zz

mongo/mongocwarler.py

@@ -39,23 +39,35 @@ def calculate_checksums(text):
         sizes.append(sz)
     return checksums, sizes
 
-def filter_links(links,language=LANGUAGE,rules=None):
+def is_robot_good(link,rules):
+    # check robots.txt rules
+    if rules is not None and not rules.can_fetch("*", llink):
+        return False
+    return True
+
+def is_link_good(link):
+    r = courlan.check_url(link,strict=True,language=language)
+    if r is None:
+        return None
+    llink,ldomain = r
+    print(llink,ldomain)
+    # domain rules
+    if not ldomain.endswith("sk"):
+        print("bad domain")
+        return None
+    if courlan.is_not_crawlable(llink):
+        print("not crawlable")
+        return None
+    return None
+
+def filter_links(links,rules=None):
     out = set()
     for link in links:
-        r = courlan.check_url(link,strict=True,language=language)
+        r = is_link_good(link)
         if r is None:
             continue
-        llink,ldomain = r
-        print(llink,ldomain)
-        # domain rules
-        if not ldomain.endswith("sk"):
-            print("bad domain")
-            continue
-        if courlan.is_not_crawlable(llink):
-            print("not crawlable")
-            continue
         # check robots.txt rules
-        if rules is not None and not rules.can_fetch("*", llink):
+        if rules is not None and not rules.can_fetch("*", r):
             continue
         out.add(llink)
     return out
@@ -93,7 +105,12 @@ def fetch_pages(link_batch):
     for link in link_batch:
         print("fetching:::::")
         print(link)
-        htmls.append(trafilatura.fetch_url(link,decode=False))
+        response = trafilatura.fetch_url(link,decode=False)
+        htmlstring, homepage = trafilatura.spider.refresh_detection(response.data, link)
+        # is there a meta-refresh on the page?
+        if homepage is None:  # malformed or malicious content
+            response = None
+        htmls.append(response)
     return htmls
 
 def fetch_rules(base_url):
@@ -107,43 +124,24 @@ def fetch_rules(base_url):
         rules = None
     return rules
 
-def fetch_front_links(start_links,rules):
-    responses = fetch_pages(start_links)
-
-    # decode response
-    htmlstring = trafilatura.utils.decode_response(response.data)
-    # is there a meta-refresh on the page?
-    htmlstring, homepage = trafilatura.spider.refresh_detection(htmlstring, homepage)
-    if homepage is None:  # malformed or malicious content
-        return None, None, None
-    visit_links = courlan.extract_links(htmlstring)
-    visit_links,known_links = trafilatura.spider.focused_crawler(start_link,lang=LANGUAGE)
-    print(visit_links,known_links)
-    filtered_links = filter_links(visit_links,LANGUAGE,rules=rules)
-    return filtered_links
 
 def extract_pages(link_batch,responses):
     out = []
+    extracted_links = set()
     for link,response in zip(link_batch,responses):
         doc = None
         assert link is not None
         html = None
         response_link = None
         if response is not None:
-            # is reponse link good?
-            # filter and normalize
-            rl = list(filter_links([response.url]))
-            if len(rl) == 1:
-                response_link = rl[0]
-                html = trafilatura.utils.decode_response(response) 
+            html = trafilatura.utils.decode_response(response) 
         if html is not None:
             doc = trafilatura.bare_extraction(html,url=link,with_metadata=True,include_formatting=True,target_language=LANGUAGE)
         out.append((link,response_link,html,doc))
     return out
 
 
-
-def index_pages(db,domain,extracted_pages,rules):
+def index_pages(db,domain,extracted_pages):
     extracted_links = set()
     linkcol = db["links"]
     htmlcol = db["html"]
@@ -163,25 +161,46 @@ def index_pages(db,domain,extracted_pages,rules):
             # todo extract links
             print(doc)
             contentcol.insert_one(doc)
-            # todo extract links
-        doc = get_link_doc(original_link,state)
-        linkcol.replace_one({"url":final_link},doc,upsert=True)
-        if original_link != final_link:
-            rl = list(filter_links([final_link],rules))
-            if len(rl) == 1:
-                doc = get_link_doc(rl[0],"redirect")
-                linkcol.replace_one({"url":rl[0]},doc,upsert=True)
 
-def extract_links(extracted_pages,domain,rules,status="frontlink"):
-    filtered_links = filter_links(extracted_links,rules)
-    for original_link,final_link,html,doc in extracted_pages:
-        doc = get_link_doc(link,status)
+
+def extract_links(link_batch,responses,domain,rules,default_status="frontlink"):
+    links = {}
+    for original_link,response in zip(link_batch,resposnes):
+        final_link = response.url
+        status = default_status
+        link = original_link
+        if original_link != final_link:
+            links[original_link] = "redirect"
+            link = final_link
         if courlan.is_external(link,domain):
-            doc["status"]= "frontlink"
+            status = "frontlink"
         elif courlan.is_navigation(link):
-            doc["status"] = "navigation"
+            status = "navigation"
+        links[link] = status
+        extracted_links = courlan.extract_links(response.content)
+        for link in extracted_links
+            if courlan.is_external(link,domain):
+                status = "frontlink"
+            elif courlan.is_navigation(link):
+                status = "navigation"
+            links[link] = status
+    outlinks = []
+    for link,status in links.items():
+        if  not is_robot_good(rules):
+            continue
+        link = is_link_good(link)
+        if link is None:
+            continue
+        outlinks.append((link,status))
+    return outlinks
+
+def index_links(db,extracted_links):
+    linkcol=db["links"]
+    for link,status in extracted_links:
+        doc = get_link_doc(link,status)
         linkcol.insert_one(doc)
 
+
 def get_links(db,domain,status,batch_size=BATCHSIZE):
     linkcol = db["links"]
     res  = linkcol.find({"status":status,"domain":domain},{"url":1},limit=batch_size)
@@ -190,7 +209,7 @@ def get_links(db,domain,status,batch_size=BATCHSIZE):
     for doc in res:
         print(doc)
         front_links.append(doc["url"])
-    return filter_links(front_links)
+    return front_links
 
 
 def index_front_links(db,filtered_links):
@@ -207,13 +226,25 @@ def index_front_links(db,filtered_links):
     #    document = trafilatura.bare_extraction(content)
     #    print(content)
 
+def process_links(status,domain,links=[],rules=None):
+    links += get_links(db,domain,status)
+    responses = fetch_pages(links)
+    extracted_pages = extract_pages(links,responses)
+    extracted_links = extract_links(links,responses,domain,rules,status)
+    index_links(db,extracted_links)
+    index_pages(db,domain,extracted_pages)
+
 def simple_visit(start_link):
     start_link,domain = courlan.check_url(start_link)
     myclient = pymongo.MongoClient("mongodb://root:example@localhost:27017/")
     db=myclient["crawler"]
+    rules = fetch_robot(domain)
     navigation_links =[start_link]
-    navigation_links += get_links(db,domain,"navigation")
     print(navigation_links)
+    process_links(db,domain,"navigation",navigation_links,rules)
+    process_links(db,domain,front_links,rules)
+    back_links = get_links(db,domain,"backlink")
+    process_links(db,domain,front_links,rules)
     #new_front_links = fetch_front_links(navigation_links)
     print("NEW FRONT LINKS")
     #print(new_front_links)
@@ -223,7 +254,5 @@ def simple_visit(start_link):
     visit_links = front_links
     print(visit_links)
     responses = fetch_pages(visit_links)
-    extracted_pages = extract_pages(visit_links,responses)
-    index_pages(db,domain,extracted_pages)
 
 simple_visit(sys.argv[1])