Facebook scraping

1 files changed, 182 insertions, 0 deletions

diff --git a/twitter/main.py b/twitter/main.py
new file mode 100644
index 0000000..be565a1
--- /dev/null
+++ b/twitter/main.py
@@ -0,0 +1,182 @@
+from bottle import route, run, request, static_file
+
+from scraper import Driver
+from api import RequestHandler
+
+from multiprocessing import Process, Queue
+from Queue import Empty
+from time import sleep
+from json import dumps
+from uuid import uuid1
+import tarfile
+import os.path
+import os
+from glob import glob
+
+
+long_queue = Queue()
+short_queue = Queue()
+lookup_queue = Queue()
+short_lookup_queue = Queue()
+processes = []
+done_queue = Queue()
+
+
+def start():
+    global long_queue, short_queue, lookup_queue, done_queue, processes,\
+        short_lookup_queue
+    processes = []
+    long_queue = Queue()
+    short_queue = Queue()
+    short_lookup_queue = Queue()
+    lookup_queue = Queue()
+    with open("api_accounts.txt") as f:
+        for line in f:
+            credentials = line.strip().split()[2:]
+            handler = RequestHandler(*credentials)
+            p = Process(target=api_target, args=(handler, long_queue,
+                                                 short_queue,
+                                                 lookup_queue, done_queue))
+            processes.append(p)
+            p.daemon = True
+            p.start()
+
+    with open("scraping_accounts.txt") as f:
+        for line in f:
+            credentials = line.strip().split()[:2]
+            driver = Driver(*credentials)
+            p = Process(target=scraper_target, args=(driver, short_queue,
+                                                     done_queue))
+            processes.append(p)
+            p.daemon = True
+            p.start()
+
+
+@route('/short_lookup', method='POST')
+def short_lookup():
+    query_list = request.forms.list.split(",")
+    user_list = zip(*[iter(query_list)] * 2)  # this is dark magic
+    short_lookup_queue.put(user_list)
+
+
+@route('/restart')
+def restart():
+    global processes
+    for p in processes:
+        p.terminate()
+    start()
+
+
+@route('/long')
+def long():
+    user_id = request.query.id
+    long_queue.put(user_id)
+
+
+@route('/short')
+def short():
+    user_id = request.query.id
+    user_name = request.query.user_name
+    short_queue.put((user_id, user_name))
+
+
+@route('/lookup', method='POST')
+def lookup():
+    id_list = request.forms.list.split(",")
+    lookup_queue.put(id_list)
+
+
+@route('/status')
+def status():
+    answer_dict = {
+        "long": long_queue.qsize(),
+        "short": short_queue.qsize(),
+        "lookup": lookup_queue.qsize(),
+        "short_lookup": short_lookup_queue.qsize(),
+        "done": done_queue.qsize(),
+        "processes": len([p for p in processes if p.is_alive()]),
+        "users": len(glob("data/users/[0-9]*.txt")),
+        "lookups": len(glob("data/users/lookup*.txt"))
+    }
+    return dumps(answer_dict)
+
+
+@route('/fetch')
+def fetch():
+    for filename in glob("data/users/*.tar.gz"):
+        os.remove(filename)
+
+    def get_filenames():
+        try:
+            while True:
+                yield done_queue.get(False)
+        except Empty:
+            pass
+
+    filename = os.path.join("data", "users", "archive-"
+                            + str(uuid1()) + ".tar.gz")
+    with tarfile.open(filename, "w:gz") as tar:
+        for name in get_filenames():
+            tar.add(name)
+    return static_file(filename, root=".")
+
+
+def scraper_target(driver, short_queue, done_queue):
+    while True:
+        try:
+            user_id, user_name = short_queue.get(False)
+        except Empty:
+            pass
+        else:
+            filename = driver.get_followers(user_id, user_name)
+            done_queue.put(filename)
+        finally:
+            sleep(0.5)
+
+
+def api_target(handler, long_queue, short_queue, lookup_queue, done_queue):
+    while True:
+        if handler.ready("followers"):
+            try:
+                user_id = long_queue.get(False)
+            except Empty:
+                try:
+                    user_id = short_queue.get(False)[0]
+                except Empty:
+                    pass
+                else:
+                    filename = handler.get_followers(user_id)
+                    done_queue.put(filename)
+                    continue
+            else:
+                filename = handler.get_followers(user_id)
+                done_queue.put(filename)
+                continue
+        if handler.ready("lookup"):
+            try:
+                users_list = lookup_queue.get(False)
+            except Empty:
+                try:
+                    user_list = short_lookup_queue.get(False)
+                except Empty:
+                    pass
+                else:
+                    filename = handler.lookup(user[0] for user in user_list)
+                    done_queue.put(filename)
+            else:
+                filename = handler.lookup(users_list)
+                done_queue.put(filename)
+        else:
+            try:
+                user_list = short_lookup_queue.get(False)
+            except Empty:
+                pass
+            else:
+                filename = handler.short_lookup(user_list)
+                done_queue.put(filename)
+
+
+if __name__ == "__main__":
+    import sys
+    start()
+    run(host="0.0.0.0", port=int(sys.argv[1]))