#!/usr/bin/perl
# Copyright 2014 Jeffrey Kegler
# This file is part of Marpa::R2.  Marpa::R2 is free software: you can
# redistribute it and/or modify it under the terms of the GNU Lesser
# General Public License as published by the Free Software Foundation,
# either version 3 of the License, or (at your option) any later version.
#
# Marpa::R2 is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser
# General Public License along with Marpa::R2.  If not, see
# http://www.gnu.org/licenses/.

=head1 Name

html_fmt - Reformat HTML, indented according to structure

=head1 Synopsis

=for Marpa::R2::Display
ignore: 1

    html_fmt [--avoid-whitespace=[yes|comment|no]]
      [--no-ws-ok-before-end-tag|--no-ws-ok-before-end-tag]
      [--no-ws-ok-after-start-tag|--no-ws-ok-after-start-tag]
      [--added-tag-comment|--no-added-tag-comment]
      [uri|file]

=for Marpa::R2::Display::End

=head1 Description

Writes its input to C<STDOUT>,
reformatted and
indented according to the HTML structure.
With no arguments,
C<html_fmt> looks for its input
on C<STDIN>.
If it has an argument that looks like a URI,
C<html_fmt> treats that argument
as a URI.
If it has an argument that does not
look like a URI,
C<html_fmt> uses that argument
as the name
of its input file.
C<html_fmt> considers its argument to
"look like a URI",
if it
starts with alphanumerics followed by a colon.

When reformatting and indenting,
C<html_fmt> takes measures
to avoid introducing
whitespace that will affect the way in which the HTML
displays.
The standards differ on whitespace treatment;
when the standards agree
they often allow considerable
latitude to the browsers;
and specific implementions will
exploit the allowed latitude in different ways,
or exceed it.
All of which is to say that
C<html_fmt>'s attempts to avoid introducing
whitespace are not always successful.

C<html_fmt> supplies missing start and end tags.
If C<html_fmt> cannot find required start and end tags,
it supplies them.
C<html_fmt>'s grammar is very liberal in its interpretation
of what is valid HTML.
Finally, if this is not sufficient to turn the input
into valid HTML,
C<html_fmt>
will treat problem sections of the input
as noise or "cruft",
ignoring them in determining
the structure of the document.
C<html_fmt> never rejects its input,
whatever its contents.

When
C<html_fmt> adds
a missing start tag,
it follows the new start tag with a comment.
When
C<html_fmt> adds
a missing end tag,
it preceeds the new end tag with a comment.
When C<html_fmt> classifies characters
as "cruft",
it adds a comment to that effect before the "cruft".

C<pre> and C<script>
elements receive special treatment.
The contents of
C<pre> and C<script> elements are not reformatted.
When missing tags or cruft occur inside a C<pre> element,
the comments to that effect are placed
before the C<< <pre> >> start tag.
The contents of C<script> elements are not
examined.

=head1 Options

No single set of reformatting choices is anywhere
near adequate to the variety of standards,
renderers and application requirements out there.
Ideally,
C<html_fmt> would have many options customizing its behavior.
As of this writing, there are only a few.

=head2 avoid-whitespace

When adding indentation,
C<html_fmt> runs the risk of introducing
whitespace that will change
the way in which the document is displayed.
How C<html_fmt> deals with this risk
is controlled by the
the C<--avoid-whitespace> option,
which
may take one of three values:
C<yes>, C<no> and C<comment>.

If the value
of the C<--avoid-whitespace> option
is C<yes>,
C<html_fmt> does not add indentation
in whitespace-sensitive places.
This avoids clutter but,
because of the missed indentation,
the structure of the 
document becomes harder to follow.
C<yes> is the default value.

If the value is C<comment>,
HTML (SGML) comments are used to achieve proper
indentation without introducing whitespace.
This works, but does get a bit cluttered.

If the value is C<no>,
C<html_fmt> 
adds whitespace whenever it is convenient
for formatting, regardless of whether or
not it may change the way in which
the document displays.

=head2 added-tag-comment

The C<--added-tag-comment> option is a Boolean.
If set, whenever C<html_fmt> supplies a missing start
or end tag, C<html_fmt> will include a comment to that effect.
If the boolean is unset with
C<--no-added-tag-comment>,
C<html_fmt> will be silent when supply missing start
and end tags.
C<--added-tag-comment> is the default.

=head2 ws-ok-after-start-tag

The C<--ws-ok-after-start-tag> option is a Boolean.
If set, C<html_fmt> will add whitespace after an start tag when
that is useful.
If the SGML standards
are adhered to, the added whitespace will not
be rendered.
However, the HTML standards, while they allow SGML-adherent behavior
for whitespace after start tags,
do not require it.
In some case in practice,
browsers will render whitespace found after
start tags.

If C<--ws-ok-after-start-tag> is false,
C<html_fmt> will not add whitespace after
start tags.
But it increases the ability of
C<html_fmt> to make its output readable,
but may result in a change in what some
browsers display.

=head2 ws-ok-before-end-tag

The C<--ws-ok-before-end-tag> option is a Boolean.
If set, C<html_fmt> will add whitespace before an end tag when
that is useful.
If the SGML standards
are adhered to, the added whitespace will not
be rendered.
However, the HTML standards, while they allow SGML-adherent behavior
for whitespace before end tags,
do not require it.
In practice,
browsers will often render whitespace found before
end tags.

If C<--ws-ok-before-end-tag> is false,
C<html_fmt> will not add whitespace before
end tags.
C<--ws-ok-before-end-tag> is false by default.
This is always safe,
in the sense that it will not change
what is there to be rendered.
But it limits the ability of
C<html_fmt> to make its output readable.

=head1 Example

Given this input:

=for Marpa::R2::Display
ignore: 1

    <title>Test page<tr>x<head attr="I am cruft"><p>Final graf

=for Marpa::R2::Display::End

to this command:

=for Marpa::R2::Display
ignore: 1

    html_fmt --avoid-whitespace=comment --ws-ok-before-end-tag

=for Marpa::R2::Display::End

the output is

=for Marpa::R2::Display
ignore: 1

  <html>
    <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
    <head>
      <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
      <title>
        Test page
        <!-- html_fmt: Following end tag is replacement for a missing one -->
      </title>
      <!-- html_fmt: Following end tag is replacement for a missing one -->
    </head><!--
      html_fmt: this comment is to avoid introducing whitespace
    --><body>
      <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
      <table>
        <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
        <tbody>
          <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
          <tr>
            <td>
              <!-- html_fmt: Preceeding start tag is replacement for a missing one -->
              x<!--
                html_fmt: this comment is to avoid introducing whitespace
              --><!-- html_fmt: Next line is cruft --><!--
                html_fmt: this comment is to avoid introducing whitespace
              --><head attr="I am cruft"><!--
                html_fmt: this comment is to avoid introducing whitespace
              --><p>
                Final graf
                <!-- html_fmt: Following end tag is replacement for a missing one -->
              </p>
              <!-- html_fmt: Following end tag is replacement for a missing one -->
            </td>
            <!-- html_fmt: Following end tag is replacement for a missing one -->
          </tr>
          <!-- html_fmt: Following end tag is replacement for a missing one -->
        </tbody>
        <!-- html_fmt: Following end tag is replacement for a missing one -->
      </table>
      <!-- html_fmt: Following end tag is replacement for a missing one -->
    </body>
    <!-- html_fmt: Following end tag is replacement for a missing one -->
  </html>

=for Marpa::R2::Display::End

=head1 Acknowledgements

The starting template for this code was
L<HTML::TokeParser|HTML::TokeParser>, by Gisle Aas.

=head1 Copyright and License

=for Marpa::R2::Display
ignore: 1

  Copyright 2014 Jeffrey Kegler
  This file is part of Marpa::R2.  Marpa::R2 is free software: you can
  redistribute it and/or modify it under the terms of the GNU Lesser
  General Public License as published by the Free Software Foundation,
  either version 3 of the License, or (at your option) any later version.

  Marpa::R2 is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  Lesser General Public License for more details.

  You should have received a copy of the GNU Lesser
  General Public License along with Marpa::R2.  If not, see
  http://www.gnu.org/licenses/.

=for Marpa::R2::Display::End

=cut

# vim: set expandtab shiftwidth=4: