download & save image metadata

chan.py

@@ -87,8 +87,6 @@ class HtmlChanHelper(ChanHelper):
 
     @staticmethod
     def item_mtime(item):
-        if item is None:
-            return int(datetime.datetime.now().timestamp())
         print(item)
         exit(0)
         return 0  # TODO
@@ -114,12 +112,12 @@ class HtmlChanHelper(ChanHelper):
 
         op_el = soup.find("div", attrs={"class": "innerOP"})
         yield {
-            "id": int(soup.find("div", attrs={"class": "opCell"}).get("id")),
+            "id": int(soup.find("div", class_="opCell").get("id")),
             "type": "thread",
             "html": str(op_el),
         }
 
-        for post_el in soup.find_all("div", attrs={"class": "postCell"}):
+        for post_el in soup.find_all("div", class_="postCell"):
             yield {
                 "id": int(post_el.get("id")),
                 "type": "post",

post_process.py

@@ -1,10 +1,77 @@
+import base64
+import hashlib
 import re
+import zlib
+from io import BytesIO
+
+import imagehash
+from PIL import Image
+
+from util import logger
 
 LINK_RE = re.compile(r"(https?://[\w\-_.]+\.[a-z]{2,4}([^\s<'\"]*|$))")
 
+IMAGE_FILETYPES = (
+    # :orig for twitter cdn
+    '.jpg',
+    '.jpg:orig',
+    '.jpeg',
+    '.jpeg:orig',
+    '.png',
+    '.png:orig',
+    '.gif',
+    '.gif:orig',
+    '.tiff',
+    '.bmp',
+    '.webp'
+)
 
-def post_process(item, board, helper):
+
-    item["_v"] = 1.3
+def _is_image(url):
+    return url.lower().endswith(IMAGE_FILETYPES)
+
+
+def b64hash(imhash, bcount):
+    return base64.b64encode(
+        sum(1 << i for i, b in enumerate(imhash.hash.flatten()) if b).to_bytes(bcount, "big")
+    ).decode("ascii")
+
+
+def image_meta(url, url_idx, web):
+    r = web.get(url)
+    if not r:
+        logger.warning("")
+        return None
+    buf = r.content
+
+    try:
+        f = BytesIO(buf)
+        im = Image.open(f)
+    except Exception as e:
+        logger.warning("exception during image open: " + str(e))
+        return None
+
+    meta = {
+        "url": url_idx,
+        "size": len(buf),
+        "width": im.width,
+        "height": im.height,
+        "sha1": hashlib.sha1(buf).hexdigest(),
+        "md5": hashlib.md5(buf).hexdigest(),
+        "crc32": format(zlib.crc32(buf), "x"),
+        "dhash": b64hash(imagehash.dhash(im, hash_size=12), 18),
+        "phash": b64hash(imagehash.phash(im, hash_size=12), 18),
+        "ahash": b64hash(imagehash.average_hash(im, hash_size=12), 18),
+        "whash": b64hash(imagehash.whash(im, hash_size=8), 8),
+    }
+
+    del im, r, buf
+
+    return meta
+
+
+def post_process(item, board, helper, web):
+    item["_v"] = 1.4
     item["_id"] = helper.item_unique_id(item, board)
 
     item["_board"] = board
@@ -12,6 +79,8 @@ def post_process(item, board, helper):
 
     item["_urls"] = helper.item_urls(item, board)
 
+    item["_img"] = [image_meta(url, i, web) for i, url in enumerate(item["_urls"]) if _is_image(url)]
+
     return item
 
 

requirements.txt

@@ -1,3 +1,5 @@
+imagehash
+Pillow
 requests
 requests[socks]
 stem

run.py

@@ -14,7 +14,7 @@ from chan import CHANS
 from post_process import post_process
 from util import logger, Web
 
-MONITORING = False
+MONITORING = True
 
 
 class ChanScanner:
@@ -58,9 +58,9 @@ class ChanScanner:
 
 
 def once(func):
-    def wrapper(item, board, helper):
+    def wrapper(item, board, helper, channel, web):
         if not state.has_visited(helper.item_unique_id(item, board), helper):
-            func(item, board, helper)
+            func(item, board, helper, channel, web)
             state.mark_visited(helper.item_unique_id(item, board), helper)
 
     return wrapper
@@ -139,10 +139,13 @@ class ChanState:
 
 
 def publish_worker(queue: Queue, helper):
+    channel = connect()
+    web = Web(monitoring if MONITORING else None)
+
     while True:
         try:
             item, board = queue.get()
-            publish(item, board, helper)
+            publish(item, board, helper, channel, web)
 
         except Exception as e:
             logger.error(str(e) + ": " + traceback.format_exc())
@@ -151,16 +154,16 @@ def publish_worker(queue: Queue, helper):
 
 
 @once
-def publish(item, board, helper):
+def publish(item, board, helper, channel, web):
     item_type = helper.item_type(item)
-    post_process(item, board, helper)
+    post_process(item, board, helper, web)
 
     while True:
         try:
-            chan_channel.basic_publish(
+            channel.basic_publish(
                 exchange='chan',
                 routing_key="%s.%s.%s" % (chan, item_type, board),
-                body=json.dumps(item)
+                body=json.dumps(item, separators=(',', ':'), ensure_ascii=False, sort_keys=True)
             )
 
             if MONITORING:
@@ -179,14 +182,14 @@ def publish(item, board, helper):
         except Exception as e:
             logger.debug(traceback.format_exc())
             logger.error(str(e))
-            connect()
+            channel = connect()
 
 
 def connect():
-    global chan_channel
     rabbit = pika.BlockingConnection(pika.ConnectionParameters(host=rabbitmq_host))
-    chan_channel = rabbit.channel()
+    channel = rabbit.channel()
-    chan_channel.exchange_declare(exchange="chan", exchange_type="topic")
+    channel.exchange_declare(exchange="chan", exchange_type="topic")
+    return channel
 
 
 if __name__ == "__main__":
@@ -204,11 +207,10 @@ if __name__ == "__main__":
     state = ChanState()
 
     publish_q = Queue()
+    for _ in range(5):
         publish_thread = Thread(target=publish_worker, args=(publish_q, chan_helper))
         publish_thread.start()
 
-    connect()
-
     s = ChanScanner(chan_helper)
     while True:
         for p, b in s.all_posts():