Mercurial > hg > lukkari
view update.sh @ 28:ab06d838d916
Cosmetics & comments.
author | Matti Hamalainen <ccr@tnsp.org> |
---|---|
date | Mon, 05 Sep 2011 10:55:16 +0300 |
parents | 1b8260151e99 |
children | ac51fc10414f |
line wrap: on
line source
#!/bin/sh URLSUFFIX=".htm" CACHEDIR="cache/" function parse() { URLPREFIX="$1" CLASSFILE="$2" LISTFILE="$2.tmp" if wget -q -O "$LISTFILE" "$URLPREFIX"; then perl -ne 'if (/<a href="OR_([A-Z]{3}\d\S+)\.htm">/) { print "$1\n"; }' < "$LISTFILE" > "$CLASSFILE" fi if test -e "$CLASSFILE"; then cat "$CLASSFILE" | while read i; do parse=no INFILE="${CACHEDIR}${i}.html" wget -q -O "$INFILE.new" "${URLPREFIX}OR_${i}${URLSUFFIX}" if test -e "$INFILE.new"; then # New data fetched, does old file exist? if test -e "$INFILE"; then # Yes, do a diff if ! diff -u "$INFILE" "$INFILE.new" > "$INFILE.diff"; then # There were differences, do a parse parse=yes mv "$INFILE" "$INFILE.old" mv "$INFILE.new" "$INFILE" fi else # No old file, parse new data mv "$INFILE.new" "$INFILE" parse=yes fi else # No new file fetched, does datafile exist? if test ! -e "$i.data"; then # No, try to parse it if old file input exists parse=yes fi fi # Parsing of old data requested? if test "x$parse" = "xyes" -a -e "$INFILE"; then OUTFILE="${CACHEDIR}/$i.data" if test -e "$OUTFILE"; then mv "$OUTFILE" "$OUTFILE.old" fi echo "parsing $i" perl parsedata.pl -php "$INFILE" -o "$OUTFILE" fi done fi } parse "http://www.oamk.fi/tyojarjestykset/otek/luokat/" "luokat.txt" parse "http://www.oamk.fi/~heikkim/riihi4/luokat/" "luokat_next.txt"