forumscraper aims to be an universal, automatic and extensive scraper for forums.
pip install forumscraper
- Invision Power Board (only 4.x version)
- PhpBB (currently excluding 1.x version)
- Simple Machines Forum
- XenForo
- XMB
- Hacker News (has aggressive protection, should be used with cookies from logged in account)
- StackExchange
Are created by create-format-examples script and contained in examples directory where they're grouped based on scraper and version. Files are in json format.
Download any kind of supported forums from URLs into DIR, creating json files for threads named by their id's, same for users but beginning with m- e.g. 24 29 m-89 m-125.
forumscraper --directory DIR URL1 URL2 URL3
Above behaviour is set by default --names id, and can be changed with --names hash which names files by sha256 sum of their source urls.
forumscraper --names hash --directory DIR URL
By default if files to be created are found and are not empty, function exits not overwriting them. This can be changed using --force option.
forumscraper output logging information to stdout (can be changed with --log FILE) and information about failures to stderr (can be changed with --failed FILE)
Failures are generally ignored but setting --pedantic flag stops the execution if any failure is encountered.
Download URLs into DIR using 8 threads and log failures into failures.txt (note that this option is specified before the --directory option, otherwise it would create it in the specified directory if relative path is used)
forumscraper --failures failures.txt --threads 8 --directory DIR URL1 URL2 URL3
Download URLs with different scrapers
forumscraper URL1 smf URL2 URL3 .thread URL4 xenforo2.forum URL5 URL6
Type of scrapers can be defined inbetween URLs, where all following URLs are assigned to previous type i.e. URL1 to default, URL2 and URL3 to smf and so on.
Type consists of scraper_name followed by . and function_name.
scraper_name can be: all, invision, phpbb, hackernews, stackexchange, smf, smf1, smf2, xenforo, xenforo1, xenforo2, xmb where all, xenforo and smf are instances of identification class meaning that they have to download the URL to identify its type which may cause redownloading of existing content if many URLs are passed as arguments i.e. all resources extracted from once identified type are assumed to have the same type, but passing thousands of thread URLs as arguments will always download them before scraping. smf1, smf2, xenforo1, xenforo2 are just scrapers with assumed version.
function_name can be: guess, findroot, thread, user, forum, tag, board (board being the main page of the forum where subforums are listed). guess guesses the other types based on the URLs alone, findroot find the main page of forum from any link on site (useful for downloading the whole forum from random urls), other names are self explainatory.
all, xenforo and smf have also identify function that identifies site type.
findroot and identify write results to file specified by the --output (by default set to stdout) option which is made specifically for these functions. findroot return url to board and url from which it was found, separated by \t. identify return name of scraper and url from which it was identified, separated by \t.
Default type is set to all.guess and it is so efective that the only reason to not use it is to avoid redownloading from running the same command many times which is caused by identification process when using --names id.
Types can also be shortened e.g. . is equivalent to all.guess, .thread is equivalent to all.thread and xenforo is equivalent to xenforo.guess.
Get version
forumscraper --version
Get some help (you might discover that many options are abbreviated to single letter)
forumscraper --help
Download URL with waiting 0.8 seconds and randomly waiting up to 400 miliseconds for each request
forumscraper --wait 0.8 --wait-random 400 URL
Download URL using 5 retries and waiting 120 seconds between them
forumscraper --retries 5 --retry-wait 120 URL
By default when encountered a non fatal failure (e.g. status code 301 and not 404) forumscraper tries 3 times waiting 60 seconds before the next attempt, setting --retries 0 would disable retries and it's a valid (if not better) method assuming that one handles the --failures option correctly.
Download URL ignoring ssl errors with timeout set to 60 seconds and custom user-agent
forumscraper --insecure --timeout 60 --user-agent 'why are we still here?'
--proxies DICT, --headers DICT and --cookies DICT (where DICT is python stringified dictionary) are directly passed to requests library.
--nothreads doesn't download threads unless url passed is a thread.
--users download users.
--reactions download reactions.
--boards creates board files.
--tags creates tags files.
--forums creates forums files.
--compression ALGO compresses created files with ALGO, that can be none, gzip, bzip2, lzma.
--only-urls-forums writes found forum urls to output, doesn't scrape.
--only-urls-threads writes found thread urls to output, doesn't scrape.
--thread-pages-max NUM and --pages-max NUM set max number of pages traversed in each thread and forum respectively.
--pages-max-depth NUM sets recursion limit for forums.
--pages-forums-max NUM limits number of forums that are processed from every page in forum.
--pages-threads-max NUM limits number of threads that are processed from every page in forum.
Combining some of the above you get:
forumscraper --thread-pages-max 1 --pages-max 1 --pages-forums-max 1 --pages-threads-max 1 URL1 URL2 URL3
which downloads only one page in one thread from one forum found from every URL which is very useful for debugging.
import os
import sys
import forumscraper
ex = forumscraper.Extractor(timeout=90)
thread = ex.guess('https://xenforo.com/community/threads/forum-data-breach.180995/',output=forumscraper.Outputs.data|forumscraper.Outputs.threads,timeout=60,retries=0) #automatically identify forum and type of page and save results
thread['data']['threads'][0] #access the result
thread['data']['users'] #found users are also saved into an array
forum = ex.get_forum('https://xenforo.com/community/forums/off-topic.7/',output=forumscraper.Outputs.data|forumscraper.Outputs.urls|forumscraper.Outputs.threads,retries=0) #get list of all threads and urls from forum
forum['data']['threads'] #access the results
forums['urls']['threads'] #list of urls to found threads
forums['urls']['forums'] #list of urls to found forums
threads = ex.smf.get_forum('https://www.simplemachines.org/community/index.php?board=1.0',output=forumscraper.Outputs.only_urls_threads) #gather only urls to threads without scraping data
threads['urls']['threads']
threads['urls']['forums'] #is also created
forums = ex.smf.get_board('https://www.simplemachines.org/community/index.php',output=forumscraper.Outputs.only_urls_forums) #only get a list of urls to all forums
threads['urls']['forums']
threads['urls']['boards']
threads['urls']['tags'] #tags and boards are also gathered
ex.smf.get_thread('https://www.simplemachines.org/community/index.php?topic=578496.0',output=forumscraper.Outputs.only_urls_forums) #returns None
os.mkdir('xenforo')
os.chdir('xenforo')
xen = forumscraper.xenforo2(timeout=30,retries=3,retry_wait=10,wait=0.4,random_wait=400,max_workers=8,output=forumscraper.Outputs.write_by_id|forumscraper.Outputs.threads)
#specifies global config, writes output in files by their id (beginning with m- in case of users) in current directory
#ex.xenforo.v2 is an initialized instance of forumscraper.xenforo2 with the same settings as ex
#output by default is set to forumscraper.Outputs.write_by_id|forumscraper.Outputs.threads anyway
failures = []
files = xen.guess('https://xenforo.com/community/',logger=sys.stdout,failed=failures, undisturbed=True)
#failed=failures writes all the failed requests to be saved in failures array or file
for i in failures: #try to download failed one last time
x = i.split(' ')
if len(x) == 4 and x[1] == 'failed':
xen.get_thread(x[0],state=files) #append results
files['files']['threads']
files['files']['users'] #lists of created files
#the above uses scraper that is an instance of ForumExtractor
#if the instance of ForumExtractorIdentify before checking if the files already exist based on url the page has to be downloaded to be indentified. Because of that any getters from this class returns results with 'scraper' field pointing to the indentified scraper type and further requests should be done through that object.
xen = forumscraper.xenforo2(timeout=30,retries=3,retry_wait=10,wait=0.4,random_wait=400,max_workers=8,output=forumscraper.Outputs.write_by_hash|forumscraper.Outputs.threads,undisturbed=True)
#specifies global config, writes output in files by sha256 hash of their url in current directory
#ex.xenforo is also an initialized forumscraper.xenforo
failures = []
files = xen.guess('https://xenforo.com/community/',logger=sys.stdout,failed=failures)
scraper = files['scraper'] #identified ForumScraper instance
for i in failures: #try to download failed one last time
x = i.split(' ')
if len(x) == 4 and x[1] == 'failed':
scraper.get_thread(x[0],state=files) #use of already identified class
os.chdir('..')forumscraper defines:
invision
phpbb
smf1
smf2
xenforo1
xenforo2
xmb
hackernews
stackexchange
scrapers that are instances of ForumExtractor class and also:
Extractor
smf
xenforo
that are instances of ForumExtractorIdentify.
Instances of ForumExtractorIdentify identify and pass requests to ForumExtractor instances in them. This means that content from the first link is downloaded regardless if files with finished work exist. (So running get_thread method on failures using these scrapers will cause needless redownloading, unless forumscraper.Outputs.write_by_hash is used)
Extractor scraper has invision, phpbb, smf, xenforo, xmb, hackernews, stackexchange fields that are already initialized scrapers of declared type.
xenforo and smf have v1 and v2 fields that are already initialized scrapers of declared versions.
Initialization of scrapers allows to specify **kwargs as settings that are kept for requests made from these scrapers.
All scrapers have the following methods:
guess
findroot
get_thread
get_user
get_forum
get_tag
get_board
ForumExtractorIdentify scrapers additionally have identify method.
which take as argument url, optionally already downloaded html either as str, bytes or reliq and state which allows to append output to previous results, and the same type of settings used on initialization of class, e.g.
ex = forumscraper.Extractor(headers={"Referer":"https://xenforo.com/community/"},timeout=20)
state = ex.guess('https://xenforo.com/community/threads/selling-and-buying-second-hand-licenses.131205/',timeout=90)
html = requests.get('https://xenforo.com/community/threads/is-it-possible-to-set-up-three-websites-with-a-second-hand-xenforo-license.222507/').text
ex.guess('https://xenforo.com/community/threads/is-it-possible-to-set-up-three-websites-with-a-second-hand-xenforo-license.222507/',html,state,timeout=40)guess method identifies based only on the url what kind of page is being passed and calls other methods so other methods are needed mostly for exceptions.
For most cases using Extractor and guess is preferred since they work really well. The only exceptions are if site has irregular urls so that guess doesn't work, or if you make a lot of calls to the same site with output=forumscraper.Outputs.write_by_id e.g. trying to scraper failed urls.
guess method creates scraper-method field in output that is pointing to function used.
Methods called from instances of ForumExtractorIdentify do the same, but also create scraper field pointing to instance of ForumExtractor used. This allows to circumvent the need of redownloading for each call just for identification.
failures = []
results = ex.guess('https://www.simplemachines.org/community/index.php',output=forumscraper.Outputs.urls|forumscraper.Outputs.data,failed=failures,undisturbed=True)
#results['scraper-method'] points to ex.smf.v2.get_board
scraper = results['scraper'] #points to ex.smf.v2
for i in failures: #try to download failed one last time
x = i.split(' ')
if len(x) == 4 and x[1] == 'failed':
scraper.get_thread(x[0],state=results) #save results in 'results'identify and findroot methods ignore state, even though they can take it as argument.
findroot method returns None on failure or url to the root of the site (i.e. board) from any link of site, that is very useful when having some random urls and wanting to automatically download the whole forum.
identify methods returns None on failure or initialized ForumExtractor that can scrape given url.
The get functions and guess return None in case of failure or dict defined as
{
'data': {
"boards": [],
"tags": [],
"forums": [],
'threads': [],
'users': []
},
'urls': {
'threads': [],
'users': [],
'reactions':[]
'forums': [],
'tags': [],
'boards': []
}
'files': {
"boards": [],
"tags": [],
"forums": [],
'threads': [],
'users': []
},
'visited': set(),
"scraper": None,
"scraper-method": None,
}
Where data field contains resulting dictionaries of data.
urls field contains found urls of specific type.
file field contains created files with results.
visited field contains every url visited by scraper, which will refuse to visit them again, see force setting for more info.
At initialization of scrapers and use of get_ methods you can specify the same settings.
output=forumscraper.Outputs.write_by_id|forumscraper.Outputs.urls|forumscraper.Outputs.threads changes behaviour of scraper and results returned by id. It takes flags from forumscraper.Outputs:
write_by_id- write results in json in files named by their id (beginning withm-in case of users) e.g2129m-24m-281write_by_hash- write results in json in files named by sha256 hash of their source urlonly_urls_threads- do not scrape, just get urls to threads and things above themonly_urls_forums- ignore everything logging only urls to found forums, tags and boardsurls- save url from which resources were scrapeddata- save results in python dictionarythreads- scrape threadsusers- scrape usersreactions- scrape reactions in threadsboards- scrape boardsforums- scrape forumstags- scrape tags
Disabling users and reactions greatly speeds up getting xenforo and invision threads.
boards forums and tags create files with names beginning with respectively b-, f-, t- followed by sha256 hash of source url. These options may be useful for getting basic information about threads without downloading them.
logger=None, failed=None can be set to list or file to which information will be logged.
logger logs only urls that are downloaded.
failed logs failures in format:
RESOURCE_URL failed STATUS_CODE FAILED_URL
RESOURCE_URL failed completely STATUS_CODE FAILED_URL
Resource fails completely only because of STATUS_CODE e.g. 404.
undisturbed=False if set, scraper doesn't care about standard errors.
pedantic=False if set, scraper fails because of errors in scraping resources related to currently scraped e.g. if getting users of reactions fails.
force=False if set, scraper overwrites files, but will still refuse to scrape urls found in visited field of state, if you are passing state between functions and you want to redownload them you will have to set it to empty set e.g. state['visited'] = set() before every function call.
max_workers=1 set number of threads used for scraping.
compress_func=None set compression function that will be called when writing to files, function should accept data in bytes as the first argument, e.g. gzip.compress.
verify=True if set to False ignore ssl errors.
timeout=120 request timeout.
proxies={} requests library proxies dictionary.
headers={} requests library headers dictionary.
cookies={} requests library cookies dictionary.
user_agent=None custom user-agent.
wait=0 waiting time for each request.
wait_random=0 random waiting time up to specified miliseconds.
retries=3 number of retries attempted in case of failure.
retry_wait=60 waiting time between retries.
thread_pages_max=0 if greater than 0 limits number of pages traversed in threads.
pages_max=0 limits number of pages traversed in each forum, tag or board.
pages_max_depth=0 sets recursion limit for forums, tags and boards.
pages_forums_max=0 limits number of forums that are processed from every page in forum or board.
pages_threads_max=0 limits number of threads that are processed from every page in forum or tag.