Merge pull request #10291 from jdufresne/html5lib

Author: pradyunsg

news/10291.feature.rst

@@ -0,0 +1,5 @@
+Changed ``PackageFinder`` to parse HTML documents using the stdlib
+:class:`html.parser.HTMLParser` class instead of the ``html5lib`` package. For
+now, the deprecated ``html5lib`` code remains and can be used with the
+``--use-deprecated=html5lib`` command line option, but it will be removed in a
+future pip release.

src/pip/_internal/cli/cmdoptions.py

@@ -964,7 +964,12 @@ def check_list_path_option(options: Values) -> None:
     metavar="feature",
     action="append",
     default=[],
-    choices=["legacy-resolver", "out-of-tree-build", "backtrack-on-build-failures"],
+    choices=[
+        "legacy-resolver",
+        "out-of-tree-build",
+        "backtrack-on-build-failures",
+        "html5lib",
+    ],
     help=("Enable deprecated functionality, that will be removed in the future."),
 )
 

src/pip/_internal/cli/req_command.py

@@ -502,4 +502,5 @@ def _build_package_finder(
             link_collector=link_collector,
             selection_prefs=selection_prefs,
             target_python=target_python,
+            use_deprecated_html5lib="html5lib" in options.deprecated_features_enabled,
         )

src/pip/_internal/commands/index.py

@@ -97,6 +97,7 @@ def _build_package_finder(
             link_collector=link_collector,
             selection_prefs=selection_prefs,
             target_python=target_python,
+            use_deprecated_html5lib="html5lib" in options.deprecated_features_enabled,
         )
 
     def get_available_package_versions(self, options: Values, args: List[Any]) -> None:

src/pip/_internal/commands/list.py

@@ -149,6 +149,7 @@ def _build_package_finder(
         return PackageFinder.create(
             link_collector=link_collector,
             selection_prefs=selection_prefs,
+            use_deprecated_html5lib="html5lib" in options.deprecated_features_enabled,
         )
 
     def run(self, options: Values, args: List[str]) -> int:

src/pip/_internal/index/collector.py

@@ -12,15 +12,20 @@
 import urllib.parse
 import urllib.request
 import xml.etree.ElementTree
+from html.parser import HTMLParser
 from optparse import Values
 from typing import (
+    TYPE_CHECKING,
+    Any,
     Callable,
+    Dict,
     Iterable,
     List,
     MutableMapping,
     NamedTuple,
     Optional,
     Sequence,
+    Tuple,
     Union,
 )
 
@@ -39,6 +44,11 @@
 
 from .sources import CandidatesFromPage, LinkSource, build_source
 
+if TYPE_CHECKING:
+    from typing import Protocol
+else:
+    Protocol = object
+
 logger = logging.getLogger(__name__)
 
 HTMLElement = xml.etree.ElementTree.Element
@@ -163,6 +173,8 @@ def _determine_base_url(document: HTMLElement, page_url: str) -> str:
     :param document: An HTML document representation. The current
         implementation expects the result of ``html5lib.parse()``.
     :param page_url: The URL of the HTML document.
+
+    TODO: Remove when `html5lib` is dropped.
     """
     for base in document.findall(".//base"):
         href = base.get("href")
@@ -234,20 +246,20 @@ def _clean_link(url: str) -> str:
 
 
 def _create_link_from_element(
-    anchor: HTMLElement,
+    element_attribs: Dict[str, Optional[str]],
     page_url: str,
     base_url: str,
 ) -> Optional[Link]:
     """
-    Convert an anchor element in a simple repository page to a Link.
+    Convert an anchor element's attributes in a simple repository page to a Link.
     """
-    href = anchor.get("href")
+    href = element_attribs.get("href")
     if not href:
         return None
 
     url = _clean_link(urllib.parse.urljoin(base_url, href))
-    pyrequire = anchor.get("data-requires-python")
-    yanked_reason = anchor.get("data-yanked")
+    pyrequire = element_attribs.get("data-requires-python")
+    yanked_reason = element_attribs.get("data-yanked")
 
     link = Link(
         url,
@@ -271,32 +283,40 @@ def __hash__(self) -> int:
         return hash(self.page.url)
 
 
-def with_cached_html_pages(
-    fn: Callable[["HTMLPage"], Iterable[Link]],
-) -> Callable[["HTMLPage"], List[Link]]:
+class ParseLinks(Protocol):
+    def __call__(
+        self, page: "HTMLPage", use_deprecated_html5lib: bool
+    ) -> Iterable[Link]:
+        ...
+
+
+def with_cached_html_pages(fn: ParseLinks) -> ParseLinks:
     """
     Given a function that parses an Iterable[Link] from an HTMLPage, cache the
     function's result (keyed by CacheablePageContent), unless the HTMLPage
     `page` has `page.cache_link_parsing == False`.
     """
 
     @functools.lru_cache(maxsize=None)
-    def wrapper(cacheable_page: CacheablePageContent) -> List[Link]:
-        return list(fn(cacheable_page.page))
+    def wrapper(
+        cacheable_page: CacheablePageContent, use_deprecated_html5lib: bool
+    ) -> List[Link]:
+        return list(fn(cacheable_page.page, use_deprecated_html5lib))
 
     @functools.wraps(fn)
-    def wrapper_wrapper(page: "HTMLPage") -> List[Link]:
+    def wrapper_wrapper(page: "HTMLPage", use_deprecated_html5lib: bool) -> List[Link]:
         if page.cache_link_parsing:
-            return wrapper(CacheablePageContent(page))
-        return list(fn(page))
+            return wrapper(CacheablePageContent(page), use_deprecated_html5lib)
+        return list(fn(page, use_deprecated_html5lib))
 
     return wrapper_wrapper
 
 
-@with_cached_html_pages
-def parse_links(page: "HTMLPage") -> Iterable[Link]:
+def _parse_links_html5lib(page: "HTMLPage") -> Iterable[Link]:
     """
     Parse an HTML document, and yield its anchor elements as Link objects.
+
+    TODO: Remove when `html5lib` is dropped.
     """
     document = html5lib.parse(
         page.content,
@@ -307,6 +327,31 @@ def parse_links(page: "HTMLPage") -> Iterable[Link]:
     url = page.url
     base_url = _determine_base_url(document, url)
     for anchor in document.findall(".//a"):
+        link = _create_link_from_element(
+            anchor.attrib,
+            page_url=url,
+            base_url=base_url,
+        )
+        if link is None:
+            continue
+        yield link
+
+
+@with_cached_html_pages
+def parse_links(page: "HTMLPage", use_deprecated_html5lib: bool) -> Iterable[Link]:
+    """
+    Parse an HTML document, and yield its anchor elements as Link objects.
+    """
+    if use_deprecated_html5lib:
+        return _parse_links_html5lib(page)
+
+    parser = HTMLLinkParser()
+    encoding = page.encoding or "utf-8"
+    parser.feed(page.content.decode(encoding))
+
+    url = page.url
+    base_url = parser.base_url or url
+    for anchor in parser.anchors:
         link = _create_link_from_element(
             anchor,
             page_url=url,
@@ -343,6 +388,49 @@ def __str__(self) -> str:
         return redact_auth_from_url(self.url)
 
 
+class HTMLLinkParser(HTMLParser):
+    """
+    HTMLParser that keeps the first base HREF and a list of all anchor
+    elements' attributes.
+    """
+
+    def __init__(self, *args: Any, **kwargs: Any) -> None:
+        super().__init__(*args, **kwargs)
+        self._seen_decl = False
+        self.base_url: Optional[str] = None
+        self.anchors: List[Dict[str, Optional[str]]] = []
+
+    def handle_decl(self, decl: str) -> None:
+        if decl != "DOCTYPE html":
+            self._raise_error()
+        self._seen_decl = True
+
+    def handle_starttag(self, tag: str, attrs: List[Tuple[str, Optional[str]]]) -> None:
+        if not self._seen_decl:
+            self._raise_error()
+
+        if tag == "base" and self.base_url is None:
+            href = self.get_href(attrs)
+            if href is not None:
+                self.base_url = href
+        elif tag == "a":
+            self.anchors.append(dict(attrs))
+
+    def get_href(self, attrs: List[Tuple[str, Optional[str]]]) -> Optional[str]:
+        for name, value in attrs:
+            if name == "href":
+                return value
+        return None
+
+    def _raise_error(self) -> None:
+        raise ValueError(
+            "HTML doctype missing or incorrect. Expected <!DOCTYPE html>.\n\n"
+            "If you believe this error to be incorrect, try passing the "
+            "command line option --use-deprecated=html5lib and please leave "
+            "a comment on the pip issue at https://github.com/pypa/pip/issues/10825."
+        )
+
+
 def _handle_get_page_fail(
     link: Link,
     reason: Union[str, Exception],

src/pip/_internal/index/package_finder.py

@@ -580,6 +580,7 @@ def __init__(
         link_collector: LinkCollector,
         target_python: TargetPython,
         allow_yanked: bool,
+        use_deprecated_html5lib: bool,
         format_control: Optional[FormatControl] = None,
         candidate_prefs: Optional[CandidatePreferences] = None,
         ignore_requires_python: Optional[bool] = None,
@@ -604,6 +605,7 @@ def __init__(
         self._ignore_requires_python = ignore_requires_python
         self._link_collector = link_collector
         self._target_python = target_python
+        self._use_deprecated_html5lib = use_deprecated_html5lib
 
         self.format_control = format_control
 
@@ -620,6 +622,8 @@ def create(
         link_collector: LinkCollector,
         selection_prefs: SelectionPreferences,
         target_python: Optional[TargetPython] = None,
+        *,
+        use_deprecated_html5lib: bool,
     ) -> "PackageFinder":
         """Create a PackageFinder.
 
@@ -644,6 +648,7 @@ def create(
             allow_yanked=selection_prefs.allow_yanked,
             format_control=selection_prefs.format_control,
             ignore_requires_python=selection_prefs.ignore_requires_python,
+            use_deprecated_html5lib=use_deprecated_html5lib,
         )
 
     @property
@@ -765,7 +770,7 @@ def process_project_url(
         if html_page is None:
             return []
 
-        page_links = list(parse_links(html_page))
+        page_links = list(parse_links(html_page, self._use_deprecated_html5lib))
 
         with indent_log():
             package_links = self.evaluate_links(

src/pip/_internal/self_outdated_check.py

@@ -141,6 +141,9 @@ def pip_self_version_check(session: PipSession, options: optparse.Values) -> Non
             finder = PackageFinder.create(
                 link_collector=link_collector,
                 selection_prefs=selection_prefs,
+                use_deprecated_html5lib=(
+                    "html5lib" in options.deprecated_features_enabled
+                ),
             )
             best_candidate = finder.find_best_candidate("pip").best_candidate
             if best_candidate is None:

tests/data/indexes/datarequire/fakepackage/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html><head><title>Links for fakepackage</title><meta name="api-version" value="2" /></head><body><h1>Links for fakepackage</h1>
     <a data-requires-python=''                href="/fakepackage-1.0.0.tar.gz#md5=00000000000000000000000000000000" rel="internal">fakepackage-1.0.0.tar.gz</a><br/>
     <a data-requires-python='&lt;2.7'         href="/fakepackage-2.6.0.tar.gz#md5=00000000000000000000000000000000" rel="internal">fakepackage-2.6.0.tar.gz</a><br/>

tests/data/indexes/dev/bar/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a href="bar-1.0.tar.gz">bar-1.0.tar.gz</a>

tests/data/indexes/in dex/simple/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a href="../../../packages/simple-1.0.tar.gz#md5=4bdf78ebb7911f215c1972cf71b378f0">simple-1.0.tar.gz</a>

tests/data/indexes/pre/bar/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a href="bar-1.0.tar.gz">bar-1.0.tar.gz</a>

tests/data/indexes/simple/simple/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a href="../../../packages/simple-1.0.tar.gz#md5=4bdf78ebb7911f215c1972cf71b378f0">simple-1.0.tar.gz</a>

tests/data/indexes/yanked/simple/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a href="../../../packages/simple-1.0.tar.gz">simple-1.0.tar.gz</a>

tests/data/indexes/yanked_all/simple/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html>
   <body>
     <a data-yanked="test reason message" href="../../../packages/simple-1.0.tar.gz">simple-1.0.tar.gz</a>

tests/data/packages3/dinner/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html><head><title>PyPI Mirror</title></head>
 <body>
   <h1>PyPI Mirror</h1>

tests/data/packages3/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html><head><title>PyPI Mirror</title></head>
 <body>
   <h1>PyPI Mirror</h1>

tests/data/packages3/requiredinner/index.html

@@ -1,3 +1,4 @@
+<!DOCTYPE html>
 <html><head><title>PyPI Mirror</title></head>
 <body>
   <h1>PyPI Mirror</h1>

tests/functional/test_build_env.py

@@ -48,6 +48,7 @@ def run_with_build_env(
             finder = PackageFinder.create(
                 link_collector=link_collector,
                 selection_prefs=selection_prefs,
+                use_deprecated_html5lib=False,
             )
 
             with global_tempdir_manager():

tests/functional/test_new_resolver_hashes.py

@@ -26,6 +26,7 @@ def _create_find_links(script: PipTestEnvironment) -> _FindLinks:
     index_html = script.scratch_path / "index.html"
     index_html.write_text(
         """
+        <!DOCTYPE html>
         <a href="{sdist_url}#sha256={sdist_hash}">{sdist_path.stem}</a>
         <a href="{wheel_url}#sha256={wheel_hash}">{wheel_path.stem}</a>
         """.format(

tests/lib/__init__.py

@@ -141,6 +141,7 @@ def make_test_finder(
     allow_all_prereleases: bool = False,
     session: Optional[PipSession] = None,
     target_python: Optional[TargetPython] = None,
+    use_deprecated_html5lib: bool = False,
 ) -> PackageFinder:
     """
     Create a PackageFinder for testing purposes.
@@ -159,6 +160,7 @@ def make_test_finder(
         link_collector=link_collector,
         selection_prefs=selection_prefs,
         target_python=target_python,
+        use_deprecated_html5lib=use_deprecated_html5lib,
     )
 
 

tests/unit/resolution_resolvelib/conftest.py

@@ -26,7 +26,7 @@ def finder(data: TestData) -> Iterator[PackageFinder]:
     scope = SearchScope([str(data.packages)], [])
     collector = LinkCollector(session, scope)
     prefs = SelectionPreferences(allow_yanked=False)
-    finder = PackageFinder.create(collector, prefs)
+    finder = PackageFinder.create(collector, prefs, use_deprecated_html5lib=False)
     yield finder