ThirdParty/Pygments/pygments/formatters/html.py

# -*- coding: utf-8 -*-
"""
    pygments.formatters.html
    ~~~~~~~~~~~~~~~~~~~~~~~~

    Formatter for HTML output.

    :copyright: Copyright 2006-2009 by the Pygments team, see AUTHORS.
    :license: BSD, see LICENSE for details.
"""
import sys, os
import StringIO

try:
    set
except NameError:
    from sets import Set as set

from pygments.formatter import Formatter
from pygments.token import Token, Text, STANDARD_TYPES
from pygments.util import get_bool_opt, get_int_opt, get_list_opt, bytes


__all__ = ['HtmlFormatter']


def escape_html(text):
    """Escape &, <, > as well as single and double quotes for HTML."""
    return text.replace('&', '&amp;').  \
                replace('<', '&lt;').   \
                replace('>', '&gt;').   \
                replace('"', '&quot;'). \
                replace("'", '&#39;')


def get_random_id():
    """Return a random id for javascript fields."""
    from random import random
    from time import time
    try:
        from hashlib import sha1 as sha
    except ImportError:
        import sha
        sha = sha.new
    return sha('%s|%s' % (random(), time())).hexdigest()


def _get_ttype_class(ttype):
    fname = STANDARD_TYPES.get(ttype)
    if fname:
        return fname
    aname = ''
    while fname is None:
        aname = '-' + ttype[-1] + aname
        ttype = ttype.parent
        fname = STANDARD_TYPES.get(ttype)
    return fname + aname


CSSFILE_TEMPLATE = '''\
td.linenos { background-color: #f0f0f0; padding-right: 10px; }
span.lineno { background-color: #f0f0f0; padding: 0 5px 0 5px; }
pre { line-height: 125%%; }
%(styledefs)s
'''

DOC_HEADER = '''\
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
   "http://www.w3.org/TR/html4/strict.dtd">

<html>
<head>
  <title>%(title)s</title>
  <meta http-equiv="content-type" content="text/html; charset=%(encoding)s">
  <style type="text/css">
''' + CSSFILE_TEMPLATE + '''
  </style>
</head>
<body>
<h2>%(title)s</h2>

'''

DOC_HEADER_EXTERNALCSS = '''\
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN"
   "http://www.w3.org/TR/html4/strict.dtd">

<html>
<head>
  <title>%(title)s</title>
  <meta http-equiv="content-type" content="text/html; charset=%(encoding)s">
  <link rel="stylesheet" href="%(cssfile)s" type="text/css">
</head>
<body>
<h2>%(title)s</h2>

'''

DOC_FOOTER = '''\
</body>
</html>
'''


class HtmlFormatter(Formatter):
    r"""
    Format tokens as HTML 4 ``<span>`` tags within a ``<pre>`` tag, wrapped
    in a ``<div>`` tag. The ``<div>``'s CSS class can be set by the `cssclass`
    option.

    If the `linenos` option is set to ``"table"``, the ``<pre>`` is
    additionally wrapped inside a ``<table>`` which has one row and two
    cells: one containing the line numbers and one containing the code.
    Example:

    .. sourcecode:: html

        <div class="highlight" >
        <table><tr>
          <td class="linenos" title="click to toggle"
            onclick="with (this.firstChild.style)
                     { display = (display == '') ? 'none' : '' }">
            <pre>1
            2</pre>
          </td>
          <td class="code">
            <pre><span class="Ke">def </span><span class="NaFu">foo</span>(bar):
              <span class="Ke">pass</span>
            </pre>
          </td>
        </tr></table></div>

    (whitespace added to improve clarity).

    Wrapping can be disabled using the `nowrap` option.

    A list of lines can be specified using the `hl_lines` option to make these
    lines highlighted (as of Pygments 0.11).

    With the `full` option, a complete HTML 4 document is output, including
    the style definitions inside a ``<style>`` tag, or in a separate file if
    the `cssfile` option is given.

    The `get_style_defs(arg='')` method of a `HtmlFormatter` returns a string
    containing CSS rules for the CSS classes used by the formatter. The
    argument `arg` can be used to specify additional CSS selectors that
    are prepended to the classes. A call `fmter.get_style_defs('td .code')`
    would result in the following CSS classes:

    .. sourcecode:: css

        td .code .kw { font-weight: bold; color: #00FF00 }
        td .code .cm { color: #999999 }
        ...

    If you have Pygments 0.6 or higher, you can also pass a list or tuple to the
    `get_style_defs()` method to request multiple prefixes for the tokens:

    .. sourcecode:: python

        formatter.get_style_defs(['div.syntax pre', 'pre.syntax'])

    The output would then look like this:

    .. sourcecode:: css

        div.syntax pre .kw,
        pre.syntax .kw { font-weight: bold; color: #00FF00 }
        div.syntax pre .cm,
        pre.syntax .cm { color: #999999 }
        ...

    Additional options accepted:

    `nowrap`
        If set to ``True``, don't wrap the tokens at all, not even inside a ``<pre>``
        tag. This disables most other options (default: ``False``).

    `full`
        Tells the formatter to output a "full" document, i.e. a complete
        self-contained document (default: ``False``).

    `title`
        If `full` is true, the title that should be used to caption the
        document (default: ``''``).

    `style`
        The style to use, can be a string or a Style subclass (default:
        ``'default'``). This option has no effect if the `cssfile`
        and `noclobber_cssfile` option are given and the file specified in
        `cssfile` exists.

    `noclasses`
        If set to true, token ``<span>`` tags will not use CSS classes, but
        inline styles. This is not recommended for larger pieces of code since
        it increases output size by quite a bit (default: ``False``).

    `classprefix`
        Since the token types use relatively short class names, they may clash
        with some of your own class names. In this case you can use the
        `classprefix` option to give a string to prepend to all Pygments-generated
        CSS class names for token types.
        Note that this option also affects the output of `get_style_defs()`.

    `cssclass`
        CSS class for the wrapping ``<div>`` tag (default: ``'highlight'``).
        If you set this option, the default selector for `get_style_defs()`
        will be this class.

        *New in Pygments 0.9:* If you select the ``'table'`` line numbers, the
        wrapping table will have a CSS class of this string plus ``'table'``,
        the default is accordingly ``'highlighttable'``.

    `cssstyles`
        Inline CSS styles for the wrapping ``<div>`` tag (default: ``''``).

    `prestyles`
        Inline CSS styles for the ``<pre>`` tag (default: ``''``).  *New in
        Pygments 0.11.*

    `cssfile`
        If the `full` option is true and this option is given, it must be the
        name of an external file. If the filename does not include an absolute
        path, the file's path will be assumed to be relative to the main output
        file's path, if the latter can be found. The stylesheet is then written
        to this file instead of the HTML file. *New in Pygments 0.6.*

    `noclobber_cssfile`
        If `cssfile` is given and the specified file exists, the css file will
        not be overwritten. This allows the use of the `full` option in
        combination with a user specified css file. Default is ``False``.
        *New in Pygments 1.1.*

    `linenos`
        If set to ``'table'``, output line numbers as a table with two cells,
        one containing the line numbers, the other the whole code.  This is
        copy-and-paste-friendly, but may cause alignment problems with some
        browsers or fonts.  If set to ``'inline'``, the line numbers will be
        integrated in the ``<pre>`` tag that contains the code (that setting
        is *new in Pygments 0.8*).

        For compatibility with Pygments 0.7 and earlier, every true value
        except ``'inline'`` means the same as ``'table'`` (in particular, that
        means also ``True``).

        The default value is ``False``, which means no line numbers at all.

        **Note:** with the default ("table") line number mechanism, the line
        numbers and code can have different line heights in Internet Explorer
        unless you give the enclosing ``<pre>`` tags an explicit ``line-height``
        CSS property (you get the default line spacing with ``line-height:
        125%``).

    `hl_lines`
        Specify a list of lines to be highlighted.  *New in Pygments 0.11.*

    `linenostart`
        The line number for the first line (default: ``1``).

    `linenostep`
        If set to a number n > 1, only every nth line number is printed.

    `linenospecial`
        If set to a number n > 0, every nth line number is given the CSS
        class ``"special"`` (default: ``0``).

    `nobackground`
        If set to ``True``, the formatter won't output the background color
        for the wrapping element (this automatically defaults to ``False``
        when there is no wrapping element [eg: no argument for the
        `get_syntax_defs` method given]) (default: ``False``). *New in
        Pygments 0.6.*

    `lineseparator`
        This string is output between lines of code. It defaults to ``"\n"``,
        which is enough to break a line inside ``<pre>`` tags, but you can
        e.g. set it to ``"<br>"`` to get HTML line breaks. *New in Pygments
        0.7.*

    `lineanchors`
        If set to a nonempty string, e.g. ``foo``, the formatter will wrap each
        output line in an anchor tag with a ``name`` of ``foo-linenumber``.
        This allows easy linking to certain lines. *New in Pygments 0.9.*

    `anchorlinenos`
        If set to `True`, will wrap line numbers in <a> tags. Used in
        combination with `linenos` and `lineanchors`.


    **Subclassing the HTML formatter**

    *New in Pygments 0.7.*

    The HTML formatter is now built in a way that allows easy subclassing, thus
    customizing the output HTML code. The `format()` method calls
    `self._format_lines()` which returns a generator that yields tuples of ``(1,
    line)``, where the ``1`` indicates that the ``line`` is a line of the
    formatted source code.

    If the `nowrap` option is set, the generator is the iterated over and the
    resulting HTML is output.

    Otherwise, `format()` calls `self.wrap()`, which wraps the generator with
    other generators. These may add some HTML code to the one generated by
    `_format_lines()`, either by modifying the lines generated by the latter,
    then yielding them again with ``(1, line)``, and/or by yielding other HTML
    code before or after the lines, with ``(0, html)``. The distinction between
    source lines and other code makes it possible to wrap the generator multiple
    times.

    The default `wrap()` implementation adds a ``<div>`` and a ``<pre>`` tag.

    A custom `HtmlFormatter` subclass could look like this:

    .. sourcecode:: python

        class CodeHtmlFormatter(HtmlFormatter):

            def wrap(self, source, outfile):
                return self._wrap_code(source)

            def _wrap_code(self, source):
                yield 0, '<code>'
                for i, t in source:
                    if i == 1:
                        # it's a line of formatted code
                        t += '<br>'
                    yield i, t
                yield 0, '</code>'

    This results in wrapping the formatted lines with a ``<code>`` tag, where the
    source lines are broken using ``<br>`` tags.

    After calling `wrap()`, the `format()` method also adds the "line numbers"
    and/or "full document" wrappers if the respective options are set. Then, all
    HTML yielded by the wrapped generator is output.
    """

    name = 'HTML'
    aliases = ['html']
    filenames = ['*.html', '*.htm']

    def __init__(self, **options):
        Formatter.__init__(self, **options)
        self.title = self._decodeifneeded(self.title)
        self.nowrap = get_bool_opt(options, 'nowrap', False)
        self.noclasses = get_bool_opt(options, 'noclasses', False)
        self.classprefix = options.get('classprefix', '')
        self.cssclass = self._decodeifneeded(options.get('cssclass', 'highlight'))
        self.cssstyles = self._decodeifneeded(options.get('cssstyles', ''))
        self.prestyles = self._decodeifneeded(options.get('prestyles', ''))
        self.cssfile = self._decodeifneeded(options.get('cssfile', ''))
        self.noclobber_cssfile = get_bool_opt(options, 'noclobber_cssfile', False)

        linenos = options.get('linenos', False)
        if linenos == 'inline':
            self.linenos = 2
        elif linenos:
            # compatibility with <= 0.7
            self.linenos = 1
        else:
            self.linenos = 0
        self.linenostart = abs(get_int_opt(options, 'linenostart', 1))
        self.linenostep = abs(get_int_opt(options, 'linenostep', 1))
        self.linenospecial = abs(get_int_opt(options, 'linenospecial', 0))
        self.nobackground = get_bool_opt(options, 'nobackground', False)
        self.lineseparator = options.get('lineseparator', '\n')
        self.lineanchors = options.get('lineanchors', '')
        self.anchorlinenos = options.get('anchorlinenos', False)
        self.hl_lines = set()
        for lineno in get_list_opt(options, 'hl_lines', []):
            try:
                self.hl_lines.add(int(lineno))
            except ValueError:
                pass

        self._class_cache = {}
        self._create_stylesheet()

    def _get_css_class(self, ttype):
        """Return the css class of this token type prefixed with
        the classprefix option."""
        if ttype in self._class_cache:
            return self._class_cache[ttype]
        return self.classprefix + _get_ttype_class(ttype)

    def _create_stylesheet(self):
        t2c = self.ttype2class = {Token: ''}
        c2s = self.class2style = {}
        cp = self.classprefix
        for ttype, ndef in self.style:
            name = cp + _get_ttype_class(ttype)
            style = ''
            if ndef['color']:
                style += 'color: #%s; ' % ndef['color']
            if ndef['bold']:
                style += 'font-weight: bold; '
            if ndef['italic']:
                style += 'font-style: italic; '
            if ndef['underline']:
                style += 'text-decoration: underline; '
            if ndef['bgcolor']:
                style += 'background-color: #%s; ' % ndef['bgcolor']
            if ndef['border']:
                style += 'border: 1px solid #%s; ' % ndef['border']
            if style:
                t2c[ttype] = name
                # save len(ttype) to enable ordering the styles by
                # hierarchy (necessary for CSS cascading rules!)
                c2s[name] = (style[:-2], ttype, len(ttype))

    def get_style_defs(self, arg=None):
        """
        Return CSS style definitions for the classes produced by the current
        highlighting style. ``arg`` can be a string or list of selectors to
        insert before the token type classes.
        """
        if arg is None:
            arg = ('cssclass' in self.options and '.'+self.cssclass or '')
        if isinstance(arg, basestring):
            args = [arg]
        else:
            args = list(arg)

        def prefix(cls):
            if cls:
                cls = '.' + cls
            tmp = []
            for arg in args:
                tmp.append((arg and arg + ' ' or '') + cls)
            return ', '.join(tmp)

        styles = [(level, ttype, cls, style)
                  for cls, (style, ttype, level) in self.class2style.iteritems()
                  if cls and style]
        styles.sort()
        lines = ['%s { %s } /* %s */' % (prefix(cls), style, repr(ttype)[6:])
                 for (level, ttype, cls, style) in styles]
        if arg and not self.nobackground and \
           self.style.background_color is not None:
            text_style = ''
            if Text in self.ttype2class:
                text_style = ' ' + self.class2style[self.ttype2class[Text]][0]
            lines.insert(0, '%s { background: %s;%s }' %
                         (prefix(''), self.style.background_color, text_style))
        if self.style.highlight_color is not None:
            lines.insert(0, '%s.hll { background-color: %s }' %
                         (prefix(''), self.style.highlight_color))
        return '\n'.join(lines)

    def _decodeifneeded(self, value):
        if isinstance(value, bytes):
            if self.encoding:
                return value.decode(self.encoding)
            return value.decode()
        return value

    def _wrap_full(self, inner, outfile):
        if self.cssfile:
            if os.path.isabs(self.cssfile):
                # it's an absolute filename
                cssfilename = self.cssfile
            else:
                try:
                    filename = outfile.name
                    if not filename or filename[0] == '<':
                        # pseudo files, e.g. name == '<fdopen>'
                        raise AttributeError
                    cssfilename = os.path.join(os.path.dirname(filename),
                                               self.cssfile)
                except AttributeError:
                    print >>sys.stderr, 'Note: Cannot determine output file name, ' \
                          'using current directory as base for the CSS file name'
                    cssfilename = self.cssfile
            # write CSS file only if noclobber_cssfile isn't given as an option.
            try:
                if not os.path.exists(cssfilename) or not self.noclobber_cssfile:
                    cf = open(cssfilename, "w")
                    cf.write(CSSFILE_TEMPLATE %
                            {'styledefs': self.get_style_defs('body')})
                    cf.close()
            except IOError, err:
                err.strerror = 'Error writing CSS file: ' + err.strerror
                raise

            yield 0, (DOC_HEADER_EXTERNALCSS %
                      dict(title     = self.title,
                           cssfile   = self.cssfile,
                           encoding  = self.encoding))
        else:
            yield 0, (DOC_HEADER %
                      dict(title     = self.title,
                           styledefs = self.get_style_defs('body'),
                           encoding  = self.encoding))

        for t, line in inner:
            yield t, line
        yield 0, DOC_FOOTER

    def _wrap_tablelinenos(self, inner):
        dummyoutfile = StringIO.StringIO()
        lncount = 0
        for t, line in inner:
            if t:
                lncount += 1
            dummyoutfile.write(line)

        fl = self.linenostart
        mw = len(str(lncount + fl - 1))
        sp = self.linenospecial
        st = self.linenostep
        la = self.lineanchors
        aln = self.anchorlinenos
        if sp:
            lines = []

            for i in range(fl, fl+lncount):
                if i % st == 0:
                    if i % sp == 0:
                        if aln:
                            lines.append('<a href="#%s-%d" class="special">%*d</a>' %
                                         (la, i, mw, i))
                        else:
                            lines.append('<span class="special">%*d</span>' % (mw, i))
                    else:
                        if aln:
                            lines.append('<a href="#%s-%d">%*d</a>' % (la, i, mw, i))
                        else:
                            lines.append('%*d' % (mw, i))
                else:
                    lines.append('')
            ls = '\n'.join(lines)
        else:
            lines = []
            for i in range(fl, fl+lncount):
                if i % st == 0:
                    if aln:
                        lines.append('<a href="#%s-%d">%*d</a>' % (la, i, mw, i))
                    else:
                        lines.append('%*d' % (mw, i))
                else:
                    lines.append('')
            ls = '\n'.join(lines)

        # in case you wonder about the seemingly redundant <div> here: since the
        # content in the other cell also is wrapped in a div, some browsers in
        # some configurations seem to mess up the formatting...
        yield 0, ('<table class="%stable">' % self.cssclass +
                  '<tr><td class="linenos"><div class="linenodiv"><pre>' +
                  ls + '</pre></div></td><td class="code">')
        yield 0, dummyoutfile.getvalue()
        yield 0, '</td></tr></table>'

    def _wrap_inlinelinenos(self, inner):
        # need a list of lines since we need the width of a single number :(
        lines = list(inner)
        sp = self.linenospecial
        st = self.linenostep
        num = self.linenostart
        mw = len(str(len(lines) + num - 1))

        if sp:
            for t, line in lines:
                yield 1, '<span class="lineno%s">%*s</span> ' % (
                    num%sp == 0 and ' special' or '', mw,
                    (num%st and ' ' or num)) + line
                num += 1
        else:
            for t, line in lines:
                yield 1, '<span class="lineno">%*s</span> ' % (
                    mw, (num%st and ' ' or num)) + line
                num += 1

    def _wrap_lineanchors(self, inner):
        s = self.lineanchors
        i = 0
        for t, line in inner:
            if t:
                i += 1
                yield 1, '<a name="%s-%d"></a>' % (s, i) + line
            else:
                yield 0, line

    def _wrap_div(self, inner):
        yield 0, ('<div' + (self.cssclass and ' class="%s"' % self.cssclass)
                  + (self.cssstyles and ' style="%s"' % self.cssstyles) + '>')
        for tup in inner:
            yield tup
        yield 0, '</div>\n'

    def _wrap_pre(self, inner):
        yield 0, ('<pre'
                  + (self.prestyles and ' style="%s"' % self.prestyles) + '>')
        for tup in inner:
            yield tup
        yield 0, '</pre>'

    def _format_lines(self, tokensource):
        """
        Just format the tokens, without any wrapping tags.
        Yield individual lines.
        """
        nocls = self.noclasses
        lsep = self.lineseparator
        # for <span style=""> lookup only
        getcls = self.ttype2class.get
        c2s = self.class2style

        lspan = ''
        line = ''
        for ttype, value in tokensource:
            if nocls:
                cclass = getcls(ttype)
                while cclass is None:
                    ttype = ttype.parent
                    cclass = getcls(ttype)
                cspan = cclass and '<span style="%s">' % c2s[cclass][0] or ''
            else:
                cls = self._get_css_class(ttype)
                cspan = cls and '<span class="%s">' % cls or ''

            parts = escape_html(value).split('\n')

            # for all but the last line
            for part in parts[:-1]:
                if line:
                    if lspan != cspan:
                        line += (lspan and '</span>') + cspan + part + \
                                (cspan and '</span>') + lsep
                    else: # both are the same
                        line += part + (lspan and '</span>') + lsep
                    yield 1, line
                    line = ''
                elif part:
                    yield 1, cspan + part + (cspan and '</span>') + lsep
                else:
                    yield 1, lsep
            # for the last line
            if line and parts[-1]:
                if lspan != cspan:
                    line += (lspan and '</span>') + cspan + parts[-1]
                    lspan = cspan
                else:
                    line += parts[-1]
            elif parts[-1]:
                line = cspan + parts[-1]
                lspan = cspan
            # else we neither have to open a new span nor set lspan

        if line:
            yield 1, line + (lspan and '</span>') + lsep

    def _highlight_lines(self, tokensource):
        """
        Highlighted the lines specified in the `hl_lines` option by
        post-processing the token stream coming from `_format_lines`.
        """
        hls = self.hl_lines

        for i, (t, value) in enumerate(tokensource):
            if t != 1:
                yield t, value
            if i + 1 in hls: # i + 1 because Python indexes start at 0
                yield 1, '<span class="hll">%s</span>' % value
            else:
                yield 1, value

    def wrap(self, source, outfile):
        """
        Wrap the ``source``, which is a generator yielding
        individual lines, in custom generators. See docstring
        for `format`. Can be overridden.
        """
        return self._wrap_div(self._wrap_pre(source))

    def format_unencoded(self, tokensource, outfile):
        """
        The formatting process uses several nested generators; which of
        them are used is determined by the user's options.

        Each generator should take at least one argument, ``inner``,
        and wrap the pieces of text generated by this.

        Always yield 2-tuples: (code, text). If "code" is 1, the text
        is part of the original tokensource being highlighted, if it's
        0, the text is some piece of wrapping. This makes it possible to
        use several different wrappers that process the original source
        linewise, e.g. line number generators.
        """
        source = self._format_lines(tokensource)
        if self.hl_lines:
            source = self._highlight_lines(source)
        if not self.nowrap:
            if self.linenos == 2:
                source = self._wrap_inlinelinenos(source)
            if self.lineanchors:
                source = self._wrap_lineanchors(source)
            source = self.wrap(source, outfile)
            if self.linenos == 1:
                source = self._wrap_tablelinenos(source)
            if self.full:
                source = self._wrap_full(source, outfile)

        for t, piece in source:
            outfile.write(piece)