Mercurial > hg > cc > cirrus_work
view lib/python/cc/test_warc.py @ 301:4981c41628dd trim
works
| author | Henry S. Thompson <ht@inf.ed.ac.uk> |
|---|---|
| date | Tue, 13 May 2025 12:04:01 +0100 |
| parents | 12d13a1d387f |
| children | 90f541382b0e |
line wrap: on
line source
import warc,sys OUT=open(sys.stdout.fileno(),'wb') if (debug:=(sys.argv[1]=='-d')): sys.argv.pop(1) tt=int(sys.argv.pop(1)) def showme(wtype,buf,part): # This should exactly reproduce a complete warc file if called # as per version 1 below if debug: OUT.write(b"----start %d-----\n"%part) OUT.write(buf) if buf[-1]!=10: OUT.write(b'\r\n') if part==7: OUT.write(b'\r\n') # to match complete file formatting if debug: OUT.write(b"----end %d-----\n"%part) return OUT if tt==1: warc.warc(sys.argv[1],showme,[1,2,3,0],parts=int(sys.argv[2]),debug=debug) elif tt==2: warc.warc(sys.argv[1],showme,[0],parts=int(sys.argv[2]),debug=debug) elif tt==3: warc.warc(sys.argv[1],showme,[0],whole=True,debug=debug) elif tt==4: warc.warc(sys.argv[1],showme,[1,2,3,0,4],whole=True,debug=debug) elif tt==5: warc.warc(sys.argv[1],showme,[1],parts=int(sys.argv[2]),debug=debug) elif tt==6: warc.warc(sys.argv[1],showme,[4],parts=int(sys.argv[2]),debug=debug)
