Skip to content

Release History

Unreleased

2.0.0a1 -- 2026-06-24

Breaking Changes

  • removed support for Python versions before 3.10
  • datastructures.LinkHeader and its exposed parameter sequences are now immutable
  • converted positional Boolean parameters to keyword-only parameters
Function Parameter
parse_accept strict
parse_content_type normalize_parameter_values
parse_forwarded only_standard_parameters
parse_link strict

Added

  • ietfparse.constants module -- this contains constant ContentType instances
  • datastructures.LinkHeader.rel property
  • indexed parameter lookup in datastructures.LinkHeader
  • datastructures.ImmutableSequence helper class
  • public errors.MalformedContentType, errors.MalformedListSegment, and errors.StrictHeaderParsingFailure exceptions for malformed parser inputs. MalformedContentType remains a ValueError subclass for compatibility.
  • default parameter to algorithms.select_content_type
  • packaged ietfparse.test benchmark utility for the directly supported HTTP header parsers, including the ietfparse-test console script, the python -m ietfparse.test entry point, implementation benchmarks, adjacent library comparisons, and saved-result diff reporting
  • curated adjacent-library comparison suites and documentation for Accept, Cache-Control, and Link parsing behavior

Changed

  • algorithms.select_content_type now accepts parsed ContentType instances, content-type strings, or a raw Accept header string as requested, and it also accepts string entries in available
  • headers.parse_link changed to allow multiple media and type parameters as described in RFC-8288.
  • headers.parse_link now preserves parentheses and escaped delimiters inside link targets and quoted parameter values, preserves valueless parameters, and raises MalformedLinkValue for malformed parameter syntax instead of misparsing the remainder of the header.
  • datastructures.LinkHeader changed to combine multiple relationship type (rel) parameters into a single space-separated parameter as described in RFC-8288. Note that this is only relevant if you disable strict mode parsing.
  • headers.parse_content_type changed to raise MalformedContentType error instead of ValueError.
  • headers.parse_content_type now strips nested HTTP comments without treating parentheses inside quoted parameter values as comments.
  • datastructures.ContentType instances can now be compared to strings
  • datastructures.ContentType.quality is now normalized metadata. Missing and effectively maximal values normalize to 1.0, rejected values normalize to 0.0, intermediate values are rounded to three decimal places, and the Accept* parsers now share the same quality-handling rules while preserving the preference for explicitly declared maximum quality values.
  • list-based header parsers now ignore stray empty items produced by extra commas, while headers.parse_list preserves those empty items and headers.parse_accept(..., strict=True) treats them as malformed.
  • list-style header parsing now preserves escaped quotes and commas inside quoted values for headers.parse_list, headers.parse_cache_control, headers.parse_forwarded, and headers.parse_accept.
  • headers.parse_list now raises errors.MalformedListSegment when a quoted item is followed by additional non-delimited content instead of silently merging it into a malformed value.
  • parser internals were simplified and optimized for token, quoted-string, and list parsing hot paths, reducing overhead in the public header parsers

Removed

  • previously deprecated functions: parse_http_accept_header, parse_link_header, parse_list_header, remove_url_auth, rewrite_url

Development environment changes

1.9.0 -- 2022-07-08

Deprecated

  • using len() on the return value from algorithms.remove_url_auth
  • rewrite_url and remove_url_auth. Use yarl instead. It is an awesome library and a more general solution.

Changed

  • algorithms.RemoveUrlAuthResult from a named tuple to a proper class.
  • Replace type hints with annotations.

Removed

  • ietfparse.compat module.
  • universal wheels

1.8.0 -- 2021-08-11

Removed

  • support for Python versions before 3.7

1.7.0 -- 2020-11-04

Behavioural Change

headers.parse_accept used to fail with a ValueError when it encountered an invalid content type value in the header. Now it skips the invalid value instead. If you want the previous behaviour, then pass strict=True.

Added

  • support for Python 3.7-3.9

Removed

  • support for 3.4 and 3.5

Changed

  • clarified that headers.parse_content_type raises a ValueError when it encounters an invalid content type header
  • headers.parse_accept skips unparseable content types unless the new strict keyword parameter is truthy

1.6.1 -- 2020-01-26

Fixed

  • corrected packaging metadata

1.6.0 -- 2020-01-25

Added

  • typestubs

Changed

  • switched from travis-ci to circle-ci
  • allow "bad whitespace" around = in link header parameter lists as indicated in RFC-8288
  • replaced nosetests with python -m unittest

1.5.1 -- 2018-03-04

Added

  • RFC-6839 content suffix support in datastructures.ContentType and headers.parse_content_type()

1.5.0 -- 2017-12-24

Deprecated

  • the normalized_parameter_values keyword parameter to the headers._parse_parameter_list function

Added

  • normalize_parameter_names keyword parameter to the headers._parse_parameter_list function. This replaces the normalized_parameter_values keyword parameter.
  • support for parsing Forwarded headers with headers.parse_forwarded
  • algorithms.remove_url_auth function

Removed

  • support for Python 2.6 and 3.3

Changed

  • headers.parse_accept() now prefers explicit highest quality preferences over the inferred highest quality preferences

1.4.3 -- (2017-10-30)

Changed

The parsing of qualified lists now retains the initial ordering whenever possible. The algoritm prefers explicit highest quality preferences (1.0) over inferred highest quality preferences. It also retains the initial ordering in the presense of multiple highest quality matches. This affects the headers.parse_accept_charset, headers.parse_accept_encoding and headers.parse_accept_language functions.

1.4.2 -- (2017-07-04)

Added

  • formatting of Link headers using str(header)

1.4.1 -- (2017-04-03)

Added

  • error handling documentation for header parsing functions

1.4.0 -- (2016-10-18)

Added

Fixed

  • parsing of parameter lists values ending in a quote character. For example, max-age=5, x-foo="prune" was parsed as ['max-age=5', 'x-foo="prune']

1.3.0 -- (2016-08-11)

Added

Deprecated

  • headers.parse_http_accept_header renamed to headers.parse_accept
  • headers.parse_link_header renamed to headers.parse_link
  • headers.parse_list_header renamed to headers.parse_list

1.2.2 -- (2015-05-27)

Added

  • headers.parse_list_header which parses generic comma-separated list headers with support for quoted parts
  • headers.parse_accept_charset which parses the Accept-Charset header into a sorted list

1.2.1 -- (2015-05-25)

Fixed

algorithms.select_content_type only worked with the augmented datastructures.ContentType values returned from the algorithms.parse_http_accept_header. IOW, the algorithm required that the quality attribute existed on instances. RFC-9110 requires that missing quality values are treated as 1

1.2.0 -- (2015-04-19)

Added

  • headers.parse_link_header which parses Link headers.
  • datastructures.LinkHeader class

1.1.1 -- (2015-02-10)

Removed

  • the setupext module since it was causing problems with source distributions

1.1.0 -- (2014-10-26)

Added

  • algorithms.rewrite_url function

1.0.0 -- (2014-09-21)

Initial implementation containing the following functionality: - algorithms.select_content_type function - datastructures.ContentType class - errors.NoMatch class - errors.RootException class - headers.parse_content_type function - headers.parse_http_accept_header function