* Refactored GetCommand, BrowseCommand to have execute method

* Fixed type name format issues in __issue_command

* Fixed everything I broke

* Changed import style so tests can run

* Added BrowseCommad to imports

* Added some more self

* Added logging to explain failing test

* Added one more self

* attempt at refactoring save_screenshot

* fixed indentation, attempt at refactoring save_screenshot

* refactored SaveScreenshot command to have execute method

* reformatted code using black

* Ported SaveScreenshotCommand

It now uses the new command.execute(...) syntax

* refactored savefullscreenshot command to follow command sequence

* formatted files with black

* removed extraneous commands

* Ported SaveScreenshotFullPage #763

* refactored dump page source and formatted code with black

* reformatted recursive dump page source command and formatted code w black

* formatted files using isort

* formatted all files with isort

* Ported DumpPageSource and RecursiveDumpPageSource (#767)

* refactor finalize command

* refactored initalize command and formatted with black and isort

* missed a conflict

* Command refactoring (#770)

* attempt at refactoring save_screenshot

* fixed indentation, attempt at refactoring save_screenshot

* refactored SaveScreenshot command to have execute method

* reformatted code using black

* refactored savefullscreenshot command to follow command sequence

* formatted files with black

* removed extraneous commands

* refactored dump page source and formatted code with black

* reformatted recursive dump page source command and formatted code w black

* formatted files using isort

* formatted all files with isort

* refactor finalize command

* refactored initalize command and formatted with black and isort

* missed a conflict

* Ran isort

* Added append_command

* remove custom function command and format code

* Refactored GetCommand, BrowseCommand to have execute method

* Fixed type name format issues in __issue_command

* Fixed everything I broke

* Changed import style so tests can run

* Added BrowseCommad to imports

* Added some more self

* Added logging to explain failing test

* Added one more self

* Ported SaveScreenshotCommand

It now uses the new command.execute(...) syntax

* Ported SaveScreenshotFullPage #763

* Ported DumpPageSource and RecursiveDumpPageSource (#767)

* Command refactoring (#770)

* attempt at refactoring save_screenshot

* fixed indentation, attempt at refactoring save_screenshot

* refactored SaveScreenshot command to have execute method

* reformatted code using black

* refactored savefullscreenshot command to follow command sequence

* formatted files with black

* removed extraneous commands

* refactored dump page source and formatted code with black

* reformatted recursive dump page source command and formatted code w black

* formatted files using isort

* formatted all files with isort

* refactor finalize command

* refactored initalize command and formatted with black and isort

* missed a conflict

* Ran isort

* Added append_command

* remove duplicate append_command

* Refactored GetCommand, BrowseCommand to have execute method

* Fixed type name format issues in __issue_command

* Fixed everything I broke

* Changed import style so tests can run

* Added BrowseCommad to imports

* Added some more self

* Added logging to explain failing test

* Added one more self

* Ported SaveScreenshotCommand

It now uses the new command.execute(...) syntax

* Ported SaveScreenshotFullPage #763

* Ported DumpPageSource and RecursiveDumpPageSource (#767)

* Command refactoring (#770)

* attempt at refactoring save_screenshot

* fixed indentation, attempt at refactoring save_screenshot

* refactored SaveScreenshot command to have execute method

* reformatted code using black

* refactored savefullscreenshot command to follow command sequence

* formatted files with black

* removed extraneous commands

* refactored dump page source and formatted code with black

* reformatted recursive dump page source command and formatted code w black

* formatted files using isort

* formatted all files with isort

* refactor finalize command

* refactored initalize command and formatted with black and isort

* missed a conflict

* Ran isort

* Added append_command

* generate new xpi

* Fixing tests

* Fixing tests

* Fixing up more tests

* Removed type annotations

* Fixing tests

* Fixing tests

* Removed command_executor

* Moved Commands to commands

* Fixing imports

* Fixed skipped test

* Removed duplicate append_command

* docs: update adding command in usingOpenWPM

* Forgot to save

* Removed datadir

* Cleaning up imports

* Implemented simple command

* Added documentation to simple_command.py

* Renamed to custom_command.py

* Moved docs around

* Referencing BaseCommand.execute

* Update docs/Using_OpenWPM.md

Co-authored-by: Steven Englehardt <senglehardt@mozilla.com>

Co-authored-by: Cyrus <cyruskarsan@gmail.com>
Co-authored-by: cyruskarsan <55566678+cyruskarsan@users.noreply.github.com>
Co-authored-by: Steven Englehardt <senglehardt@mozilla.com>
This commit is contained in:
Stefan Zabka 2021-01-09 11:15:01 +01:00 коммит произвёл GitHub
Родитель fc7c772986
Коммит e338bb29f0
Не найден ключ, соответствующий данной подписи
Идентификатор ключа GPG: 4AEE18F83AFDEB23
20 изменённых файлов: 679 добавлений и 693 удалений

Просмотреть файл

@ -13,7 +13,7 @@ import sentry_sdk
from openwpm.command_sequence import CommandSequence
from openwpm.config import BrowserParams, ManagerParams
from openwpm.mp_logger import parse_config_from_env
from openwpm.task_manager import TaskManager, load_default_params
from openwpm.task_manager import TaskManager
from openwpm.utilities import rediswq
from test.utilities import LocalS3Session, local_s3_bucket

43
custom_command.py Normal file
Просмотреть файл

@ -0,0 +1,43 @@
""" This file aims to demonstrate how to write custom commands in OpenWPM
Steps to have a custom command run as part of a CommandSequence
1. Create a class that derives from BaseCommand
2. Implement the execute method
3. Append it to the CommandSequence
4. Execute the CommandSequence
"""
import logging
from selenium.webdriver import Firefox
from selenium.webdriver.common.by import By
from openwpm.commands.types import BaseCommand
from openwpm.config import BrowserParams, ManagerParams
from openwpm.socket_interface import ClientSocket
class LinkCountingCommand(BaseCommand):
"""This command logs how many links it found on any given page"""
def __init__(self) -> None:
self.logger = logging.getLogger("openwpm")
# While this is not strictly necessary, we use the repr of a command for logging
# So not having a proper repr will make your logs a lot less useful
def __repr__(self) -> str:
return "LinkCountingCommand"
# Have a look at openwpm.commands.types.BaseCommand.execute to see
# an explanation of each parameter
def execute(
self,
webdriver: Firefox,
browser_params: BrowserParams,
manager_params: ManagerParams,
extension_socket: ClientSocket,
) -> None:
current_url = webdriver.current_url
link_count = len(webdriver.find_elements(By.TAG_NAME, "a"))
self.logger.info("There are %d links on %s", link_count, current_url)

Просмотреть файл

@ -1,4 +1,6 @@
from custom_command import LinkCountingCommand
from openwpm.command_sequence import CommandSequence
from openwpm.commands.browser_commands import GetCommand
from openwpm.config import BrowserParams, ManagerParams
from openwpm.task_manager import TaskManager
@ -10,9 +12,9 @@ sites = [
"http://citp.princeton.edu/",
]
# Loads the default ManagerParams
# and NUM_BROWSERS copies of the default BrowserParams
manager_params = ManagerParams(
num_browsers=NUM_BROWSERS
) # num_browsers is necessary to let TaskManager know how many browsers to spawn
@ -58,7 +60,9 @@ for site in sites:
)
# Start by visiting the page
command_sequence.get(sleep=3, timeout=60)
command_sequence.append_command(GetCommand(url=site, sleep=3), timeout=60)
# Have a look at custom_command.py to see how to implement your own command
command_sequence.append_command(LinkCountingCommand())
# Run commands across the three browsers (simple parallelization)
manager.execute_command_sequence(command_sequence)

Просмотреть файл

@ -6,7 +6,7 @@ In this section, we present three basic development demos for working on the Ope
Have a look at [demo.py](../demo.py)
Generally, measurement crawls should be able to be run using scripts with lengths on the order of 100 lines of code.
Even within this short script, there are several different options that a user can change.
Even within this short script, there are several options that a user can change.
Users can change the settings for task manager and individual browsers so that, for instance, certain browsers can run headless while others do not. We provide a method to read the default configuration settings into a classes that can be passed to the `TaskManager` instance. Note that browser configuration is **per-browser**, so this command will return a list of class<BrowserParams>.
@ -18,97 +18,28 @@ browser_params = [BrowserParams() for _ in range(manager_params.num_browsers)]
```
#### Loading Custom Browser or Manager configs
Users can load custom Browser and Platform/Manager configuration by writing them into a json file and then loading then into respective dataclasses. For example:
Users can load custom Browser and Platform/Manager configuration by writing them into a json file and then loading them into respective dataclasses. For example:
```py
from openwpm.config import BrowserParams, ManagerParams
with open('<custom_manager_params>.json', 'r') as f:
with open("<custom_manager_params>.json", 'r') as f:
manager_params = ManagerParams.from_json(f.read())
browser_params = list()
for _ in range(num_browsers):
with open('<custom_browser_params>.json', 'r') as file:
with open("<custom_browser_params>.json", 'r') as file:
browser_params.append(BrowserParams.from_json(file.read()))
```
## Adding a new command
## Defining a new command
OpenWPM commands exist as part of a command sequence object, which allows one to string together a sequence of actions for a browser to take and deploy that sequence to the first available browser from the manager. Adding a command to a `CommandSequence` object will cause the browser to execute it immediately after the previously added command as long as the previous command does not time out or fail.
Suppose we want to add a top-level command to cause the browser to jiggle the mouse a few times. We may want to have the browser visit a site, jiggle the mouse, and extract the links from the site.
To add a new command you need to modify the following four files:
1. Define all required paramters in a type in `openwpm/commands/types.py`
In our case this looks like this:
```python
class JiggleCommand(BaseCommand):
def __init__(self, num_jiggles):
self.num_jiggles = num_jiggles
def __repr__(self):
return "JiggleCommand({})".format(self.num_jiggles)
```
2. Define the behaviour of our new command in `*_commands.py` in `openwpm/commands/`,
e.g. `browser_commands.py`.
Feel free to add a new module within `openwpm/commands/` for your own custom commands
In our case this looks like this:
```python
from selenium.webdriver.common.action_chains import ActionChains
def jiggle_mouse(webdriver, number_jiggles):
for i in xrange(0, number_jiggles):
x = random.randrange(0, 500)
y = random.randrange(0, 500)
action = ActionChains(webdriver)
action.move_by_offset(x, y)
action.perform()
```
3. Make our function be called when the command_sequence reaches our Command, by adding it to the
`execute_command` function in `openwpm/commands/command_executer.py`
In our case this looks like this:
```python
elif type(command) is JiggleCommand:
browser_commands.jiggle_mouse(
webdriver=webdriver,
number_jiggles=self.num_jiggles)
```
4. Lastly we change ```openwpm/CommandSequence.py``` by adding a `jiggle_mouse` method to the `CommandSequence`
so we can add our command to the commands list
In our case this looks like this:
```python
def jiggle_mouse(self, num_jiggles, timeout=60):
""" jiggles mouse <num_jiggles> times """
self.total_timeout += timeout
if not self.contains_get_or_browse:
raise CommandExecutionError("No get or browse request preceding "
"the jiggle_mouse command", self)
command = JiggleCommand(num_jiggles)
self.commands_with_timeout.append((command, timeout))
```
A timeout is given and set by default to 60 seconds. This is added to the overall sequence timeout. Finally, we check that the `CommandSequence` instance contains a `get` or a `browse` command prior to this command being added by checking `self.contains_get_or_browse`. This is necessary as it wouldn't make sense to have selenium jiggle the mouse before loading a page.
Please have a look at [`custom_command.py`](../custom_command.py). Note that custom commands must be
defined in a separate module and imported. They can't be defined within the main crawl script.
See [#837](https://github.com/mozilla/OpenWPM/issues/837).
Notice that any arguments to the command are added both to the command sequence top-level method, and are then stored in the `Command` object to be serialized and sent across the process boundary between the task manager and browser manager.
Finally, the command sequence given to the Task Manager to visit a site, sleep for 10 seconds, jiggle the mouse 10 times, and takes a screenshot would look like this:
```python
site = 'http://www.example.com'
command_sequence = CommandSequence.CommandSequence(site)
command_sequence.get(sleep=10)
command_sequence.jiggle_mouse(10)
command_sequence.screenshot_full_page()
manager.execute_command_sequence(command_sequence)
```
## Running a simple analysis
Suppose that we ran the platform over some set of sites while logged into several sites while using a particular email. During the crawl, we turned on the proxy option to log HTTP traffic. One possible threat is, perhaps due to sloppy coding, the first-party leaks the user's email as plaintext over HTTP traffic. Given an OpenWPM database, the following script logs the first-party sites on which such a leakage occurs.
@ -133,7 +64,7 @@ for url, top_url in cur.execute("SELECT DISTINCT h.url, v.site_url "
fp_sites.add(top_url)
# outputs the results
print list(fp_sites)
print(list(fp_sites))
````
The variety of data stored in OpenWPM databases (with all instrumentation enabled) allows the above script to easily be expanded into a larger study. For instance, one step would be to see which parties are the recipients of the email address. Do these recipients later place cookies containing the email? Besides the site on which the original email leak was made, on which other first parties do these recipients appear as a third party? All of these questions are answerable through OpenWPM database instances.

Просмотреть файл

@ -5,7 +5,7 @@ dependencies:
- beautifulsoup4=4.9.3
- black=20.8b1
- click=7.1.2
- codecov=2.1.10
- codecov=2.1.11
- dill=0.3.3
- geckodriver=0.28.0
- ipython=7.19.0
@ -13,20 +13,20 @@ dependencies:
- localstack=0.11.1.1
- multiprocess=0.70.11.1
- nodejs=14.15.1
- pandas=1.1.4
- pandas=1.1.5
- pillow=8.0.1
- pip=20.2.4
- pre-commit=2.9.2
- psutil=5.7.3
- pip=20.3.3
- pre-commit=2.9.3
- psutil=5.8.0
- pyarrow=2.0.0
- pytest-cov=2.10.1
- pytest=6.1.2
- pytest=6.2.1
- python=3.8.6
- pyvirtualdisplay=0.2.5
- redis-py=3.5.3
- s3fs=0.4.0
- selenium=3.141.0
- sentry-sdk=0.19.4
- sentry-sdk=0.19.5
- tabulate=0.8.7
- tblib=1.6.0
- wget=1.20.1

Просмотреть файл

@ -9,15 +9,14 @@ import threading
import time
import traceback
from queue import Empty as EmptyQueue
from typing import Optional
from typing import Optional, Union
import psutil
from multiprocess import Queue
from selenium.common.exceptions import WebDriverException
from tblib import pickling_support
from .commands import command_executor
from .commands.types import ShutdownCommand
from .commands.types import BaseCommand, ShutdownSignal
from .deploy_browsers import deploy_firefox
from .errors import BrowserConfigError, BrowserCrashError, ProfileLoadError
from .socket_interface import ClientSocket
@ -286,7 +285,7 @@ class Browser:
return
# Send the shutdown command
command = ShutdownCommand()
command = ShutdownSignal()
self.command_queue.put((command))
# Verify that webdriver has closed (30 second timeout)
@ -487,9 +486,9 @@ def BrowserManager(
time.sleep(0.001)
continue
command = command_queue.get()
command: Union[ShutdownSignal, BaseCommand] = command_queue.get()
if type(command) is ShutdownCommand:
if type(command) is ShutdownSignal:
# Geckodriver creates a copy of the profile (and the original
# temp file created by FirefoxProfile() is deleted).
# We clear the profile attribute here to prevent prints from:
@ -509,8 +508,7 @@ def BrowserManager(
# if command fails for whatever reason, tell the TaskManager to
# kill and restart its worker processes
try:
command_executor.execute_command(
command,
command.execute(
driver,
browser_params,
manager_params,

Просмотреть файл

@ -1,18 +1,16 @@
from typing import Callable, List, Tuple
from .commands.types import (
BaseCommand,
from .commands.browser_commands import (
BrowseCommand,
DumpPageSourceCommand,
DumpProfCommand,
FinalizeCommand,
GetCommand,
InitializeCommand,
RecursiveDumpPageSourceCommand,
RunCustomFunctionCommand,
SaveScreenshotCommand,
ScreenshotFullPageCommand,
)
from .commands.types import BaseCommand
from .errors import CommandExecutionError
@ -39,7 +37,7 @@ class CommandSequence:
retry_number: int = None,
site_rank: int = None,
callback: Callable[[bool], None] = None,
):
) -> None:
"""Initialize command sequence.
Parameters
@ -179,15 +177,7 @@ class CommandSequence:
command = RecursiveDumpPageSourceCommand(suffix)
self._commands_with_timeout.append((command, timeout))
def run_custom_function(self, function_handle, func_args=(), timeout=30):
"""Run a custom by passing the function handle"""
self.total_timeout += timeout
if not self.contains_get_or_browse:
raise CommandExecutionError(
"No get or browse request preceding " "the dump page source command",
self,
)
command = RunCustomFunctionCommand(function_handle, func_args)
def append_command(self, command: BaseCommand, timeout: int = 30) -> None:
self._commands_with_timeout.append((command, timeout))
def mark_done(self, success: bool):

Просмотреть файл

@ -16,11 +16,10 @@ from selenium.common.exceptions import (
WebDriverException,
)
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.remote.webdriver import WebDriver
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.ui import WebDriverWait
from ..socket_interface import ClientSocket
from .types import BaseCommand
from .utils.webdriver_utils import (
execute_in_all_frames,
execute_script_with_retry,
@ -31,6 +30,7 @@ from .utils.webdriver_utils import (
)
# Constants for bot mitigation
NUM_MOUSE_MOVES = 10 # Times to randomly move the mouse
RANDOM_SLEEP_LOW = 1 # low (in sec) for random sleep between page loads
RANDOM_SLEEP_HIGH = 7 # high (in sec) for random sleep between page loads
@ -111,93 +111,141 @@ def tab_restart_browser(webdriver):
webdriver.switch_to_window(webdriver.window_handles[0])
def get_website(
url, sleep, visit_id, webdriver, browser_params, extension_socket: ClientSocket
):
class GetCommand(BaseCommand):
"""
goes to <url> using the given <webdriver> instance
"""
tab_restart_browser(webdriver)
def __init__(self, url, sleep):
self.url = url
self.sleep = sleep
if extension_socket is not None:
extension_socket.send(visit_id)
def __repr__(self):
return "GetCommand({},{})".format(self.url, self.sleep)
# Execute a get through selenium
try:
webdriver.get(url)
except TimeoutException:
pass
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
tab_restart_browser(webdriver)
# Sleep after get returns
time.sleep(sleep)
# Close modal dialog if exists
try:
WebDriverWait(webdriver, 0.5).until(EC.alert_is_present())
alert = webdriver.switch_to_alert()
alert.dismiss()
time.sleep(1)
except (TimeoutException, WebDriverException):
pass
close_other_windows(webdriver)
if browser_params.bot_mitigation:
bot_mitigation(webdriver)
def browse_website(
url,
num_links,
sleep,
visit_id,
webdriver,
browser_params,
manager_params,
extension_socket,
):
"""Calls get_website before visiting <num_links> present on the page.
Note: the site_url in the site_visits table for the links visited will
be the site_url of the original page and NOT the url of the links visited.
"""
# First get the site
get_website(url, sleep, visit_id, webdriver, browser_params, extension_socket)
# Then visit a few subpages
for _ in range(num_links):
links = [x for x in get_intra_links(webdriver, url) if is_displayed(x) is True]
if not links:
break
r = int(random.random() * len(links))
logger.info(
"BROWSER %i: visiting internal link %s"
% (browser_params.browser_id, links[r].get_attribute("href"))
)
if extension_socket is not None:
extension_socket.send(self.visit_id)
# Execute a get through selenium
try:
links[r].click()
wait_until_loaded(webdriver, 300)
time.sleep(max(1, sleep))
if browser_params.bot_mitigation:
bot_mitigation(webdriver)
webdriver.back()
wait_until_loaded(webdriver, 300)
except Exception:
webdriver.get(self.url)
except TimeoutException:
pass
# Sleep after get returns
time.sleep(self.sleep)
def save_screenshot(visit_id, browser_id, driver, manager_params, suffix=""):
""" Save a screenshot of the current viewport"""
if suffix != "":
suffix = "-" + suffix
# Close modal dialog if exists
try:
WebDriverWait(webdriver, 0.5).until(EC.alert_is_present())
alert = webdriver.switch_to_alert()
alert.dismiss()
time.sleep(1)
except (TimeoutException, WebDriverException):
pass
urlhash = md5(driver.current_url.encode("utf-8")).hexdigest()
outname = os.path.join(
manager_params.screenshot_path, "%i-%s%s.png" % (visit_id, urlhash, suffix)
)
driver.save_screenshot(outname)
close_other_windows(webdriver)
if browser_params.bot_mitigation:
bot_mitigation(webdriver)
class BrowseCommand(BaseCommand):
def __init__(self, url, num_links, sleep):
self.url = url
self.num_links = num_links
self.sleep = sleep
def __repr__(self):
return "BrowseCommand({},{},{})".format(self.url, self.num_links, self.sleep)
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
"""Calls get_website before visiting <num_links> present on the page.
Note: the site_url in the site_visits table for the links visited will
be the site_url of the original page and NOT the url of the links visited.
"""
# First get the site
get_command = GetCommand(self.url, self.sleep)
get_command.set_visit_browser_id(self.visit_id, self.browser_id)
get_command.execute(
webdriver,
browser_params,
manager_params,
extension_socket,
)
# Then visit a few subpages
for _ in range(self.num_links):
links = [
x
for x in get_intra_links(webdriver, self.url)
if is_displayed(x) is True
]
if not links:
break
r = int(random.random() * len(links))
logger.info(
"BROWSER %i: visiting internal link %s"
% (browser_params.browser_id, links[r].get_attribute("href"))
)
try:
links[r].click()
wait_until_loaded(webdriver, 300)
time.sleep(max(1, self.sleep))
if browser_params.bot_mitigation:
bot_mitigation(webdriver)
webdriver.back()
wait_until_loaded(webdriver, 300)
except Exception as e:
logger.error(
"BROWSER %i: Error visitit internal link %s",
browser_params.browser_id,
links[r].get_attribute("href"),
exc_info=e,
)
pass
class SaveScreenshotCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "SaveScreenshotCommand({})".format(self.suffix)
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
if self.suffix != "":
self.suffix = "-" + self.suffix
urlhash = md5(webdriver.current_url.encode("utf-8")).hexdigest()
outname = os.path.join(
manager_params.screenshot_path,
"%i-%s%s.png" % (self.visit_id, urlhash, self.suffix),
)
webdriver.save_screenshot(outname)
def _stitch_screenshot_parts(visit_id, browser_id, manager_params):
@ -262,127 +310,209 @@ def _stitch_screenshot_parts(visit_id, browser_id, manager_params):
pass
def screenshot_full_page(visit_id, browser_id, driver, manager_params, suffix=""):
class ScreenshotFullPageCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
outdir = os.path.join(manager_params.screenshot_path, "parts")
if not os.path.isdir(outdir):
os.mkdir(outdir)
if suffix != "":
suffix = "-" + suffix
urlhash = md5(driver.current_url.encode("utf-8")).hexdigest()
outname = os.path.join(
outdir, "%i-%s%s-part-%%i-%%i.png" % (visit_id, urlhash, suffix)
)
def __repr__(self):
return "ScreenshotFullPageCommand({})".format(self.suffix)
try:
part = 0
max_height = execute_script_with_retry(
driver, "return document.body.scrollHeight;"
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
self.outdir = os.path.join(manager_params.screenshot_path, "parts")
if not os.path.isdir(self.outdir):
os.mkdir(self.outdir)
if self.suffix != "":
self.suffix = "-" + self.suffix
urlhash = md5(webdriver.current_url.encode("utf-8")).hexdigest()
outname = os.path.join(
self.outdir,
"%i-%s%s-part-%%i-%%i.png" % (self.visit_id, urlhash, self.suffix),
)
inner_height = execute_script_with_retry(driver, "return window.innerHeight;")
curr_scrollY = execute_script_with_retry(driver, "return window.scrollY;")
prev_scrollY = -1
driver.save_screenshot(outname % (part, curr_scrollY))
while (
curr_scrollY + inner_height
) < max_height and curr_scrollY != prev_scrollY:
# Scroll down to bottom of previous viewport
try:
driver.execute_script("window.scrollBy(0, window.innerHeight)")
except WebDriverException:
logger.info(
"BROWSER %i: WebDriverException while scrolling, "
"screenshot may be misaligned!" % browser_id
try:
part = 0
max_height = execute_script_with_retry(
webdriver, "return document.body.scrollHeight;"
)
inner_height = execute_script_with_retry(
webdriver, "return window.innerHeight;"
)
curr_scrollY = execute_script_with_retry(
webdriver, "return window.scrollY;"
)
prev_scrollY = -1
webdriver.save_screenshot(outname % (part, curr_scrollY))
while (
curr_scrollY + inner_height
) < max_height and curr_scrollY != prev_scrollY:
# Scroll down to bottom of previous viewport
try:
webdriver.execute_script("window.scrollBy(0, window.innerHeight)")
except WebDriverException:
logger.info(
"BROWSER %i: WebDriverException while scrolling, "
"screenshot may be misaligned!" % self.browser_id
)
pass
# Update control variables
part += 1
prev_scrollY = curr_scrollY
curr_scrollY = execute_script_with_retry(
webdriver, "return window.scrollY;"
)
pass
# Update control variables
part += 1
prev_scrollY = curr_scrollY
curr_scrollY = execute_script_with_retry(driver, "return window.scrollY;")
# Save screenshot
driver.save_screenshot(outname % (part, curr_scrollY))
except WebDriverException:
excp = traceback.format_exception(*sys.exc_info())
logger.error(
"BROWSER %i: Exception while taking full page screenshot \n %s"
% (browser_id, "".join(excp))
)
return
_stitch_screenshot_parts(visit_id, browser_id, manager_params)
def dump_page_source(visit_id, driver, manager_params, suffix=""):
if suffix != "":
suffix = "-" + suffix
outname = md5(driver.current_url.encode("utf-8")).hexdigest()
outfile = os.path.join(
manager_params.source_dump_path, "%i-%s%s.html" % (visit_id, outname, suffix)
)
with open(outfile, "wb") as f:
f.write(driver.page_source.encode("utf8"))
f.write(b"\n")
def recursive_dump_page_source(visit_id, driver, manager_params, suffix=""):
"""Dump a compressed html tree for the current page visit"""
if suffix != "":
suffix = "-" + suffix
outname = md5(driver.current_url.encode("utf-8")).hexdigest()
outfile = os.path.join(
manager_params.source_dump_path,
"%i-%s%s.json.gz" % (visit_id, outname, suffix),
)
def collect_source(driver, frame_stack, rv={}):
is_top_frame = len(frame_stack) == 1
# Gather frame information
doc_url = driver.execute_script("return window.document.URL;")
if is_top_frame:
page_source = rv
else:
page_source = dict()
page_source["doc_url"] = doc_url
source = driver.page_source
if type(source) != str:
source = str(source, "utf-8")
page_source["source"] = source
page_source["iframes"] = dict()
# Store frame info in correct area of return value
if is_top_frame:
# Save screenshot
webdriver.save_screenshot(outname % (part, curr_scrollY))
except WebDriverException:
excp = traceback.format_exception(*sys.exc_info())
logger.error(
"BROWSER %i: Exception while taking full page screenshot \n %s"
% (self.browser_id, "".join(excp))
)
return
out_dict = rv["iframes"]
for frame in frame_stack[1:-1]:
out_dict = out_dict[frame.id]["iframes"]
out_dict[frame_stack[-1].id] = page_source
page_source = dict()
execute_in_all_frames(driver, collect_source, {"rv": page_source})
with gzip.GzipFile(outfile, "wb") as f:
f.write(json.dumps(page_source).encode("utf-8"))
_stitch_screenshot_parts(self.visit_id, self.browser_id, manager_params)
def finalize(
visit_id: int, webdriver: WebDriver, extension_socket: ClientSocket, sleep: int
) -> None:
""" Informs the extension that a visit is done """
tab_restart_browser(webdriver)
# This doesn't immediately stop data saving from the current
# visit so we sleep briefly before unsetting the visit_id.
time.sleep(sleep)
msg = {"action": "Finalize", "visit_id": visit_id}
extension_socket.send(msg)
class DumpPageSourceCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "DumpPageSourceCommand({})".format(self.suffix)
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
if self.suffix != "":
self.suffix = "-" + self.suffix
outname = md5(webdriver.current_url.encode("utf-8")).hexdigest()
outfile = os.path.join(
manager_params.source_dump_path,
"%i-%s%s.html" % (self.visit_id, outname, self.suffix),
)
with open(outfile, "wb") as f:
f.write(webdriver.page_source.encode("utf8"))
f.write(b"\n")
def initialize(visit_id: int, extension_socket: ClientSocket) -> None:
msg = {"action": "Initialize", "visit_id": visit_id}
extension_socket.send(msg)
class RecursiveDumpPageSourceCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "RecursiveDumpPageSourceCommand({})".format(self.suffix)
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
"""Dump a compressed html tree for the current page visit"""
if self.suffix != "":
self.suffix = "-" + self.suffix
outname = md5(webdriver.current_url.encode("utf-8")).hexdigest()
outfile = os.path.join(
manager_params.source_dump_path,
"%i-%s%s.json.gz" % (self.visit_id, outname, self.suffix),
)
def collect_source(webdriver, frame_stack, rv={}):
is_top_frame = len(frame_stack) == 1
# Gather frame information
doc_url = webdriver.execute_script("return window.document.URL;")
if is_top_frame:
page_source = rv
else:
page_source = dict()
page_source["doc_url"] = doc_url
source = webdriver.page_source
if type(source) != str:
source = str(source, "utf-8")
page_source["source"] = source
page_source["iframes"] = dict()
# Store frame info in correct area of return value
if is_top_frame:
return
out_dict = rv["iframes"]
for frame in frame_stack[1:-1]:
out_dict = out_dict[frame.id]["iframes"]
out_dict[frame_stack[-1].id] = page_source
page_source = dict()
execute_in_all_frames(webdriver, collect_source, {"rv": page_source})
with gzip.GzipFile(outfile, "wb") as f:
f.write(json.dumps(page_source).encode("utf-8"))
class FinalizeCommand(BaseCommand):
"""This command is automatically appended to the end of a CommandSequence
It's apperance means there won't be any more commands for this
visit_id
"""
def __init__(self, sleep):
self.sleep = sleep
def __repr__(self):
return f"FinalizeCommand({self.sleep})"
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
""" Informs the extension that a visit is done """
tab_restart_browser(webdriver)
# This doesn't immediately stop data saving from the current
# visit so we sleep briefly before unsetting the visit_id.
time.sleep(self.sleep)
msg = {"action": "Finalize", "visit_id": self.visit_id}
extension_socket.send(msg)
class InitializeCommand(BaseCommand):
"""The command is automatically prepended to the beginning of a
CommandSequence
It initializes state both in the extensions as well in as the
Aggregator
"""
def __repr__(self):
return "IntitializeCommand()"
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
):
msg = {"action": "Initialize", "visit_id": self.visit_id}
extension_socket.send(msg)

Просмотреть файл

@ -1,118 +0,0 @@
from ..errors import CommandExecutionError
from . import browser_commands, profile_commands
from .types import (
BrowseCommand,
DumpPageSourceCommand,
DumpProfCommand,
FinalizeCommand,
GetCommand,
InitializeCommand,
RecursiveDumpPageSourceCommand,
RunCustomFunctionCommand,
SaveScreenshotCommand,
ScreenshotFullPageCommand,
)
def execute_command(
command,
webdriver,
browser_params,
manager_params,
extension_socket,
):
"""Executes BrowserManager commands
commands are of form (COMMAND, ARG0, ARG1, ...)
"""
if type(command) is GetCommand:
browser_commands.get_website(
url=command.url,
sleep=command.sleep,
visit_id=command.visit_id,
webdriver=webdriver,
browser_params=browser_params,
extension_socket=extension_socket,
)
elif type(command) is BrowseCommand:
browser_commands.browse_website(
url=command.url,
num_links=command.num_links,
sleep=command.sleep,
visit_id=command.visit_id,
webdriver=webdriver,
browser_params=browser_params,
manager_params=manager_params,
extension_socket=extension_socket,
)
elif type(command) is DumpProfCommand:
profile_commands.dump_profile(
browser_profile_folder=browser_params.profile_path,
manager_params=manager_params,
browser_params=browser_params,
tar_location=command.dump_folder,
close_webdriver=command.close_webdriver,
webdriver=webdriver,
compress=command.compress,
)
elif type(command) is DumpPageSourceCommand:
browser_commands.dump_page_source(
visit_id=command.visit_id,
driver=webdriver,
manager_params=manager_params,
suffix=command.suffix,
)
elif type(command) is RecursiveDumpPageSourceCommand:
browser_commands.recursive_dump_page_source(
visit_id=command.visit_id,
driver=webdriver,
manager_params=manager_params,
suffix=command.suffix,
)
elif type(command) is SaveScreenshotCommand:
browser_commands.save_screenshot(
visit_id=command.visit_id,
browser_id=command.browser_id,
driver=webdriver,
manager_params=manager_params,
suffix=command.suffix,
)
elif type(command) is ScreenshotFullPageCommand:
browser_commands.screenshot_full_page(
visit_id=command.visit_id,
browser_id=command.browser_id,
driver=webdriver,
manager_params=manager_params,
suffix=command.suffix,
)
elif type(command) is RunCustomFunctionCommand:
arg_dict = {
"command": command,
"driver": webdriver,
"browser_params": browser_params,
"manager_params": manager_params,
"extension_socket": extension_socket,
}
command.function_handle(*command.func_args, **arg_dict)
elif type(command) is FinalizeCommand:
browser_commands.finalize(
visit_id=command.visit_id,
sleep=command.sleep,
webdriver=webdriver,
extension_socket=extension_socket,
)
elif type(command) is InitializeCommand:
browser_commands.initialize(
visit_id=command.visit_id, extension_socket=extension_socket
)
else:
raise CommandExecutionError("Invalid Command", command)

Просмотреть файл

@ -1,113 +1,129 @@
import logging
import os
import pickle
import shutil
import tarfile
from selenium.webdriver import Firefox
from openwpm.config import BrowserParams, ManagerParams
from ..errors import ProfileLoadError
from ..socket_interface import ClientSocket
from .types import BaseCommand
from .utils.firefox_profile import sleep_until_sqlite_checkpoint
logger = logging.getLogger("openwpm")
def dump_profile(
browser_profile_folder,
manager_params,
browser_params,
tar_location,
close_webdriver,
webdriver=None,
compress=False,
):
class DumpProfileCommand(BaseCommand):
"""
dumps a browser profile currently stored in <browser_profile_folder> to
<tar_location> in which both folders are absolute paths.
"""
logger.debug(
"BROWSER %i: Profile dumping is currently unsupported. "
"See: https://github.com/mozilla/OpenWPM/projects/2."
% browser_params.browser_id
)
return
# ensures that folder paths end with slashes
if browser_profile_folder[-1] != "/":
browser_profile_folder = browser_profile_folder + "/"
if tar_location[-1] != "/":
tar_location = tar_location + "/"
def __init__(self, dump_folder, close_webdriver, compress):
self.dump_folder = dump_folder
self.close_webdriver = close_webdriver
self.compress = compress
if not os.path.exists(tar_location):
os.makedirs(tar_location)
if compress:
tar_name = "profile.tar.gz"
else:
tar_name = "profile.tar"
# see if this file exists first
# if it does, delete it before we try to save the current session
if os.path.isfile(tar_location + tar_name):
os.remove(tar_location + tar_name)
# if this is a dump on close, close the webdriver and wait for checkpoint
if close_webdriver:
webdriver.close()
sleep_until_sqlite_checkpoint(browser_profile_folder)
# backup and tar profile
if compress:
tar = tarfile.open(tar_location + tar_name, "w:gz", errorlevel=1)
else:
tar = tarfile.open(tar_location + tar_name, "w", errorlevel=1)
logger.debug(
"BROWSER %i: Backing up full profile from %s to %s"
% (
browser_params.browser_id,
browser_profile_folder,
tar_location + tar_name,
def __repr__(self):
return "DumpProfCommand({},{},{})".format(
self.dump_folder, self.close_webdriver, self.compress
)
)
storage_vector_files = [
"cookies.sqlite", # cookies
"cookies.sqlite-shm",
"cookies.sqlite-wal",
"places.sqlite", # history
"places.sqlite-shm",
"places.sqlite-wal",
"webappsstore.sqlite", # localStorage
"webappsstore.sqlite-shm",
"webappsstore.sqlite-wal",
]
storage_vector_dirs = [
"webapps", # related to localStorage?
"storage", # directory for IndexedDB
]
for item in storage_vector_files:
full_path = os.path.join(browser_profile_folder, item)
if (
not os.path.isfile(full_path)
and full_path[-3:] != "shm"
and full_path[-3:] != "wal"
):
logger.critical(
"BROWSER %i: %s NOT FOUND IN profile folder, skipping."
% (browser_params.browser_id, full_path)
def execute(
self,
webdriver: Firefox,
browser_params: BrowserParams,
manager_params: ManagerParams,
extension_socket: ClientSocket,
) -> None:
logger.debug(
"BROWSER %i: Profile dumping is currently unsupported. "
"See: https://github.com/mozilla/OpenWPM/projects/2."
% browser_params.browser_id
)
return
browser_profile_folder = browser_params.profile_path
# ensures that folder paths end with slashes
if browser_profile_folder[-1] != "/":
browser_profile_folder = browser_profile_folder + "/"
if tar_location[-1] != "/":
tar_location = tar_location + "/"
if not os.path.exists(tar_location):
os.makedirs(tar_location)
if compress:
tar_name = "profile.tar.gz"
else:
tar_name = "profile.tar"
# see if this file exists first
# if it does, delete it before we try to save the current session
if os.path.isfile(tar_location + tar_name):
os.remove(tar_location + tar_name)
# if this is a dump on close, close the webdriver and wait for checkpoint
if close_webdriver:
webdriver.close()
sleep_until_sqlite_checkpoint(browser_profile_folder)
# backup and tar profile
if compress:
tar = tarfile.open(tar_location + tar_name, "w:gz", errorlevel=1)
else:
tar = tarfile.open(tar_location + tar_name, "w", errorlevel=1)
logger.debug(
"BROWSER %i: Backing up full profile from %s to %s"
% (
browser_params.browser_id,
browser_profile_folder,
tar_location + tar_name,
)
elif not os.path.isfile(full_path) and (
full_path[-3:] == "shm" or full_path[-3:] == "wal"
):
continue # These are just checkpoint files
tar.add(full_path, arcname=item)
for item in storage_vector_dirs:
full_path = os.path.join(browser_profile_folder, item)
if not os.path.isdir(full_path):
logger.warning(
"BROWSER %i: %s NOT FOUND IN profile folder, skipping."
% (browser_params.browser_id, full_path)
)
continue
tar.add(full_path, arcname=item)
tar.close()
)
storage_vector_files = [
"cookies.sqlite", # cookies
"cookies.sqlite-shm",
"cookies.sqlite-wal",
"places.sqlite", # history
"places.sqlite-shm",
"places.sqlite-wal",
"webappsstore.sqlite", # localStorage
"webappsstore.sqlite-shm",
"webappsstore.sqlite-wal",
]
storage_vector_dirs = [
"webapps", # related to localStorage?
"storage", # directory for IndexedDB
]
for item in storage_vector_files:
full_path = os.path.join(browser_profile_folder, item)
if (
not os.path.isfile(full_path)
and full_path[-3:] != "shm"
and full_path[-3:] != "wal"
):
logger.critical(
"BROWSER %i: %s NOT FOUND IN profile folder, skipping."
% (browser_params.browser_id, full_path)
)
elif not os.path.isfile(full_path) and (
full_path[-3:] == "shm" or full_path[-3:] == "wal"
):
continue # These are just checkpoint files
tar.add(full_path, arcname=item)
for item in storage_vector_dirs:
full_path = os.path.join(browser_profile_folder, item)
if not os.path.isdir(full_path):
logger.warning(
"BROWSER %i: %s NOT FOUND IN profile folder, skipping."
% (browser_params.browser_id, full_path)
)
continue
tar.add(full_path, arcname=item)
tar.close()
def load_profile(browser_profile_folder, manager_params, browser_params, tar_location):

Просмотреть файл

@ -1,4 +1,20 @@
class BaseCommand:
from abc import ABC, abstractmethod
from selenium.webdriver import Firefox
from ..config import BrowserParams, ManagerParams
from ..socket_interface import ClientSocket
class BaseCommand(ABC):
"""
Base class for all Commands in OpenWPM
See `custom_command.py` for instructions on how
to implement your own and `openwpm/commands` for
all commands that are already implemented
"""
def set_visit_browser_id(self, visit_id, browser_id):
self.visit_id = visit_id
self.browser_id = browser_id
@ -6,105 +22,34 @@ class BaseCommand:
def set_start_time(self, start_time):
self.start_time = start_time
@abstractmethod
def execute(
self,
webdriver: Firefox,
browser_params: BrowserParams,
manager_params: ManagerParams,
extension_socket: ClientSocket,
) -> None:
"""
This method gets called in the Browser process
:parameter webdriver:
WebDriver is a Selenium class used to control
browser.
You can simulate arbitrary interactions and extract almost all browser state
with the tools that Selenium gives you
:parameter browser_params:
Contains the per browser configuration
E.g. which instruments are enabled
:parameter manager_params:
Per crawl parameters
E.g. where to store files
:parameter extension_socket: Communication channel to the storage provider
TODO: Further document this once the StorageProvider PR has landed
This allows you to send data to be persisted to storage.
"""
pass
class GetCommand(BaseCommand):
def __init__(self, url, sleep):
self.url = url
self.sleep = sleep
class ShutdownSignal:
def __repr__(self):
return "GetCommand({},{})".format(self.url, self.sleep)
class BrowseCommand(BaseCommand):
def __init__(self, url, num_links, sleep):
self.url = url
self.num_links = num_links
self.sleep = sleep
def __repr__(self):
return "BrowseCommand({},{},{})".format(self.url, self.num_links, self.sleep)
class DumpProfCommand(BaseCommand):
def __init__(self, dump_folder, close_webdriver, compress):
self.dump_folder = dump_folder
self.close_webdriver = close_webdriver
self.compress = compress
def __repr__(self):
return "DumpProfCommand({},{},{})".format(
self.dump_folder, self.close_webdriver, self.compress
)
class DumpPageSourceCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "DumpPageSourceCommand({})".format(self.suffix)
class RecursiveDumpPageSourceCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "RecursiveDumpPageSourceCommand({})".format(self.suffix)
class SaveScreenshotCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "SaveScreenshotCommand({})".format(self.suffix)
class ScreenshotFullPageCommand(BaseCommand):
def __init__(self, suffix):
self.suffix = suffix
def __repr__(self):
return "ScreenshotFullPageCommand({})".format(self.suffix)
class RunCustomFunctionCommand(BaseCommand):
def __init__(self, function_handle, func_args):
self.function_handle = function_handle
self.func_args = func_args
def __repr__(self):
return "RunCustomFunctionCommand({},{})".format(
self.function_handle, self.func_args
)
class ShutdownCommand(BaseCommand):
def __repr__(self):
return "ShutdownCommand()"
class FinalizeCommand(BaseCommand):
"""This command is automatically appended to the end of a CommandSequence
It's apperance means there won't be any more commands for this
visit_id
"""
def __init__(self, sleep):
self.sleep = sleep
def __repr__(self):
return f"FinalizeCommand({self.sleep})"
class InitializeCommand(BaseCommand):
"""The command is automatically prepended to the beginning of a
CommandSequence
It initializes state both in the extensions as well in as the
Aggregator
"""
def __repr__(self):
return "IntitializeCommand()"
return "ShutdownSignal()"

Просмотреть файл

@ -57,6 +57,14 @@ ALL_RESOURCE_TYPES = {
@dataclass_json
@dataclass
class BrowserParams:
"""
Configuration that might differ per browser
OpenWPM allows you to run multiple browsers with different
configurations in parallel and this class allows you
to customize behaviour of an individual browser
"""
extension_enabled: bool = True
cookie_instrument: bool = True
js_instrument: bool = False
@ -83,6 +91,15 @@ class BrowserParams:
@dataclass_json
@dataclass
class ManagerParams:
"""
Configuration for the TaskManager
The configuration will be the same for all browsers running on the same
TaskManager.
It can be used to control storage locations or which watchdogs should
run
"""
data_directory: str = "~/openwpm/"
log_directory: str = "~/openwpm/"
screenshot_path: Optional[str] = None

Просмотреть файл

@ -1,4 +1,3 @@
import copy
import json
import logging
import os
@ -418,7 +417,7 @@ class TaskManager:
self, browser: Browser, command_sequence: CommandSequence
) -> None:
"""
sends command tuple to the BrowserManager
Sends CommandSequence to the BrowserManager one command at a time
"""
browser.is_fresh = False
@ -508,7 +507,7 @@ class TaskManager:
{
"browser_id": browser.browser_id,
"visit_id": browser.curr_visit_id,
"command": type(command),
"command": type(command).__name__,
"arguments": json.dumps(
command.__dict__, default=lambda x: repr(x)
).encode("utf-8"),

Просмотреть файл

@ -1,5 +1,4 @@
import atexit
import json
import subprocess
from os.path import dirname, join, realpath

Просмотреть файл

@ -1,4 +1,9 @@
from selenium.webdriver import Firefox
from openwpm import command_sequence, task_manager
from openwpm.commands.types import BaseCommand
from openwpm.config import BrowserParams, ManagerParams
from openwpm.socket_interface import ClientSocket
from openwpm.utilities import db_utils
from . import utilities
@ -22,6 +27,54 @@ PAGE_LINKS = {
}
class CollectLinksCommand(BaseCommand):
""" Collect links with `scheme` and save in table `table_name` """
def __init__(self, scheme, table_name) -> None:
self.scheme = scheme
self.table_name = table_name
def execute(
self,
webdriver: Firefox,
browser_params: BrowserParams,
manager_params: ManagerParams,
extension_socket: ClientSocket,
) -> None:
link_urls = [
x
for x in (
element.get_attribute("href")
for element in webdriver.find_elements_by_tag_name("a")
)
if x.startswith(self.scheme + "://")
]
current_url = webdriver.current_url
sock = ClientSocket()
sock.connect(*manager_params.aggregator_address)
query = (
"CREATE TABLE IF NOT EXISTS %s ("
"top_url TEXT, link TEXT, "
"visit_id INTEGER, browser_id INTEGER);" % self.table_name
)
sock.send(("create_table", query))
for link in link_urls:
query = (
self.table_name,
{
"top_url": current_url,
"link": link,
"visit_id": self.visit_id,
"browser_id": self.browser_id,
},
)
sock.send(query)
sock.close()
class TestCustomFunctionCommand(OpenWPMTest):
"""Test `custom_function` command's ability to handle inline functions"""
@ -31,52 +84,11 @@ class TestCustomFunctionCommand(OpenWPMTest):
def test_custom_function(self):
""" Test `custom_function` with an inline func that collects links """
from openwpm.socket_interface import ClientSocket
def collect_links(table_name, scheme, **kwargs):
""" Collect links with `scheme` and save in table `table_name` """
driver = kwargs["driver"]
manager_params = kwargs["manager_params"]
browser_id = kwargs["command"].browser_id
visit_id = kwargs["command"].visit_id
link_urls = [
x
for x in (
element.get_attribute("href")
for element in driver.find_elements_by_tag_name("a")
)
if x.startswith(scheme + "://")
]
current_url = driver.current_url
sock = ClientSocket()
sock.connect(*manager_params.aggregator_address)
query = (
"CREATE TABLE IF NOT EXISTS %s ("
"top_url TEXT, link TEXT, "
"visit_id INTEGER, browser_id INTEGER);" % table_name
)
sock.send(("create_table", query))
for link in link_urls:
query = (
table_name,
{
"top_url": current_url,
"link": link,
"visit_id": visit_id,
"browser_id": browser_id,
},
)
sock.send(query)
sock.close()
manager_params, browser_params = self.get_config()
manager = task_manager.TaskManager(manager_params, browser_params)
cs = command_sequence.CommandSequence(url_a)
cs.get(sleep=0, timeout=60)
cs.run_custom_function(collect_links, ("page_links", "http"))
cs.append_command(CollectLinksCommand("http", "page_links"))
manager.execute_command_sequence(cs)
manager.close()
query_result = db_utils.query_db(

Просмотреть файл

@ -9,8 +9,6 @@ from openwpm.config import (
)
from openwpm.errors import ConfigError
from .openwpmtest import OpenWPMTest
def test_display_mode():
browser_params = BrowserParams()

Просмотреть файл

@ -11,6 +11,7 @@ from urllib.parse import urlparse
import pytest
from openwpm import command_sequence, task_manager
from openwpm.commands.types import BaseCommand
from openwpm.utilities import db_utils
from . import utilities
@ -997,21 +998,12 @@ class TestPOSTInstrument(OpenWPMTest):
img_file_path = os.path.abspath("test_pages/shared/test_image.png")
css_file_path = os.path.abspath("test_pages/shared/test_style.css")
def type_filenames_into_form(**kwargs):
"""Simulate typing into the file upload input fields."""
driver = kwargs["driver"]
img_file_upload_element = driver.find_element_by_id("upload-img")
css_file_upload_element = driver.find_element_by_id("upload-css")
img_file_upload_element.send_keys(img_file_path)
css_file_upload_element.send_keys(css_file_path)
sleep(5) # wait for the form submission (3 sec after onload)
manager_params, browser_params = self.get_config()
manager = task_manager.TaskManager(manager_params, browser_params)
test_url = utilities.BASE_TEST_URL + "/post_file_upload.html"
cs = command_sequence.CommandSequence(test_url)
cs.get(sleep=0, timeout=60)
cs.run_custom_function(type_filenames_into_form, ())
cs.append_command(FilenamesIntoFormCommand(img_file_path, css_file_path))
manager.execute_command_sequence(cs)
manager.close()
@ -1029,3 +1021,22 @@ class TestPOSTInstrument(OpenWPMTest):
u"upload-img": img_file_content,
}
assert expected_body == post_body_decoded
class FilenamesIntoFormCommand(BaseCommand):
def __init__(self, img_file_path, css_file_path) -> None:
self.img_file_path = img_file_path
self.css_file_path = css_file_path
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
) -> None:
img_file_upload_element = webdriver.find_element_by_id("upload-img")
css_file_upload_element = webdriver.find_element_by_id("upload-css")
img_file_upload_element.send_keys(self.img_file_path)
css_file_upload_element.send_keys(self.css_file_path)
sleep(5) # wait for the form submission (3 sec after onload)

Просмотреть файл

@ -2,9 +2,10 @@ from os.path import isfile, join
import pytest
from openwpm import task_manager
from openwpm.command_sequence import CommandSequence
from openwpm.commands.types import BaseCommand
from openwpm.errors import CommandExecutionError, ProfileLoadError
from openwpm.task_manager import TaskManager
from openwpm.utilities import db_utils
from .openwpmtest import OpenWPMTest
@ -23,7 +24,7 @@ class TestProfile(OpenWPMTest):
@pytest.mark.xfail(run=False)
def test_saving(self):
manager_params, browser_params = self.get_config()
manager = task_manager.TaskManager(manager_params, browser_params)
manager = TaskManager(manager_params, browser_params)
manager.get("http://example.com")
manager.close()
assert isfile(join(browser_params[0].profile_archive_dir, "profile.tar.gz"))
@ -32,7 +33,7 @@ class TestProfile(OpenWPMTest):
def test_crash(self):
manager_params, browser_params = self.get_config()
manager_params.failure_limit = 0
manager = task_manager.TaskManager(manager_params, browser_params)
manager = TaskManager(manager_params, browser_params)
with pytest.raises(CommandExecutionError):
manager.get("http://example.com") # So we have a profile
manager.get("example.com") # Selenium requires scheme prefix
@ -42,7 +43,7 @@ class TestProfile(OpenWPMTest):
def test_crash_profile(self):
manager_params, browser_params = self.get_config()
manager_params.failure_limit = 2
manager = task_manager.TaskManager(manager_params, browser_params)
manager = TaskManager(manager_params, browser_params)
try:
manager.get("http://example.com") # So we have a profile
manager.get("example.com") # Selenium requires scheme prefix
@ -58,14 +59,14 @@ class TestProfile(OpenWPMTest):
manager_params, browser_params = self.get_config()
browser_params[0].seed_tar = "/tmp/NOTREAL"
with pytest.raises(ProfileLoadError):
task_manager.TaskManager(manager_params, browser_params) # noqa
TaskManager(manager_params, browser_params) # noqa
@pytest.mark.skip(reason="proxy no longer supported, need to update")
def test_profile_saved_when_launch_crashes(self):
manager_params, browser_params = self.get_config()
browser_params[0].proxy = True
browser_params[0].save_content = "script"
manager = task_manager.TaskManager(manager_params, browser_params)
manager = TaskManager(manager_params, browser_params)
manager.get("http://example.com")
# Kill the LevelDBAggregator
@ -84,32 +85,15 @@ class TestProfile(OpenWPMTest):
assert isfile(join(browser_params[0].profile_archive_dir, "profile.tar.gz"))
def test_seed_persistance(self):
def test_config_is_set(*args, **kwargs):
driver = kwargs["driver"]
driver.get("about:config")
result = driver.execute_script(
"""
var prefs = Components
.classes["@mozilla.org/preferences-service;1"]
.getService(Components.interfaces.nsIPrefBranch);
try {
return prefs.getBoolPref("test_pref")
} catch (e) {
return false;
}
"""
)
assert result
manager_params, browser_params = self.get_test_config(num_browsers=1)
browser_params[0].seed_tar = "."
command_sequences = []
for _ in range(2):
cs = CommandSequence(url="https://example.com", reset=True)
cs.get()
cs.run_custom_function(test_config_is_set)
cs.append_command(TestConfigSetCommand("test_pref", True))
command_sequences.append(cs)
manager = task_manager.TaskManager(manager_params, browser_params)
manager = TaskManager(manager_params, browser_params)
for cs in command_sequences:
manager.execute_command_sequence(cs)
manager.close()
@ -120,3 +104,31 @@ class TestProfile(OpenWPMTest):
assert len(query_result) > 0
for row in query_result:
assert row["command_status"] == "ok", f"Command {tuple(row)} was not ok"
class TestConfigSetCommand(BaseCommand):
def __init__(self, pref_name, expected_value) -> None:
self.pref_name = pref_name
self.expected_value = expected_value
def execute(
self,
webdriver,
browser_params,
manager_params,
extension_socket,
) -> None:
webdriver.get("about:config")
result = webdriver.execute_script(
f"""
var prefs = Components
.classes["@mozilla.org/preferences-service;1"]
.getService(Components.interfaces.nsIPrefBranch);
try {{
return prefs.getBoolPref("{self.pref_name}")
}} catch (e) {{
return false;
}}
"""
)
assert result == self.expected_value

Просмотреть файл

@ -17,10 +17,9 @@ class TestCommandDuration(OpenWPMTest):
manager = task_manager.TaskManager(manager_params, browser_params)
manager.get(url=TEST_URL, sleep=5)
manager.close()
get_command = db_utils.query_db(
manager_params.database_name,
"SELECT duration FROM crawl_history WHERE command = \"<class 'openwpm.commands.types.GetCommand'>\"",
"SELECT duration FROM crawl_history WHERE command = 'GetCommand'",
as_tuple=True,
)[0]

Просмотреть файл

@ -28,7 +28,7 @@ class TestCustomFunctionCommand(OpenWPMTest):
get_command = db_utils.query_db(
manager_params.database_name,
"SELECT command_status, error FROM crawl_history WHERE command = \"<class 'openwpm.commands.types.GetCommand'>\"",
"SELECT command_status, error FROM crawl_history WHERE command = 'GetCommand'",
as_tuple=True,
)[0]