zz

mongo/cli.py

@@ -32,9 +32,10 @@ def classify(start_link):
     mongocrawler.classify(start_link)
 
 @cli.command()
-@click.argument("hostname")
-def visit(hostname):
-    mongocrawler.visit(hostname)
+@click.argument("hostname",help="Hostname to crawl")
+@click.option("--filter_content",default=True,help="Filter content")
+def visit(hostname,filter_content=True):
+    mongocrawler.visit(hostname,filter_content=filter_content)
 
 @cli.command()
 def summary():

mongo/mongocrawler.py

@@ -24,24 +24,28 @@ import hashlib
 from bs4 import BeautifulSoup
 import urllib.parse
 import os.path
+import binascii
+import json
 
-LANGUAGE= os.getenv("SUCKER_LANGUAGE","sk")
-DOMAIN = os.getenv("SUCKER_DOMAIN","sk")
-BATCHSIZE=int(os.getenv("SUCKER_BATCHSIZE","10"))
+# database options
 CONNECTION=os.getenv("SUCKER_CONNECTION","mongodb://root:example@localhost:27017/")
 DBNAME=os.getenv("SUCKER_DBNAME","crawler")
-MINFILESIZE=300
-MAXFILESIZE=10000000
-MINTEXTSIZE=200
-CHECK_PARAGRAPH_SIZE=200
-TEXT_TRASH_SIZE=200
-TEXT_TRASH_RATIO=0.6
-DISCOVER_LINK_RATIO = 0.3
-DISCOVER_DOMAIN_RATIO = 0.5
-SAMPLE_SET_SIZE =10000
-CLASSIFIER_SET_SIZE = 200
-STOP_PATHS=["xml","rss","login","admin"]
-
+# retrieving filter
+BATCH_SIZE = int(os.getenv("SUCKER_BATCH_SIZE","10"))
+MIN_FILE_SIZE=int(os.getenv("SUCKER_MIN_FILE_SIZE","300"))
+MAX_FILE_SIZE=int(os.getenv("SUCKER_MAX_FILE_SIZE","10000000"))
+# document originality filter
+MIN_TEXT_SIZE=int(os.getenv("SUCKER_MIN_TEXT_SIZE","200"))
+CHECK_PARAGRAPH_SIZE=int(os.getenv("SUCKER_CHECK_PARAGRAPH_SIZE","150"))
+TEXT_TRASH_RATIO=float(os.getenv("SUCKER_TEXT_TRASH_RATIO","0.6"))
+# link and domain sampling
+DISCOVER_LINK_RATIO = float(os.getenv("SUCKER_DISCOVER_LINK_RATIO","0.3"))
+SAMPLE_SET_SIZE = int(os.getenv("SUCKER_DISCOVER_LINK_RATIO","10000"))
+CLASSIFIER_SET_SIZE = int(os.getenv("SUCKER_DISCOVER_LINK_RATIO","200"))
+# link filter
+LANGUAGE= os.getenv("SUCKER_LANGUAGE","sk")
+DOMAIN = os.getenv("SUCKER_DOMAIN","sk")
+STOP_PATHS=os.getenv("SUCKER_STOP_PATHS","xml,rss,login,admin").split(",")
 
 def get_bs_links(link,html):
     # Extrakcia linkov zo stranky
@@ -165,14 +169,15 @@ def fetch_page(link:str)->(str,str):
     html = None
     if response is not None :
         good = True
+        print(response)
         if response.status != 200:
             good = False
             LOGGER.error('not a 200 response: %s for URL %s', response.status, url)
-        elif response.data is None or len(response.data) < MINFILESIZE:
+        elif response.data is None or len(response.data) < MIN_FILE_SIZE:
             LOGGER.error('too small/incorrect for URL %s', link)
             good = False
         # raise error instead?
-        elif len(response.data) > MAXFILESIZE:
+        elif len(response.data) > MAX_FILE_SIZE:
             good = False
             LOGGER.error('too large: length %s for URL %s', len(response.data), link)
         if good:
@@ -204,7 +209,7 @@ def extract_page(final_link,html):
     if html is not None:
         doc = trafilatura.bare_extraction(html,url=final_link,with_metadata=True,include_formatting=False,target_language=LANGUAGE,favor_precision=True)
         if doc is not None:
-            if not "text" in doc or len(doc["text"]) < MINTEXTSIZE:
+            if not "text" in doc or len(doc["text"]) < MIN_TEXT_SIZE:
                 # text too small
                 doc = None
     return doc
@@ -226,7 +231,7 @@ def set_content_checksums(doc):
             sentences += 1
     doc["sentences_count"] = sentences
 
-def index_page(db,original_link,final_link,html,doc):
+def index_page(db,original_link,final_link,html,doc,filter_content=True):
     linkcol = db["links"]
     htmlcol = db["html"]
     contentcol = db["content"]
@@ -244,7 +249,7 @@ def index_page(db,original_link,final_link,html,doc):
         set_content_checksums(doc)
         tsz = doc["text_size"]
         psz = doc["paragraph_sizes_sum"]
-        if tsz < TEXT_TRASH_SIZE or psz/tsz < TEXT_TRASH_RATIO:
+        if filter_content and (tsz < MIN_TEXT_SIZE or psz/tsz < TEXT_TRASH_RATIO):
             state = "small"
     # check copy
     if state == "good":
@@ -256,7 +261,7 @@ def index_page(db,original_link,final_link,html,doc):
                 origsz += paragraph_size
         doc["original_text_size"] = origsz
 
-        if (1 - (origsz / tsz)) > TEXT_TRASH_RATIO:
+        if filter_content and (1 - (origsz / tsz)) > TEXT_TRASH_RATIO:
             state = "copy"
     if state == "good":
         htdoc = get_link_doc(link,state)
@@ -662,10 +667,10 @@ def classify(start_link):
     cl.train(trainset)
     cl.test(testset)
 
-def visit(hostname):
+def visit(hostname,filter_content=True):
     myclient = pymongo.MongoClient(CONNECTION)
     db=myclient[DBNAME]
-    batch_size = BATCHSIZE
+    batch_size = BATCH_SIZE
     rules = fetch_robot(hostname)
     start_link = "https://" + hostname
     # renew front links
@@ -696,7 +701,7 @@ def visit(hostname):
     final_states = []
     docs = []
     for original_link,final_link,html,doc in extracted_pages:
-        status = index_page(db,original_link,final_link,html,doc)
+        status = index_page(db,original_link,final_link,html,doc,filter_content)
         final_states.append(status)
         docs.append(doc)
     save_batch_info(db,hostname,final_states,docs)
@@ -726,8 +731,6 @@ def crawl_summary():
         values = [str(item[x]) for x in headers]
         print("\t".join(values))
 
-import binascii
-import json
 
 def import_html():
     myclient= pymongo.MongoClient(CONNECTION)
@@ -751,7 +754,7 @@ def sample_domains():
     all_domains = []
     for domain in domains:
         all_domains.append(domain)
-    sample_size = min(int(DISCOVER_DOMAIN_RATIO* BATCHSIZE), len(all_domains))
+    sample_size = min(int(DISCOVER_LINK_RATIO* BATCH_SIZE), len(all_domains))
     print(">>> Discover domains {}".format(sample_size))
     sample_domains = random.sample(all_domains,sample_size)
     domaincol = db["domains"]
@@ -760,7 +763,7 @@ def sample_domains():
     all_domains = []
     for item in res:
         all_domains.append(item["host"])
-    sample_size = min(int((1 - DISCOVER_DOMAIN_RATIO) * BATCHSIZE),len(all_domains))
+    sample_size = min(int((1 - DISCOVER_LINK_RATIO) * BATCH_SIZE),len(all_domains))
     print(">>>> Best domains {}".format(sample_size))
     sample_domains += random.sample(all_domains,sample_size)
     for domain in sample_domains: