diff options
Diffstat (limited to 'bulk-audio/get')
| -rwxr-xr-x | bulk-audio/get | 88 |
1 files changed, 60 insertions, 28 deletions
diff --git a/bulk-audio/get b/bulk-audio/get index b7791aa..bc3839d 100755 --- a/bulk-audio/get +++ b/bulk-audio/get @@ -1,59 +1,91 @@ #!/bin/sh -KANJI="$1" -KANA="$2" +SCRIPT_NAME="$0" +SOURCES="lp101,lp101_alt,jisho" +KANJI="" +KANA="" -if [ -z "$KANJI" -o -z "$KANA" ]; then - cat << EOF -usage: $0 <kanji> <kana> > <output> - -return value is 0 if <output> was succesfully written, 1 if the word could not -be found. this script searches languagepod101, languagepod101 (alt) and -jisho.org. -EOF - exit 1 -fi - -get_languagepod101() { +lp101() { URL="https://assets.languagepod101.com/dictionary/japanese/audiomp3.php?kanji=$KANJI&kana=$KANA" # 52288 is the content-length of the "the audio for this clip is currently # not available. it will be recorded and uploaded shortly. thank you for your # patience" message (404, but server sends 200 anyways) curl -X HEAD -iso - "$URL" | awk '/^Content-length:/ { exit $2 == 52288 }' - [ $? -ne 0 ] && return 1 + [ $? -ne 0 ] && return curl -so - "$URL" + exit 0 } -get_languagepod101_alt() { +lp101_alt() { HTML="$(curl -s -X POST \ "https://www.japanesepod101.com/learningcenter/reference/dictionary_post" \ -H "Content-Type: application/x-www-form-urlencoded" \ -d "post=dictionary_reference&match_type=exact&search_query=$KANJI&vulgar=true" \ )" - [ $? -ne 0 ] && return 1 + [ $? -ne 0 ] && return URL="$(echo "$HTML" | pup "audio source attr{src}" | head -n1)" - [ -z "$URL" ] && return 1 + [ -z "$URL" ] && return curl -so - "$URL" + exit 0 } -get_jisho() { +jisho() { HTML="$(curl -s "https://jisho.org/search/$KANJI")" - [ $? -ne 0 ] && return 1 + [ $? -ne 0 ] && return URL="$(echo "$HTML" | pup "audio[id=\"audio_$KANJI:$KANA\"] source attr{src}" | head -n1)" - [ -z "$URL" ] && return 1 + [ -z "$URL" ] && return URL="https:$URL" curl -so - "$URL" + exit 0 +} + +usage() { + cat << EOF +usage: $SCRIPT_NAME [OPTIONS] <KANJI> <KANA> > <OUTPUT> + +attempt to download a native Japanese recording of word written as KANJI and +read as KANA. outputs mp3 to stdout. return value is 0 if OUTPUT was written +(clip was found), and 1 if no clip could be found. + +options: + -s <source1[,source2,...]> set source order (default: $SOURCES) + -h show help + +sources: + lp101 JapanesePod101 + lp101_alt JapanesePod101 (Alternate) + jisho Jisho.org +EOF } -# get_languagepod101_alt -# [ $? -eq 0 ] && exit 0 -# -# get_jisho -# [ $? -eq 0 ] && exit 0 +while getopts 'hs:' OPT; do + case $OPT in + h) + usage + exit 0 + ;; + s) + SOURCES="$OPTARG" + ;; + \?|*) + usage > /dev/stderr + exit 1 + ;; + esac +done -get_languagepod101 -[ $? -eq 0 ] && exit 0 +# invalid argument count +if [ $(( $# - $OPTIND + 1 )) -ne 2 ]; then + usage > /dev/stderr + exit 1 +fi + +KANJI=${@:$OPTIND:1} +KANA=${@:$OPTIND+1:1} + +$(printf '%s;' "$SOURCES" | sed -z 's/[;,\n]/ ; /g') # if none were succesful, delete output file and exit with error rm -f "$OUTPUT" exit 1 + |