v0.38.1

Performance

This version contains major performance improvements in the following functions:

• Floki.filter_out/2.
• Floki.find/2 - with some improvements to specific selectors, like classes
  and attribute selectors.
• Floki.text/2.

Those functions are not only faster, but are now using less memory. Please check
the PRs related to this release if you want to better understand the numbers.

• Speed up do_classes_matches? - #649
• Make filter_out faster - #650
• Speed up attribute lookup - #651
• Enable tuple traversal optimization for multiple selectors - #652
• Optimize HTMLTree.to_tuple conversion using Enum.reduce - #657
• Optimize Finder.get_descendant_ids/2 memory usage and speed - #660
• Optimize Finder.get_siblings/2 memory usage and speed - #663
• Optimize FlatText.get/3 memory usage and speed - #664
• Optimize class matching in Floki.Selector - #665

All the improvements in this version were made by Barna Kovacs - @preciz,
so shout out and thanks to him!

Pull requests

• Fix build status badge by @sgerrand in #632
• Update v0.38.0 changelog item for breaking changes by @kianmeng in #634
• Bump earmark from 1.4.47 to 1.4.48 by @dependabot[bot] in #635
• Speed up do_classes_matches? by @preciz in #649
• Make filter_out faster by @preciz in #650
• Speed up attribute lookup by @preciz in #651
• Enable tuple traversal optimization for multiple selectors by @preciz in #652
• Optimize HTMLTree to_tuple conversion using Enum.reduce by @preciz in #657
• Optimize Finder.get_descendant_ids/2 memory usage and speed by @preciz in #660
• Fix unused require Logger warning in parser_test.exs by @preciz in #661
• Optimize Finder.get_siblings/2 memory usage and speed by @preciz in #663
• Optimize FlatText.get/3 memory usage and speed by @preciz in #664
• Optimize class matching in Floki.Selector by @preciz in #665
• Release v0.38.1 by @philss in #662

New Contributors

• @sgerrand made their first contribution in #632
• @preciz made their first contribution in #649

Full Changelog: v0.38.0...v0.38.1

v0.38.0

Added

• This version adds initial support for the :has pseudo-selector.
  It is a great addition that enables finding elements containing
  matching children.

  Examples for selectors:

  • "div:has(h1)"
  • "div:has(h1, p, span)"
  • "div:has(p.foo)"
  • "div:has(img[src='https://example.com'])"
  • "tr:has(*:fl-contains('TEST'))"

  Note that combinators like ">" are not allowed yet.

  Thank you @bvobart for this feature!

Fixed

• Add :style option documentation to Floki.text/2.
  Thanks @s3cur3 for the fix.

• Fix deprecation warnings for upcoming Elixir 1.19.

• Prevent from crashing when selector is an empty string.

Removed

• Remove support for Elixir 1.14 and OTP 23.

• Remove deprecated functions and function clauses
  that were accepting strings (binaries).

  Affected functions:

  • parse/1 - removed function
  • map/2 - removed function
  • attr/4 - removed clause
  • find/2 - removed clause
  • text/3 - removed clause
  • text/3 - removed clause
  • attribute/2 - removed clause
  • filter_out/2 - removed clause

  HTML must be parsed before searching. Functions like Floki.find/2,
  Floki.attribute/2, and other HTML manipulation functions no longer work
  directly with HTML strings. The HTML must be parsed first using
  Floki.parse_fragment/2 or Floki.parse_document/2.

  Before:

  html = "<div class='foobar'><p>Hello</p></div>"
  Floki.find(html, "p")
  Floki.attribute(html, "div", "class")

  After:

  html = "<div class='foobar'><p>Hello</p></div>"
  parsed_html = Floki.parse_fragment!(html)
  Floki.find(parsed_html, "p")
  Floki.attribute(parsed_html, "div", "class")

Pull requests

• Implement parsing rules for the :has pseudo class selector by @philss in #623
• feat: implement :has pseudo-selector functionality by @bvobart in #624
• Bump ex_doc from 0.37.3 to 0.38.2 by @dependabot in #625
• Bump credo from 1.7.11 to 1.7.12 by @dependabot in #619
• Bump benchee from 1.3.1 to 1.4.0 by @dependabot in #618
• Add :style flag to text/2` docs by @s3cur3 in #627
• Remove support for Elixir 1.14 by @philss in #626
• Remove deprecations by @philss in #628
• Remove deprecation warnings for the upcoming Elixir 1.19 by @philss in #630
• Prevent find/2 from crashing with empty selector by @philss in #631
• Prepare to release v0.38 by @philss in #629

New Contributors

• @bvobart made their first contribution in #624
• @s3cur3 made their first contribution in #627

Full Changelog: v0.37.1...v0.38.0

v0.37.1

Fixed

Move regex declaration from module tag to inside function. This is a fix to be compatible with the upcoming OTP 28.

Pull requests

• Add Elixir 1.18 to the CI workflow by @philss in #607
• Bump ex_doc from 0.35.1 to 0.36.1 by @dependabot in #606
• Bump ex_doc from 0.36.1 to 0.37.1 by @dependabot in #611
• Fix versions we describe in README.md by @philss in #616
• Bump credo from 1.7.10 to 1.7.11 by @dependabot in #608
• Bump ex_doc from 0.37.1 to 0.37.3 by @dependabot in #615
• Bump fast_html from 2.4.0 to 2.4.1 by @dependabot in #609

Full Changelog: v0.37.0...v0.37.1

v0.37.0

Added

• Add Floki.css_escape/1 - thanks @SteffenDE.

Fixed

• Fix bug propagating identity encoder in raw_html/2 - thanks @andyleclair.

Removed

• Remove support for Elixir 1.13 and OTP 22.

Pull requests

• Drop support for Elixir 1.13 by @philss in #595
• Bump credo from 1.7.8 to 1.7.9 by @dependabot in #596
• Bump credo from 1.7.9 to 1.7.10 by @dependabot in #597
• Bump fast_html from 2.3.0 to 2.4.0 by @dependabot in #599
• Bump dialyxir from 1.4.4 to 1.4.5 by @dependabot in #600
• Bump ex_doc from 0.34.2 to 0.35.1 by @dependabot in #602
• Fix bug propagating identity encoder in raw_html/2 by @andyleclair in #603

New Contributors

• @andyleclair made their first contribution in #603

Full Changelog: v0.36.3...v0.37.0

v0.36.3

This release contains some performance improvements, thanks to @ypconstante.

Fixed

• Stop Floki.get_by_id/2 traversal on first match. Thanks @ypconstante.

• Remove extra whitespaces from nodes without attributes on Floki.raw_html/1.
  Thank you @ypconstante.

• Fix Floki.raw_html/1 typespecs. Thanks @davydog187.

Pull requests

• chore: fix some typos by @tianzedavid in #564
• Bump ex_doc from 0.32.1 to 0.32.2 by @dependabot in #566
• Bump credo from 1.7.5 to 1.7.6 by @dependabot in #565
• Bump html5ever from 0.15.0 to 0.16.1 by @dependabot in #567
• Find without html tree for some pseudo classes by @ypconstante in #568
• Remove stack usage on Finder by @ypconstante in #569
• Remove extra whitespace on node without attributes by @ypconstante in #571
• Stop get_by_id traversal on first match by @ypconstante in #572
• Bump ex_doc from 0.32.2 to 0.33.0 by @dependabot in #570
• Bump ex_doc from 0.33.0 to 0.34.0 by @dependabot in #573
• Bump benchee from 1.3.0 to 1.3.1 by @dependabot in #574
• Bump credo from 1.7.6 to 1.7.7 by @dependabot in #575
• Add OTP 27 and Elixir 1.17 to the CI matrix by @philss in #576
• Bump ex_doc from 0.34.0 to 0.34.1 by @dependabot in #577
• Bump jason from 1.4.1 to 1.4.3 by @dependabot in #579
• Optimize traverse_and_update without accumulator by @ypconstante in #584
• Bump ex_doc from 0.34.1 to 0.34.2 by @dependabot in #581
• Bump jason from 1.4.3 to 1.4.4 by @dependabot in #585
• Bump earmark from 1.4.46 to 1.4.47 by @dependabot in #582
• Copy and paste the benchmark commands by @kianmeng in #587
• Ignore .cover folder by @kianmeng in #586
• Use accumulator on raw_html by @ypconstante in #588
• Optimize raw html padding for small depths by @ypconstante in #589
• Bump credo from 1.7.7 to 1.7.8 by @dependabot in #590
• Bump dialyxir from 1.4.3 to 1.4.4 by @dependabot in #591
• Fix raw_html/2 typespec by @davydog187 in #593

New Contributors

• @tianzedavid made their first contribution in #564

Full Changelog: v0.36.2...v0.36.3

v0.36.2

Added

• Implement the Inspect protocol for the Floki.HTMLTree struct.
  This struct is currently private. Thank you @vittoriabitton.

Fixed

• Fix regression to respect config option :encode in Floki.raw_html/2.
  Thanks @Sgoettschkes.

• Make the Floki.raw_html/2 treat the contents of the <title> tag as plain text.
  The idea is to align with parse_document/2.
  Thank you @aymanosman.

Pull requests

• fix typespec of get_by_id/2 by @SteffenDE in #549
• Bump ex_doc from 0.31.1 to 0.31.2 by @dependabot in #553
• raw_html treats the content of title tags as plain text by @aymanosman in #555
• Implement Inspect protocol for HTMLTree by @vittoriabitton in #547
• Bump ex_doc from 0.31.2 to 0.32.0 by @dependabot in #559
• Bump ex_doc from 0.32.0 to 0.32.1 by @dependabot in #560
• fix: read encode_raw_html config as default for raw_html encode option by @Sgoettschkes in #561
• Release 0.36.2 by @JohnnyCurran in #563

New Contributors

• @aymanosman made their first contribution in #555
• @Sgoettschkes made their first contribution in #561
• @JohnnyCurran made their first contribution in #563

Full Changelog: v0.36.0...v0.36.2

v0.36.1

Fixed

• Fix typespec of get_by_id/2.

Pull requests

• fix typespec of get_by_id/2 by @SteffenDE in #549
• Bump ex_doc from 0.31.1 to 0.31.2 by @dependabot in #553

Full Changelog: v0.36.0...v0.36.1

v0.36.0

Added

• Add Floki.get_by_id/1 that returns one element by ID or nil.
  Thanks @SteffenDE.

Changed

• Improve options validation with Keyword.validate!/2.
  This is not a change in APIs, but the error messages and opts validation should be standardized now.
  Thanks @vittoriabitton.

Removed

• Drop support for Elixir v1.12.

Pull requests

• Use adjacent_sibling instead of sibling by @ypconstante in #544
• Update Elixir version requirement to 1.13 by @vittoriabitton in #541
• Improve opts validation with Keyword.validate!/2 by @vittoriabitton in #542
• Bump credo from 1.7.4 to 1.7.5 by @dependabot in #546
• add Floki.find_by_id/2 by @SteffenDE in #548
• Find without html tree for the remaining combinators by @ypconstante in #545

New Contributors

• @vittoriabitton made their first contribution in #541
• @SteffenDE made their first contribution in #548

Full Changelog: v0.35.4...v0.36.0

v0.35.4

Fixed

• Fix regression in the order of elements in Floki.find/2

Pull requests

• Polymorphism on Finder.find by @ypconstante in #522
• Run all selector test with tuple list and html tree by @ypconstante in #523
• Bump fast_html from 2.2.0 to 2.3.0 by @dependabot in #530
• Reduce number of function call on traverse by @ypconstante in #531
• Optimize Floki.children by @ypconstante in #533
• Find without build html tree by @ypconstante in #534
• Bump credo from 1.7.3 to 1.7.4 by @dependabot in #535
• Optimize type selector by using pattern match by @ypconstante in #536
• Raw HTML code clean up by @ypconstante in #538
• Always return find elements in the correct order by @ypconstante in #540
• Find using descendant selector without html tree by @ypconstante in #537
• Release v0.35.4 by @philss in #543

Full Changelog: v0.35.3...v0.35.4

v0.35.3

This release has great performance improvements, thanks to the PRs
from @ypconstante!

Most of the main functions, such as Floki.raw_html/2 and Floki.find/2 are
faster and are using less memory. It's something like twice as fast, and half
usage of memory for find/2, for example.

Fixed

• Add :leex to Mix compilers. Fixes the build when running with dev version of Elixir.
  Thanks @wojtekmach.

• Fix Floki.raw_html/2 when a tree using attributes as maps is given.
  Thanks @SupaMic.

• Add a guard to Floki.find/2 so people can have a better error message when an
  invalid input is given. Thanks @Hajto.

• Fix parsers to consider IO data as inputs. This may change in the next version
  of Floki, as I plan to drop support for IO data.
  Thanks @ypconstante.

Removed

• Remove outdated Gleam wrapper code. The external functions syntax in Gleam
  has changed. So now
  the wrapper is not needed anymore.
  Thanks @michallepicki.

Pull requests

• Add :leex to Mix compilers by @wojtekmach in #502
• Update raw_html.ex to handle :attributes_as_maps option by @SupaMic in #498
• Bump benchee from 1.1.0 to 1.2.0 by @dependabot in #499
• Remove outdated gleam wrapper code by @michallepicki in #500
• find/2 input protection proposal. by @Hajto in #497
• Bump ex_doc from 0.30.9 to 0.31.0 by @dependabot in #503
• Enable parse of IO data html by @ypconstante in #504
• Optimize RawHTML.build_attrs/2 by @ypconstante in #505
• Optimize parse_flag by @ypconstante in #506
• Use recursion instead of Enum.flat_map on find_selectors by @ypconstante in #508
• Optimize HTMLTree.build by @ypconstante in #511
• Always use optimal encoding function by @ypconstante in #512
• Bump credo from 1.7.1 to 1.7.2 by @dependabot in #513
• Optimize selectors matching by @ypconstante in #510
• Bump benchee from 1.2.0 to 1.3.0 by @dependabot in #514
• Use stack on Finder by @ypconstante in #518
• Call self_closing_tags only once on raw_html by @ypconstante in #517
• Optimize id matching by @ypconstante in #519
• Bump dialyxir from 1.4.2 to 1.4.3 by @dependabot in #516
• Include Elixir v1.16 in the build matrix by @philss in #521
• Bump credo from 1.7.2 to 1.7.3 by @dependabot in #520
• Bump ex_doc from 0.31.0 to 0.31.1 by @dependabot in #527
• Optimize leftpad on raw_html by @ypconstante in #526
• Fix pretty raw_html with encoded text by @ypconstante in #525
• Move data extraction for selector matching by @ypconstante in #524
• Improve benchmark files by @philss in #528
• Prepare release v0.35.3 by @philss in #529

New Contributors

• @SupaMic made their first contribution in #498
• @michallepicki made their first contribution in #500
• @Hajto made their first contribution in #497
• @ypconstante made their first contribution in #504

Full Changelog: v0.35.2...v0.35.3