From 58d81d1a6d4dd09a8f19d4c42ab45caae7db030d Mon Sep 17 00:00:00 2001
From: dvs1
Date: Sun, 19 Jan 2025 21:41:47 +1000
Subject: Use the wiki print button, filter the wiki HTML early, and use
 cmark-gfm.

---
 SuckItFos | 43 ++++++++++++++++++++++++++++---------------
 1 file changed, 28 insertions(+), 15 deletions(-)

(limited to 'SuckItFos')

diff --git a/SuckItFos b/SuckItFos
index 54fc376..b59f6b0 100755
--- a/SuckItFos
+++ b/SuckItFos
@@ -21,16 +21,34 @@ do
     mkdir -p Foswiki/${base}/`dirname ${file}`
     mkdir -p combined/$base
     mkdir -p combined/${base}/`dirname ${file}`
-    echo "Converting ${URL}/${base}/${file} -> Foswiki/${base}/${file}.md"
+    echo "Converting ${URL}/${base}/${file}?cover=print -> Foswiki/${base}/${file}.md"
 #    pandoc -f html -t markdown --self-contained ${URL}/${base}/${file} >Foswiki/${base}/${file}.md
     # TODO - try curl, to see what is actually downloaded, and maybe not download unchanged pages.  curl to .HTM
     # 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.
-    curl --silent --no-progress-meter ${URL}/${base}/${file} -o Foswiki/${base}/${file}.HTM
+    curl --silent --no-progress-meter ${URL}/${base}/${file}?cover=print -o Foswiki/${base}/${file}.HTM
+
+    cp Foswiki/${base}/${file}.HTM Foswiki/${base}/${file}.HTM_ORIGINAL
+    csplit -ks Foswiki/${base}/${file}.HTM '%<div id="patternMainContents">%' '/<div class="foswikiAttachments foswikiFormStep" style="overflow:auto">/'
+    if [ -f xx00 ]; then
+	rm Foswiki/${base}/${file}.HTM
+	mv xx00 Foswiki/${base}/${file}.HTM
+    fi
+    sed -i -E Foswiki/${base}/${file}.HTM \
+	-e "s/rel='nofollow'//g" \
+	-e 's/rel="nofollow"//g' \
+	-e "s/target='_blank'//g" \
+	-e "s/class='foswiki[[:alpha:]]*'//g" \
+	-e 's/class="foswikiTopic"/class="FoswikiTopic"/g' \
+	-e 's/class="foswiki[[:alpha:]]*"//g' \
+	-e "s/style='.*;'//g"
+#	-e "s/style='background-color: #.*;'//g" \
+#	-e "s/style='font-size: .*;'//g"
+
     pandoc -f html -t commonmark_x --self-contained Foswiki//${base}/${file}.HTM >Foswiki/${base}/${file}.md
-    ln -frs Foswiki/${base}/${file}.md combined/${base}/${file}.md
     cp Foswiki/${base}/${file}.md Foswiki/${base}/${file}.md_ORIGINAL
 
-    csplit -ks Foswiki/${base}/${file}.md '%::: {.foswikiTopic}%' '/::: {.foswikiContentFooter}/'
+#    csplit -ks Foswiki/${base}/${file}.md '%::: {.foswikiTopic}%' '/::: {.foswikiContentFooter}/'
+    csplit -ks Foswiki/${base}/${file}.md '%::: {.FoswikiTopic}%' '/::: {.patternInfo}/'
     if [ -f xx00 ]; then
 	rm Foswiki/${base}/${file}.md
 	mv xx00 Foswiki/${base}/${file}.md
@@ -40,21 +58,16 @@ do
     sed -i -E Foswiki/${base}/${file}.md \
 	-e 's/\$/\$dlr\$/g' \
 	-e 's/\{#.*\}//g' \
-	-e 's/\{\.foswiki.*\}//g' \
-	-e 's/\{\.foswiki.*//g' \
-	-e 's/\{\.foswikiNewLink rel=“nofollow”\}//g' \
-	-e 's/\{\.foswikiNewLink$//g' \
-	-e 's/^\.foswiki.*\}//g' \
 	-e 's/\{\.pattern.*\}//g' \
 	-e 's/\{\.pattern.*//g' \
-	-e 's/\{rel="nofollow"\}//g' \
-	-e 's/^rel="nofollow"\}//g' \
-	-e 's/rel=“nofollow”\}$//g' \
+	-e '/^<!-- -->/d' \
 	-e '/^:::/d'
-    echo -e "****\n[Original page](${URL}/${base}/${file}) where you can edit it." >> Foswiki/${base}/${file}.md
+    echo -e "****\n[Original page](${URL}/${base}/${file}) where maybe you can edit it." >> Foswiki/${base}/${file}.md
 
-#    pandoc -t html -f commonmark_x --self-contained		Foswiki/${base}/${file}.md > Foswiki//${base}/${file}.htm
-    cmark-gfm -t html -e footnotes -e table -e strikethrough	Foswiki/${base}/${file}.md > Foswiki//${base}/${file}.body
+#    pandoc -t html -f commonmark_x --self-contained		Foswiki/${base}/${file}.md > Foswiki/${base}/${file}.htm
+#    cmark-gfm -t html -e footnotes -e table -e strikethrough	Foswiki/${base}/${file}.md > Foswiki/${base}/${file}.body
+#    ln -frs Foswiki/${base}/${file}.body combined/${base}/${file}.body
+    ln -frs Foswiki/${base}/${file}.md combined/${base}/${file}.md
 done
 
 notYetAnotherWiki.lua
-- 
cgit v1.1