diff options
author | Jonas Smedegaard <dr@jones.dk> | 2014-12-25 13:50:24 +0100 |
---|---|---|
committer | Jonas Smedegaard <dr@jones.dk> | 2014-12-25 13:51:03 +0100 |
commit | a8d873cea1672a2341d2f57d4e0057056a5fc983 (patch) | |
tree | fcb9c7a374ec5b7b3d681381186b843bb435fcc3 | |
parent | fd5f0ad6473c84787c87cfa7637e6e283f970171 (diff) |
Work around modest URL parsing.
-rw-r--r-- | Makefile | 6 |
1 files changed, 5 insertions, 1 deletions
@@ -5,6 +5,9 @@ source_basename = EUT/2nd-edition # work around in mediawiki bug parsing blockquote tags on same line re_blockquote = s|\n*(</?blockquote>)\n*|\n$$1\n|g +# work around modest URL parsing <https://github.com/jgm/pandoc/issues/1838> +re_urlencode = s|\bhttp[^\s\"]+?\K(--+)|uri_escape($$1, "-")|eg + templates = template.tex header.tex before.tex after.tex filters = ./pandoc-filter-sections ./pandoc-filter-bib @@ -35,12 +38,13 @@ download: wget -o $(stem).raw '$(source_baseurl)index.php?title=$(source_basename)&action=edit' $(stem).mediawiki: $(stem).raw - perl -0777 -MHTML::Entities -p \ + perl -0777 -MHTML::Entities -MURI::Escape -p \ -e 's|.*<textarea[^>]*>||s; s|</textarea.*||s;' \ -e 'decode_entities($$_);' \ -e 's|.*?\n= |= |s;' \ -e 's|<!--.*-->||s;' \ -e '$(re_blockquote);' \ + -e '$(re_urlencode);' \ < $< > $@ $(stem).pdf: $(stem).mediawiki $(templates) $(filters) |