changeset 29:ac51fc10414f

Add support for URL prefix pattern in updates.
author Matti Hamalainen <ccr@tnsp.org>
date Mon, 05 Sep 2011 10:55:45 +0300
parents ab06d838d916
children 6148ce4a83ed
files update.sh
diffstat 1 files changed, 6 insertions(+), 4 deletions(-) [+]
line wrap: on
line diff
--- a/update.sh	Mon Sep 05 10:55:16 2011 +0300
+++ b/update.sh	Mon Sep 05 10:55:45 2011 +0300
@@ -7,16 +7,17 @@
 URLPREFIX="$1"
 CLASSFILE="$2"
 LISTFILE="$2.tmp"
+PATPREFIX="$3"
 
 if wget -q -O "$LISTFILE" "$URLPREFIX"; then
-	perl -ne 'if (/<a href="OR_([A-Z]{3}\d\S+)\.htm">/) { print "$1\n"; }' < "$LISTFILE" > "$CLASSFILE"
+	perl -ne "if (/<a href=\"${PATPREFIX}([A-Z]{3}\d\S+)${URLSUFFIX}\">/) { print \"\$1\n\"; }" < "$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}"
+	wget -q -O "$INFILE.new" "${URLPREFIX}${PATPREFIX}${i}${URLSUFFIX}"
 
 	if test -e "$INFILE.new"; then
 		# New data fetched, does old file exist?
@@ -54,5 +55,6 @@
 fi
 }
 
-parse "http://www.oamk.fi/tyojarjestykset/otek/luokat/" "luokat.txt"
-parse "http://www.oamk.fi/~heikkim/riihi4/luokat/" "luokat_next.txt"
+#parse "http://www.oamk.fi/tyojarjestykset/otek/luokat/" "luokat.txt" ""
+parse "http://www.oamk.fi/~heikkim/Luhti1/Ryhm%84t/" "luokat.txt" "Ryh._"
+