Esempi in Python per crawlingSites

Esempio n. 1

File: zeroboard.py Progetto: nathnatica/bin

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys
import commonImgCrawler

imgUrlPattern = r'zb_target_resize.*src="?(https?://[0-9a-zA-Z\_\.\/]+)"?.*?>'


if len(sys.argv) == 4 :
        commonImgCrawler.crawlingSite('zeroboard.py', sys.argv[1], int(sys.argv[2]), int(sys.argv[3]), imgUrlPattern)
elif len(sys.argv) == 1 :
	commonImgCrawler.crawlingSites('zeroboard.py', imgUrlPattern)
        pass

Esempio n. 2

File: tistory2.py Progetto: nathnatica/bin

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys
import commonImgCrawler

imgUrlPattern = r'src="(http://[0-9a-z\.]+/(image|original)/[0-9A-Z]+)"'

if len(sys.argv) == 4:
    commonImgCrawler.crawlingSite("tistory2.py", sys.argv[1], int(sys.argv[2]), int(sys.argv[3]), imgUrlPattern)
elif len(sys.argv) == 1:
    commonImgCrawler.crawlingSites("tistory2.py", imgUrlPattern)
    pass

Esempio n. 3

File: tistory3.py Progetto: nathnatica/bin

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys
import commonImgCrawler

imgUrlPattern = r'src="(http://[0-9a-z\.]+/(image|original)/[0-9A-Z]+)"'

if len(sys.argv) == 4 :
	commonImgCrawler.crawlingSite('tistory3.py', sys.argv[1], int(sys.argv[2]), int(sys.argv[3]), imgUrlPattern)
elif len(sys.argv) == 1 :
	commonImgCrawler.crawlingSites('tistory3.py', imgUrlPattern)
	pass