annotate update.sh @ 46:9b5fa0f3812b

Improvements in the update logic.
author Matti Hamalainen <ccr@tnsp.org>
date Sat, 13 Oct 2012 11:58:26 +0300
parents 76900cdfc066
children ceca06576918
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
1 #!/bin/sh
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
2 URLSUFFIX=".htm"
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
3
24
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
4 function parse()
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
5 {
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
6 # Create cache directories, if they do not exist
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
7 if test ! -d "${OLDCACHEDIR}"; then
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
8 mkdir -p "${OLDCACHEDIR}"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
9 fi
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
10
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
11 if test ! -d "${CACHEDIR}"; then
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
12 mkdir -p "${CACHEDIR}"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
13 fi
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
14
24
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
15 URLPREFIX="$1"
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
16 CLASSFILE="$2"
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
17 LISTFILE="$2.tmp"
29
ac51fc10414f Add support for URL prefix pattern in updates.
Matti Hamalainen <ccr@tnsp.org>
parents: 24
diff changeset
18 PATPREFIX="$3"
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
19
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
20 if wget -q -O "$LISTFILE" "$URLPREFIX"; then
29
ac51fc10414f Add support for URL prefix pattern in updates.
Matti Hamalainen <ccr@tnsp.org>
parents: 24
diff changeset
21 perl -ne "if (/<a href=\"${PATPREFIX}([A-Z]{3}\d\S+)${URLSUFFIX}\">/) { print \"\$1\n\"; }" < "$LISTFILE" > "$CLASSFILE"
34
1f6be8457912 Script output cosmetics.
Matti Hamalainen <ccr@tnsp.org>
parents: 31
diff changeset
22 echo -n "* Fetched classfile $CLASSFILE: "
1f6be8457912 Script output cosmetics.
Matti Hamalainen <ccr@tnsp.org>
parents: 31
diff changeset
23 cat "$CLASSFILE" | wc -l
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
24 fi
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
25
24
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
26 if test -e "$CLASSFILE"; then
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
27 cat "$CLASSFILE" | while read i; do
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
28 parse=no
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
29 INFILE="${CACHEDIR}${i}.html"
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
30 ONFILE="${OLDCACHEDIR}${i}.html"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
31 DATAFILE="${i}.data"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
32 wget -q -O "${INFILE}.new" "${URLPREFIX}${PATPREFIX}${i}${URLSUFFIX}"
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
33
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
34 if test -e "${INFILE}.new"; then
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
35 # New data fetched, does old file exist?
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
36 if test -e "$INFILE"; then
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
37 # Yes, do a diff
22
f025d36926a1 Make backup copies of old versions of data files when updating.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
38 if ! diff -u "$INFILE" "$INFILE.new" > "$INFILE.diff"; then
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
39 # There were differences, do a parse
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
40 parse=yes
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
41 mv "$INFILE" "$ONFILE" && \
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
42 mv "$INFILE.new" "$INFILE"
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
43 else
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
44 # No changes, apparently .. remove the new one
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
45 rm -f "$INFILE.new" "$INFILE.diff"
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
46 fi
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
47 else
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
48 # No old file, parse new data
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
49 mv "$INFILE.new" "$INFILE"
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
50 parse=yes
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
51 fi
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
52 fi
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
53 # No new file fetched, does datafile exist?
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
54 if test ! -e "${CACHEDIR}${DATAFILE}"; then
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
55 # No, try to parse it if old file input exists
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
56 parse=yes
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
57 fi
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
58
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
59 # Parsing of old data requested?
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
60 if test "x$parse" = "xyes" -a -e "$INFILE"; then
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
61 if test -e "${CACHEDIR}${DATAFILE}"; then
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
62 mv "${CACHEDIR}${DATAFILE}" "${OLDCACHEDIR}${DATAFILE}"
22
f025d36926a1 Make backup copies of old versions of data files when updating.
Matti Hamalainen <ccr@tnsp.org>
parents: 21
diff changeset
63 fi
34
1f6be8457912 Script output cosmetics.
Matti Hamalainen <ccr@tnsp.org>
parents: 31
diff changeset
64 echo "Parsing $i"
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
65 perl parsedata.pl -php "$INFILE" -o "${CACHEDIR}${DATAFILE}"
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
66 fi
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
67 done
24
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
68 fi
1b8260151e99 Get updates / data from upcoming work-in-progress timetables for next period also.
Matti Hamalainen <ccr@tnsp.org>
parents: 22
diff changeset
69 }
6
7fca87c41e17 Added data fetching and updating shellscript.
Matti Hamalainen <ccr@tnsp.org>
parents:
diff changeset
70
35
4d9354abda73 Update fetching URLs and parameters to match the latest changes on OAMK's web.
Matti Hamalainen <ccr@tnsp.org>
parents: 34
diff changeset
71 CACHEDIR="cache/"
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
72 OLDCACHEDIR="cache-old/"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
73 parse "http://www.oamk.fi/tyojarjestykset/otek/luokat/" "luokat.txt" "OR_"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
74 #parse "http://www.oamk.fi/~heikkim/riihi2/Oppilaat/" "luokat.txt" "Ryh._"
29
ac51fc10414f Add support for URL prefix pattern in updates.
Matti Hamalainen <ccr@tnsp.org>
parents: 24
diff changeset
75
31
dbe7ff545293 Add support for fetching and showing data for next/upcoming period.
Matti Hamalainen <ccr@tnsp.org>
parents: 29
diff changeset
76 CACHEDIR="cache-next/"
46
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
77 OLDCACHEDIR="cache-next-old/"
9b5fa0f3812b Improvements in the update logic.
Matti Hamalainen <ccr@tnsp.org>
parents: 42
diff changeset
78 parse "http://www.oamk.fi/~heikkim/riihi2/Oppilaat/" "luokat_next.txt" "Ryh._"