Selenium是一个支持多平台+多浏览器+多版本的浏览器自动化测试框。
Selenium在启动浏览器时可以通过desired_capbilities参数来指定需要启动的平台、浏览器、版本及浏览器启动参数。
ChromeOptions是chromedriver支持的浏览器启动选项。
这是我用到的一些配置
python
options = webdriver.ChromeOptions()
options.add_argument('--headless')
options.add_argument("--remote-debugging-port=9222")
options.add_argument("--no-sandbox")
options.add_argument("--disable-setuid-sandbox")
options.add_argument("--disable-dev-shm-using")
options.add_argument("--disable-extensions")
options.add_argument("--disable-gpu")
options.add_argument("--start-maximized")
options.add_argument("--disable-infobars")
driver = webdriver.Chrome(options=options)
driver.get(url)常用配置
python
# 无界面模式
options.add_argument('--headless')
# 指定用户客户端-模拟手机浏览
options.add_argument('user-agent="MQQBrowser/26 Mozilla/5.0 (Linux; U; Android 2.3.7; zh-cn; MB200 Build/GRJ22; CyanogenMod-7) AppleWebKit/533.1 (KHTML, like Gecko) Version/4.0 Mobile Safari/533.1"')
# 禁用图片加载
options.add_argument('--blink-settings=imagesEnabled=false')
# 隐身模式
options.add_argument('--incognito')
# 自动打开开发者工具
options.add_argument("--auto-open-devtools-for-tabs")
# 设置窗口尺寸，注意宽高之间使用逗号而不是x
options.add_argument('--window-size=300,600')
# 设置窗口启动位置（左上角坐标）
options.add_argument('--window-position=120,0')
# 禁用gpu渲染
options.add_argument('--disable-gpu')
# 全屏启动
options.add_argument('--start-fullscreen')
# 全屏启动，无地址栏
options.add_argument('--kiosk')
# 启动时，不激活（前置）窗口
options.add_argument('--no-startup-window')
...下面是官方提供的全部配置
python
全部:(下面的没有进行翻译,仔细看下应该都看得懂. ~ ~)
--disable-hang-monitor
Suppresses hang monitor dialogs in renderer processes.
--disable-metrics
Completely disables UMA metrics system.
--disable-metrics-reporting
Disables only the sending of metrics reports. In contrast to kDisableMetrics, this executes all the code that a normal client would use for reporting, except the report is dropped rather than sent to the server. This is useful for finding issues in the metrics code during UI and performance tests.
--assert-test
Causes the browser process to throw an assertion on startup.
--renderer-assert-test
Causes the renderer process to throw an assertion on launch.
--crash-test
Performs a crash test when the browser is starte.
--renderer-crash-test
Causes the renderer process to crash on launch.
--renderer-startup-dialog
Use this argument when you want to see the child processes as soon as Chrome start.
--plugin-startup-dialog
Causes the plugin process to display a dialog on launch.
--testshell-startup-dialog
Causes the test shell process to display a dialog on launch.
--plugin-launcher
Specifies a command that should be used to launch the plugin process. Useful
for running the plugin process through purify or quantify. Ex:
--plugin-launcher="path\to\purify /Run=yes.
--plugin-launche.
--channel
The value of this switch tells the child process which
IPC channel the browser expects to use to communicate with it.
--testing-channel
The value of this switch tells the app to listen for and broadcast
testing-related messages on IPC channel with the given ID.
--homepage
The value of this switch specifies which page will be displayed
in newly-opened tabs. We need this for testing purposes so
that the UI tests don't depend on what comes up for http://google.com.
--start-renderers-manually
When this switch is present, the browser will throw up a dialog box
asking the user to start a renderer process independently rather
than launching the renderer itself. (This is useful for debugging..
--renderer
Causes the process to run as renderer instead of as browser.
--renderer-path
Path to the executable to run for the renderer subproces.
--plugin
Causes the process to run as plugin hos.
--single-process
Runs the renderer and plugins in the same process as the browse.
--process-per-tab
Runs each set of script-connected tabs (i.e., a BrowsingInstance) in its own
renderer process. We default to using a renderer process for each
site instance (i.e., group of pages from the same registered domain with
script connections to each other).
--process-per-site
Runs a single process for each site (i.e., group of pages from the same
registered domain) the user visits. We default to using a renderer process
for each site instance (i.e., group of pages from the same registered
domain with script connections to each other).
--in-process-plugins
Runs plugins inside the renderer proces.
--no-sandbox
Runs the renderer outside the sandbox.
--safe-plugins
Runs the plugin processes inside the sandbox.
--trusted-plugins
Excludes these plugins from the plugin sandbox.
This is a comma separated list of plugin dlls name and activex clsid.
--test-sandbox
Runs the security test for the sandbox.
--user-data-dir
Specifies the user data directory, which is where the browser will look
for all of its state.
--app
Specifies that the associated value should be launched in "application" mode.
--upload-file
Specifies the file that should be uploaded to the provided application. This
switch is expected to be used with --app option.
--dom-automation
Specifies if the dom_automation_controller_ needs to be bound in the
renderer. This binding happens on per-frame basis and hence can potentially
be a performance bottleneck. One should only enable it when automating
dom based tests.
--plugin-path
Tells the plugin process the path of the plugin to loa.
--js-flags
Specifies the flags passed to JS engin.
--geoid
The GeoID we should use. This is normally obtained from the operating system
during first run and cached in the preferences afterwards. This is a numeric
value; see http://msdn.microsoft.com/en-us/library/ms776390.aspx .
--lang
The language file that we want to try to open. Of the form
language[-country] where language is the 2 letter code from ISO-639.
--debug-children
Will add kDebugOnStart to every child processes. If a value is passed, it
will be used as a filter to determine if the child process should have the
kDebugOnStart flag passed on or not.
--debug-on-start
Causes the process to start the JIT debugger on itself (mainly used by --debug-children.
--wait-for-debugger-children
Will add kWaitForDebugger to every child processes. If a value is passed, it
will be used as a filter to determine if the child process should have the
kWaitForDebugger flag passed on or not.
--wait-for-debugger
Waits for a debugger for 60 second.
--log-filter-prefix
Will filter log messages to show only the messages that are prefixed
with the specified valu.
--enable-logging
Force logging to be enabled. Logging is disabled by default in release
builds.
--dump-histograms-on-exit
Dump any accumualted histograms to the log when browser terminates (requires
logging to be enabled to really do anything). Used by developers and test
scripts.
--disable-logging
Force logging to be disabled. Logging is enabled by default in debug
builds.
--log-level
Sets the minimum log level. Valid values are from 0 to 3:
INFO = 0, WARNING = 1, LOG_ERROR = 2, LOG_FATAL = 3.
--remote-shell-port
Enable remote debug / automation shell on the specified por.
--uninstall
Runs un-installation steps that were done by chrome first-run.
--omnibox-popup-count
Number of entries to show in the omnibox popup.
--uninstallomnibox-popup-count
Removes the previous set suggestion coun.
--automation-channel
The value of this switch tells the app to listen for and broadcast
automation-related messages on IPC channel with the given ID.
--restore-last-session
Indicates the last session should be restored on startup. This overrides
the preferences value and is primarily intended for testing.
--record-mode
--playback-mode
Chrome supports a playback and record mode. Record mode saves *everything*
to the cache. Playback mode reads data exclusively from the cache. This
allows us to record a session into the cache and then replay it at will.
--no-events
Don't record/playback events when using record & playback.
--hide-icons
--show-icons
Make Windows happy by allowing it to show "Enable access to this program"
checkbox in Add/Remove Programs->Set Program Access and Defaults. This
only shows an error box because the only way to hide Chrome is by
uninstalling it.
--make-default-browser
Make Chrome default browse.
--proxy-server
Use a specified proxy server, overrides system settings. This switch only
affects HTTP and HTTPS requests.
--dns-log-details
--dns-prefetch-disable
Chrome will support prefetching of DNS information. Until this becomes
the default, we'll provide a command line switch.
--debug-print
Enables support to debug printing subsystem.
--allow-all-activex
Allow initialization of all activex controls. This is only to help website
developers test their controls to see if they are compatible in Chrome.
Note there's a duplicate value in activex_shared.cc (to avoid
dependency on chrome module). Please change both locations at the same time.
--disable-dev-tools
Browser flag to disable the web inspector for all renderers.
--always-enable-dev-tools
Enable web inspector for all windows, even if they're part of the browser.
Allows us to use our dev tools to debug browser windows itself.
--memory-model
Configure Chrome's memory model.
Does chrome really need multiple memory models? No. But we get a lot
of concerns from individuals about how the changes work on *their*
system, and we need to be able to experiment with a few choices.
--tab-count-to-load-on-session-restore
Used to set the value of SessionRestore::num_tabs_to_load_. See session_restore.h for details.
const wchar_t kTabCountToLoadOnSessionRestore[] .
--memory-profile
Enable dynamic loading of the Memory Profiler DLL, which will trace
all memory allocations during the run.
--enable-file-cookies
By default, cookies are not allowed on file://. They are needed in for
testing, for example page cycler and layout tests. See bug 1157243.
--start-maximized
Start the browser maximized, regardless of any previous settings.
TODO(pjohnson): Remove this once bug 1137420 is fixed. We are using this
as a workaround for not being able to use moveTo and resizeTo on a
top-level window.
--enable-watchdog
Spawn threads to watch for excessive delays in specified message loops.
User should set breakpoints on Alarm() to examine problematic thread.
Usage: -enable-watchdog=[ui][io]
Order of the listed sub-arguments does not matter.
--first-run
Display the First Run experience when the browser is started, regardless of
whether or not it's actually the first run.
--message-loop-strategy
--message-loop-histogrammer
Enable histograming of tasks served by MessageLoop. See about:histograms/Loop
for results, which show frequency of messages on each thread, including APC
count, object signalling count, etc.
--import
Perform importing from another browser. The value associated with this
setting encodes the target browser and what items to import.
--silent-dump-on-dcheck
Change the DCHECKS to dump memory and continue instead of crashing.
This is valid only in Release mode when --enable-dcheck is specified.
--disable-prompt-on-repost
Normally when the user attempts to navigate to a page that was the result of
a post we prompt to make sure they want to. This switch may be used to
disable that check. This switch is used during automated testing.
--disable-popup-blocking
Disable pop-up blocking.
--disable-javascript
Don't execute JavaScript (browser JS like the new tab page still runs).
--disable-java
Prevent Java from running.
--disable-plugins
Prevent plugins from running.
--disable-images
Prevent images from loading.
--use-lf-heap
Use the low fragmentation heap for the CRT.
--gears-plugin-path
Debug only switch to specify which gears plugin dll to load.
--gears-in-renderer
Switch to load Gears in the renderer process.
--enable-p13n
--javascript-debugger-path
Allow loading of the javascript debugger UI from the filesystem.
--new-http
Enable new HTTP stack.
全部