diff options
Diffstat (limited to '')
-rwxr-xr-x | SuckItPm | 72 |
1 files changed, 0 insertions, 72 deletions
diff --git a/SuckItPm b/SuckItPm deleted file mode 100755 index 246f485..0000000 --- a/SuckItPm +++ /dev/null | |||
@@ -1,72 +0,0 @@ | |||
1 | #!/bin/bash | ||
2 | |||
3 | URL="https://wiki.devuan.org" | ||
4 | |||
5 | filter=" | ||
6 | -not -name "*~" -a \ | ||
7 | -not -name ".flock" -a \ | ||
8 | -not -name ".htaccess" -a \ | ||
9 | -not -name ".lastmod" -a \ | ||
10 | -not -name ".pageindex" -a \ | ||
11 | " | ||
12 | |||
13 | pushd /opt/mergedWork | ||
14 | |||
15 | find /opt/pmwiki/wiki.d ${filter} \ | ||
16 | -name "*.*" -type f,l -printf "%P\n" | while read line | ||
17 | do | ||
18 | base=`echo "${line}" | cut -d '.' -f 1` | ||
19 | file=`echo "${line}" | cut -d '.' -f 2` | ||
20 | time=`date --rfc-3339=seconds -ur /opt/pmwiki/wiki.d/${base}.${file} | cut -d '+' -f 1` | ||
21 | mkdir -p PmWiki/$base | ||
22 | mkdir -p combined/$base | ||
23 | echo "Converting ${URL}/?n=${base}.${file}?action=print -> PmWiki/${base}/${file}.md" | ||
24 | echo -e "ogWiki=PmWiki\nogURL=${URL}\nogBase=${base}\nogFile=${file}\ntimestamp=${time}\n" > PmWiki/${base}/${file}.md.md | ||
25 | # Doesn't help with redownloads, coz natch a dynamic site isn't cached. But I can at least comment out the curl command during testing to save time. | ||
26 | # curl --no-progress-meter ${URL}/?n=${base}.${file}?action=markdown -o PmWiki/${base}/${file}.MD | ||
27 | curl --no-progress-meter ${URL}/?n=${base}.${file}?action=print -o PmWiki/${base}/${file}.HTM | ||
28 | cp PmWiki/${base}/${file}.HTM PmWiki/${base}/${file}.HTM_ORIGINAL | ||
29 | csplit -ks PmWiki/${base}/${file}.HTM '%<!--PageText-->%' '/<!--HTMLFooter-->/' | ||
30 | if [ -f xx00 ]; then | ||
31 | rm PmWiki/${base}/${file}.HTM | ||
32 | mv xx00 PmWiki/${base}/${file}.HTM | ||
33 | fi | ||
34 | if [ -f xx01 ]; then | ||
35 | rm xx01 | ||
36 | fi | ||
37 | sed -i -E PmWiki/${base}/${file}.HTM \ | ||
38 | -e "s/rel='nofollow'//g" \ | ||
39 | -e "s/target='_blank'//g" \ | ||
40 | -e "s/class='categorylink'//g" \ | ||
41 | -e "s/class='createlink'//g" \ | ||
42 | -e "s/class='createlinktext'//g" \ | ||
43 | -e "s/class='escaped'//g" \ | ||
44 | -e "s/class='diffmarkup'//g" \ | ||
45 | -e "s/class='selflink'//g" \ | ||
46 | -e "s/class='urllink'//g" \ | ||
47 | -e "s/class='vspace'//g" \ | ||
48 | -e "s/class='wikilink'//g" \ | ||
49 | -e "s/style='.*;'//g" | ||
50 | |||
51 | # pandoc -f markdown -t commonmark_x --self-contained PmWiki//${base}/${file}.MD >PmWiki/${base}/${file}.md | ||
52 | pandoc -f html -t commonmark_x --self-contained PmWiki//${base}/${file}.HTM >PmWiki/${base}/${file}.md | ||
53 | cp PmWiki/${base}/${file}.md PmWiki/${base}/${file}.md_ORIGINAL | ||
54 | |||
55 | # Attempt to clean things up, badly. | ||
56 | sed -i -E PmWiki/${base}/${file}.md \ | ||
57 | -e 's/\$/\$dlr\$/g' \ | ||
58 | -e 's/\{#.*\}//g' \ | ||
59 | -e '/^:::/d' \ | ||
60 | # -e '/\[Site$/d' \ | ||
61 | # -e '/^\[\]/d' \ | ||
62 | # -e "s/\`<a id='trailstart'>\`\{=html\}\`<\/a>\`\{=html\}//g" \ | ||
63 | # -e "s/^\`<img /<img /g" \ | ||
64 | # -e "s/^\`\`\`\{=html\}//g" \ | ||
65 | # -e "s/^\`\`\`//g" \ | ||
66 | # -e "s/\`\{=html\}//g" | ||
67 | |||
68 | ln -frs PmWiki/${base}/${file}.md combined/${base}/${file}.md | ||
69 | ln -frs PmWiki/${base}/${file}.md.md combined/${base}/${file}.md.md | ||
70 | done | ||
71 | |||
72 | popd | ||