ladybird/Meta/import-wpt-test.py
Shannon Booth 06f4ac367e Meta: Support importing WPT reftests with multiple references
Collect all match and mismatch links when identifying WPT reftests
instead of rejecting tests with more than one reference.

Map and rewrite every imported reference URL so tests with both match
and mismatch expectations can be imported.
2026-06-14 22:12:00 +02:00

432 lines
16 KiB
Python
Executable file

#!/usr/bin/env python3
import argparse
import os
import re
from collections import namedtuple
from dataclasses import dataclass
from enum import Enum
from html.parser import HTMLParser
from pathlib import Path
from posixpath import normpath
from urllib.error import HTTPError
from urllib.parse import urljoin
from urllib.parse import urlparse
from urllib.parse import urlsplit
from urllib.parse import urlunsplit
from urllib.request import urlopen
download_exclude_list = {
# This relies on a dynamic path that cannot be rewitten by the importer.
"/resources/idlharness.js",
# This file has been modified to improve the performance of our test runner.
"/resources/testharness.js",
# We have modified this file to output text files compatible with our test runner.
"/resources/testharnessreport.js",
# We have modified this file to use our internals object instead of relying on WebDriver.
"/resources/testdriver-vendor.js",
# Modified to use a relative path, which the importer can't rewrite.
"/fonts/ahem.css",
}
visited_paths = set()
class TestType(Enum):
TEXT = 1, "Tests/LibWeb/Text/input/wpt-import", "Tests/LibWeb/Text/expected/wpt-import"
REF = 2, "Tests/LibWeb/Ref/input/wpt-import", "Tests/LibWeb/Ref/expected/wpt-import"
CRASH = 3, "Tests/LibWeb/Crash/wpt-import", ""
def __new__(cls, *args, **kwds):
obj = object.__new__(cls)
obj._value_ = args[0]
return obj
def __init__(self, _: int, input_path: str, expected_path: str):
self.input_path = input_path
self.expected_path = expected_path
PathMapping = namedtuple("PathMapping", ["source", "destination"])
class ResourceType(Enum):
INPUT = 1
EXPECTED = 2
@dataclass
class ResourceAndType:
resource: str
type: ResourceType
test_type = TestType.TEXT
raw_reference_paths = [] # As specified in the test HTML
reference_paths = [] # With parent directories
class LinkedResourceFinder(HTMLParser):
def __init__(self):
super().__init__()
self._tag_stack_ = []
self._match_css_url_ = re.compile(r"url\(['\"]?(?P<url>[^'\")]+)['\"]?\)")
self._match_css_import_string_ = re.compile(r"@import\s+\"(?P<url>[^\")]+)\"")
self._match_fetch_import_path = re.compile(r"fetch\((\"|\')(?P<url>.*)(\"|\')\)")
self._match_import_call = re.compile(r"import\(['\"](?P<url>.*?)['\"].*\)")
self._match_import_statement = re.compile(r"import (.*? from )?['\"](?P<url>.*?)['\"]")
self._match_src_assignment = re.compile(r"\.src ?= ?['\"](?P<url>.*?)['\"]")
self._match_worker_import_path = re.compile(r"Worker\(['\"](?P<url>.*)['\"]\)")
self._resources = set()
@property
def resources(self) -> set:
return self._resources
def handle_starttag(self, tag, attrs):
self._tag_stack_.append(tag)
if ":" in tag:
tag = tag.split(":", 1)[-1]
if tag in ["script", "img", "iframe"]:
attr_dict = dict(attrs)
if "src" in attr_dict:
self._resources.add(attr_dict["src"])
if tag == "link":
attr_dict = dict(attrs)
if "rel" in attr_dict and attr_dict["rel"] == "stylesheet":
self._resources.add(attr_dict["href"])
if tag == "form":
attr_dict = dict(attrs)
if "action" in attr_dict:
self._resources.add(attr_dict["action"])
def handle_endtag(self, tag):
self._tag_stack_.pop()
def handle_data(self, data):
if self._tag_stack_ and self._tag_stack_[-1] == "style":
# Look for uses of url()
url_iterator = self._match_css_url_.finditer(data)
for match in url_iterator:
self._resources.add(match.group("url"))
# Look for @imports that use plain strings - we already found the url() ones
import_iterator = self._match_css_import_string_.finditer(data)
for match in import_iterator:
self._resources.add(match.group("url"))
elif self._tag_stack_ and self._tag_stack_[-1] == "script":
# Look for uses of fetch()
fetch_iterator = self._match_fetch_import_path.finditer(data)
for match in fetch_iterator:
self._resources.add(match.group("url"))
# Look for uses of import()
import_call_iterator = self._match_import_call.finditer(data)
for match in import_call_iterator:
self._resources.add(match.group("url"))
# Look for uses of import statements
import_statement_iterator = self._match_import_statement.finditer(data)
for match in import_statement_iterator:
self._resources.add(match.group("url"))
# Look for uses of .src = "..."
src_assignment_iterator = self._match_src_assignment.finditer(data)
for match in src_assignment_iterator:
self._resources.add(match.group("url"))
# Look for uses of Worker()
filepath_iterator = self._match_worker_import_path.finditer(data)
for match in filepath_iterator:
self._resources.add(match.group("url"))
class TestTypeIdentifier(HTMLParser):
"""Identifies what kind of test the page is, and stores it in self.test_type
For reference tests, the URLs of the reference pages are saved as self.reference_paths
"""
def __init__(self, url):
super().__init__()
self.url = url
self.test_type = TestType.TEXT
self.reference_paths = []
def handle_starttag(self, tag, attrs):
if ":" in tag:
tag = tag.split(":", 1)[-1]
if tag == "link":
attr_dict = dict(attrs)
rel_tokens = (attr_dict.get("rel") or "").split()
href = attr_dict.get("href")
if href is not None and ("match" in rel_tokens or "mismatch" in rel_tokens):
self.test_type = TestType.REF
self.reference_paths.append(href)
def map_to_path(
sources: list[ResourceAndType], wpt_base_url: str, is_resource=True, resource_path=None
) -> list[PathMapping]:
filepaths: list[PathMapping] = []
for source in sources:
base_directory = test_type.input_path if source.type == ResourceType.INPUT else test_type.expected_path
if source.resource.startswith("/") or not is_resource:
file_path = Path(base_directory, source.resource.lstrip("/"))
else:
assert resource_path is not None
parsed_url = urlparse(source.resource)
if parsed_url.scheme != "":
print(f"Skipping '{source.resource}'. Downloading external resources is not supported.")
continue
# Add it as a sibling path if it's a relative resource
sibling_location = Path(resource_path).parent
parent_directory = Path(base_directory, sibling_location)
file_path = Path(parent_directory, source.resource)
# Map to source and destination
output_path = wpt_base_url + str(file_path).replace(base_directory, "")
filepaths.append(PathMapping(output_path, file_path.absolute()))
return filepaths
def is_crash_test(url_string):
# https://web-platform-tests.org/writing-tests/crashtest.html
# A test file is treated as a crash test if they have -crash in their name before the file extension, or they are
# located in a folder named crashtests
parsed_url = urlparse(url_string)
path_segments = parsed_url.path.strip("/").split("/")
if len(path_segments) > 1 and "crashtests" in path_segments[::-1]:
return True
file_name = path_segments[-1]
file_name_parts = file_name.split(".")
if len(file_name_parts) > 1 and any([part.endswith("-crash") for part in file_name_parts[:-1]]):
return True
return False
def modify_sources(files, resources: list[ResourceAndType]) -> None:
for file in files:
# Get the distance to the wpt-imports folder
folder_index = str(file).find(test_type.input_path)
if folder_index == -1:
folder_index = str(file).find(test_type.expected_path)
non_prefixed_path = str(file)[folder_index + len(test_type.expected_path) :]
else:
non_prefixed_path = str(file)[folder_index + len(test_type.input_path) :]
parent_folder_count = len(Path(non_prefixed_path).parent.parts) - 1
parent_folder_path = "../" * parent_folder_count
# Open the file in binary mode. Some WPT test files (e.g,
# encoding-detection tests) are intentionally stored in a legacy
# encoding rather than UTF-8. We only rewrite ASCII path references
# below, so byte-level replacements preserve the original encoding of
# the non-ASCII content.
with open(file, "rb") as f:
page_source = f.read()
# Iterate all scripts and overwrite the src attribute
for resource in map(lambda r: r.resource, resources):
if resource.startswith("/"):
new_src_value = parent_folder_path + resource[1::]
page_source = page_source.replace(resource.encode(), new_src_value.encode())
# Look for mentions of reference pages, and update their hrefs
for raw_reference_path, reference_path in zip(raw_reference_paths, reference_paths):
new_reference_path = parent_folder_path + "../../expected/wpt-import/" + reference_path[::]
page_source = page_source.replace(raw_reference_path.encode(), new_reference_path.encode())
with open(file, "wb") as f:
f.write(page_source)
def normalize_url(url):
parts = urlsplit(url)
normalized_path = normpath(parts.path)
normalized_url = urlunsplit((parts.scheme, parts.netloc, normalized_path, parts.query, parts.fragment))
return normalized_url
def remove_repeated_url_slashes(url):
parsed = urlparse(url)
return "/" + "/".join(segment for segment in parsed.path.split("/") if segment)
def download_files(filepaths, wpt_base_url, skip_existing):
downloaded_files = []
for file in filepaths:
normalized_path = remove_repeated_url_slashes(file.source)
print(f"Source {normalized_path}, Destination {file.destination}")
if normalized_path in visited_paths:
continue
if normalized_path in download_exclude_list:
print(f"Skipping {file.source} as it is in the exclude list")
visited_paths.add(normalized_path)
continue
source = urljoin(wpt_base_url, normalized_path)
destination = Path(os.path.normpath(file.destination))
if skip_existing and destination.exists():
print(f"Skipping {destination} as it already exists")
visited_paths.add(normalized_path)
continue
visited_paths.add(normalized_path)
print(f"Downloading {source} to {destination}")
connection = urlopen(source)
if connection.status != 200:
print(f"Failed to download {file.source}")
continue
os.makedirs(destination.parent, exist_ok=True)
with open(destination, "wb") as f:
f.write(connection.read())
downloaded_files.append(destination)
return downloaded_files
def download_headers_files(filepaths, wpt_base_url, skip_existing):
for file in filepaths:
normalized_path = remove_repeated_url_slashes(file.source)
headers_path = normalized_path + ".headers"
if headers_path in visited_paths:
continue
source = urljoin(wpt_base_url, headers_path)
destination = Path(os.path.normpath(str(file.destination) + ".headers"))
if skip_existing and destination.exists():
print(f"Skipping {destination} as it already exists")
visited_paths.add(headers_path)
continue
visited_paths.add(headers_path)
try:
connection = urlopen(source)
except HTTPError:
continue
if connection.status != 200:
continue
os.makedirs(destination.parent, exist_ok=True)
with open(destination, "wb") as f:
f.write(connection.read())
print(f"Downloaded headers file: {destination}")
def create_expectation_files(files, skip_existing):
# Ref tests don't have an expectation text file
if test_type in [TestType.REF, TestType.CRASH]:
return
for file in files:
new_path = str(file.destination).replace(test_type.input_path, test_type.expected_path)
new_path = new_path.rsplit(".", 1)[0] + ".txt"
expected_file = Path(new_path)
if skip_existing and expected_file.exists():
print(f"Skipping {expected_file} as it already exists")
continue
os.makedirs(expected_file.parent, exist_ok=True)
expected_file.touch()
def main():
parser = argparse.ArgumentParser(description="Import a WPT test into LibWeb")
parser.add_argument("--force", action="store_true", help="Force download of files even if they already exist")
parser.add_argument("--wpt-base-url", type=urlparse, default="https://wpt.live/")
parser.add_argument("url", type=str, help="The URL of the WPT test to import")
args = parser.parse_args()
skip_existing = not args.force
url_to_import = args.url
wpt_base_url = args.wpt_base_url.geturl()
print(f"Importing WPT test from: {wpt_base_url}")
resource_path = "/".join(Path(url_to_import).parts[2::])
with urlopen(url_to_import) as response:
# Some WPT test files (e.g. `encoding-detection` tests) are intentionally stored in a
# legacy encoding rather than UTF-8. We only parse the ASCII tag structure here, so
# replacing undecodable bytes with U+FFFD is safe and avoids a UnicodeDecodeError.
page = response.read().decode("utf-8", errors="replace")
global test_type, reference_paths, raw_reference_paths
if is_crash_test(url_to_import):
test_type = TestType.CRASH
else:
identifier = TestTypeIdentifier(url_to_import)
identifier.feed(page)
test_type = identifier.test_type
raw_reference_paths = identifier.reference_paths
print(f"Identified {url_to_import} as type {test_type}, refs {raw_reference_paths}")
main_file = [ResourceAndType(resource_path, ResourceType.INPUT)]
main_paths = map_to_path(main_file, wpt_base_url, False)
if test_type == TestType.REF and not raw_reference_paths:
raise RuntimeError("Failed to file reference path in ref test")
for raw_reference_path in raw_reference_paths:
if raw_reference_path.startswith("/"):
reference_path = raw_reference_path
main_paths.append(
PathMapping(
normalize_url(wpt_base_url + raw_reference_path),
Path(test_type.expected_path + raw_reference_path).absolute(),
)
)
else:
reference_path = Path(resource_path).parent.joinpath(raw_reference_path).__str__()
main_paths.append(
PathMapping(
normalize_url(wpt_base_url + "/" + reference_path),
Path(test_type.expected_path + "/" + reference_path).absolute(),
)
)
reference_paths.append(reference_path)
files_to_modify = download_files(main_paths, wpt_base_url, skip_existing)
download_headers_files(main_paths, wpt_base_url, skip_existing)
create_expectation_files(main_paths, skip_existing)
input_parser = LinkedResourceFinder()
input_parser.feed(page)
additional_resources = list(map(lambda s: ResourceAndType(s, ResourceType.INPUT), input_parser.resources))
expected_parser = LinkedResourceFinder()
for path in main_paths[1:]:
with urlopen(path.source) as response:
# Some WPT test files (e.g. `encoding-detection` tests) are intentionally stored in a
# legacy encoding rather than UTF-8. We only parse the ASCII tag structure here, so
# replacing undecodable bytes with U+FFFD is safe and avoids a UnicodeDecodeError.
page = response.read().decode("utf-8", errors="replace")
expected_parser.feed(page)
additional_resources.extend(
list(map(lambda s: ResourceAndType(s, ResourceType.EXPECTED), expected_parser.resources))
)
modify_sources(files_to_modify, additional_resources)
script_paths = map_to_path(additional_resources, wpt_base_url, True, resource_path)
download_files(script_paths, wpt_base_url, skip_existing)
download_headers_files(script_paths, wpt_base_url, skip_existing)
if __name__ == "__main__":
main()