mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 08:39:59 -05:00
Fix naming conventions.
This commit is contained in:
parent
2029964a74
commit
77dabd4057
@ -129,26 +129,22 @@ class Browser:
|
||||
def __exit__(self, *args):
|
||||
self.stop()
|
||||
|
||||
def start(self, proxy=None, cookieDb=None):
|
||||
def start(self, proxy=None, cookie_db=None):
|
||||
if not self._chrome_instance:
|
||||
# these can raise exceptions
|
||||
self.chrome_port = self._find_available_port()
|
||||
self._work_dir = tempfile.TemporaryDirectory()
|
||||
if cookieDb is not None:
|
||||
cookieDir = os.sep.join([self._work_dir.name, "chrome-user-data","Default"])
|
||||
cookieLocation = os.sep.join([cookieDir,"Cookies"])
|
||||
self.logger.debug("Cookie DB provided. Writing to: %s", cookieLocation)
|
||||
try:
|
||||
os.makedirs(cookieDir)
|
||||
except OSError as ex:
|
||||
if ex.errno!=errno.EEXIST:
|
||||
self.logger.error("Error creating cookie directory: %s", cookieDir, exc_info=True)
|
||||
if cookie_db is not None:
|
||||
cookie_dir = os.sep.join([self._work_dir.name, "chrome-user-data","Default"])
|
||||
cookie_location = os.sep.join([cookie_dir,"Cookies"])
|
||||
self.logger.debug("Cookie DB provided. Writing to: %s", cookie_location)
|
||||
os.makedirs(cookie_dir, exist_ok=True)
|
||||
|
||||
try:
|
||||
with open(cookieLocation,'wb') as cookieFile:
|
||||
cookieFile.write(cookieDb)
|
||||
with open(cookie_location,'wb') as cookie_file:
|
||||
cookie_file.write(cookie_db)
|
||||
except EnvironmentError:
|
||||
self.logger.error("exception writing cookie file at: %s", cookieLocation, exc_info=True)
|
||||
self.logger.error("exception writing cookie file at: %s", cookie_location, exc_info=True)
|
||||
|
||||
self._chrome_instance = Chrome(
|
||||
port=self.chrome_port, executable=self.chrome_exe,
|
||||
@ -178,23 +174,23 @@ class Browser:
|
||||
except:
|
||||
self.logger.error("problem stopping", exc_info=True)
|
||||
|
||||
def read_cookie_db(self):
|
||||
cookieLocation = os.sep.join([self._work_dir.name, "chrome-user-data","Default","Cookies"])
|
||||
self.logger.debug("Saving Cookie DB from: %s", cookieLocation)
|
||||
def persist_and_read_cookie_db(self):
|
||||
cookie_location = os.sep.join([self._work_dir.name, "chrome-user-data","Default","Cookies"])
|
||||
self.logger.debug("Saving Cookie DB from: %s", cookie_location)
|
||||
try:
|
||||
with sqlite3.connect(cookieLocation) as conn:
|
||||
with sqlite3.connect(cookie_location) as conn:
|
||||
cur = conn.cursor()
|
||||
cur.execute("UPDATE cookies SET persistent = 1")
|
||||
except sqlite3.Error:
|
||||
self.logger.error("exception updating cookie DB", exc_info=True)
|
||||
|
||||
cookieDb=None
|
||||
cookie_db=None
|
||||
try:
|
||||
with open(cookieLocation, "rb") as cookieFile:
|
||||
cookieDb=bytearray(cookieFile.read())
|
||||
with open(cookie_location, "rb") as cookie_file:
|
||||
cookie_db=cookie_file.read()
|
||||
except EnvironmentError:
|
||||
self.logger.error("exception reading from cookie DB file at: %s", cookieLocation, exc_info=True)
|
||||
return cookieDb
|
||||
self.logger.error("exception reading from cookie DB file at: %s", cookie_location, exc_info=True)
|
||||
return cookie_db
|
||||
|
||||
def _find_available_port(self):
|
||||
port_available = False
|
||||
|
@ -91,7 +91,7 @@ class Site(brozzler.BaseDictable):
|
||||
enable_warcprox_features=False, reached_limit=None,
|
||||
status="ACTIVE", claimed=False, start_time=None,
|
||||
last_disclaimed=_EPOCH_UTC, last_claimed_by=None,
|
||||
last_claimed=_EPOCH_UTC, metadata={}, remember_outlinks=None, cookieDb=None):
|
||||
last_claimed=_EPOCH_UTC, metadata={}, remember_outlinks=None, cookie_db=None):
|
||||
|
||||
self.seed = seed
|
||||
self.id = id
|
||||
@ -110,7 +110,7 @@ class Site(brozzler.BaseDictable):
|
||||
self.last_claimed = last_claimed
|
||||
self.metadata = metadata
|
||||
self.remember_outlinks = remember_outlinks
|
||||
self.cookieDb = bytearray(cookieDb) if cookieDb is not None else None
|
||||
self.cookie_db = cookie_db
|
||||
|
||||
self.scope = scope or {}
|
||||
if not "surt" in self.scope:
|
||||
|
@ -257,7 +257,7 @@ class BrozzlerWorker:
|
||||
if self._needs_browsing(page, ydl_spy):
|
||||
self.logger.info('needs browsing: %s', page)
|
||||
if not browser.is_running():
|
||||
browser.start(proxy=self._proxy(site), cookieDb=site.cookieDb)
|
||||
browser.start(proxy=self._proxy(site), cookie_db=site.cookie_db)
|
||||
outlinks = browser.browse_page(
|
||||
page.url, extra_headers=site.extra_headers(),
|
||||
on_screenshot=_on_screenshot,
|
||||
@ -312,7 +312,7 @@ class BrozzlerWorker:
|
||||
page = self._frontier.claim_page(site, "%s:%s" % (
|
||||
socket.gethostname(), browser.chrome_port))
|
||||
outlinks = self.brozzle_page(browser, site, page)
|
||||
site.cookieDb=browser.read_cookie_db()
|
||||
site.cookie_db=browser.persist_and_read_cookie_db()
|
||||
self._frontier.completed_page(site, page)
|
||||
self._frontier.scope_and_schedule_outlinks(
|
||||
site, page, outlinks)
|
||||
|
Loading…
x
Reference in New Issue
Block a user