From 36af0104d8af10bf90053f4a24d61e6e7d3f174e Mon Sep 17 00:00:00 2001
From: trizen <trizen@protonmail.com>
Date: Sat, 30 Jul 2022 19:18:26 +0300
Subject: [PATCH] new file:   Converters/code2pdf.pl

---
 Converters/code2pdf.pl    | 144 ++++++++++++++++++++++++++++++++++++++
 Converters/gitbook2pdf.pl |   4 +-
 README.md                 |   1 +
 3 files changed, 147 insertions(+), 2 deletions(-)
 create mode 100755 Converters/code2pdf.pl
 mode change 100644 => 100755 Converters/gitbook2pdf.pl

diff --git a/Converters/code2pdf.pl b/Converters/code2pdf.pl
new file mode 100755
index 00000000..5945f79d
--- /dev/null
+++ b/Converters/code2pdf.pl
@@ -0,0 +1,144 @@
+#!/usr/bin/perl
+
+# Author: Trizen
+# Date: 30 July 2022
+# https://github.com/trizen
+
+# Code to PDF converter, with syntax highlighting, given a summary file.
+
+# Uses the following tools:
+#   md2html         -- for converting markdown to HTML
+#   markdown2pdf.pl -- for converting markdown to PDF (with syntax highlighting)
+
+use 5.010;
+use strict;
+use warnings;
+
+use open IO => ':utf8', ':std';
+use HTML::TreeBuilder 5 ('-weak');
+
+use Encode       qw(decode_utf8 encode_utf8);
+use Getopt::Long qw(GetOptions);
+use URI::Escape  qw(uri_unescape);
+use Digest::MD5  qw(md5_hex);
+
+my $markdown2pdf = "markdown2pdf.pl";    # path to the `markdown2pdf.pl` script
+
+my $style = 'github';
+my $title = 'Document';
+my $lang  = 'perl';
+
+sub usage {
+    my ($exit_code) = @_;
+    $exit_code //= 0;
+
+    print <<"EOT";
+usage: $0 [options] [SUMMARY.md] [output.pdf]
+
+options:
+
+    --style=s   : style theme for `highlight` (default: $style)
+    --title=s   : title of the PDF file (default: $title)
+    --lang=s    : language code used for highlighting (default: $lang)
+
+EOT
+
+    exit($exit_code);
+}
+
+GetOptions(
+           "style=s" => \$style,
+           "title=s" => \$title,
+           "lang=s"  => \$lang,
+           "h|help"  => sub { usage(0) },
+          )
+  or die("Error in command line arguments\n");
+
+my $input_markdown_file = $ARGV[0] // usage(2);
+my $output_pdf_file     = $ARGV[1] // "OUTPUT.pdf";
+
+say ":: Converting $input_markdown_file to HTML...";
+my $html = `md2html $input_markdown_file`;
+
+if ($? != 0) {
+    die "`md2html` failed with code: $?";
+}
+
+my $tree = HTML::TreeBuilder->new();
+$tree->parse($html);
+$tree->eof();
+
+#my @nodes = $tree->guts();
+my @nodes = $tree->disembowel();
+
+say ":: Reading Markdown files...";
+my $markdown_content = '';
+
+sub expand_ul {
+    my ($ul, $depth) = @_;
+
+    foreach my $t (@{$ul->content}) {
+        if ($t->tag eq 'li') {
+            foreach my $x (@{$t->content}) {
+
+                if (!ref($x)) {
+                    $markdown_content .= ("#" x $depth) . ' ' . $x . "\n\n";
+                    next;
+                }
+
+                if ($x->tag eq 'ul') {
+                    expand_ul($x, $depth + 1);
+                }
+                else {
+                    if ($x->tag eq 'a') {
+
+                        my $href = $x->attr('href');
+                        my $file = decode_utf8(uri_unescape($href));
+
+                        if (not -e $file) {
+                            warn ":: File <<$file>> does not exist. Skipping...\n";
+                            next;
+                        }
+
+                        if (open my $fh, '<:utf8', $file) {
+                            local $/;
+                            $markdown_content .= ("#" x $depth) . ' ' . $x->content->[0] . "\n\n";
+                            $markdown_content .= "```$lang\n";
+                            $markdown_content .= <$fh>;
+                            if (substr($markdown_content, -1) ne "\n") {
+                                $markdown_content .= "\n";
+                            }
+                            $markdown_content .= "```\n\n";
+                        }
+                        else {
+                            warn ":: Cannot open file <<$file>> for reading: $!\n";
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
+
+foreach my $entry (@nodes) {
+    if ($entry->tag eq 'ul') {
+        expand_ul($entry, 1);
+    }
+}
+
+my $markdown_file = "$output_pdf_file.md";
+
+open my $fh, '>:utf8', $markdown_file
+  or die "Can't open file <<$markdown_file>> for writing: $!";
+
+print $fh $markdown_content;
+close $fh;
+
+say ":: Converting Markdown to PDF...";
+system($markdown2pdf, "--style", $style, "--title", $title, $markdown_file, $output_pdf_file);
+
+unlink($markdown_file);
+
+if ($? != 0) {
+    die "`$markdown2pdf` failed with code: $?";
+}
diff --git a/Converters/gitbook2pdf.pl b/Converters/gitbook2pdf.pl
old mode 100644
new mode 100755
index d5d2b032..6203f623
--- a/Converters/gitbook2pdf.pl
+++ b/Converters/gitbook2pdf.pl
@@ -1,4 +1,4 @@
-#!/usr/bin/perl
+#~ #!/usr/bin/perl
 
 # Author: Trizen
 # Date: 30 July 2022
@@ -113,7 +113,7 @@ sub expand_ul {
 
 foreach my $entry (@nodes) {
     if ($entry->tag eq 'ul') {
-        expand_ul($entry, 0);
+        expand_ul($entry, 1);
     }
 }
 
diff --git a/README.md b/README.md
index b44fd722..aa6a14df 100644
--- a/README.md
+++ b/README.md
@@ -38,6 +38,7 @@ A nice collection of day-to-day Perl scripts.
 * Converters
     * [Any to 3gp](./Converters/any_to_3gp.pl)
     * [Ass2srt](./Converters/ass2srt.pl)
+    * [Code2pdf](./Converters/code2pdf.pl)
     * [Euler2pdf](./Converters/euler2pdf.pl)
     * [From hex](./Converters/from_hex.pl)
     * [Gitbook2pdf](./Converters/gitbook2pdf.pl)