API Reference

Parsing

tinycss2 is “low-level” in that it doesn’t parse all of CSS: it doesn’t know about the syntax of any specific properties or at-rules. Instead, it provides a set of functions that can be composed to support exactly the parts of CSS you’re interested in, including new or non-standard rules or properties, without modifying tinycss2 or having a complex hook/plugin system.

In many cases, parts of the parsed values (such as the content of a AtRule) is given as component values that can be parsed further with other functions.

tinycss2.parse_stylesheet_bytes(css_bytes, protocol_encoding=None, environment_encoding=None, skip_comments=False, skip_whitespace=False)

Parse stylesheet from bytes, determining the character encoding as web browsers do.

This is used when reading a file or fetching a URL. The character encoding is determined from the initial bytes (a BOM or a @charset rule) as well as the parameters. The ultimate fallback is UTF-8.

Parameters

  • css_bytes (bytes) – A CSS byte string.

  • protocol_encoding (str) – The encoding label, if any, defined by HTTP or equivalent protocol. (e.g. via the charset parameter of the Content-Type header.)

  • environment_encoding (webencodings.Encoding) – The environment encoding, if any.

  • skip_comments (bool) – Ignore CSS comments at the top-level of the stylesheet. If the input is a string, ignore all comments.

  • skip_whitespace (bool) – Ignore whitespace at the top-level of the stylesheet. Whitespace is still preserved in the prelude and the content of rules.

Returns

A (rules, encoding) tuple.

response = urlopen('http://example.net/foo.css')
rules, encoding = parse_stylesheet_bytes(
    css_bytes=response.read(),
    # Python 3.x
    protocol_encoding=response.info().get_content_type().get_param('charset'),
    # Python 2.x
    protocol_encoding=response.info().gettype().getparam('charset'),
)
for rule in rules:
    ...
tinycss2.parse_stylesheet(input, skip_comments=False, skip_whitespace=False)

Parse stylesheet from text.

This is used e.g. for a <style> HTML element.

This differs from parse_rule_list() in that top-level <!-- and --> tokens are ignored. This is a legacy quirk for the <style> HTML element.

Parameters

  • input (str or iterable) – A string or an iterable of component values.

  • skip_comments (bool) – Ignore CSS comments at the top-level of the stylesheet. If the input is a string, ignore all comments.

  • skip_whitespace (bool) – Ignore whitespace at the top-level of the stylesheet. Whitespace is still preserved in the prelude and the content of rules.

Returns

A list of QualifiedRule, AtRule, Comment (if skip_comments is false), WhitespaceToken (if skip_whitespace is false), and ParseError objects.

tinycss2.parse_rule_list(input, skip_comments=False, skip_whitespace=False)

Parse a non-top-level rule list.

This is used for parsing the content of nested rules like @media. This differs from parse_stylesheet() in that top-level <!-- and --> tokens are not ignored.

Parameters

  • input (str or iterable) – A string or an iterable of component values.

  • skip_comments (bool) – Ignore CSS comments at the top-level of the list. If the input is a string, ignore all comments.

  • skip_whitespace (bool) – Ignore whitespace at the top-level of the list. Whitespace is still preserved in the prelude and the content of rules.

Returns

A list of QualifiedRule, AtRule, Comment (if skip_comments is false), WhitespaceToken (if skip_whitespace is false), and ParseError objects.

tinycss2.parse_one_rule(input, skip_comments=False)

Parse a single qualified rule or at-rule.

This would be used e.g. by insertRule() in an implementation of CSSOM.

Parameters
Returns

A QualifiedRule, AtRule, or ParseError objects.

Any whitespace or comment before or after the rule is dropped.

tinycss2.parse_declaration_list(input, skip_comments=False, skip_whitespace=False)

Parse a declaration list (which may also contain at-rules).

This is used e.g. for the content of a style rule or @page rule, or for the style attribute of an HTML element.

In contexts that don’t expect any at-rule, all AtRule objects should simply be rejected as invalid.

Parameters

  • input (str or iterable) – A string or an iterable of component values.

  • skip_comments (bool) – Ignore CSS comments at the top-level of the list. If the input is a string, ignore all comments.

  • skip_whitespace (bool) – Ignore whitespace at the top-level of the list. Whitespace is still preserved in the value of declarations and the prelude and content of at-rules.

Returns

A list of Declaration, AtRule, Comment (if skip_comments is false), WhitespaceToken (if skip_whitespace is false), and ParseError objects

tinycss2.parse_one_declaration(input, skip_comments=False)

Parse a single declaration.

This is used e.g. for a declaration in an @supports test.

Parameters
Returns

A Declaration or ParseError.

Any whitespace or comment before the : colon is dropped.

tinycss2.parse_component_value_list(css, skip_comments=False)

Parse a list of component values.

Parameters

  • css (str) – A CSS string.

  • skip_comments (bool) – Ignore CSS comments. The return values (and recursively its blocks and functions) will not contain any Comment object.

Returns

A list of component values.

tinycss2.parse_one_component_value(input, skip_comments=False)

Parse a single component value.

This is used e.g. for an attribute value referred to by attr(foo length).

Parameters
Returns

A component value (that is neither whitespace or comment), or a ParseError.

Serialization

In addition to each node’s a serialize() method, some serialization-related functions are available:

tinycss2.serialize(nodes)

Serialize nodes to CSS syntax.

This should be used for component values instead of just tinycss2.ast.Node.serialize() on each node as it takes care of corner cases such as ; between declarations, and consecutive identifiers that would otherwise parse back as the same token.

Parameters

nodes (iterable) – An iterable of tinycss2.ast.Node objects.

Returns

A string representing the nodes.

tinycss2.serialize_identifier(value)

Serialize any string as a CSS identifier

Parameters

value (str) – A string representing a CSS value.

Returns

A string that would parse as an tinycss2.ast.IdentToken whose tinycss2.ast.IdentToken.value attribute equals the passed value argument.

Color

tinycss2.color3.parse_color(input)

Parse a color value as defined in CSS Color Level 3.

Parameters

input (str or iterable) – A string or an iterable of component values.

Returns

  • None if the input is not a valid color value. (No exception is raised.)

  • The string 'currentColor' for the currentColor keyword

  • Or a RGBA object for every other values (including keywords, HSL and HSLA.) The alpha channel is clipped to [0, 1] but red, green, or blue can be out of range (eg. rgb(-10%, 120%, 0%) is represented as (-0.1, 1.2, 0, 1).)

class tinycss2.color3.RGBA(red, green, blue, alpha)

An RGBA color.

A tuple of four floats in the 0..1 range: (red, green, blue, alpha).

red

Convenience access to the red channel. Same as rgba[0].

green

Convenience access to the green channel. Same as rgba[1].

blue

Convenience access to the blue channel. Same as rgba[2].

alpha

Convenience access to the alpha channel. Same as rgba[3].

<An+B>

tinycss2.nth.parse_nth(input)

Parse <An+B>, as found in :nth-child() and related Selector pseudo-classes.

Although tinycss2 does not include a full Selector parser, this bit of syntax is included as it is particularly tricky to define on top of a CSS tokenizer.

Parameters

input (str or iterable) – A string or an iterable of component values.

Returns

A (a, b) tuple of integers, or None if the input is invalid.

AST nodes

Various parsing functions return a node or a list of nodes. Some types of nodes contain nested nodes which may in turn contain more nodes, forming together an abstract syntax tree.

Although you typically don’t need to import it, the tinycss2.ast module defines a class for every type of node.

class tinycss2.ast.Node

Every node type inherits from this class, which is never instantiated directly.

type

Each child class has a type class attribute with a unique string value. This allows checking for the node type with code like:

if node.type == 'whitespace':

instead of the more verbose:

from tinycss2.ast import WhitespaceToken
if isinstance(node, WhitespaceToken):

Every node also has these attributes and methods, which are not repeated for brevity:

source_line

The line number of the start of the node in the CSS source. Starts at 1.

source_column

The column number within source_line of the start of the node in the CSS source. Starts at 1.

serialize()

Serialize this node to CSS syntax and return a Unicode string.

class tinycss2.ast.QualifiedRule

A qualified rule.

<prelude> '{' <content> '}'

The interpretation of qualified rules depend on their context. At the top-level of a stylesheet or in a conditional rule such as @media, they are style rules where the prelude is Selectors list and the content is a list of property declarations.

type = 'qualified-rule'
prelude

The rule’s prelude, the part before the {} block, as a list of component values.

content

The rule’s content, the part inside the {} block, as a list of component values.

class tinycss2.ast.AtRule

An at-rule.

@<at_keyword> <prelude> '{' <content> '}'
@<at_keyword> <prelude> ';'

The interpretation of at-rules depend on their at-keyword as well as their context. Most types of at-rules (ie. at-keyword values) are only allowed in some context, and must either end with a {} block or a semicolon.

type = 'at-rule'
at_keyword

The unescaped value of the rule’s at-keyword, without the @ symbol, as a Unicode string.

lower_at_keyword

Same as at_keyword but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS at-keyword.

if node.type == 'at-rule' and node.lower_at_keyword == 'import':
prelude

The rule’s prelude, the part before the {} block or semicolon, as a list of component values.

content

The rule’s content, if any. The block’s content as a list of component values for at-rules with a {} block, or None for at-rules ending with a semicolon.

class tinycss2.ast.Declaration

A (property or descriptor) declaration.

<name> ':' <value>
<name> ':' <value> '!important'
type = 'declaration'
name

The unescaped name, as a Unicode string.

lower_name

Same as name but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS property or descriptor name.

if node.type == 'declaration' and node.lower_name == 'color':
value

The declaration value as a list of component values: anything between : and the end of the declaration, or !important.

important

A boolean, true if the declaration had an !important marker. It is up to the consumer to reject declarations that do not accept this flag, such as non-property descriptor declarations.

class tinycss2.ast.ParseError

A syntax error of some sort. May occur anywhere in the tree.

Syntax errors are not fatal in the parser to allow for different error handling behaviors. For example, an error in a Selector list makes the whole rule invalid, but an error in a Media Query list only replaces one comma-separated query with not all.

type = 'error'
kind

Machine-readable string indicating the type of error. Example: 'bad-url'.

message

Human-readable explanation of the error, as a string. Could be translated, expanded to include details, etc.

class tinycss2.ast.Comment

A CSS comment.

Comments can be ignored by passing skip_comments=True to functions such as parse_component_value_list().

type = 'comment'
value

The content of the comment, between /* and */, as a string.

class tinycss2.ast.WhitespaceToken

A <whitespace-token>.

type = 'whitespace'
value

The whitespace sequence, as a string, as in the original CSS source.

class tinycss2.ast.LiteralToken

Token that represents one or more characters as in the CSS source.

type = 'literal'
value

A string of one to four characters.

Instances compare equal to their value, so that these are equivalent:

if node == ';':
if node.type == 'literal' and node.value == ';':

This regroups what the specification defines as separate token types:

  • <colon-token> :

  • <semicolon-token> ;

  • <comma-token> ,

  • <cdc-token> -->

  • <cdo-token> <!--

  • <include-match-token> ~=

  • <dash-match-token> |=

  • <prefix-match-token> ^=

  • <suffix-match-token> $=

  • <substring-match-token> *=

  • <column-token> ||

  • <delim-token> (a single ASCII character not part of any another token)

class tinycss2.ast.IdentToken

An <ident-token>.

type = 'ident'
value

The unescaped value, as a Unicode string.

lower_value

Same as value but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS keyword.

class tinycss2.ast.AtKeywordToken

An <at-keyword-token>.

'@' <value>
type = 'at-keyword'
value

The unescaped value, as a Unicode string, without the preceding @.

lower_value

Same as value but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS at-keyword.

if node.type == 'at-keyword' and node.lower_value == 'import':
class tinycss2.ast.HashToken

A <hash-token>.

'#' <value>
type = 'hash'
value

The unescaped value, as a Unicode string, without the preceding #.

is_identifier

A boolean, true if the CSS source for this token was # followed by a valid identifier. (Only such hash tokens are valid ID selectors.)

class tinycss2.ast.StringToken

A <string-token>.

'"' <value> '"'
type = 'string'
value

The unescaped value, as a Unicode string, without the quotes.

class tinycss2.ast.URLToken

An <url-token>.

'url(' <value> ')'
type = 'url'
value

The unescaped URL, as a Unicode string, without the url( and ) markers.

class tinycss2.ast.UnicodeRangeToken

A unicode-range token.

type = 'unicode-range'
start

The start of the range, as an integer between 0 and 1114111.

end

The end of the range, as an integer between 0 and 1114111. Same as start if the source only specified one value.

class tinycss2.ast.NumberToken

A <number-token>.

type = 'number'
value

The numeric value as a float.

int_value

The numeric value as an int if is_integer is true, None otherwise.

is_integer

Whether the token was syntactically an integer, as a boolean.

representation

The CSS representation of the value, as a Unicode string.

class tinycss2.ast.PercentageToken

A <percentage-token>.

<representation> '%'
type = 'percentage'
value

The value numeric as a float.

int_value

The numeric value as an int if the token was syntactically an integer, or None.

is_integer

Whether the token’s value was syntactically an integer, as a boolean.

representation

The CSS representation of the value without the unit, as a Unicode string.

class tinycss2.ast.DimensionToken

A <dimension-token>.

<representation> <unit>
type = 'dimension'
value

The value numeric as a float.

int_value

The numeric value as an int if the token was syntactically an integer, or None.

is_integer

Whether the token’s value was syntactically an integer, as a boolean.

representation

The CSS representation of the value without the unit, as a Unicode string.

unit

The unescaped unit, as a Unicode string.

lower_unit

Same as unit but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS unit.

if node.type == 'dimension' and node.lower_unit == 'px':
class tinycss2.ast.ParenthesesBlock

A <()-block>.

'(' <content> ')'
type = '() block'
content

The content of the block, as list of component values. The ( and ) markers themselves are not represented in the list.

class tinycss2.ast.SquareBracketsBlock

A <[]-block>.

'[' <content> ']'
type = '[] block'
content

The content of the block, as list of component values. The [ and ] markers themselves are not represented in the list.

class tinycss2.ast.CurlyBracketsBlock

A <{}-block>.

'{' <content> '}'
type = '{} block'
content

The content of the block, as list of component values. The [ and ] markers themselves are not represented in the list.

class tinycss2.ast.FunctionBlock

A <function-block>.

<name> '(' <arguments> ')'
type = 'function'
name

The unescaped name of the function, as a Unicode string.

lower_name

Same as name but normalized to ASCII lower case, see ascii_lower(). This is the value to use when comparing to a CSS function name.

arguments

The arguments of the function, as list of component values. The ( and ) markers themselves are not represented in the list. Commas are not special, but represented as LiteralToken objects in the list.