ats_cache_scan.py
ActivePublic
Actions

Authored by • ema on Mar 15 2019, 10:33 AM.

Tags

None

Referenced Files

	F28392982: raw.txt
	Mar 15 2019, 2:38 PM

	F28392422: raw.txt
	Mar 15 2019, 10:33 AM

Subscribers

None

	#!/usr/bin/python3 -OO

	import os
	import sys
	import psutil

	from scan import span

	def all_urls(span_path="/dev/sda3"):
	s = span.Span(span_path)

	for counter, obj in enumerate(s.storedObjects()):
	#print("=>", obj)
	if counter % 1000 == 0:
	print(counter)

	print("Finished, total =", counter)

	def first_docs(span_path="/dev/sda3", first=1000):
	max_length = 0

	s = span.Span(span_path)
	for counter, block in enumerate(s.blocks):
	block.readDir()
	for docnum, doc in enumerate(block.firstDocs):
	# doc.totalLength is Content-Length
	# doc.data is the actual content
	# str(doc.urls()[0]) is the remapped URL
	# doc.alternates[0].requestHeaders (as bytearray)
	# doc.alternates[0].responseHeaders
	if doc.totalLength > max_length:
	max_length = doc.totalLength

	if docnum == first:
	break

	print("Max length out of {} objects: {} bytes".format(docnum, max_length))

	if __name__ == "__main__":
	# Lowest possible I/O priority to avoid interfering with actual ATS read/writes
	p = psutil.Process(os.getpid())
	p.ionice(psutil.IOPRIO_CLASS_IDLE)

	print("DEBUG ->", __debug__)

	first_docs("/dev/sdb3", first=10000000)

ats_cache_scan.pyActivePublicActions