Move the processing/sed into a clean perl script

14a8e33e · Arnaud Legrand · 9e816cab · 14a8e33e · 14a8e33e
Commit 14a8e33e authored Mar 21, 2019 by Arnaud Legrand
Show whitespace changes
Inline Side-by-side

Showing with 41 additions and 6 deletions

.gitlab-ci.yml .gitlab-ci.yml +3 -6

pandoc_fixer.pl bin/pandoc_fixer.pl +38 -0

No files found.
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -6,12 +6,9 @@ pages:
  - pandoc --help
  # - pandoc --list-input-formats    # Broken as pandoc dates from 2013 on this image! :(
  - for file in $(find -name "*.md"); do
-        pandoc -s -f markdown_github -t html -o ${file%.*}.html $file && echo "Exported ${file%.*}.html" && echo "<li><a href='https://learninglab.gitlabpages.inria.fr/mooc-rr/mooc-rr-ressources/${file%.*}.html'>${file%.*}.html</a></li>" >> index.html || echo "ERROR $file";
+        pandoc -s -f gfm -t html -o ${file%.*}.html $file && echo "Exported ${file%.*}.html" && echo "<li><a href='https://learninglab.gitlabpages.inria.fr/mooc-rr/mooc-rr-ressources/${file%.*}.html'>${file%.*}.html</a></li>" >> index.html || echo "ERROR $file";
-        sed -i 's|https://gitlab.inria.fr/learninglab/|https://learninglab.gitlabpages.inria.fr/|g' ${file%.*}.html;
+        bin/pandoc_fixer.pl ${file} > ${file}.new;
-        sed -i "s|<body>|<body>The <a href='https://gitlab.inria.fr/learninglab/mooc-rr/mooc-rr-ressources/blob/master/$file'>original version of this document is available on gitlab</a>.|g" ${file%.*}.html;
+        mv ${file}.new ${file};
-        sed -i "s|\-\-\-</p>|<hr/>|g" ${file%.*}.html;
-        sed -i "s|<p>TITLE:\(.*\)<br|<b>TITLE:\1</b><br|g" ${file%.*}.html;
-        sed -i "s|Date\(.*\)<br|<i>Date\1</i><br|g" ${file%.*}.html;
    done
  # The next part is broken because pandoc is way too old on this image :(
  # - for file in $(find -name "*.org"); do 

--- a/bin/pandoc_fixer.pl
+++ b/bin/pandoc_fixer.pl
+#!/usr/bin/perl -w
+use strict;
+if($#ARGV!=0) {
+    die "Usage: pandox_fixer.pl input.md";
+}
+my($input)=shift(@ARGV);
+if(!($input =~ /.md$/)) { 
+    die "Usage: pandox_fixer.pl input.md";
+}
+my($line);
+my($gitdate)=`git log $input | grep Date | head -n 1`;
+chomp($gitdate);
+$gitdate =~ s/Date: *//g;
+$gitdate =~ s/\s*\+.*$//g;
+$input =~ s/.md/.html/;
+open INPUT, $input or die;
+while(defined($line=<INPUT>)) {
+    $line =~ s|https://gitlab.inria.fr/learninglab/|https://learninglab.gitlabpages.inria.fr/|g; ## Not such a good idea!
+    if($input=~ /_fr.html/) {
+	$line =~ s|<body>|<body>Les <a href='https://gitlab.inria.fr/learninglab/mooc-rr/mooc-rr-ressources/blob/master/$input'>sources de ce document sont disponibles sur gitlab</a>.|g;
+    } else {
+	$line =~ s|<body>|<body>The <a href='https://gitlab.inria.fr/learninglab/mooc-rr/mooc-rr-ressources/blob/master/$input'>source of this this document is available on gitlab</a>.|g;
+    }
+    $line =~ s|---</p>|<hr/>|g;
+    $line =~ s|Date:.*<br|<i>Date: $gitdate</i><br|g;
+    $line =~ s|<p>TITLE:\(.*\)<br|<b>TITLE:$1</b><br|g;
+    $line =~ s|href=" ---</p>|<hr/>|g;
+    if($line =~ /<p>AUTHOR:/) { next; }
+    print $line;
+}