From 4d3abe74035d7d3c2c5d6927cc0cbb9032486b30 Mon Sep 17 00:00:00 2001 From: Jonas Smedegaard Date: Fri, 31 Aug 2012 15:12:31 +0200 Subject: Add tidying scripts. --- report/mkmd.sh | 18 ++++++++++++++++++ report/normalize-html.pl | 29 +++++++++++++++++++++++++++++ 2 files changed, 47 insertions(+) create mode 100755 report/mkmd.sh create mode 100755 report/normalize-html.pl (limited to 'report') diff --git a/report/mkmd.sh b/report/mkmd.sh new file mode 100755 index 0000000..39ec83d --- /dev/null +++ b/report/mkmd.sh @@ -0,0 +1,18 @@ +#!/bin/sh + +set -e + +#stem="Greens-IslandsofResiliences" +stem="Greens-IslandsofResilience" + +pdftohtml -q -c -s -i -noframes -stdout "$stem.pdf" tmp + +infile=tmp.html + +outfile="${2:-$(basename "$infile" .html).md}" + +#perl ./normalize-html.pl tmp.html | less +perl ./normalize-html.pl tmp.html \ + |pandoc --normalize --no-wrap --parse-raw -f html -t markdown -o "$stem.md" + +pandoc --standalone --toc -f markdown -t html5 -i "$stem.md" -o "$stem.html" diff --git a/report/normalize-html.pl b/report/normalize-html.pl new file mode 100755 index 0000000..7245a4e --- /dev/null +++ b/report/normalize-html.pl @@ -0,0 +1,29 @@ +#!/usr/bin/perl + +use File::Slurp; + +my $_ = read_file( $ARGV[0] ) ; + +#s!( )! !g; # filler horisontal space +s!( )+! !g; # filler horisontal space +#s!(\s*
\s*)+!
!g; # filler vertical space +s!(\s*
\s*)+!
!g; # filler vertical space +s,
(?=(?:)?

),,ig; # filler horisontal space +s,

\s*

,,ig; # filler horisontal space +s!\s*\s*\s*!!g; # filler vertical space +s!!!gi; # page dividers +#s!]+>\K
(.+?)!

$1

!i; # H1 +#s!
(.+?)!

$1

!i; # H1 +s!\.(ft\d+){font\-size:52px;.*?\K]*\bclass="\1"[^>]*>(.*?)

$2

!is; +s!\.(ft\d+){font\-size:19px;.*?\K]*\bclass="\1"[^>]*>(.*?)

$2

!isg; +s!\.(ft\d+){font\-size:16px;.*?\K]*\bclass="\1"[^>]*>(.*?)

$2

!isg; +s!]*>([^<\s][^<]*?)

$1

!isg; +s,]*>(?!),
,isg; +s,,,isg; +s,,,isg; +s,