splitqtd.py

#!/bin/python3
import os
import os.path
import re
import argparse

def formatSplitFilepath(filepath, splitIndex):
	splitQtdFormat = filepath.replace('.qtd', '-{}{}.qtd')
	assert splitIndex <= 26*26
	a1 = chr(ord('a') + (splitIndex // 26))
	a2 = chr(ord('a') + (splitIndex % 26))
	return splitQtdFormat.format(a1, a2)

# Using tabs shaves off 2-3 Mb per file
def useTabs(line):
	return re.sub('    ', '\t', line)

def splitFile(filepath, splitIndex, eventData, events, ranges, heapSize, heapUsed):
	# Set <trace> traceStart and traceEnd
	firstRange = ranges[0]
	m = re.search(r'startTime="(\d+)"', firstRange)
	traceStart = m.group(1)

	lastRange = firstRange = ranges[-1]
	m = re.search(r'startTime="(\d+)"', lastRange)
	traceEnd = m.group(1)

	trace = '<trace version="1.02" traceStart="{}" traceEnd="{}">\n'
	trace = trace.format(traceStart, traceEnd)

	xmlSchema = '<?xml version="1.0" encoding="UTF-8"?>\n'
	split = '\t</eventData>\n\t<profilerDataModel>\n'
	foot = '\t</profilerDataModel>\n</trace>\n'

	if splitIndex != 0: # First split should already have starting heap size+used events.
		# We need to add starting heap size/usage or it'll break Qt Creator if more memory
		# is freed than allocated and the graph value goes negative.
		heapRanges = [
			'		<range startTime="{}" eventIndex="0" amount="{}"/>\n'.format(traceStart, heapSize),
			'		<range startTime="{}" eventIndex="1" amount="{}"/>\n'.format(traceStart, heapUsed),
			'		<range startTime="{}" eventIndex="2" amount="0"/>\n'.format(traceStart),
		]
		ranges = heapRanges + ranges

	splitFilepath = formatSplitFilepath(filepath, splitIndex)
	print(splitFilepath)
	with open(splitFilepath, 'w') as fout:
		fout.write(xmlSchema)
		fout.write(trace)
		fout.write(eventData)
		fout.write(events)
		fout.write(split)
		for r in ranges:
			fout.write(r)
		fout.write(foot)

def splitQtd(filepath, rangesPerFile):
	eventData=''
	events=''
	ranges=[]
	with open(filepath, 'r') as fin:
		captureEvents = False
		captureRanges = False
		rangeCount = 0
		splitIndex = 0
		heapSize = 0
		heapUsed = 0
		for line in fin.readlines():
			if captureEvents:
				if '</eventData>' in line:
					print(line.rstrip())
					captureEvents = False
				else:
					events += useTabs(line)
			elif captureRanges:
				if '</profilerDataModel>' in line:
					print(line.rstrip())
					captureRanges = False
				else:
					# https://github.com/qt/qtdeclarative/blob/src/qml/jsruntime/qv4profiling_p.h
					# eventIndex="0" is MemoryAllocation(type=0) aka MemoryType::HeapPage (shown as Memory/Heap Allocation in Qt Creator)
					# eventIndex="1" is MemoryAllocation(type=2) aka MemoryType::LargeItem (shown as Memory/Heap Usage in Qt Creator)
					# eventIndex="2" is MemoryAllocation(type=1) aka MemoryType::SmallItem (never used?)
					if 'eventIndex="0"' in line: # Heap Allocation
						m = re.search(r'amount="((-)?\d+)"', line)
						amount = int(m.group(1))
						heapSize += amount
					elif 'eventIndex="1"' in line: # Heap Used
						m = re.search(r'amount="((-)?\d+)"', line)
						amount = int(m.group(1))
						heapUsed += amount

					ranges.append(useTabs(line))
					rangeCount += 1
					if rangeCount >= rangesPerFile:
						splitFile(filepath, splitIndex, eventData, events, ranges, heapSize, heapUsed)
						splitIndex += 1
						rangeCount = 0
						ranges = []

			else:
				if '<?xml ' in line:
					print(line.rstrip())
				elif '<trace ' in line:
					print(line.rstrip())
				elif '<eventData ' in line:
					print(line.rstrip())
					eventData = useTabs(line)
					captureEvents = True
				elif '<profilerDataModel>' in line:
					print(line.rstrip())
					captureRanges = True

	if rangeCount > 0:
		splitFile(filepath, splitIndex, eventData, events, ranges, heapSize, heapUsed)


def main():
	eventsPerFileDefault = 200000
	parser = argparse.ArgumentParser(description='Process some integers.')
	parser.add_argument('filepath', help='The .qtd filepath')
	parser.add_argument('-n', '--eventsperfile', type=int, default=eventsPerFileDefault, dest='eventsPerFile', help='Number of events per split file.')
	args = parser.parse_args()
	splitQtd(args.filepath, args.eventsPerFile)

	# filepath = 'plasmashell1649564180-1.qtd'
	# filepath = 'plasmashell1649368701-1.qtd'
	# splitQtd(filepath, eventsPerFileDefault)

if __name__ == '__main__':
	main()