#!/usr/bin/perl -n
# hevea-retarget-crossrefs
# Michael Ernst
# Last updated: February 20, 2013

# To use:
#   hevea-retarget-crossrefs < orig.html > new.html

# This script replaces HTML cross-references of the form
#    <a href="#htoc1">
# by cross-refenences to named labels, such as
#    <a href="#introduction">
# This script is not necessary with Hevea version 2.04 and later.

# It is required that the original .tex source file contained a \label
# command at the end of each \chapter or \[sub]section command, like so:
#   \chapter{Introduction\label{introduction}}
# The given label that will replace the "htoc" one in the .html file.

# Rationale:
# In the table of contents, Hevea (versions before 2.04) creates HTML
# cross-references that use Hevea-generated labels of the form "htoc99",
# even when a \label already exists.  This leads to users following a link
# from the table of contents, then bookmarking or mentioning that link.
# The "htoc99" link may point to a completely different section if the
# manual is reordered or even if a new section is added.  So, it is better
# for webpages not to contain the easy-to-misuse "htoc99" cross-references.

# Note:  Hevea 2.00-2.03 breaks the \label command within \section-like
# headings.  (Hevea 1.10 works properly; 2.03 is the latest as of this
# writing.)  No matter what descriptive text you use in the \label command,
# Hevea 2.0 changes it to an inscrutible label like "sec99", which makes it
# impossible to create stable links into a document from outside.  (Hevea
# 2.0 does not break other uses of \label, but as described in section
# B.4.1 of the manual (http://hevea.inria.fr/doc/), it is necessary to put
# the \label within the \section command.)  This message on the Hevea
# mailing list gives a workaround:  replace \label by \hlabel, within
# \section commands:
# https://sympa.inria.fr/sympa/arc/hevea/2012-09/msg00006.html


# This script does not work with in-place editing (perl's -i argument).

# use strict;
# use English;
# $WARNING = 1;

$debug = 0;
# $debug = 1;

# if (scalar(@ARGV) != 1) {
#   die "Expected exactly 1 argument, got " . scalar(@ARGV);
# }
# my $filename = $ARGV[0];

push @lines, $_;

END {

  for (my $i = 0; $i<scalar(@lines); $i++) {
    # Handle lines *with* htoc, substituting it by the first other anchor and moving others forward.
    if ($lines[$i] =~ s:<A NAME="(htoc[0-9]+)">(((Chapter&#XA0;)?([0-9]+|[A-Z]))(\.[0-9]+)*)(</A>)(.*?)(<A NAME="(.*?)">)</A>((<A NAME=".*"></A>)*)(</H[0-9]+>):$9$2$7$11$8$13:) {
      $mapping{$1} = $10;
      if ($debug) { print STDERR "$1 => $mapping{$1}\n"; }
    }
    # Move around the "<A NAME=" for sections *without* htoc (anything not in
    # tocdepth, which is not in a table of contents).  If the anchor comes
    # within but at the end of a header, then when going to that URL, some browsers
    # will position the header off the top of the screen.  Putting the
    # anchor at the beginning of the header fixes this problem.
    $lines[$i] =~ s:(<(H[345]) CLASS="((sub)*section|paragraph)">)(.*?)(<A NAME=".*">(</A><A NAME=".*">)*)(</A></\2>):$1$6$5$8:;
  }

  foreach my $line (@lines) {
    if ($line =~ /<A HREF="#(htoc[0-9]+)">/) {
      my $htoc = $1;
      my $replacement = $mapping{$htoc};
      if (defined($replacement)) {
        if ($debug) { print STDERR $line; }
        # Also remove "Chapter" if present, for brevity
        $line =~ s/$htoc(">)(Chapter&#XA0;)?/$replacement$1/;
        if ($debug) { print STDERR $line; }
      } else {
        print STDERR "No symbolic name for section $htoc\n";
      }
    }
    $line =~ s/(<IMG SRC="([^"]+\.[^".]+)")>/$1 ALT="$2">/g;
    print $line;
  }

}

# Local Variables:
# time-stamp-start: "^# Last updated: "
# time-stamp-end: "\\.?$"
# time-stamp-format: "%:b %:d, %:y"
# time-stamp-line-limit: 10
# End: