsl:stylesheet> EOF - echo "`echo \`wc -l < \"$outdir-list\" \`` file(s) available on server" - #output file: $outdir-list + echo "`echo \`wc -l < \"${SITE_DIR}-${TARGET_DIR}-list\" \`` file(s) available on server" + #output file: ${TARGET_DIR}-list } -CHECK() -{ - if [ "$ISNEW" -ne 1 ];then +# check tools availability +Check_Tools() { + # verify all programs required do indeed exist + #MD5 + if [ ! "${MD5}" ]; then + case `uname` in + *BSD) MD5="md5 -r";; + Linux|SunOS) MD5="md5sum";; + *) Fatal_Err "No known md5 tool for this platform. Please specify manually" + esac + fi + MD5_COMMAND=`echo ${MD5} | cut -d' ' -f1` + # basic tools + COMMANDS="cut sed wc wget xsltproc xargs rm mkdir chown comm grep ${MD5_COMMAND}" + for COMMAND in ${COMMANDS} + do + COMMAND_CHECK=`command -v "${COMMAND}"` + [ "${COMMAND_CHECK}" ] || Err_Fatal "${COMMAND} doesn't exist in ${PATH}" + done + + # grep checking + # originally created for workaround on solaris + #if [ `uname` = "SunOS" ]; then + FAIL="" + echo "blah" > superrandomtestfile + echo "blah" > superrandomtestfile.2 + grep -f superrandomtestfile.2 superrandomtestfile > /dev/null 2>&1 || FAIL=1 + rm -f superrandomtestfile superrandomtestfile.2 + [ "${FAIL}" ] && Err_Fatal "Your grep is not compatible. Please install or set path of correct grep" +} + +# verify required folders exist and writeable +Check_Folders(){ + [ -O "${BASE_DIR}" ] || Err_Fatal "You don't own ${BASE_DIR}. Please fix ${BASE_DIR}." + for FOLDER in temp trash deleted ${TARGET_DIR} + do + if [ ! -d "${BASE_DIR}/${FOLDER}" ]; then + mkdir "${BASE_DIR}/${FOLDER}" || Err_Fatal "${FOLDER} folder creation failed" + fi + if [ ! -O "${BASE_DIR}/${FOLDER}" ]; then + echo "You don't own the ${BASE_DIR}/{$FOLDER}, applying globally writeable permission on it" + chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${FOLDER}" || Err_Fatal "Error changing ownership. This shouldn't happen" + fi + done + [ `echo \`ls "${BASE_DIR}/${TARGET_DIR}" | wc -l\`` -eq 0 ] && ISNEW=1 + # let's move to workdir + cd "${BASE_DIR}/temp" + for i in error ok list newlist; do + touch "${TARGET_DIR}-${i}" || Fatal_Err "Error creating ${TARGET_DIR}-${i}. This shouldn't happen" + done + # +} + +# check files correctness +Check_Files() { + if [ "$ISNEW" -ne 1 ]; then echo "Checking for errors..." # THE FILES - printf "" > "$outdir-error" - cd "../$outdir" - for file in `ls` + + # current dir: ${BASE_DIR}/${SITE_DIR}/${TARGET_DIR} + cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" + for TRASH in `ls * | sed -e 's/\([0-9a-f]\{32\}.*\)//g' | grep -v ^$` do - if [ `$md5 "$file" | cut -d ' ' -f1 -` != `echo "$file" | cut -d '.' -f1` ] + mv -f "${TRASH}" "${BASE_DIR}/trash" + echo "Moved ${TRASH} to ${BASE_DIR}/trash" + done + printf "" > "${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}-error" + for FILE in * + do + if [ `${MD5} "${FILE}" | cut -d ' ' -f1 -` != `echo "${FILE}" | cut -d '.' -f1` ] then - echo "$file" >> "../temp/$outdir-error" - echo "Error: $file" + echo + echo "${FILE}" >> "${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}-error" + echo "Error: ${FILE}" fi printf "." done echo - cd ../temp - totalerr=`echo \`wc -l < "$outdir-error"\`` - echo "$totalerr file(s) error" + + # current dir: ${BASE_DIR}/temp + cd ${BASE_DIR}/temp + TOTAL_ERROR=`echo \`wc -l < "${SITE_DIR}-${TARGET_DIR}-error"\`` + echo "${TOTAL_ERROR} file(s) error" echo "Generating list of new files..." # THE FILES - #ls "../$outdir" | grep -vf "$outdir-error" > "$outdir-ok" + #ls "../${TARGET_DIR}" | grep -vf "${TARGET_DIR}-error" > "${TARGET_DIR}-ok" # - ls "../$outdir" | comm -1 -3 "$outdir-error" - > "$outdir-ok" - cat "$outdir-list" | grep -vf "$outdir-ok" > "$outdir-newlist" - echo "`echo \`wc -l < \"$outdir-newlist\"\`` file(s) to be downloaded" - cd "../$outdir" + ls "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" | comm -1 -3 "${SITE_DIR}-${TARGET_DIR}-error" - > "${SITE_DIR}-${TARGET_DIR}-ok" + cat "${SITE_DIR}-${TARGET_DIR}-list" | grep -vf "${SITE_DIR}-${TARGET_DIR}-ok" > "${SITE_DIR}-${TARGET_DIR}-newlist" + echo "`echo \`wc -l < \"${SITE_DIR}-${TARGET_DIR}-newlist\"\`` file(s) to be downloaded" + + # back to target dir + cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" echo "Removing error files" - if [ $totalerr -eq 0 ]; then + if [ "${TOTAL_ERROR}" -eq 0 ]; then echo "No error file. 0 file removed" else - cat "../temp/$outdir-error" | xargs rm - echo "$totalerr file(s) removed" + cat "${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}-error" | xargs rm + echo "${TOTAL_ERROR} file(s) removed" fi echo "`echo \`ls | wc -l\`` file(s) available locally" cd ../temp else - if [ $ISQUICK -eq 1 ]; then + if [ "${ISQUICK}" ]; then echo "quick mode selected. Skipping check" else echo "Empty local repository" fi - cat "$outdir-list" > "$outdir-newlist" + cd "${BASE_DIR}/temp" + cat "${SITE_DIR}-${TARGET_DIR}-list" > "${TARGET_DIR}-newlist" fi } -FETCH() -{ - if [ `echo \`wc -l < "$outdir-newlist"\`` -eq 0 ]; then +# start downloading the images +Fetch_Images() { + if [ `echo \`wc -l < "${TARGET_DIR}-newlist"\`` -eq 0 ]; then echo "No new file" else echo "Starting wget" - cd "../$outdir" - wget -e continue=on -bi "../temp/$outdir-newlist" -o "../temp/$outdir.log" + cd "../${TARGET_DIR}" + wget -e continue=on -bi "../temp/${TARGET_DIR}-newlist" -o "../temp/${TARGET_DIR}.log" fi } - -# path initialization -export PATH=$extrapath:${PATH} - -# verify all programs required do indeed exist +# initialize base variables and initial command check +Init(){ + # Get base folder - current folder or fallback to ${HOME} + [ "${BASE_DIR}" ] || BASE_DIR="${PWD}" + [ "${BASE_DIR}" ] || BASE_DIR="{$HOME}" + [ "`echo ${BASE_DIR} | cut -c1 | grep \/`" ] || BASE_DIR="/${BASE_DIR}" + # path initialization + [ "${ADDITIONAL_PATH}" ] && PATH=${ADDITIONAL_PATH}:${PATH} + export PATH + # misc variables + ISQUICK="" + ISNEW="" -# basic tools -commands="cut sed wc wget xsltproc xargs rm mkdir chown comm grep" -cmderr=" " -for cmd in $commands -do - command -v "$cmd" >/dev/null || cmderr="$cmderr $cmd" -done -if [ x"$cmderr" != x" " ]; then - echo "$cmderr doesn't exist in $PATH" - exit 1 -fi -#md5 -md5base=`echo $md5 | cut -d ' ' -f 1 -` -if [ x`command -v "$md5base" >/dev/null || echo x` != "x" ]; then - echo "$md5base doesn't exist in $PATH" - exit 1 -fi -mdtest= -if [ `echo test | $md5 | cut -d ' ' -f 1 -` != "d8e8fca2dc0f896fd7cb4cb0031ba249" ]; then - echo "$md5 doesn't produce wanted output" - exit 1 -fi + [ $# -lt 2 ] && Err_Help + case "$1" in + status|fetch|quickfetch) + echo "Starting..." + JOB="$1" + ;; + *) + Err_Help + ;; + esac + shift + SITE= + case "$1" in + -s|--site) + shift + SITE="$1" + ;; + *) + SITE=DEFAULT_SITE + ;; + esac + shift + TAGS="$@" + echo "Tags: ${TAGS}" + # slash is not wanted for folder name + TARGET_DIR=`echo "${TAGS}" | sed -e 's/\//_/g'` + SITE_DIR=`echo "${SITE}" | sed -e 's/\//_/g'` +} -# grep checking -# originally created for workaround on solaris -#if [ `uname` = "SunOS" ]; then - FAIL=0 - echo "blah" > superrandomtestfile - echo "blah" > superrandomtestfile.2 - grep -f superrandomtestfile.2 superrandomtestfile > /dev/null 2>&1 || FAIL=1 - rm superrandomtestfile superrandomtestfile.2 superrandomtestfile.3 > /dev/null 2>&1 - if [ $FAIL = 1 ]; then - echo "Your grep is not compatible. Please install or set path of correct grep" - exit 1 - fi -#fi +Msg_Welcome +Init "$@" +Check_Tools +Check_Folders -# all green (part 1)! let's go (until we check the tag) - -# initialization -# are we really doing it? -HELP="Usage: moefetch (quick)fetch|status <tags>" - -if [ $# -lt 2 ]; then - echo "$HELP" - exit 1 -fi - -case "$1" in - status|fetch|quickfetch) - echo "Starting..." +# let's do the job! +case "${JOB}" in + status) + Generate_Link + Check_Files ;; - *) - echo "$HELP" - exit 1 + fetch) + Generate_Link + Check_Files + Fetch_Images + ;; + quickfetch) + ISNEW=1 + ISQUICK=1 + Generate_Link + Check_Files + Fetch_Images ;; esac - -# we did it indeed -# get started - -# do we own the files -tags=`echo "$@" | cut -d ' ' -f 2- -` -echo "Tags: $tags" -# slash do not want -outdir=`echo "$tags" | sed -e 's/\//_/g'` -ISNEW=0 -if [ -O "$basedir" ]; then - if [ ! -d "$basedir/$outdir" ]; then - ISNEW=1 - mkdir "$basedir/$outdir" - fi - if [ ! -O "$basedir/$outdir" ]; then - echo "You don't own the $basedir/$outdir, applying globally writeable permission on it" - chmod -R u=rwX,g=rwX,o=rwX "$basedir/$outdir" - fi - if [ `echo \`ls "$basedir/$outdir" | wc -l\`` -eq 0 ]; then - ISNEW=1 - fi - if [ ! -d "$basedir/temp" ]; then - mkdir "$basedir/temp" - fi - if [ ! -O "$basedir/temp" ]; then - echo "You don't own the $basedir/temp, applying globally writeable permission on it" - chmod -R u=rwX,g=rwX,o=rwX "$basedir/temp" - fi -else - echo "Fatal error: you don't own ${basedir}. Please fix ${basedir}. Stopping" - exit 1 -fi -# let's move to workdir -cd "$basedir/temp" -touch "$outdir-error" -touch "$outdir-ok" -touch "$outdir-list" -touch "$outdir-newlist" -# - -# let's do the job! -ISQUICK=0 -case "$1" in - status) - GENERATE - CHECK - ;; - fetch) - GENERATE - CHECK - FETCH - ;; - quickfetch) - GENERATE - ISNEW=1 - ISQUICK=1 - CHECK - FETCH -esac