Available formatters

This page lists all builtin formatters.

Common options

All formatters support these options:

encoding

If given, must be an encoding name (such as "utf-8"). This will be used to convert the token strings (which are Unicode strings) to byte strings in the output (default: None). It will also be written in an encoding declaration suitable for the document format if the full option is given (e.g. a meta content-type directive in HTML or an invocation of the inputenc package in LaTeX).

If this is "" or None, Unicode strings will be written to the output file, which most file-like objects do not support. For example, pygments.highlight() will return a Unicode string if called with no outfile argument and a formatter that has encoding set to None because it uses a StringIO.StringIO object that supports Unicode arguments to write(). Using a regular file object wouldn’t work.

New in version 0.6.

outencoding

When using Pygments from the command line, any encoding option given is passed to the lexer and the formatter. This is sometimes not desirable, for example if you want to set the input encoding to "guess". Therefore, outencoding has been introduced which overrides encoding for the formatter if given.

New in version 0.7.

Formatter classes

All these classes are importable from pygments.formatters.

class BBCodeFormatter
Short names:B, B, C, o, d, e
Filenames:bbcode, bb

Format tokens with BBcodes. These formatting codes are used by many bulletin boards, so you can highlight your sourcecode with pygments before posting it there.

This formatter has no support for background colors and borders, as there are no common BBcode tags for that.

Some board systems (e.g. phpBB) don’t support colors in their [code] tag, so you can’t use the highlighting together with that tag. Text in a [code] tag usually is shown with a monospace font (which this formatter can do with the monofont option) and no spaces (which you need for indentation) are removed.

Additional options accepted:

style
The style to use, can be a string or a Style subclass (default: 'default').
codetag
If set to true, put the output into [code] tags (default: false)
monofont
If set to true, add a tag to show the code with a monospace font (default: false).
class BmpImageFormatter
Short names:i, m, g, _, b, m, p
Filenames:bmp, bitmap

Create a bitmap image from source code. This uses the Python Imaging Library to generate a pixmap from the source code.

New in version 1.0.

class GifImageFormatter
Short names:i, m, g, _, g, i, f
Filenames:gif

Create a GIF image from source code. This uses the Python Imaging Library to generate a pixmap from the source code.

New in version 1.0.

class HtmlFormatter
Short names:H, T, M, L
Filenames:html

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:

<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.

When tagsfile is set to the path of a ctags index file, it is used to generate hyperlinks from names to their definition. You must enable anchorlines and run ctags with the -n option for this to work. The python-ctags module from PyPI must be installed to use this feature; otherwise a RuntimeError will be raised.

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:

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:

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

The output would then look like this:

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 version 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 version 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 version 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 version 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 version 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 version 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 version 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 version 0.9.

linespans

If set to a nonempty string, e.g. foo, the formatter will wrap each output line in a span tag with an id of foo-linenumber. This allows easy access to lines via javascript.

New in version 1.6.

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

If set to the path of a ctags file, wrap names in anchor tags that link to their definitions. lineanchors should be used, and the tags file should specify line numbers (see the -n option to ctags).

New in version 1.6.

tagurlformat

A string formatting pattern used to generate links to ctags definitions. Available variables are %(path)s, %(fname)s and %(fext)s. Defaults to an empty string, resulting in just #prefix-number links.

New in version 1.6.

Subclassing the HTML formatter

New in version 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:

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.

class ImageFormatter
Short names:i, m, g
Filenames:img, IMG, png

Create a PNG image from source code. This uses the Python Imaging Library to generate a pixmap from the source code.

New in version 0.10.

Additional options accepted:

image_format

An image format to output to that is recognised by PIL, these include:

  • “PNG” (default)
  • “JPEG”
  • “BMP”
  • “GIF”
line_pad

The extra spacing (in pixels) between each line of text.

Default: 2

font_name

The font name to be used as the base font from which others, such as bold and italic fonts will be generated. This really should be a monospace font to look sane.

Default: “Bitstream Vera Sans Mono”

font_size

The font size in points to be used.

Default: 14

image_pad

The padding, in pixels to be used at each edge of the resulting image.

Default: 10

line_numbers

Whether line numbers should be shown: True/False

Default: True

line_number_start

The line number of the first line.

Default: 1

line_number_step

The step used when printing line numbers.

Default: 1

line_number_bg

The background colour (in “#123456” format) of the line number bar, or None to use the style background color.

Default: “#eed”

line_number_fg

The text color of the line numbers (in “#123456”-like format).

Default: “#886”

line_number_chars

The number of columns of line numbers allowable in the line number margin.

Default: 2

line_number_bold

Whether line numbers will be bold: True/False

Default: False

line_number_italic

Whether line numbers will be italicized: True/False

Default: False

line_number_separator

Whether a line will be drawn between the line number area and the source code area: True/False

Default: True

line_number_pad

The horizontal padding (in pixels) between the line number margin, and the source code area.

Default: 6

hl_lines

Specify a list of lines to be highlighted.

New in version 1.2.

Default: empty list

hl_color

Specify the color for highlighting lines.

New in version 1.2.

Default: highlight color of the selected style

class JpgImageFormatter
Short names:i, m, g, _, j, p, g
Filenames:jpg, jpeg

Create a JPEG image from source code. This uses the Python Imaging Library to generate a pixmap from the source code.

New in version 1.0.

class LatexFormatter
Short names:L, a, T, e, X
Filenames:latex, tex

Format tokens as LaTeX code. This needs the fancyvrb and color standard packages.

Without the full option, code is formatted as one Verbatim environment, like this:

\begin{Verbatim}[commandchars=\\\{\}]
\PY{k}{def }\PY{n+nf}{foo}(\PY{n}{bar}):
    \PY{k}{pass}
\end{Verbatim}

The special command used here (\PY) and all the other macros it needs are output by the get_style_defs method.

With the full option, a complete LaTeX document is output, including the command definitions in the preamble.

The get_style_defs() method of a LatexFormatter returns a string containing \def commands defining the macros needed inside the Verbatim environments.

Additional options accepted:

style
The style to use, can be a string or a Style subclass (default: 'default').
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: '').
docclass
If the full option is enabled, this is the document class to use (default: 'article').
preamble
If the full option is enabled, this can be further preamble commands, e.g. \usepackage (default: '').
linenos
If set to True, output line numbers (default: False).
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.
verboptions
Additional options given to the Verbatim environment (see the fancyvrb docs for possible values) (default: '').
commandprefix

The LaTeX commands used to produce colored output are constructed using this prefix and some letters (default: 'PY').

New in version 0.7.

Changed in version 0.10: The default is now 'PY' instead of 'C'.

texcomments

If set to True, enables LaTeX comment lines. That is, LaTex markup in comment tokens is not escaped so that LaTeX can render it (default: False).

New in version 1.2.

mathescape

If set to True, enables LaTeX math mode escape in comments. That is, '$...$' inside a comment will trigger math mode (default: False).

New in version 1.2.

escapeinside

If set to a string of length 2, enables escaping to LaTeX. Text delimited by these 2 characters is read as LaTeX code and typeset accordingly. It has no effect in string literals. It has no effect in comments if texcomments or mathescape is set. (default: '').

New in version 2.0.

envname

Allows you to pick an alternative environment name replacing Verbatim. The alternate environment still has to support Verbatim’s option syntax. (default: 'Verbatim').

New in version 2.0.

class NullFormatter
Short names:T, e, x, t, , o, n, l, y
Filenames:text, null

Output the text unchanged without any formatting.

class RawTokenFormatter
Short names:R, a, w, , t, o, k, e, n, s
Filenames:raw, tokens

Format tokens as a raw representation for storing token streams.

The format is tokentype<TAB>repr(tokenstring)\n. The output can later be converted to a token stream with the RawTokenLexer, described in the lexer list.

Only two options are accepted:

compress
If set to 'gz' or 'bz2', compress the output with the given compression algorithm after encoding (default: '').
error_color

If set to a color name, highlight error tokens using that color. If set but with no value, defaults to 'red'.

New in version 0.11.

class RtfFormatter
Short names:R, T, F
Filenames:rtf

Format tokens as RTF markup. This formatter automatically outputs full RTF documents with color information and other useful stuff. Perfect for Copy and Paste into Microsoft(R) Word(R) documents.

Please note that encoding and outencoding options are ignored. The RTF format is ASCII natively, but handles unicode characters correctly thanks to escape sequences.

New in version 0.6.

Additional options accepted:

style
The style to use, can be a string or a Style subclass (default: 'default').
fontface
The used font famliy, for example Bitstream Vera Sans. Defaults to some generic font which is supposed to have fixed width.
fontsize

Size of the font used. Size is specified in half points. The default is 24 half-points, giving a size 12 font.

New in version 2.0.

class SvgFormatter
Short names:S, V, G
Filenames:svg

Format tokens as an SVG graphics file. This formatter is still experimental. Each line of code is a <text> element with explicit x and y coordinates containing <tspan> elements with the individual token styles.

By default, this formatter outputs a full SVG document including doctype declaration and the <svg> root element.

New in version 0.9.

Additional options accepted:

nowrap
Don’t wrap the SVG <text> elements in <svg><g> elements and don’t add a XML declaration and a doctype. If true, the fontfamily and fontsize options are ignored. Defaults to False.
fontfamily
The value to give the wrapping <g> element’s font-family attribute, defaults to "monospace".
fontsize
The value to give the wrapping <g> element’s font-size attribute, defaults to "14px".
xoffset
Starting offset in X direction, defaults to 0.
yoffset
Starting offset in Y direction, defaults to the font size if it is given in pixels, or 20 else. (This is necessary since text coordinates refer to the text baseline, not the top edge.)
ystep
Offset to add to the Y coordinate for each subsequent line. This should roughly be the text size plus 5. It defaults to that value if the text size is given in pixels, or 25 else.
spacehack
Convert spaces in the source to &#160;, which are non-breaking spaces. SVG provides the xml:space attribute to control how whitespace inside tags is handled, in theory, the preserve value could be used to keep all whitespace as-is. However, many current SVG viewers don’t obey that rule, so this option is provided as a workaround and defaults to True.
class Terminal256Formatter
Short names:T, e, r, m, i, n, a, l, 2, 5, 6
Filenames:terminal256, console256, 256

Format tokens with ANSI color sequences, for output in a 256-color terminal or console. Like in TerminalFormatter color sequences are terminated at newlines, so that paging the output works correctly.

The formatter takes colors from a style defined by the style option and converts them to nearest ANSI 256-color escape sequences. Bold and underline attributes from the style are preserved (and displayed).

New in version 0.9.

Options accepted:

style
The style to use, can be a string or a Style subclass (default: 'default').
class TerminalFormatter
Short names:T, e, r, m, i, n, a, l
Filenames:terminal, console

Format tokens with ANSI color sequences, for output in a text console. Color sequences are terminated at newlines, so that paging the output works correctly.

The get_style_defs() method doesn’t do anything special since there is no support for common styles.

Options accepted:

bg
Set to "light" or "dark" depending on the terminal’s background (default: "light").
colorscheme
A dictionary mapping token types to (lightbg, darkbg) color names or None (default: None = use builtin colorscheme).
linenos
Set to True to have line numbers on the terminal output as well (default: False = no line numbers).
class TestcaseFormatter
Short names:T, e, s, t, c, a, s, e
Filenames:testcase

Format tokens as appropriate for a new testcase.

New in version 2.0.