|
@ -181,12 +181,16 @@ def getNameFromURL(url): |
|
|
def getInterestedLinks(): |
|
|
def getInterestedLinks(): |
|
|
links = [] |
|
|
links = [] |
|
|
|
|
|
|
|
|
# # cyber security |
|
|
|
|
|
|
|
|
# cybersecurity |
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/CyberSecurity') |
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/CyberSecurity') |
|
|
# # services |
|
|
|
|
|
# links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/Services') |
|
|
|
|
|
# # programming |
|
|
|
|
|
# links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/Programming') |
|
|
|
|
|
|
|
|
# services |
|
|
|
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/Services') |
|
|
|
|
|
# programming |
|
|
|
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/Programming') |
|
|
|
|
|
# jobs for crypto |
|
|
|
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/JobsforCypto') |
|
|
|
|
|
# darknet markets |
|
|
|
|
|
links.append('http://libreeunomyly6ot7kspglmbd5cvlkogib6rozy43r2glatc6rmwauqd.onion/c/DarkNetMarkets') |
|
|
|
|
|
|
|
|
return links |
|
|
return links |
|
|
|
|
|
|
|
@ -226,9 +230,9 @@ def crawlForum(driver): |
|
|
driver.refresh() |
|
|
driver.refresh() |
|
|
savePage(driver, driver.page_source, topic + f"page{counter}") # very important |
|
|
savePage(driver, driver.page_source, topic + f"page{counter}") # very important |
|
|
|
|
|
|
|
|
# comment out |
|
|
|
|
|
if counter == 2: |
|
|
|
|
|
break |
|
|
|
|
|
|
|
|
# # comment out |
|
|
|
|
|
# if counter == 2: |
|
|
|
|
|
# break |
|
|
|
|
|
|
|
|
try: |
|
|
try: |
|
|
page = "" # no next page so far may have some later on |
|
|
page = "" # no next page so far may have some later on |
|
@ -242,12 +246,12 @@ def crawlForum(driver): |
|
|
for j in range(counter): |
|
|
for j in range(counter): |
|
|
driver.back() |
|
|
driver.back() |
|
|
|
|
|
|
|
|
# comment out |
|
|
|
|
|
# break |
|
|
|
|
|
|
|
|
|
|
|
# comment out |
|
|
|
|
|
if count == 1: |
|
|
|
|
|
break |
|
|
|
|
|
|
|
|
# # comment out |
|
|
|
|
|
# break |
|
|
|
|
|
# |
|
|
|
|
|
# # comment out |
|
|
|
|
|
# if count == 1: |
|
|
|
|
|
# break |
|
|
|
|
|
|
|
|
try: |
|
|
try: |
|
|
link = driver.find_element(by=By.LINK_TEXT, value='>').get_attribute('href') |
|
|
link = driver.find_element(by=By.LINK_TEXT, value='>').get_attribute('href') |
|
|