diff --git a/.idea/.gitignore b/.gitignore similarity index 100% rename from .idea/.gitignore rename to .gitignore diff --git a/Forums/AbyssForum/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/AbyssForum/__pycache__/crawler_selenium.cpython-311.pyc index e7576b0..d08a6e5 100644 Binary files a/Forums/AbyssForum/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/AbyssForum/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/AbyssForum/__pycache__/parser.cpython-311.pyc b/Forums/AbyssForum/__pycache__/parser.cpython-311.pyc index 861faf6..55a8281 100644 Binary files a/Forums/AbyssForum/__pycache__/parser.cpython-311.pyc and b/Forums/AbyssForum/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/Altenens/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/Altenens/__pycache__/crawler_selenium.cpython-311.pyc index ee63b76..38b5dbc 100644 Binary files a/Forums/Altenens/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/Altenens/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/Altenens/__pycache__/parser.cpython-311.pyc b/Forums/Altenens/__pycache__/parser.cpython-311.pyc index 5432220..3f1c5d5 100644 Binary files a/Forums/Altenens/__pycache__/parser.cpython-311.pyc and b/Forums/Altenens/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/Altenens/crawler_selenium.py b/Forums/Altenens/crawler_selenium.py index 547b765..dbbdd24 100644 --- a/Forums/Altenens/crawler_selenium.py +++ b/Forums/Altenens/crawler_selenium.py @@ -31,18 +31,18 @@ baseURL = 'https://altenens.is/' # Opens Tor Browser, crawls the website def startCrawling(): opentor() - # forumName = getForumName() + forumName = getForumName() driver = getAccess() if driver != 'down': try: - login(driver) + # login(driver) crawlForum(driver) except Exception as e: print(driver.current_url, e) closetor(driver) - # new_parse(forumName, False) + new_parse(forumName, baseURL, False) # Opens Tor Browser diff --git a/Forums/BestCardingWorld/__pycache__/__init__.cpython-311.pyc b/Forums/BestCardingWorld/__pycache__/__init__.cpython-311.pyc index 09c70d9..d26a083 100644 Binary files a/Forums/BestCardingWorld/__pycache__/__init__.cpython-311.pyc and b/Forums/BestCardingWorld/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/BestCardingWorld/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/BestCardingWorld/__pycache__/crawler_selenium.cpython-311.pyc index 6cc8cc7..5c8aa08 100644 Binary files a/Forums/BestCardingWorld/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/BestCardingWorld/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/BestCardingWorld/__pycache__/parser.cpython-311.pyc b/Forums/BestCardingWorld/__pycache__/parser.cpython-311.pyc index 7b3b7d7..82c26c2 100644 Binary files a/Forums/BestCardingWorld/__pycache__/parser.cpython-311.pyc and b/Forums/BestCardingWorld/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/Classifier/__pycache__/__init__.cpython-311.pyc b/Forums/Classifier/__pycache__/__init__.cpython-311.pyc index fb9d31a..5ab0323 100644 Binary files a/Forums/Classifier/__pycache__/__init__.cpython-311.pyc and b/Forums/Classifier/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/Classifier/__pycache__/classify_product.cpython-311.pyc b/Forums/Classifier/__pycache__/classify_product.cpython-311.pyc index a2ac40a..b1adb58 100644 Binary files a/Forums/Classifier/__pycache__/classify_product.cpython-311.pyc and b/Forums/Classifier/__pycache__/classify_product.cpython-311.pyc differ diff --git a/Forums/Classifier/__pycache__/transformer.cpython-311.pyc b/Forums/Classifier/__pycache__/transformer.cpython-311.pyc index 96bb77c..38d1f58 100644 Binary files a/Forums/Classifier/__pycache__/transformer.cpython-311.pyc and b/Forums/Classifier/__pycache__/transformer.cpython-311.pyc differ diff --git a/Forums/CryptBB/__pycache__/__init__.cpython-311.pyc b/Forums/CryptBB/__pycache__/__init__.cpython-311.pyc index 9fe9c55..c09fee6 100644 Binary files a/Forums/CryptBB/__pycache__/__init__.cpython-311.pyc and b/Forums/CryptBB/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/CryptBB/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/CryptBB/__pycache__/crawler_selenium.cpython-311.pyc index 416fe08..3f5473f 100644 Binary files a/Forums/CryptBB/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/CryptBB/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/CryptBB/__pycache__/parser.cpython-311.pyc b/Forums/CryptBB/__pycache__/parser.cpython-311.pyc index 340289d..dc04000 100644 Binary files a/Forums/CryptBB/__pycache__/parser.cpython-311.pyc and b/Forums/CryptBB/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/DB_Connection/__pycache__/__init__.cpython-311.pyc b/Forums/DB_Connection/__pycache__/__init__.cpython-311.pyc index 6e497b1..b180464 100644 Binary files a/Forums/DB_Connection/__pycache__/__init__.cpython-311.pyc and b/Forums/DB_Connection/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/DB_Connection/__pycache__/db_connection.cpython-311.pyc b/Forums/DB_Connection/__pycache__/db_connection.cpython-311.pyc index d97297e..eb9492d 100644 Binary files a/Forums/DB_Connection/__pycache__/db_connection.cpython-311.pyc and b/Forums/DB_Connection/__pycache__/db_connection.cpython-311.pyc differ diff --git a/Forums/HiddenAnswers/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/HiddenAnswers/__pycache__/crawler_selenium.cpython-311.pyc index 655b46b..80d392b 100644 Binary files a/Forums/HiddenAnswers/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/HiddenAnswers/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/HiddenAnswers/__pycache__/parser.cpython-311.pyc b/Forums/HiddenAnswers/__pycache__/parser.cpython-311.pyc index 15564de..d9c9fb5 100644 Binary files a/Forums/HiddenAnswers/__pycache__/parser.cpython-311.pyc and b/Forums/HiddenAnswers/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/Initialization/__pycache__/__init__.cpython-311.pyc b/Forums/Initialization/__pycache__/__init__.cpython-311.pyc index 0bdadbf..c6c10a4 100644 Binary files a/Forums/Initialization/__pycache__/__init__.cpython-311.pyc and b/Forums/Initialization/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/Initialization/__pycache__/forums_mining.cpython-311.pyc b/Forums/Initialization/__pycache__/forums_mining.cpython-311.pyc index 26d18ea..294aebc 100644 Binary files a/Forums/Initialization/__pycache__/forums_mining.cpython-311.pyc and b/Forums/Initialization/__pycache__/forums_mining.cpython-311.pyc differ diff --git a/Forums/Initialization/__pycache__/prepare_parser.cpython-311.pyc b/Forums/Initialization/__pycache__/prepare_parser.cpython-311.pyc index a02526d..121809c 100644 Binary files a/Forums/Initialization/__pycache__/prepare_parser.cpython-311.pyc and b/Forums/Initialization/__pycache__/prepare_parser.cpython-311.pyc differ diff --git a/Forums/Initialization/forumsList.txt b/Forums/Initialization/forumsList.txt index f1320f9..801a104 100644 --- a/Forums/Initialization/forumsList.txt +++ b/Forums/Initialization/forumsList.txt @@ -1 +1 @@ -OnniForums \ No newline at end of file +Altenens \ No newline at end of file diff --git a/Forums/Initialization/forums_mining.py b/Forums/Initialization/forums_mining.py index 85f6a61..3f0e5d5 100644 --- a/Forums/Initialization/forums_mining.py +++ b/Forums/Initialization/forums_mining.py @@ -111,6 +111,8 @@ if __name__ == '__main__': crawlerAbyssForum() elif forum == "HiddenAnswers": crawlerHiddenAnswers() + elif forum == "Altenens": + crawlerAltenensForum() diff --git a/Forums/Initialization/geckodriver.log b/Forums/Initialization/geckodriver.log index 018847f..556db20 100644 --- a/Forums/Initialization/geckodriver.log +++ b/Forums/Initialization/geckodriver.log @@ -7654,3 +7654,396 @@ JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: uncaught e JavaScript error: resource://gre/modules/PageThumbs.jsm, line 718: AbortError: IOUtils.profileBeforeChange getter: IOUtils: profileBeforeChange phase has already finished >>>>>>> 4856cd9f1f47ee8e0bf83456fc261e3c59c63faa >>>>>>> 6485ddb42552dee0b828f020640f169235a9b5a8 +1689020397937 geckodriver INFO Listening on 127.0.0.1:55884 +1689020401292 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "55885" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileCk06H2" +console.info: SecurityLevel: Listening for messages from NoScript. +console.info: SecurityLevel: Initializing security-prefs.js +console.info: SecurityLevel: security-prefs.js initialization complete +console.info: TorProtocolService: SOCKS port type: TCP +console.info: TorProtocolService: SOCKS host: 127.0.0.1 +console.info: TorProtocolService: SOCKS port: 9150 +1689020401710 Marionette INFO Marionette enabled +1689020401720 Marionette INFO Listening on port 55892 +WebDriver BiDi listening on ws://localhost:55885 +Read port: 55892 +console.debug: TorProtocolService: + TorProtocolService initialized +console.debug: TorMonitorService: + TorMonitorService initialized +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +1689020401910 RemoteAgent WARN TLS certificate errors will be ignored for this session +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.info: TorMonitorService: tor started +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +console.debug: TorMonitorService: + Control port not ready, waiting 0.05s. +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +console.debug: TorMonitorService: + Control port not ready, waiting 0.1s. +Jul 10 20:20:05.476 [notice] Tor 0.4.7.13 (git-7c1601fb6edd780f) running on Windows 8 [or later] with Libevent 2.1.12-stable, OpenSSL 1.1.1u, Zlib 1.2.13, Liblzma N/A, Libzstd N/A and Unknown N/A as libc. +console.warn: TorProcess: The watched Tor process exited with code 1. +console.info: TorProcess: Tor exited during startup. This might be due to an error in your torrc file, a bug in Tor or another program on your system, or faulty hardware. Until you fix the underlying problem and restart Tor, Tor Browser will not start. +Jul 10 20:20:05.476 [notice] Tor can't help you if you use it wrong! Learn how to be safe at https://support.torproject.org/faq/staying-anonymous/ +Jul 10 20:20:05.476 [notice] Read configuration file "C:\Users\minhkhoitran\Desktop\Tor Browser\Browser\TorBrowser\Data\Tor\torrc-defaults". +Jul 10 20:20:05.476 [notice] Read configuration file "C:\Users\minhkhoitran\Desktop\Tor Browser\Browser\TorBrowser\Data\Tor\torrc". +Jul 10 20:20:05.483 [notice] Opening Control listener on 127.0.0.1:9151 +Jul 10 20:20:05.483 [warn] Could not bind to 127.0.0.1:9151: Address already in use [WSAEADDRINUSE ]. Is Tor already running? +Jul 10 20:20:05.483 [warn] Failed to parse/validate config: Failed to bind one of the listener ports. +Jul 10 20:20:05.483 [err] Reading config failed--see warnings above. +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +openPromptSync@resource://gre/modules/Prompter.jsm:1057:17 +confirmEx@resource://gre/modules/Prompter.jsm:1465:10 +confirmEx@resource://gre/modules/Prompter.jsm:305:14 +showConfirm@resource://gre/modules/TorLauncherUtil.jsm:357:25 +_processExitedUnexpectedly@resource://gre/modules/TorProcess.jsm:217:37 +_watchProcess@resource://gre/modules/TorProcess.jsm:185:12 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +console.debug: TorMonitorService: + Control port not ready, waiting 0.2s. +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +openPromptSync@resource://gre/modules/Prompter.jsm:1057:17 +confirmEx@resource://gre/modules/Prompter.jsm:1465:10 +confirmEx@resource://gre/modules/Prompter.jsm:305:14 +showConfirm@resource://gre/modules/TorLauncherUtil.jsm:357:25 +_processExitedUnexpectedly@resource://gre/modules/TorProcess.jsm:217:37 +_watchProcess@resource://gre/modules/TorProcess.jsm:185:12 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +DevTools listening on ws://localhost:55885/devtools/browser/61e0589a-2454-4671-968c-a7e34b360c33 +console.debug: TorMonitorService: + Control port not ready, waiting 0.4s. +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +openPromptSync@resource://gre/modules/Prompter.jsm:1057:17 +confirmEx@resource://gre/modules/Prompter.jsm:1465:10 +confirmEx@resource://gre/modules/Prompter.jsm:305:14 +showConfirm@resource://gre/modules/TorLauncherUtil.jsm:357:25 +_processExitedUnexpectedly@resource://gre/modules/TorProcess.jsm:217:37 +_watchProcess@resource://gre/modules/TorProcess.jsm:185:12 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +console.debug: TorMonitorService: + Control port not ready, waiting 0.8s. +console.error: TorMonitorService: + Cannot open a control port connection + Message: Error: authenticate 31257924272e3a23472756672a495e5c -> 515 Authentication failed: Password did not match HashedControlPassword *or* authentication cookie. + Stack: + _handleCommandReply@resource://torbutton/modules/tor-control-port.js:320:19 +pushMessage@resource://torbutton/modules/tor-control-port.js:392:19 +_startMessagePump@resource://torbutton/modules/tor-control-port.js:277:29 +openPromptSync@resource://gre/modules/Prompter.jsm:1057:17 +confirmEx@resource://gre/modules/Prompter.jsm:1465:10 +confirmEx@resource://gre/modules/Prompter.jsm:305:14 +showConfirm@resource://gre/modules/TorLauncherUtil.jsm:357:25 +_processExitedUnexpectedly@resource://gre/modules/TorProcess.jsm:217:37 +_watchProcess@resource://gre/modules/TorProcess.jsm:185:12 +awaitPromise@resource://gre/modules/addons/XPIProvider.jsm:220:15 +processFileChanges@resource://gre/modules/addons/XPIDatabase.jsm:3504:19 +getNewSideloads@resource://gre/modules/addons/XPIProvider.jsm:3111:28 + +console.log: "TorConnect: Observed TorProcessExited" +1689020408717 Marionette WARN Ignoring event 'pageshow' because document has an invalid readyState of 'uninitialized'. +1689020409138 Marionette INFO Stopped listening on port 55892 +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: AbortError: Actor 'Conduits' destroyed before query 'RunListener' was resolved +[Parent 5388, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-4dcf009406d1/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: uncaught exception: SessionFileInternal.getWriter() called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\MINHKH~1\AppData\Local\Temp\rust_mozprofileCk06H2\thumbnails) because it does not exist +1689020874894 geckodriver INFO Listening on 127.0.0.1:63758 +1689020878211 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "63759" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileys7Dpv" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689020879093 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.warn: SearchSettings: "get: No settings file exists, new profile?" (new NotFoundError("Could not open the file at C:\\Users\\minhkhoitran\\AppData\\Local\\Temp\\rust_mozprofileys7Dpv\\search.json.mozlz4", (void 0), 215)) +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +DevTools listening on ws://localhost:63759/devtools/browser/22943c94-85df-47dc-937e-fab7980763fd +1689020896962 Marionette INFO Listening on port 63765 +1689020897501 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689020903377 Marionette INFO Stopped listening on port 63765 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 4132, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +[Parent 4132, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689020907472 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689020995872 geckodriver INFO Listening on 127.0.0.1:54312 +1689020999275 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "54313" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileNuHc16" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689020999882 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/ExtensionContent.jsm, line 575: TypeError: PrecompiledScript.executeInGlobal: Argument 1 is not an object. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +DevTools listening on ws://localhost:54313/devtools/browser/38ca2136-020b-4bf4-a65e-22d6d331921d +1689021007075 Marionette INFO Listening on port 54320 +1689021007524 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689021012297 Marionette INFO Stopped listening on port 54320 +JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: NotFoundError: No such JSWindowActor 'MarionetteEvents' +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileNuHc16\thumbnails) because it does not exist +1689021013873 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689021698864 geckodriver INFO Listening on 127.0.0.1:60561 +1689021702463 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "60562" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofilegDpwml" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689021704521 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:60562/devtools/browser/7b1f2f1e-e886-4136-8422-74207bddbef8 +1689021708199 Marionette INFO Listening on port 60567 +1689021708367 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689021712957 Marionette INFO Stopped listening on port 60567 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 10060, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofilegDpwml\thumbnails) because it does not exist +[Parent 10060, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689021715445 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689021919049 geckodriver INFO Listening on 127.0.0.1:62505 +1689021922644 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "62506" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileMqmHa9" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689021924036 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:62506/devtools/browser/d184b10a-4965-48a0-aa11-65dac24bc97a +1689021927418 Marionette INFO Listening on port 62518 +[Parent 8388, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689021927618 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689021932166 Marionette INFO Stopped listening on port 62518 +JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: NotFoundError: No such JSWindowActor 'MarionetteEvents' +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileMqmHa9\thumbnails) because it does not exist +[Parent 8388, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689021933984 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689022129815 geckodriver INFO Listening on 127.0.0.1:53867 +1689022133319 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "53868" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileIo1eYI" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689022134804 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:53868/devtools/browser/fb37b7cc-f948-43ad-9024-fc68549ce358 +1689022138337 Marionette INFO Listening on port 53873 +1689022138868 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: http://cryptbbtg65gibadeeo2awe3j7s6evg7eklserehqr4w4e2bis5tebid.onion/, line 2: ReferenceError: $ is not defined +JavaScript error: http://cryptbbtg65gibadeeo2awe3j7s6evg7eklserehqr4w4e2bis5tebid.onion/member.php?action=login, line 2: ReferenceError: $ is not defined +JavaScript error: http://cryptbbtg65gibadeeo2awe3j7s6evg7eklserehqr4w4e2bis5tebid.onion/member.php?action=login, line 5: ReferenceError: lang is not defined +JavaScript error: http://cryptbbtg65gibadeeo2awe3j7s6evg7eklserehqr4w4e2bis5tebid.onion/member.php?action=login, line 9: ReferenceError: use_xmlhttprequest is not defined +1689022164621 Marionette INFO Stopped listening on port 53873 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +[Parent 11240, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileIo1eYI\thumbnails) because it does not exist +[Parent 11240, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689022166722 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689022263833 geckodriver INFO Listening on 127.0.0.1:55913 +1689022267278 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "55914" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileRfxFdS" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689022268587 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:55914/devtools/browser/54f5a8ec-dbf0-4df3-84dc-6c58db34b3db +1689022271654 Marionette INFO Listening on port 55919 +[Parent 2648, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689022272175 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: https://altenens.is/js/xenmake/headroom.min.js, line 155: TypeError: this.elem is null +1689022281249 Marionette INFO Stopped listening on port 55919 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileRfxFdS\thumbnails) because it does not exist +[Parent 2648, IPC I/O Parent] WARNING: pipe error: 232: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/chrome/common/ipc_channel_win.cc:544 +1689022286198 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689022428127 geckodriver INFO Listening on 127.0.0.1:55065 +1689022431519 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "55066" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofilergxOQ0" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689022433333 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:55066/devtools/browser/2bdac81d-8172-4cd2-bce1-a8ed4a126d71 +1689022436493 Marionette INFO Listening on port 55076 +1689022436978 RemoteAgent WARN TLS certificate errors will be ignored for this session +[Parent 1084, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +[Parent 1084, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: https://altenens.is/js/xenmake/headroom.min.js, line 155: TypeError: this.elem is null +JavaScript error: https://altenens.is/js/xenmake/headroom.min.js, line 155: TypeError: this.elem is null +JavaScript error: https://altenens.is/js/xenmake/headroom.min.js, line 155: TypeError: this.elem is null +1689022528954 Marionette INFO Stopped listening on port 55076 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 1084, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofilergxOQ0\thumbnails) because it does not exist +[Parent 1084, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689022531708 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 diff --git a/Forums/OnniForums/__pycache__/crawler_selenium.cpython-311.pyc b/Forums/OnniForums/__pycache__/crawler_selenium.cpython-311.pyc index 2bc9d6a..8278bd9 100644 Binary files a/Forums/OnniForums/__pycache__/crawler_selenium.cpython-311.pyc and b/Forums/OnniForums/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/Forums/OnniForums/__pycache__/parser.cpython-311.pyc b/Forums/OnniForums/__pycache__/parser.cpython-311.pyc index 327436b..cdefc99 100644 Binary files a/Forums/OnniForums/__pycache__/parser.cpython-311.pyc and b/Forums/OnniForums/__pycache__/parser.cpython-311.pyc differ diff --git a/Forums/OnniForums/parser.py b/Forums/OnniForums/parser.py index 793609a..426e771 100644 --- a/Forums/OnniForums/parser.py +++ b/Forums/OnniForums/parser.py @@ -115,15 +115,9 @@ def onniForums_description_parser(soup: BeautifulSoup) -> tuple: def onniForums_listing_parser(soup: BeautifulSoup): -<<<<<<< HEAD boardName = "-1" # board name (the previous level of the topic in the Forum categorization tree. # For instance: Security/Malware/Tools to hack Facebook. The board here should be Malware) forum = "OnniForums" -======= - boardName = "-1" # board name (the previous level of the topic in the Forum categorization tree. - # For instance: Security/Malware/Tools to hack Facebook. The board here should be Malware) - ->>>>>>> 6485ddb42552dee0b828f020640f169235a9b5a8 nm = 0 # this variable should receive the number of topics topic : List[str] = [] # all topics user : List[str] = [] # all users of each topic diff --git a/Forums/Utilities/__pycache__/__init__.cpython-311.pyc b/Forums/Utilities/__pycache__/__init__.cpython-311.pyc index 633cca8..45a912d 100644 Binary files a/Forums/Utilities/__pycache__/__init__.cpython-311.pyc and b/Forums/Utilities/__pycache__/__init__.cpython-311.pyc differ diff --git a/Forums/Utilities/__pycache__/utilities.cpython-311.pyc b/Forums/Utilities/__pycache__/utilities.cpython-311.pyc index 5455c6c..0e77224 100644 Binary files a/Forums/Utilities/__pycache__/utilities.cpython-311.pyc and b/Forums/Utilities/__pycache__/utilities.cpython-311.pyc differ diff --git a/Forums/__pycache__/__init__.cpython-311.pyc b/Forums/__pycache__/__init__.cpython-311.pyc index af5755f..c7da357 100644 Binary files a/Forums/__pycache__/__init__.cpython-311.pyc and b/Forums/__pycache__/__init__.cpython-311.pyc differ diff --git a/MarketPlaces/AnonymousMarketplace/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/AnonymousMarketplace/__pycache__/crawler_selenium.cpython-311.pyc index 2fc80b9..a6674c4 100644 Binary files a/MarketPlaces/AnonymousMarketplace/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/AnonymousMarketplace/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/AnonymousMarketplace/__pycache__/parser.cpython-311.pyc b/MarketPlaces/AnonymousMarketplace/__pycache__/parser.cpython-311.pyc index 5491ee8..ef25fdf 100644 Binary files a/MarketPlaces/AnonymousMarketplace/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/AnonymousMarketplace/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Apocalypse/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/Apocalypse/__pycache__/crawler_selenium.cpython-311.pyc index 883fad6..9c3ad58 100644 Binary files a/MarketPlaces/Apocalypse/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/Apocalypse/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/Apocalypse/__pycache__/parser.cpython-311.pyc b/MarketPlaces/Apocalypse/__pycache__/parser.cpython-311.pyc index 7570e55..8256433 100644 Binary files a/MarketPlaces/Apocalypse/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/Apocalypse/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Apocalypse/crawler_selenium.py b/MarketPlaces/Apocalypse/crawler_selenium.py index b46e4af..28d6a0f 100644 --- a/MarketPlaces/Apocalypse/crawler_selenium.py +++ b/MarketPlaces/Apocalypse/crawler_selenium.py @@ -32,19 +32,19 @@ baseURL = 'http://apocam5hnoqskkmhr325nivjuh5phbmmggadxgcjabzzirap5iklkxad.onion # Opens Tor Browser, crawls the website, then parses, then closes tor #acts like the main method for the crawler, another function at the end of this code calls this function later def startCrawling(): - opentor() - # mktName = getMKTName() - driver = getAccess() + # opentor() + mktName = getMKTName() + # driver = getAccess() - if driver != 'down': - try: - login(driver) - crawlForum(driver) - except Exception as e: - print(driver.current_url, e) - closetor(driver) + # if driver != 'down': + # try: + # login(driver) + # crawlForum(driver) + # except Exception as e: + # print(driver.current_url, e) + # closetor(driver) - # new_parse(forumName, baseURL, False) + new_parse(mktName, baseURL, False) # Opens Tor Browser @@ -245,10 +245,10 @@ def crawlForum(driver): driver.back() # comment out - break + # break # comment out - if count == 1: + if count == 20: count = 0 break diff --git a/MarketPlaces/Apocalypse/parser.py b/MarketPlaces/Apocalypse/parser.py index f120b54..469241a 100644 --- a/MarketPlaces/Apocalypse/parser.py +++ b/MarketPlaces/Apocalypse/parser.py @@ -4,154 +4,85 @@ __author__ = 'DarkWeb' from MarketPlaces.Utilities.utilities import * # Here, we are importing BeautifulSoup to search through the HTML tree -from bs4 import BeautifulSoup +from bs4 import BeautifulSoup, ResultSet, Tag -#parses description pages, so takes html pages of description pages using soup object, and parses it for info it needs -#stores info it needs in different lists, these lists are returned after being organized -#@param: soup object looking at html page of description page -#return: 'row' that contains a variety of lists that each hold info on the description page -def darkfox_description_parser(soup): - +def apocalypse_description_parser(soup: Tag): + # Fields to be parsed - name = "-1" # 0 Product_Name - describe = "-1" # 1 Product_Description - lastSeen = "-1" # 2 Product_LastViewDate - rules = "-1" # 3 NOT USED ... - CVE = "-1" # 4 Product_CVE_Classification (Common Vulnerabilities and Exposures) - MS = "-1" # 5 Product_MS_Classification (Microsoft Security) - review = "-1" # 6 Product_Number_Of_Reviews + vendor = "-1" # 0 *Vendor_Name + success = "-1" # 1 Vendor_Successful_Transactions + rating_vendor = "-1" # 2 Vendor_Rating + name = "-1" # 3 *Product_Name + describe = "-1" # 4 Product_Description + CVE = "-1" # 5 Product_CVE_Classification (Common Vulnerabilities and Exposures) + MS = "-1" # 6 Product_MS_Classification (Microsoft Security) category = "-1" # 7 Product_Category - shipFrom = "-1" # 8 Product_ShippedFrom - shipTo = "-1" # 9 Product_ShippedTo - left = "-1" # 10 Product_QuantityLeft - escrow = "-1" # 11 Vendor_Warranty - terms = "-1" # 12 Vendor_TermsAndConditions - vendor = "-1" # 13 Vendor_Name - sold = "-1" # 14 Product_QuantitySold - addDate = "-1" # 15 Product_AddedDate - available = "-1" # 16 NOT USED ... - endDate = "-1" # 17 NOT USED ... - BTC = "-1" # 18 Product_BTC_SellingPrice - USD = "-1" # 19 Product_USD_SellingPrice - rating = "-1" # 20 Vendor_Rating - success = "-1" # 21 Vendor_Successful_Transactions - EURO = "-1" # 22 Product_EURO_SellingPrice - - # Finding Product Name - name = soup.find('h1').text - name = name.replace('\n', ' ') - name = name.replace(",", "") - name = name.strip() - - # Finding Vendor - vendor = soup.find('h3').find('a').text.strip() - - # Finding Vendor Rating - rating = soup.find('span', {'class': "tag is-dark"}).text.strip() - - # Finding Successful Transactions - success = soup.find('h3').text - success = success.replace("Vendor: ", "") - success = success.replace(vendor, "") - success = success.replace("(", "") - success = success.replace(")", "") - success = success.strip() - - bae = soup.find('div', {'class': "box"}).find_all('ul') - - # Finding Prices - USD = bae[1].find('strong').text.strip() - - li = bae[2].find_all('li') - - # Finding Escrow - escrow = li[0].find('span', {'class': "tag is-dark"}).text.strip() - - # Finding the Product Category - category = li[1].find('span', {'class': "tag is-dark"}).text.strip() - - # Finding the Product Quantity Available - left = li[3].find('span', {'class': "tag is-dark"}).text.strip() - - # Finding Number Sold - sold = li[4].find('span', {'class': "tag is-dark"}).text.strip() - - li = bae[3].find_all('li') - - # Finding Shipment Information (Origin) - if "Ships from:" in li[-2].text: - shipFrom = li[-2].text - shipFrom = shipFrom.replace("Ships from: ", "") - # shipFrom = shipFrom.replace(",", "") - shipFrom = shipFrom.strip() - - # Finding Shipment Information (Destination) - shipTo = li[-1].find('div', {'title': "List of countries is scrollable"}).text - shipTo = shipTo.replace("Ships to: ", "") - shipTo = shipTo.strip() - if "certain countries" in shipTo: - countries = "" - tags = li[-1].find_all('span', {'class': "tag"}) - for tag in tags: - country = tag.text.strip() - countries += country + ", " - shipTo = countries.strip(", ") - - # Finding the Product description - describe = soup.find('div', {'class': "pre-line"}).text - describe = describe.replace("\n", " ") - describe = describe.strip() - - '''# Finding the Number of Product Reviews - tag = soup.findAll(text=re.compile('Reviews')) - for index in tag: - reviews = index - par = reviews.find('(') - if par >=0: - reviews = reviews.replace("Reviews (","") - reviews = reviews.replace(")","") - reviews = reviews.split(",") - review = str(abs(int(reviews[0])) + abs(int(reviews[1]))) - else : - review = "-1"''' - - # Searching for CVE and MS categories - cve = soup.findAll(text=re.compile('CVE-\d{4}-\d{4}')) - if cve: - CVE = " " - for idx in cve: - CVE += (idx) - CVE += " " - CVE = CVE.replace(',', ' ') - CVE = CVE.replace('\n', '') - ms = soup.findAll(text=re.compile('MS\d{2}-\d{3}')) - if ms: - MS = " " - for im in ms: - MS += (im) - MS += " " - MS = MS.replace(',', ' ') - MS = MS.replace('\n', '') - + views = "-1" # 8 Product_Number_Of_Views + reviews = "-1" # 9 Product_Number_Of_Reviews + rating_item = "-1" # 10 Product_Rating + addDate = "-1" # 11 Product_AddedDate + BTC = "-1" # 12 Product_BTC_SellingPrice + USD = "-1" # 13 Product_USD_SellingPrice + EURO = "-1" # 14 Product_EURO_SellingPrice + sold = "-1" # 15 Product_QuantitySold + left = "-1" # 16 Product_QuantityLeft + shipFrom = "-1" # 17 Product_ShippedFrom + shipTo = "-1" # 18 Product_ShippedTo + + content: Tag = soup.find("div", {'id': "article_page"}) + + product_name = content.find("p", {"class": "list-group-item text-center mb-0 box"}).text + name = cleanString(product_name.strip()) + + product_description = content.find("pre").text + describe = cleanString(product_description.strip()) + + product_reviews_list: Tag = content.find("table", {"class": "table product_reviews"}) \ + .find_all("li") + + review = str(len(product_reviews_list)) + + product_category = content.find("a", {"class": "badge badge-danger"}).text + category = cleanString(product_category.strip()) + + product_ships_from = content.find("span", {"class": "badge badge-info"}).text + shipFrom = cleanString(product_ships_from.strip()) + + product_success_badge: ResultSet[Tag] = content.find_all("span", {"class": "badge badge-success"}) + product_ships_to = product_success_badge[1].text + shipTo = cleanString(product_ships_to.strip()) + + product_supply = content.find("span", {"class": "badge badge-warning"}).text + left = cleanString(product_supply.strip()) + + product_primary_badge: ResultSet[Tag] = content.find_all("span", {"class": "badge badge-primary"}) + + # Product vendor comes in the form of "@ vendor_name" + product_vendor = product_primary_badge[0].text.replace("@", "") + + vendor = cleanString(product_vendor.strip()) + sold = cleanString(product_primary_badge[1].text.strip()) + + product_prices: Tag = content.find("p", {"style": "border-bottom:1px solid GREY;"}) + USD = product_prices.find("span", {"class": "pr"}).text + prices_array: ResultSet[Tag] = product_prices.find_all("span", {"class": "pr1"}) + BTC = prices_array[1].text + # Populating the final variable (this should be a list with all fields scraped) - row = (name, describe, lastSeen, rules, CVE, MS, review, category, shipFrom, shipTo, left, escrow, terms, vendor, - sold, addDate, available, endDate, BTC, USD, rating, success, EURO) + row = (vendor, rating_vendor, success, name, describe, CVE, MS, category, views, reviews, rating_item, addDate, + BTC, USD, EURO, sold, left, shipFrom, shipTo) # Sending the results return row -#parses listing pages, so takes html pages of listing pages using soup object, and parses it for info it needs -#stores info it needs in different lists, these lists are returned after being organized -#@param: soup object looking at html page of listing page -#return: 'row' that contains a variety of lists that each hold info on the listing page -def darkfox_listing_parser(soup): - +def apocalypse_listing_parser(soup: Tag): + # Fields to be parsed nm = 0 # Total_Products (Should be Integer) - mktName = "DarkFox" # 0 Marketplace_Name + mktName = "Apocalypse" # 0 Marketplace_Name name = [] # 1 Product_Name CVE = [] # 2 Product_CVE_Classification (Common Vulnerabilities and Exposures) MS = [] # 3 Product_MS_Classification (Microsoft Security) @@ -173,105 +104,94 @@ def darkfox_listing_parser(soup): rating = [] # 19 Vendor_Rating success = [] # 20 Vendor_Successful_Transactions href = [] # 23 Product_Links (Urls) - - listing = soup.findAll('div', {"class": "card"}) - - # Populating the Number of Products - nm = len(listing) - - for a in listing: - bae = a.findAll('a', href=True) - - # Adding the url to the list of urls - link = bae[0].get('href') - link = cleanLink(link) - href.append(link) - - # Finding the Product - product = bae[1].find('p').text - product = product.replace('\n', ' ') - product = product.replace(",", "") - product = product.replace("...", "") - product = product.strip() - name.append(product) - - bae = a.find('div', {'class': "media-content"}).find('div').find_all('div') - - if len(bae) >= 5: - # Finding Prices - price = bae[0].text - ud = price.replace(" USD", " ") - # u = ud.replace("$","") - u = ud.replace(",", "") - u = u.strip() - USD.append(u) - # bc = (prc[1]).strip(' BTC') - # BTC.append(bc) - - # Finding the Vendor - vendor_name = bae[1].find('a').text - vendor_name = vendor_name.replace(",", "") - vendor_name = vendor_name.strip() - vendor.append(vendor_name) - - # Finding the Category - cat = bae[2].find('small').text - cat = cat.replace("Category: ", "") - cat = cat.replace(",", "") - cat = cat.strip() - category.append(cat) - - # Finding Number Sold and Quantity Left - num = bae[3].text - num = num.replace("Sold: ", "") - num = num.strip() - sold.append(num) - - quant = bae[4].find('small').text - quant = quant.replace("In stock: ", "") - quant = quant.strip() - qLeft.append(quant) - - # Finding Successful Transactions - freq = bae[1].text - freq = freq.replace(vendor_name, "") - freq = re.sub(r'Vendor Level \d+', "", freq) - freq = freq.replace("(", "") - freq = freq.replace(")", "") - freq = freq.strip() - success.append(freq) - - # Searching for CVE and MS categories - cve = a.findAll(text=re.compile('CVE-\d{4}-\d{4}')) - if not cve: - cveValue="-1" - else: - cee = " " - for idx in cve: - cee += (idx) - cee += " " - cee = cee.replace(',', ' ') - cee = cee.replace('\n', '') - cveValue=cee - CVE.append(cveValue) + + + listings: ResultSet[Tag] = soup.find("div", {"class": "col-lg-9 my-4"}).find_all("div", {"class": "col-lg-4 col-md-6 mb-1"}) + + for prod in listings: - ms = a.findAll(text=re.compile('MS\d{2}-\d{3}')) - if not ms: - MSValue="-1" - else: - me = " " - for im in ms: - me += (im) - me += " " - me = me.replace(',', ' ') - me = me.replace('\n', '') - MSValue=me - MS.append(MSValue) - - # Populate the final variable (this should be a list with all fields scraped) - return organizeProducts(mktName, nm, name, CVE, MS, category, describe, escrow, views, reviews, addDate, lastSeen, - BTC, USD, EURO, qLeft, shipFrom, shipTo, vendor, rating, success, sold, href) - + product_name = prod.find('h5', {"class": "art_title"}).text + name.append(cleanString(product_name.strip())) + + CVE.append("-1") + MS.append("-1") + describe.append("-1") + escrow.append("-1") + reviews.append("-1") + addDate.append("-1") + lastSeen.append("-1") + BTC.append("-1") + EURO.append("-1") + shipTo.append("-1") + success.append("-1") + + product_price = prod.find("span", {"class": "priceP"}).text + USD.append(cleanString(product_price.strip())) + + + + product_sold = prod.find("span", {"class": "badge badge-success"}).text + sold.append(cleanString(product_sold.strip())) + + product_statistics: ResultSet[Tag] = prod.find_all("p", {"class": "mb-0 card-text"}) + + product_category = product_statistics[0].find("a").text + category.append(cleanString(product_category.strip())) + + product_sold = product_statistics[1].find("span").text + sold.append(cleanString(product_sold.strip())) + + product_quantity_left = product_statistics[2].find("span", {"class": "badge bluebadge"}).text + qLeft.append(cleanString(product_quantity_left.strip())) + + product_views = product_statistics[3].find("span").text + views.append(cleanString(product_views.strip())) + + product_ships_from = product_statistics[4].find("span").text + shipFrom.append(cleanString(product_ships_from.strip())) + + product_vendor_tag: Tag = product_statistics[5].find("a").find("span", {"class": "badge badge-primary"}) + # Product vendors & ratings are displayed as "vender_name ★ 5.0" + # When split by the star (★), it should return a 2-value array + product_vendor, product_vendor_rating = product_vendor_tag.text.split("★") + + + try: + vendor.append(cleanString(product_vendor.strip())) + rating.append(cleanString(product_vendor_rating.strip())) + except Exception as e: + raise e + + product_href = prod.find('a').get('href') + href.append(product_href) + + nm += 1 + + + return organizeProducts( + marketplace=mktName, + nm=nm, + vendor=vendor, + rating_vendor=rating, + success_vendor=success, + nombre=name, + CVE=CVE, + MS=MS, + category=category, + describe=describe, + views=views, + reviews=reviews, + rating_item=["-1" for _ in range(nm)], + addDate=addDate, + BTC=BTC, + USD=USD, + EURO=EURO, + sold=sold, + qLeft=qLeft, + shipFrom=shipFrom, + shipTo=shipTo, + href=href + ) #called by the crawler to get description links on a listing page #@param: beautifulsoup object that is using the correct html page (listing page) diff --git a/MarketPlaces/CityMarket/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/CityMarket/__pycache__/crawler_selenium.cpython-311.pyc index 541ae49..49ba90f 100644 Binary files a/MarketPlaces/CityMarket/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/CityMarket/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/CityMarket/__pycache__/parser.cpython-311.pyc b/MarketPlaces/CityMarket/__pycache__/parser.cpython-311.pyc index d362c9c..b3ce724 100644 Binary files a/MarketPlaces/CityMarket/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/CityMarket/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Classifier/__pycache__/__init__.cpython-311.pyc b/MarketPlaces/Classifier/__pycache__/__init__.cpython-311.pyc index d346a0f..4b915db 100644 Binary files a/MarketPlaces/Classifier/__pycache__/__init__.cpython-311.pyc and b/MarketPlaces/Classifier/__pycache__/__init__.cpython-311.pyc differ diff --git a/MarketPlaces/Classifier/__pycache__/classify_product.cpython-311.pyc b/MarketPlaces/Classifier/__pycache__/classify_product.cpython-311.pyc index 7fb5747..b935f0d 100644 Binary files a/MarketPlaces/Classifier/__pycache__/classify_product.cpython-311.pyc and b/MarketPlaces/Classifier/__pycache__/classify_product.cpython-311.pyc differ diff --git a/MarketPlaces/Classifier/__pycache__/transformer.cpython-311.pyc b/MarketPlaces/Classifier/__pycache__/transformer.cpython-311.pyc index 4bd87b9..1214a29 100644 Binary files a/MarketPlaces/Classifier/__pycache__/transformer.cpython-311.pyc and b/MarketPlaces/Classifier/__pycache__/transformer.cpython-311.pyc differ diff --git a/MarketPlaces/DB_Connection/__pycache__/__init__.cpython-311.pyc b/MarketPlaces/DB_Connection/__pycache__/__init__.cpython-311.pyc index 66896ea..366f9e0 100644 Binary files a/MarketPlaces/DB_Connection/__pycache__/__init__.cpython-311.pyc and b/MarketPlaces/DB_Connection/__pycache__/__init__.cpython-311.pyc differ diff --git a/MarketPlaces/DB_Connection/__pycache__/db_connection.cpython-311.pyc b/MarketPlaces/DB_Connection/__pycache__/db_connection.cpython-311.pyc index 9bcd79c..cd3f879 100644 Binary files a/MarketPlaces/DB_Connection/__pycache__/db_connection.cpython-311.pyc and b/MarketPlaces/DB_Connection/__pycache__/db_connection.cpython-311.pyc differ diff --git a/MarketPlaces/DarkFox/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/DarkFox/__pycache__/crawler_selenium.cpython-311.pyc index 5b3ac1b..38a4bb7 100644 Binary files a/MarketPlaces/DarkFox/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/DarkFox/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/DarkFox/__pycache__/parser.cpython-311.pyc b/MarketPlaces/DarkFox/__pycache__/parser.cpython-311.pyc index ec8c0f9..70dccd6 100644 Binary files a/MarketPlaces/DarkFox/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/DarkFox/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/DarkMatter/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/DarkMatter/__pycache__/crawler_selenium.cpython-311.pyc index d37fd6f..f74efb2 100644 Binary files a/MarketPlaces/DarkMatter/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/DarkMatter/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/DarkMatter/__pycache__/parser.cpython-311.pyc b/MarketPlaces/DarkMatter/__pycache__/parser.cpython-311.pyc index ab519a1..9229629 100644 Binary files a/MarketPlaces/DarkMatter/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/DarkMatter/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/DarkTor/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/DarkTor/__pycache__/crawler_selenium.cpython-311.pyc index 61c9d72..22102ff 100644 Binary files a/MarketPlaces/DarkTor/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/DarkTor/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/DarkTor/__pycache__/parser.cpython-311.pyc b/MarketPlaces/DarkTor/__pycache__/parser.cpython-311.pyc index 069cbaf..822eabc 100644 Binary files a/MarketPlaces/DarkTor/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/DarkTor/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/DarkTor/crawler_selenium.py b/MarketPlaces/DarkTor/crawler_selenium.py index c4cbfd4..3c44dc7 100644 --- a/MarketPlaces/DarkTor/crawler_selenium.py +++ b/MarketPlaces/DarkTor/crawler_selenium.py @@ -230,10 +230,10 @@ def crawlForum(driver): driver.back() # comment out - break + # break # comment out - if count == 1: + if count == 30: count = 0 break diff --git a/MarketPlaces/DigitalThriftShop/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/DigitalThriftShop/__pycache__/crawler_selenium.cpython-311.pyc index 6b5e8c0..e87f595 100644 Binary files a/MarketPlaces/DigitalThriftShop/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/DigitalThriftShop/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/DigitalThriftShop/__pycache__/parser.cpython-311.pyc b/MarketPlaces/DigitalThriftShop/__pycache__/parser.cpython-311.pyc index 9a2cab1..6bfdaca 100644 Binary files a/MarketPlaces/DigitalThriftShop/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/DigitalThriftShop/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Initialization/__pycache__/__init__.cpython-311.pyc b/MarketPlaces/Initialization/__pycache__/__init__.cpython-311.pyc index 1c9dd27..1309640 100644 Binary files a/MarketPlaces/Initialization/__pycache__/__init__.cpython-311.pyc and b/MarketPlaces/Initialization/__pycache__/__init__.cpython-311.pyc differ diff --git a/MarketPlaces/Initialization/__pycache__/markets_mining.cpython-311.pyc b/MarketPlaces/Initialization/__pycache__/markets_mining.cpython-311.pyc index aaa6eba..d2bbff8 100644 Binary files a/MarketPlaces/Initialization/__pycache__/markets_mining.cpython-311.pyc and b/MarketPlaces/Initialization/__pycache__/markets_mining.cpython-311.pyc differ diff --git a/MarketPlaces/Initialization/__pycache__/prepare_parser.cpython-311.pyc b/MarketPlaces/Initialization/__pycache__/prepare_parser.cpython-311.pyc index 9aa29d6..b669316 100644 Binary files a/MarketPlaces/Initialization/__pycache__/prepare_parser.cpython-311.pyc and b/MarketPlaces/Initialization/__pycache__/prepare_parser.cpython-311.pyc differ diff --git a/MarketPlaces/Initialization/geckodriver.log b/MarketPlaces/Initialization/geckodriver.log index c3073be..9252e3d 100644 --- a/MarketPlaces/Initialization/geckodriver.log +++ b/MarketPlaces/Initialization/geckodriver.log @@ -12154,3 +12154,409 @@ JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't fin JavaScript error: resource://gre/modules/AsyncShutdown.jsm, line 575: uncaught exception: SessionFileInternal.getWriter() called too early! Please read the session file from disk first. JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\MINHKH~1\AppData\Local\Temp\rust_mozprofile5wtzYO\thumbnails) because it does not exist [Parent 9204, IPC I/O Parent] WARNING: pipe error: 232: file /var/tmp/build/firefox-b1c79245b9a4/ipc/chromium/src/chrome/common/ipc_channel_win.cc:554 +1689027536873 geckodriver INFO Listening on 127.0.0.1:58896 +1689027540458 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "58897" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileoBZVUy" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689027542141 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:58897/devtools/browser/1cb24560-d41a-49a9-8a36-d3872360db16 +1689027548133 Marionette INFO Listening on port 58902 +1689027549218 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689027592577 Marionette INFO Stopped listening on port 58902 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 9308, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileoBZVUy\thumbnails) because it does not exist +[Parent 9308, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689027594612 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689027762781 geckodriver INFO Listening on 127.0.0.1:60672 +1689027767350 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "60673" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofile29AjXd" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689027769668 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:60673/devtools/browser/b46f2c03-b442-4bac-b701-97d5e4c99495 +1689027774565 Marionette INFO Listening on port 60685 +1689027775418 RemoteAgent WARN TLS certificate errors will be ignored for this session +console.log: "TorConnect: beginBootstrap()" +console.log: "TorConnect: Try transitioning from Initial to Bootstrapping" +console.log: "TorConnect: Exited Initial state" +console.log: "TorConnect: Entering Bootstrapping state" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689027793585 Marionette INFO Stopped listening on port 60685 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 10780, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: moz-extension://ea77c2e9-b8db-4689-b4a8-30cbf0652171/nscl/common/log.js, line 36: TypeError: can't access dead object +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofile29AjXd\thumbnails) because it does not exist +1689027795208 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689027854442 geckodriver INFO Listening on 127.0.0.1:53627 +1689027858053 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "53628" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofilel8aw8B" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689027860047 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:53628/devtools/browser/2b102403-a270-4c00-a055-4aa2ca464549 +1689027863667 Marionette INFO Listening on port 65239 +1689027864969 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689027958937 Marionette INFO Stopped listening on port 65239 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 5784, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: moz-extension://ea77c2e9-b8db-4689-b4a8-30cbf0652171/nscl/common/log.js, line 36: TypeError: can't access dead object +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofilel8aw8B\thumbnails) because it does not exist +[Parent 5784, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689027960521 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689028040965 geckodriver INFO Listening on 127.0.0.1:61932 +1689028045731 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "61933" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofiled80SDJ" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689028052563 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:61933/devtools/browser/bd75cb83-0b67-4923-859a-e0defd477b1a +1689028059480 Marionette INFO Listening on port 61943 +1689028060321 RemoteAgent WARN TLS certificate errors will be ignored for this session +[Parent 6496, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +console.log: "TorConnect: beginBootstrap()" +console.log: "TorConnect: Try transitioning from Initial to Bootstrapping" +console.log: "TorConnect: Exited Initial state" +console.log: "TorConnect: Entering Bootstrapping state" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689028085304 Marionette INFO Stopped listening on port 61943 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 6496, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +[Parent 6496, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofiled80SDJ\thumbnails) because it does not exist +[Parent 6496, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689028087339 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689028324198 geckodriver INFO Listening on 127.0.0.1:64235 +1689028328225 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "64236" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofilei2mDK9" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689028336958 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:64236/devtools/browser/c3c1850a-220c-4aee-b4dd-ce0235633765 +[Parent 9132, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689028352323 Marionette INFO Listening on port 64248 +1689028352692 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689028364606 Marionette INFO Stopped listening on port 64248 +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofilei2mDK9\thumbnails) because it does not exist +[Parent 9132, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689028366563 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689029032462 geckodriver INFO Listening on 127.0.0.1:62083 +1689029038456 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "62084" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileFp57y8" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689029040753 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:62084/devtools/browser/9ac05d95-d613-4d68-87c3-3447fff93c27 +1689029046119 Marionette INFO Listening on port 62089 +1689029046752 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/actors/AutoCompleteChild.jsm, line 125: Error: Invalid autocomplete selectedIndex +JavaScript error: resource:///modules/UrlbarInput.jsm, line 2952: TypeError: event.target.closest is not a function +1689029623751 Marionette INFO Stopped listening on port 62089 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +JavaScript error: chrome://remote/content/marionette/cert.js, line 55: NS_ERROR_NOT_AVAILABLE: Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsICertOverrideService.setDisableAllSecurityChecksAndLetAttackersInterceptMyData] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +[Parent 10372, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: resource://gre/modules/PageThumbs.jsm, line 709: AbortError: IOUtils.profileBeforeChange getter: IOUtils: profileBeforeChange phase has already finished +1689029627939 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689029978003 geckodriver INFO Listening on 127.0.0.1:56350 +1689029984147 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "56351" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileli0ySk" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689029988350 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:56351/devtools/browser/7b71fd6a-c5d8-49e1-ae2d-748a5f602160 +1689029993231 Marionette INFO Listening on port 65309 +1689029993827 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689030288524 Marionette INFO Stopped listening on port 65309 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileli0ySk\thumbnails) because it does not exist +1689030292660 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689030525024 geckodriver INFO Listening on 127.0.0.1:59971 +1689030529958 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "59972" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileBjxvxN" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689030533635 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:59972/devtools/browser/7b60176d-d841-422b-bc49-8a8082db7c39 +1689030539135 Marionette INFO Listening on port 59978 +1689030539531 RemoteAgent WARN TLS certificate errors will be ignored for this session +1689030595909 Marionette INFO Stopped listening on port 59978 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +[Parent 8704, IPC I/O Parent] WARNING: pipe error: 232: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/chrome/common/ipc_channel_win.cc:544 +[Parent 8704, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +JavaScript error: chrome://remote/content/marionette/cert.js, line 55: NS_ERROR_NOT_AVAILABLE: Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsICertOverrideService.setDisableAllSecurityChecksAndLetAttackersInterceptMyData] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileBjxvxN\thumbnails) because it does not exist +[Parent 8704, IPC I/O Parent] WARNING: file /var/tmp/build/firefox-b6010b1466c9/ipc/chromium/src/base/process_util_win.cc:167 +1689030598070 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689030671606 geckodriver INFO Listening on 127.0.0.1:50366 +1689030675847 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "50367" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileMQLh6V" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689030677801 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:50367/devtools/browser/86fd830f-0cb1-4e11-88dc-48b626199e7d +1689030682980 Marionette INFO Listening on port 50372 +1689030683465 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileMQLh6V\thumbnails) because it does not exist +1689032941949 Marionette INFO Stopped listening on port 50372 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +JavaScript error: chrome://remote/content/marionette/cert.js, line 55: NS_ERROR_NOT_AVAILABLE: Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsICertOverrideService.setDisableAllSecurityChecksAndLetAttackersInterceptMyData] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileMQLh6V\thumbnails) because it does not exist +1689032951168 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689098476059 geckodriver INFO Listening on 127.0.0.1:49799 +1689098479402 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "49800" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofileOSrhn8" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689098479886 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:49800/devtools/browser/c0db15dd-f0dd-482e-9661-3b709ee430f3 +1689098481156 Marionette INFO Listening on port 49805 +1689098481479 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/actors/AutoCompleteChild.jsm, line 125: Error: Invalid autocomplete selectedIndex +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689098631986 Marionette INFO Stopped listening on port 49805 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofileOSrhn8\thumbnails) because it does not exist +1689098632298 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 +1689098733626 geckodriver INFO Listening on 127.0.0.1:49891 +1689098736947 mozrunner::runner INFO Running command: "C:\\Users\\minhkhoitran\\Desktop\\Tor Browser\\Browser\\firefox.exe" "--marionette" "--remote-debugging-port" "49892" "--remote-allow-hosts" "localhost" "-no-remote" "-profile" "C:\\Users\\MINHKH~1\\AppData\\Local\\Temp\\rust_mozprofilekUphMW" +console.log: "TorSettings: loadFromPrefs()" +console.log: "TorConnect: init()" +console.log: "TorConnect: Entering Initial state" +console.log: "TorConnect: Observed profile-after-change" +console.log: "TorConnect: Observing topic 'TorProcessExited'" +console.log: "TorConnect: Observing topic 'TorLogHasWarnOrErr'" +console.log: "TorConnect: Observing topic 'torsettings:ready'" +console.log: "TorSettings: Observed profile-after-change" +1689098737377 Marionette INFO Marionette enabled +console.log: "TorConnect: Will load after bootstrap => [about:blank]" +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +console.error: "Could not load engine blockchair-onion@search.mozilla.org: Error: Extension is invalid" +JavaScript error: resource://gre/modules/XPCOMUtils.jsm, line 161: TypeError: Cc[aContract] is undefined +DevTools listening on ws://localhost:49892/devtools/browser/8d693373-4ed2-4bdc-8d1f-4c58e0f2a4b4 +1689098738455 Marionette INFO Listening on port 49897 +1689098738514 RemoteAgent WARN TLS certificate errors will be ignored for this session +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +JavaScript error: resource://gre/modules/XULStore.jsm, line 66: Error: Can't find profile directory. +1689099035811 Marionette INFO Stopped listening on port 49897 +JavaScript error: resource:///modules/Interactions.jsm, line 209: NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIUserIdleService.removeIdleObserver] +!!! error running onStopped callback: TypeError: callback is not a function +JavaScript error: resource:///modules/sessionstore/SessionFile.jsm, line 375: Error: _initWorker called too early! Please read the session file from disk first. +JavaScript error: resource://gre/modules/PromiseWorker.jsm, line 106: Error: Could not get children of file(C:\Users\minhkhoitran\AppData\Local\Temp\rust_mozprofilekUphMW\thumbnails) because it does not exist + +###!!! [Parent][MessageChannel] Error: (msgtype=0x390076,name=PContent::Msg_DestroyBrowsingContextGroup) Closed channel: cannot send/recv + +1689099036125 RemoteAgent ERROR unable to stop listener: [Exception... "Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsIWindowMediator.getEnumerator]" nsresult: "0x8000ffff (NS_ERROR_UNEXPECTED)" location: "JS frame :: chrome://remote/content/cdp/observers/TargetObserver.jsm :: stop :: line 64" data: no] Stack trace: stop()@TargetObserver.jsm:64 +unwatchForTabs()@TargetList.jsm:70 +unwatchForTargets()@TargetList.jsm:37 +destructor()@TargetList.jsm:109 +stop()@CDP.jsm:104 +close()@RemoteAgent.jsm:138 diff --git a/MarketPlaces/Initialization/marketsList.txt b/MarketPlaces/Initialization/marketsList.txt index 8ce2822..559010a 100644 --- a/MarketPlaces/Initialization/marketsList.txt +++ b/MarketPlaces/Initialization/marketsList.txt @@ -1 +1 @@ -MikesGrandStore +Apocalypse \ No newline at end of file diff --git a/MarketPlaces/Initialization/markets_mining.py b/MarketPlaces/Initialization/markets_mining.py index aad5f63..664454a 100644 --- a/MarketPlaces/Initialization/markets_mining.py +++ b/MarketPlaces/Initialization/markets_mining.py @@ -73,7 +73,7 @@ if __name__ == '__main__': for mkt in mktsList: mkt = mkt.replace('\n','') - print("Creating listing and description directories ...") + print(f"Creating listing and description directories of {mkt} ...") createDirectory(mkt) time.sleep(5) input("Directories created successfully. Press ENTER to continue\n") diff --git a/MarketPlaces/Initialization/prepare_parser.py b/MarketPlaces/Initialization/prepare_parser.py index accd697..cbf0455 100644 --- a/MarketPlaces/Initialization/prepare_parser.py +++ b/MarketPlaces/Initialization/prepare_parser.py @@ -7,6 +7,7 @@ import shutil from MarketPlaces.DB_Connection.db_connection import * from MarketPlaces.DarkFox.parser import * from MarketPlaces.Tor2door.parser import * +from MarketPlaces.Apocalypse.parser import * from MarketPlaces.Classifier.classify_product import predict @@ -139,14 +140,16 @@ def new_parse(marketPlace, url, createLog): rmm = darkfox_description_parser(soup) elif marketPlace == "Tor2door": rmm = tor2door_description_parser(soup) - + elif marketPlace == "Apocalypse": + rmm = apocalypse_description_parser(soup) + # key = u"Pr:" + rmm[0].upper()[:desc_lim1] + u" Vendor:" + rmm[13].upper()[:desc_lim2] key = u"Url:" + os.path.basename(line2).replace(".html", "") # save file address with description record in memory detPage[key] = {'rmm': rmm, 'filename': os.path.basename(line2)} - except: + except: nError += 1 print("There was a problem to parse the file " + line2 + " in the Description section!") @@ -186,6 +189,8 @@ def new_parse(marketPlace, url, createLog): rw = darkfox_listing_parser(soup) elif marketPlace == "Tor2door": rw = tor2door_listing_parser(soup) + elif marketPlace == "Apocalypse": + rw = apocalypse_listing_parser(soup) else: parseError = True diff --git a/MarketPlaces/LionMarketplace/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/LionMarketplace/__pycache__/crawler_selenium.cpython-311.pyc index 9231c84..fe2d7c4 100644 Binary files a/MarketPlaces/LionMarketplace/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/LionMarketplace/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/LionMarketplace/__pycache__/parser.cpython-311.pyc b/MarketPlaces/LionMarketplace/__pycache__/parser.cpython-311.pyc index f67444e..b43a4c8 100644 Binary files a/MarketPlaces/LionMarketplace/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/LionMarketplace/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/M00nkeyMarket/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/M00nkeyMarket/__pycache__/crawler_selenium.cpython-311.pyc index ec7c591..1ea14d0 100644 Binary files a/MarketPlaces/M00nkeyMarket/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/M00nkeyMarket/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/M00nkeyMarket/__pycache__/parser.cpython-311.pyc b/MarketPlaces/M00nkeyMarket/__pycache__/parser.cpython-311.pyc index 215d3f2..77a5388 100644 Binary files a/MarketPlaces/M00nkeyMarket/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/M00nkeyMarket/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/MikesGrandStore/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/MikesGrandStore/__pycache__/crawler_selenium.cpython-311.pyc index ba35f17..0a45291 100644 Binary files a/MarketPlaces/MikesGrandStore/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/MikesGrandStore/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/MikesGrandStore/__pycache__/parser.cpython-311.pyc b/MarketPlaces/MikesGrandStore/__pycache__/parser.cpython-311.pyc index 6f1f497..804ba0c 100644 Binary files a/MarketPlaces/MikesGrandStore/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/MikesGrandStore/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/ThiefWorld/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/ThiefWorld/__pycache__/crawler_selenium.cpython-311.pyc index 42d4131..3dc0317 100644 Binary files a/MarketPlaces/ThiefWorld/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/ThiefWorld/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/ThiefWorld/__pycache__/parser.cpython-311.pyc b/MarketPlaces/ThiefWorld/__pycache__/parser.cpython-311.pyc index eff2b39..1e8dc5a 100644 Binary files a/MarketPlaces/ThiefWorld/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/ThiefWorld/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Tor2door/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/Tor2door/__pycache__/crawler_selenium.cpython-311.pyc index 3c35890..46ed74e 100644 Binary files a/MarketPlaces/Tor2door/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/Tor2door/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/Tor2door/__pycache__/parser.cpython-311.pyc b/MarketPlaces/Tor2door/__pycache__/parser.cpython-311.pyc index 3f4dd22..a6b28ec 100644 Binary files a/MarketPlaces/Tor2door/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/Tor2door/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Tor2door/crawler_selenium.py b/MarketPlaces/Tor2door/crawler_selenium.py index b0c5a6b..ea509cf 100644 --- a/MarketPlaces/Tor2door/crawler_selenium.py +++ b/MarketPlaces/Tor2door/crawler_selenium.py @@ -29,19 +29,19 @@ baseURL = 'http://yzrrne3pveltulbavydr2kiashvlnysdwclwmklo6cyjuqpxi7ku4xqd.onion # Opens Tor Browser, crawls the website def startCrawling(): - # opentor() - marketName = getMarketName() - # driver = getAccess() - # - # if driver != 'down': - # try: - # login(driver) - # crawlForum(driver) - # except Exception as e: - # print(driver.current_url, e) - # closetor(driver) - # - new_parse(marketName, baseURL, False) + opentor() + # marketName = getMarketName() + driver = getAccess() + + if driver != 'down': + try: + login(driver) + crawlForum(driver) + except Exception as e: + print(driver.current_url, e) + closetor(driver) + + # new_parse(marketName, baseURL, False) # Opens Tor Browser diff --git a/MarketPlaces/TorBay/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/TorBay/__pycache__/crawler_selenium.cpython-311.pyc index 98e1606..5d4bbfc 100644 Binary files a/MarketPlaces/TorBay/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/TorBay/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/TorBay/__pycache__/parser.cpython-311.pyc b/MarketPlaces/TorBay/__pycache__/parser.cpython-311.pyc index 430b2f0..990e55e 100644 Binary files a/MarketPlaces/TorBay/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/TorBay/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/TorMarket/__pycache__/crawler_selenium.cpython-311.pyc b/MarketPlaces/TorMarket/__pycache__/crawler_selenium.cpython-311.pyc index 5c200d1..3f6c54d 100644 Binary files a/MarketPlaces/TorMarket/__pycache__/crawler_selenium.cpython-311.pyc and b/MarketPlaces/TorMarket/__pycache__/crawler_selenium.cpython-311.pyc differ diff --git a/MarketPlaces/TorMarket/__pycache__/parser.cpython-311.pyc b/MarketPlaces/TorMarket/__pycache__/parser.cpython-311.pyc index 5cb0759..175e17a 100644 Binary files a/MarketPlaces/TorMarket/__pycache__/parser.cpython-311.pyc and b/MarketPlaces/TorMarket/__pycache__/parser.cpython-311.pyc differ diff --git a/MarketPlaces/Utilities/__pycache__/__init__.cpython-311.pyc b/MarketPlaces/Utilities/__pycache__/__init__.cpython-311.pyc index 3c8c6f8..85cec09 100644 Binary files a/MarketPlaces/Utilities/__pycache__/__init__.cpython-311.pyc and b/MarketPlaces/Utilities/__pycache__/__init__.cpython-311.pyc differ diff --git a/MarketPlaces/Utilities/__pycache__/utilities.cpython-311.pyc b/MarketPlaces/Utilities/__pycache__/utilities.cpython-311.pyc index 812e268..933ba93 100644 Binary files a/MarketPlaces/Utilities/__pycache__/utilities.cpython-311.pyc and b/MarketPlaces/Utilities/__pycache__/utilities.cpython-311.pyc differ diff --git a/MarketPlaces/__pycache__/__init__.cpython-311.pyc b/MarketPlaces/__pycache__/__init__.cpython-311.pyc index 164969b..af5e022 100644 Binary files a/MarketPlaces/__pycache__/__init__.cpython-311.pyc and b/MarketPlaces/__pycache__/__init__.cpython-311.pyc differ