From f8955236b96711e609e3f2eb5d95998de4812755 Mon Sep 17 00:00:00 2001
From: Jonas Smedegaard <dr@jones.dk>
Date: Mon, 22 Dec 2014 09:16:35 +0100
Subject: Add section and bibliography filters.

---
 Makefile               |  6 ++++-
 TODO                   |  6 +++--
 pandoc-filter-bib      | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++
 pandoc-filter-sections | 55 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 132 insertions(+), 3 deletions(-)
 create mode 100755 pandoc-filter-bib
 create mode 100755 pandoc-filter-sections

diff --git a/Makefile b/Makefile
index d1f0ca5..1cd26e5 100644
--- a/Makefile
+++ b/Makefile
@@ -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
 
+export PANDOC_CITEPROC_FILE = $(stem).bib
+#PANDOC_CITEPROC_FILE = $(stem).bib
+
 all: $(stem).pdf
 
 download:
@@ -21,7 +24,8 @@ $(stem).mediawiki: $(stem).raw
 
 $(stem).pdf: $(stem).mediawiki template.tex
 	pandoc -f mediawiki --template=template.tex --latex-engine=xelatex \
-	--toc \
+	--filter ./pandoc-filter-sections --filter ./pandoc-filter-bib \
+	--bibliography=$(PANDOC_CITEPROC_FILE) \
 	-V papersize=a4paper -V fontsize=11pt \
 	-V documentclass=memoir -V chapterstyle=demo3 \
 	-V classoption=oneside -V classoption=titlepage -V classoption=twocolumn \
diff --git a/TODO b/TODO
index 2c44301..8aa6c97 100644
--- a/TODO
+++ b/TODO
@@ -14,8 +14,10 @@ Input (last checked 2014-12-21 22-39):
 
 Processing:
   * Handle draft/final switch, and fail on warnings when in draft mode.
-  * Use pandoc filter for bugfixing (not multiple passes + perl regex).
-  * Decode and properly handle bibliography (i.e. {{cite ...}} markup).
+  * Fix avoid section filter eating subsequent content.
+  * use Header (not latex RawBlock) as header markup in section filter.
+  * Refine bibliography regarding web sources and timestamp markup.
+    <https://tex.stackexchange.com/questions/3587>
 
 Content modelling:
   * Favor british at all uses of "english" documentclass hint.
diff --git a/pandoc-filter-bib b/pandoc-filter-bib
new file mode 100755
index 0000000..27ba8d4
--- /dev/null
+++ b/pandoc-filter-bib
@@ -0,0 +1,68 @@
+#!/usr/bin/perl
+
+use warnings;
+use strict;
+
+use v5.10.1; # use switch keyword "when" (but avoid too complex "given")
+no if $] >= 5.018, warnings => "experimental::smartmatch";
+
+use Pandoc::Filter;
+use Pandoc::Elements;
+
+use Path::Tiny;
+use List::Util qw(pairmap);
+
+my $bibfilename = $ENV{'PANDOC_CITEPROC_FILE'} || 'bibliography.bib';
+my $bibfile = path($bibfilename);
+
+my (@a, $i);
+
+pandoc_filter sub {
+	my $self = shift;
+	return unless ($self->name eq 'RawInline' and $self->format eq 'mediawiki');
+
+	for ($self->content) {
+		when (/^{{cite\s+(\w+)\s*\|([^}]*)}}$/) {
+			my $id = 'ref'.++$i;
+			my @data = pairmap { $b =~ s/"/\\"/g; "$a=\"$b\"" }
+				map { /^\s*(\w+)\s*=\s*"?(.*?)"?\s*$/ }
+					split( /\|/, $2 );
+			push @a, join ",\n",
+				'@'.$1.'{'.$id,
+				@data,
+				'}';
+			return Cite(
+				[{
+					'citationId' => $id,
+					'citationPrefix' => [],
+					'citationSuffix' => [],
+					'citationMode' => {
+						t => 'NormalCitation',
+						c => [],
+					},
+					'citationNoteNum' => 0,
+					'citationHash' => 0,
+				 }],
+				[ Str $id ],
+			);
+		}
+		when (/^{{citation needed}}$/) {
+			say STDERR "WARNING: Mediawiki citation needed.";
+			return [
+				Str "citation",
+				Space,
+				Str "needed",
+			];
+		}
+		default {
+			say STDERR "WARNING: Mediawiki unknown data skipped: "
+				. '"' . $_[0]->content . '"';
+			return Str "";
+		}
+	}
+};
+
+END {
+	$bibfile->spew_utf8( join "\n\n", @a )
+		if (@a);
+}
diff --git a/pandoc-filter-sections b/pandoc-filter-sections
new file mode 100755
index 0000000..f0bb5a5
--- /dev/null
+++ b/pandoc-filter-sections
@@ -0,0 +1,55 @@
+#!/usr/bin/perl
+
+use warnings;
+use strict;
+
+use Pandoc::Filter;
+use Pandoc::Elements;
+
+# FIXME: avoid eating content past tweaked headers
+pandoc_filter(
+	\&frontmatter,
+	\&mainmatter,
+	\&backmatter,
+	\&toc,
+);
+
+# FIXME: use Header (not latex RawBlock) - why does it hang?!?
+sub header {
+	my $label = shift;
+#	Header( 1, attributes {}, [ Str $label ], );
+	RawBlock( 'latex', '\\chapter{'.$label.'}' );
+};
+
+sub frontmatter {
+	my $self = shift;
+	return [ RawBlock( 'latex', '\\frontmatter' ), header('About'), ]
+		if ( $self->name eq 'Header' and $self->level >= 1
+			and stringify($self) eq 'About' );
+	return;
+}
+
+sub mainmatter {
+	my $self = shift;
+	return [ RawBlock( 'latex', '\\mainmatter' ), header(stringify($self)), ]
+		if ( $self->name eq 'Header' and $self->level == 1
+			and stringify($self) =~ /^Scope/ );
+	return;
+}
+
+sub backmatter {
+	my $self = shift;
+	return [ RawBlock( 'latex', '\\backmatter' ), header('References'), ]
+		if ( $self->name eq 'Header' and $self->level == 1
+			and stringify($self) =~ /^Notes/ );
+	return;
+}
+
+sub toc {
+	my $self = shift;
+	return unless ( $self->name eq 'Header' and $self->level == 1
+		and stringify($self) =~ /^Table/ );
+	return RawBlock( 'latex',
+		'{\\hypersetup{linkcolor=black}\\setcounter{tocdepth}{3}\\tableofcontents}'
+	);
+}
-- 
cgit v1.2.3