feat!: Reflect changes in lxml 5.3.1 and 6.0

- More html.builder shorthands
- libxml feature constants
- DTD(external_id=...) fix
- Deprecate some Memdebug methods

Author: abelcheung

pyproject.toml

@@ -72,7 +72,7 @@ lxml49 = ["lxml == 4.9.4"]
 lxml50 = ["lxml == 5.0.2"]
 lxml51 = ["lxml == 5.1.1"]
 lxml52 = ["lxml == 5.2.2"]
-lxml53 = ["lxml == 5.3.0"]
+lxml53 = ["lxml == 5.3.1"]
 dev = [
     'tox ~= 4.22',
     # microsoft/pyright#9286, microsoft/pyright#9296, microsoft/pyright#9412

src/lxml-stubs/etree/__init__.pyi

@@ -62,7 +62,9 @@ from ._module_misc import (
     CDATA as CDATA,
     DEBUG as DEBUG,
     ICONV_COMPILED_VERSION as ICONV_COMPILED_VERSION,
+    LIBXML_COMPILED_FEATURES as LIBXML_COMPILED_FEATURES,
     LIBXML_COMPILED_VERSION as LIBXML_COMPILED_VERSION,
+    LIBXML_FEATURES as LIBXML_FEATURES,
     LIBXML_VERSION as LIBXML_VERSION,
     LXML_VERSION as LXML_VERSION,
     C14NError as C14NError,

src/lxml-stubs/etree/_dtd.pyi

@@ -85,7 +85,9 @@ class DTD(_Validator):
     @overload
     def __init__(self, file: _FileReadSource) -> None: ...
     @overload
-    def __init__(self, file: None = None, *, external_id: bytes) -> None: ...
+    def __init__(  # str only supported since 5.3.1
+        self, file: None = None, *, external_id: str | bytes | bytearray
+    ) -> None: ...
     @property
     def name(self) -> str | None: ...
     @property

src/lxml-stubs/etree/_module_func.pyi

@@ -470,6 +470,7 @@ class _MemDebug:
             The current size of the global name dictionary used by libxml2
             for the current thread.  Each thread has its own dictionary.
         """
+    @deprecated("Removed since 6.0, due to corresponding removal in libxml2")
     def dump(
         self, output_file: str | bytes | None = None, byte_count: int | None = None
     ) -> None:
@@ -482,6 +483,7 @@ class _MemDebug:
         byte_count : int, optional
             Limits number of bytes in the dump, default is None (unlimited)
         """
+    @deprecated("Removed since 6.0, due to corresponding removal in libxml2")
     def show(
         self, output_file: str | bytes | None = None, block_count: int | None = None
     ) -> None:

src/lxml-stubs/etree/_module_misc.pyi

@@ -24,7 +24,9 @@ from ._xmlerror import _BaseErrorLog, _ListErrorLog
 DEBUG: int
 ICONV_COMPILED_VERSION: tuple[int, int]
 LIBXML_VERSION: tuple[int, int, int]
+LIBXML_COMPILED_FEATURES: set[str]
 LIBXML_COMPILED_VERSION: tuple[int, int, int]
+LIBXML_FEATURES: set[str]
 LXML_VERSION: tuple[int, int, int, int]
 __version__: LiteralString
 

src/lxml-stubs/html/builder.pyi

@@ -19,31 +19,44 @@ ACRONYM = E.acronym
 ADDRESS = E.address
 APPLET = E.applet
 AREA = E.area
+ARTICLE = E.article
+ASIDE = E.aside
+AUDIO = E.audio
 B = E.b
 BASE = E.base
 BASEFONT = E.basefont
+BDI = E.bdi
 BDO = E.bdo
 BIG = E.big
 BLOCKQUOTE = E.blockquote
 BODY = E.body
 BR = E.br
 BUTTON = E.button
+CANVAS = E.canvas
 CAPTION = E.caption
 CENTER = E.center
 CITE = E.cite
 CODE = E.code
 COL = E.col
 COLGROUP = E.colgroup
+DATA = E.data
+DATALIST = E.datalist
 DD = E.dd
 DEL = E.__getattr__("del")
+DETAILS = E.details
 DFN = E.dfn
+DIALOG = E.dialog
 DIR = E.dir
 DIV = E.div
 DL = E.dl
 DT = E.dt
 EM = E.em
+EMBED = E.embed
 FIELDSET = E.fieldset
+FIGCAPTION = E.figcaption
+FIGURE = E.figure
 FONT = E.font
+FOOTER = E.footer
 FORM: partial[FormElement]
 FRAME = E.frame
 FRAMESET = E.frameset
@@ -54,6 +67,8 @@ H4 = E.h4
 H5 = E.h5
 H6 = E.h6
 HEAD = E.head
+HEADER = E.header
+HGROUP = E.hgroup
 HR = E.hr
 HTML = E.html
 I = E.i
@@ -67,43 +82,68 @@ LABEL: partial[LabelElement]
 LEGEND = E.legend
 LI = E.li
 LINK = E.link
+MAIN = E.main
 MAP = E.map
+MARK = E.mark
+MARQUEE = E.marquee
 MENU = E.menu
 META = E.meta
+METER = E.meter
+NAV = E.nav
+NOBR = E.nobr
 NOFRAMES = E.noframes
 NOSCRIPT = E.noscript
 OBJECT = E.object
 OL = E.ol
 OPTGROUP = E.optgroup
 OPTION = E.option
+OUTPUT = E.output
 P = E.p
 PARAM = E.param
+PICTURE = E.picture
+PORTAL = E.portal
 PRE = E.pre
+PROGRESS = E.progress
 Q = E.q
+RB = E.rb
+RP = E.rp
+RT = E.rt
+RTC = E.rtc
+RUBY = E.ruby
 S = E.s
 SAMP = E.samp
 SCRIPT = E.script
+SEARCH = E.search
+SECTION = E.section
 SELECT: partial[SelectElement]
+SLOT = E.slot
 SMALL = E.small
+SOURCE = E.source
 SPAN = E.span
 STRIKE = E.strike
 STRONG = E.strong
 STYLE = E.style
 SUB = E.sub
+SUMMARY = E.summary
 SUP = E.sup
 TABLE = E.table
 TBODY = E.tbody
 TD = E.td
+TEMPLATE = E.template
 TEXTAREA: partial[TextareaElement]
 TFOOT = E.tfoot
 TH = E.th
 THEAD = E.thead
+TIME = E.time
 TITLE = E.title
 TR = E.tr
+TRACK = E.track
 TT = E.tt
 U = E.u
 UL = E.ul
 VAR = E.var
+VIDEO = E.video
+WBR = E.wbr
 
 # attributes
 ATTR = dict

tests/runtime/test_constants.py

@@ -27,6 +27,10 @@ def test_ver_const(self) -> None:
         if hasattr(_e, "ICONV_COMPILED_VERSION"):
             reveal_type(_e.ICONV_COMPILED_VERSION)
 
+        if _e.LXML_VERSION >= (6, 0):
+            reveal_type(_e.LIBXML_COMPILED_FEATURES)
+            reveal_type(_e.LIBXML_FEATURES)
+
     def test_other_const(self) -> None:
         reveal_type(_e.DEBUG)
         _e.DEBUG = 1

tests/static/test-dtd.yml

@@ -8,14 +8,16 @@
     - args: filename
     - args: file=path
     - args: fileio
-    - args: external_id=dtd_id
+    - args: external_id=b
+    - args: external_id=s
   main: |
     from pathlib import PosixPath
     from lxml.etree import DTD
     filename: str
     path: PosixPath
     fileio = open('file', 'r')
-    dtd_id: bytes
+    b: bytes
+    s: str
     reveal_type(DTD({{ args }}))  # N: Revealed type is "lxml.etree._dtd.DTD"
 
 - case: constructor_noargs
@@ -24,14 +26,13 @@
     from lxml.etree import DTD
     dtd = DTD()
 
-# str not supported as of lxml 4.8
 - case: constructor_badargs
   expect_fail: true
   parametrized:
     - args: 1
     - args: "None, 1"
     - args: "None, badarg=b"
-    - args: "external_id=s"
+    - args: "external_id=1"
   main: |
     from lxml.etree import DTD
     s: str