websucker-pip/mongo/mongocwarler.py
2023-03-07 08:58:28 +01:00

147 lines
4.1 KiB
Python

import pymongo
import trafilatura
import trafilatura.feeds
import trafilatura.sitemaps
import trafilatura.spider
import sys
import courlan
def calculate_checksums(self, text):
"""
@return fingerprints of a paragraphs in text. Paragraphs are separated by a blank line
"""
checksums = []
sizes = []
hval = 0
hsz = 0
sz = 0
for c in text:
cv = ord(c)
sz += 1
if cv > 64:
hval += (hval << 3) + cv
zv = hval >> 31
hval &= 0x7fffffff
hval += zv
hsz += 1
if c == "\n" and hsz > 0:
if hsz > 100:
checksums.append(hval)
sizes.append(sz)
sz = 0
hsz = 0
if hsz > 100:
checksums.append(hval)
sizes.append(sz)
return checksums, sizes
def extract_pages(link_docs):
content = []
extracted_links = set()
for doc in link_docs:
if doc["status"] != "html_ok":
continue
extracted_doc = trafilatura.bare_extraction(doc["content"],extract_links=True)
links = extracted_doc["links"]
extracted_links += links
del extracted_doc["links"]
content.append(extracted_doc)
return content, extracted_links
def index_pages(pagedb,pages_list):
contentcol = pagedb["content"]
for page in page_list:
# get paragraph checksums
checksums,sizes = get_checksums(page["text"])
page["paragraph_checksums"] = checksums
page["paragraph_sizes"] = sizes
x = contentcol.insert_many(pages_list)
page_hashes = []
def process_pages(db,fetched_pages):
content_pages, extracted_links = extract_pages(fetched_pages)
contentcol = db["content"]
contentcol.insertMany(content_pages)
linkcol = db["links"]
extracted = []
for link in extracted_links:
extracted.append({"url":link,"status":"backlink"})
# shuld fail if link already exists
linkcol.insertMany(extracted)
html_pages = []
for page in fetched_pages:
linkcol.updateOne({"url":page["url"]},{"$set":{"status":"visited"}})
if "html" in page:
html_pages.append({"url":page["url"],"html":page["html"],"update_time":datetime.now()})
del page["html"]
htmlcol = db["html"]
htmlcol.insertMany(html_pages)
def filter_links(links,domain):
out = set()
for link in links:
r = courlan.check_url(link)
if r is None:
continue
link,ldomain = r
# domain rules
if not ldomain.endswith("sk"):
continue
if not courlan.is_crawlable(link):
continue
out.add(link)
return out
def get_visited_links(domain):
return []
def generic_visit(domain):
known_links = set(get_visited_links(domain))
visit_links = []
visit_links = trafilatura.find_feed_urls(domain)
if visit_links is None:
visit_links = trafilatura.sitemap_search(domain)
if visit_links is None:
visit_links = trafilatura.focused_crawler(dommain,known_links=known_links)
def get_new_links(domain):
known_links = []
# get seed links
# get new links
out_links = trafilatura.spider.focused_crawler(domain,known_links=known_links)
filtered_links = filter_links(out_links)
print(res)
return filtered_links
def index_links(db,link_batch):
htmls = []
for link in link_batch:
rr = trafilatura.fetch_url(page,decode=True)
htmls.append(rr)
html_docs = []
link_docs = []
for link,html in zip(link_batch,html):
status = "visited"
if html is None:
status = "html_error"
html_docs.append({"url":link,"html":html})
link_docs.append({"url":link,"status":status})
return docs
#visit_links = trafilatura.feeds.find_feed_urls(domain)
#visit_links = trafilatura.sitemaps.sitemap_search(domain)
#print(visit_links)
#for link in visit_links:
# content = trafilatura.fetch_url(link,decode=True)
# document = trafilatura.bare_extraction(content)
# print(content)
simple_visit(sys.argv[1])