mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-08-06 05:24:19 -04:00
Restructure browser.py to take screenshot after behavior script.
This commit is contained in:
parent
f285be71fb
commit
61cec15fff
1 changed files with 21 additions and 5 deletions
|
@ -174,6 +174,7 @@ class Browser:
|
||||||
self.on_screenshot = on_screenshot
|
self.on_screenshot = on_screenshot
|
||||||
self.on_url_change = on_url_change
|
self.on_url_change = on_url_change
|
||||||
|
|
||||||
|
self._waiting_on_scroll_to_top_msg_id = None
|
||||||
self._waiting_on_screenshot_msg_id = None
|
self._waiting_on_screenshot_msg_id = None
|
||||||
self._waiting_on_document_url_msg_id = None
|
self._waiting_on_document_url_msg_id = None
|
||||||
self._waiting_on_outlinks_msg_id = None
|
self._waiting_on_outlinks_msg_id = None
|
||||||
|
@ -181,6 +182,7 @@ class Browser:
|
||||||
self._reached_limit = None
|
self._reached_limit = None
|
||||||
self._aw_snap_hes_dead_jim = None
|
self._aw_snap_hes_dead_jim = None
|
||||||
self._abort_browse_page = False
|
self._abort_browse_page = False
|
||||||
|
self._has_screenshot = False
|
||||||
|
|
||||||
self._websock = websocket.WebSocketApp(self._websocket_url,
|
self._websock = websocket.WebSocketApp(self._websocket_url,
|
||||||
on_open=self._visit_page, on_message=self._wrap_handle_message)
|
on_open=self._visit_page, on_message=self._wrap_handle_message)
|
||||||
|
@ -226,6 +228,15 @@ class Browser:
|
||||||
raise BrowsingException("""chrome tab went "aw snap" or "he's dead jim"!""")
|
raise BrowsingException("""chrome tab went "aw snap" or "he's dead jim"!""")
|
||||||
elif (self._behavior != None and self._behavior.is_finished()
|
elif (self._behavior != None and self._behavior.is_finished()
|
||||||
or time.time() - self._start > Browser.HARD_TIMEOUT_SECONDS):
|
or time.time() - self._start > Browser.HARD_TIMEOUT_SECONDS):
|
||||||
|
|
||||||
|
if ( not self._has_screenshot and not self._waiting_on_scroll_to_top_msg_id and not self._waiting_on_screenshot_msg_id):
|
||||||
|
self.logger.info("Behaviors finished, requesting screenshot %s", self.url)
|
||||||
|
self._waiting_on_scroll_to_top_msg_id = self.send_to_chrome(method="Runtime.evaluate",
|
||||||
|
params={"expression":"window.scrollTo(0, 0);"})
|
||||||
|
return False
|
||||||
|
elif not self._has_screenshot and (self._waiting_on_scroll_to_top_msg_id or self._waiting_on_screenshot_msg_id):
|
||||||
|
return False
|
||||||
|
|
||||||
if self._outlinks:
|
if self._outlinks:
|
||||||
self.logger.info("got outlinks, finished browsing %s", self.url)
|
self.logger.info("got outlinks, finished browsing %s", self.url)
|
||||||
return True
|
return True
|
||||||
|
@ -246,6 +257,7 @@ class Browser:
|
||||||
return False
|
return False
|
||||||
else: # self._waiting_on_outlinks_msg_id
|
else: # self._waiting_on_outlinks_msg_id
|
||||||
return False
|
return False
|
||||||
|
|
||||||
elif time.time() - self._start > Browser.HARD_TIMEOUT_SECONDS:
|
elif time.time() - self._start > Browser.HARD_TIMEOUT_SECONDS:
|
||||||
return True
|
return True
|
||||||
elif self._reached_limit:
|
elif self._reached_limit:
|
||||||
|
@ -307,8 +319,10 @@ class Browser:
|
||||||
self.logger.info("reached limit %s", self._reached_limit)
|
self.logger.info("reached limit %s", self._reached_limit)
|
||||||
|
|
||||||
def _page_load_event_fired(self, message):
|
def _page_load_event_fired(self, message):
|
||||||
self.logger.info("Page.loadEventFired, requesting screenshot url={} message={}".format(self.url, message))
|
self.logger.info("Page.loadEventFired, moving on to starting behaviors url={}".format(self.url))
|
||||||
self._waiting_on_screenshot_msg_id = self.send_to_chrome(method="Page.captureScreenshot")
|
self._behavior = Behavior(self.url, self)
|
||||||
|
self._behavior.start()
|
||||||
|
|
||||||
self._waiting_on_document_url_msg_id = self.send_to_chrome(method="Runtime.evaluate", params={"expression":"document.URL"})
|
self._waiting_on_document_url_msg_id = self.send_to_chrome(method="Runtime.evaluate", params={"expression":"document.URL"})
|
||||||
|
|
||||||
def _console_message_added(self, message):
|
def _console_message_added(self, message):
|
||||||
|
@ -333,9 +347,11 @@ class Browser:
|
||||||
if self.on_screenshot:
|
if self.on_screenshot:
|
||||||
self.on_screenshot(base64.b64decode(message["result"]["data"]))
|
self.on_screenshot(base64.b64decode(message["result"]["data"]))
|
||||||
self._waiting_on_screenshot_msg_id = None
|
self._waiting_on_screenshot_msg_id = None
|
||||||
self.logger.info("got screenshot, moving on to starting behaviors url={}".format(self.url))
|
self._has_screenshot = True
|
||||||
self._behavior = Behavior(self.url, self)
|
self.logger.info("got screenshot, moving on to getting outlinks url={}".format(self.url))
|
||||||
self._behavior.start()
|
elif message["id"] == self._waiting_on_scroll_to_top_msg_id:
|
||||||
|
self._waiting_on_screenshot_msg_id = self.send_to_chrome(method="Page.captureScreenshot")
|
||||||
|
self._waiting_on_scroll_to_top_msg_id = None
|
||||||
elif message["id"] == self._waiting_on_outlinks_msg_id:
|
elif message["id"] == self._waiting_on_outlinks_msg_id:
|
||||||
self.logger.debug("got outlinks message=%s", message)
|
self.logger.debug("got outlinks message=%s", message)
|
||||||
self._outlinks = frozenset(message["result"]["result"]["value"].split(" "))
|
self._outlinks = frozenset(message["result"]["result"]["value"].split(" "))
|
||||||
|
|
Loading…
Add table
Add a link
Reference in a new issue