split puppeteer plugin into Puppeteer, Playwright, and Chrome

This commit is contained in:
Nick Sweeting 2024-09-21 04:00:54 -07:00
parent 33fd7fe439
commit 541cd6c5a1
No known key found for this signature in database
10 changed files with 414 additions and 124 deletions

View file

@ -6,33 +6,38 @@ from django.conf import settings
# Depends on other PyPI/vendor packages:
from pydantic import InstanceOf, Field
from pydantic_pkgr import BinProvider, BinName, BinProviderName, ProviderLookupDict, InstallArgs, HostBinPath, bin_abspath
from pydantic_pkgr import (
BinProvider,
BinName,
BinProviderName,
ProviderLookupDict,
InstallArgs,
PATHStr,
HostBinPath,
)
# Depends on other Django apps:
from plugantic.base_plugin import BasePlugin
from plugantic.base_configset import BaseConfigSet, ConfigSectionName
from plugantic.base_configset import BaseConfigSet
from plugantic.base_binary import BaseBinary, BaseBinProvider, env
# from plugantic.base_extractor import BaseExtractor
# from plugantic.base_queue import BaseQueue
from plugantic.base_hook import BaseHook
# Depends on Other Plugins:
from builtin_plugins.npm.apps import SYS_NPM_BINPROVIDER
from builtin_plugins.npm.apps import LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER
###################### Config ##########################
class PuppeteerDependencyConfigs(BaseConfigSet):
section: ConfigSectionName = 'DEPENDENCY_CONFIG'
class PuppeteerConfigs(BaseConfigSet):
# section: ConfigSectionName = 'DEPENDENCY_CONFIG'
PUPPETEER_BINARY: str = Field(default='wget')
PUPPETEER_ARGS: Optional[List[str]] = Field(default=None)
PUPPETEER_EXTRA_ARGS: List[str] = []
PUPPETEER_DEFAULT_ARGS: List[str] = ['--timeout={TIMEOUT-10}']
class PuppeteerConfigs(PuppeteerDependencyConfigs):
# section: ConfigSectionName = 'ALL_CONFIGS'
# PUPPETEER_BINARY: str = Field(default='wget')
# PUPPETEER_ARGS: Optional[List[str]] = Field(default=None)
# PUPPETEER_EXTRA_ARGS: List[str] = []
# PUPPETEER_DEFAULT_ARGS: List[str] = ['--timeout={TIMEOUT-10}']
pass
DEFAULT_GLOBAL_CONFIG = {
@ -42,17 +47,29 @@ PUPPETEER_CONFIG = PuppeteerConfigs(**DEFAULT_GLOBAL_CONFIG)
LIB_DIR_BROWSERS = settings.CONFIG.OUTPUT_DIR / "lib" / "browsers"
class PuppeteerBinary(BaseBinary):
name: BinName = "puppeteer"
binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
PUPPETEER_BINARY = PuppeteerBinary()
class PuppeteerBinProvider(BaseBinProvider):
name: BinProviderName = "puppeteer"
INSTALLER_BIN: BinName = "npx"
PATH: PATHStr = str(settings.CONFIG.BIN_DIR)
puppeteer_browsers_dir: Optional[Path] = LIB_DIR_BROWSERS
puppeteer_install_args: List[str] = ["@puppeteer/browsers", "install", "--path", str(LIB_DIR_BROWSERS)]
# packages_handler: ProviderLookupDict = {
# "chrome": lambda:
# ['chrome@stable'],
# }
packages_handler: ProviderLookupDict = Field(default={
"chrome": lambda:
['chrome@stable'],
}, exclude=True)
_browser_abspaths: ClassVar[Dict[str, HostBinPath]] = {}
@ -61,6 +78,15 @@ class PuppeteerBinProvider(BaseBinProvider):
if self.puppeteer_browsers_dir:
self.puppeteer_browsers_dir.mkdir(parents=True, exist_ok=True)
def installed_browser_bins(self, browser_name: str='*') -> List[Path]:
# if on macOS, browser binary is inside a .app, otherwise it's just a plain binary
if platform.system().lower() == 'darwin':
# /data/lib/browsers/chrome/mac_arm-129.0.6668.58/chrome-mac-arm64/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing
return sorted(self.puppeteer_browsers_dir.glob(f'{browser_name}/mac*/chrome*/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing'))
# /data/lib/browsers/chrome/linux-131.0.6730.0/chrome-linux64/chrome
return sorted(self.puppeteer_browsers_dir.glob(f"{browser_name}/linux*/chrome*/chrome"))
def on_get_abspath(self, bin_name: BinName, **context) -> Optional[HostBinPath]:
assert bin_name == 'chrome', 'Only chrome is supported using the @puppeteer/browsers install method currently.'
@ -70,21 +96,13 @@ class PuppeteerBinProvider(BaseBinProvider):
return self._browser_abspaths[bin_name]
# first time loading, find browser in self.puppeteer_browsers_dir by searching filesystem for installed binaries
browsers_present = [d.name for d in self.puppeteer_browsers_dir.glob("*")]
if bin_name in browsers_present:
candidates = []
# if on macOS, browser binary is inside a .app, otherwise it's just a plain binary
if platform.system().lower() == 'darwin':
# /data/lib/browsers/chrome/mac_arm-129.0.6668.58/chrome-mac-arm64/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing
candidates = sorted(self.puppeteer_browsers_dir.glob(f'/{bin_name}/mac*/chrome*/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing'))
else:
# /data/lib/browsers/chrome/linux-131.0.6730.0/chrome-linux64/chrome
candidates = sorted(self.puppeteer_browsers_dir.glob(f'/{bin_name}/linux*/chrome*/chrome'))
if candidates:
self._browser_abspaths[bin_name] = candidates[-1]
return self._browser_abspaths[bin_name]
matching_bins = [abspath for abspath in self.installed_browser_bins() if bin_name in str(abspath)]
if matching_bins:
newest_bin = matching_bins[-1] # already sorted alphabetically, last should theoretically be highest version number
self._browser_abspaths[bin_name] = newest_bin
return self._browser_abspaths[bin_name]
return super().on_get_abspath(bin_name, **context)
return None
def on_install(self, bin_name: str, packages: Optional[InstallArgs] = None, **context) -> str:
"""npx @puppeteer/browsers install chrome@stable"""
@ -119,64 +137,6 @@ class PuppeteerBinProvider(BaseBinProvider):
PUPPETEER_BINPROVIDER = PuppeteerBinProvider()
CHROMIUM_BINARY_NAMES = [
'chromium',
'chromium-browser',
'chromium-browser-beta',
'chromium-browser-unstable',
'chromium-browser-canary',
'chromium-browser-dev'
'/Applications/Chromium.app/Contents/MacOS/Chromium',
]
CHROME_BINARY_NAMES = [
'google-chrome',
'google-chrome-stable',
'google-chrome-beta',
'google-chrome-canary',
'google-chrome-unstable',
'google-chrome-dev',
# 'chrome',
'/Applications/Google Chrome.app/Contents/MacOS/Google Chrome',
'/Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary',
]
def autodetect_system_chrome_install(PATH=None):
for bin_name in CHROME_BINARY_NAMES + CHROMIUM_BINARY_NAMES:
abspath = bin_abspath(bin_name, PATH=env.PATH)
if abspath:
return abspath
return None
class ChromeBinary(BaseBinary):
name: BinName = 'chrome'
binproviders_supported: List[InstanceOf[BinProvider]] = [PUPPETEER_BINPROVIDER, env]
provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
env.name: {
'abspath': lambda:
autodetect_system_chrome_install(PATH=env.PATH),
},
PUPPETEER_BINPROVIDER.name: {
'packages': lambda:
['chrome@stable'],
}
}
@staticmethod
def symlink_to_lib(binary, bin_dir=settings.CONFIG.BIN_DIR) -> None:
if not (binary.abspath and binary.abspath.exists()):
return
bin_dir.mkdir(parents=True, exist_ok=True)
symlink = bin_dir / binary.name
if platform.system().lower() == 'darwin':
# if on macOS, browser binary is inside a .app, so we need to create a tiny bash script instead of a symlink
symlink.write_text(f"""#!/usr/bin/env bash\nexec '{binary.abspath}' "$@"\n""")
symlink.chmod(0o777) # make sure its executable by everyone
else:
# otherwise on linux we can symlink directly to binary executable
symlink.symlink_to(binary.abspath)
# ALTERNATIVE INSTALL METHOD using Ansible:
# install_playbook = self.plugin_dir / 'install_puppeteer.yml'
@ -192,18 +152,14 @@ class ChromeBinary(BaseBinary):
# )
CHROME_BINARY = ChromeBinary()
PLUGIN_BINARIES = [CHROME_BINARY]
class PuppeteerPlugin(BasePlugin):
app_label: str ='puppeteer'
verbose_name: str = 'SingleFile'
verbose_name: str = 'Puppeteer & Playwright'
hooks: List[InstanceOf[BaseHook]] = [
PUPPETEER_CONFIG,
PUPPETEER_BINPROVIDER,
CHROME_BINARY,
PUPPETEER_BINARY,
]