def start(cdxjFilePath=INDEX_FILE, proxy=None):
hostPort = ipwbConfig.getIPWBReplayConfig()
app.proxy = proxy
if not hostPort:
ipwbConfig.setIPWBReplayConfig(IPWBREPLAY_IP, IPWBREPLAY_PORT)
hostPort = ipwbConfig.getIPWBReplayConfig()
if ipwbConfig.isDaemonAlive():
if cdxjFilePath == INDEX_FILE:
ipwbConfig.firstRun()
ipwbConfig.setIPWBReplayIndexPath(cdxjFilePath)
app.cdxjFilePath = cdxjFilePath
else:
print('Sample data not pulled from IPFS.')
print('Check that the IPFS daemon is running.')
try:
print('IPWB replay started on http://{0}:{1}'.format(
IPWBREPLAY_IP, IPWBREPLAY_PORT))
app.run(host='0.0.0.0', port=IPWBREPLAY_PORT)
except gaierror:
print('Detected no active Internet connection.')
print('Overriding to use default IP and port configuration.')
app.run()
except socketerror:
print('Address {0}:{1} already in use!'.format(IPWBREPLAY_IP,
IPWBREPLAY_PORT))
sys.exit()
def start(cdxjFilePath, proxy=None):
hostPort = ipwbUtils.getIPWBReplayConfig()
app.proxy = proxy
if not hostPort:
ipwbUtils.setIPWBReplayConfig(IPWBREPLAY_HOST, IPWBREPLAY_PORT)
hostPort = ipwbUtils.getIPWBReplayConfig()
if ipwbUtils.isDaemonAlive():
ipwbUtils.setIPWBReplayIndexPath(cdxjFilePath)
app.cdxjFilePath = cdxjFilePath
else:
print('Sample data not pulled from IPFS.')
print('Check that the IPFS daemon is running.')
# Perform checks for CDXJ file existence, TODO: reuse cached contents
app.cdxjFileContents = getIndexFileContents(cdxjFilePath)
try:
print('IPWB replay started on http://{0}:{1}'.format(
IPWBREPLAY_HOST, IPWBREPLAY_PORT))
app.run(host='0.0.0.0', port=IPWBREPLAY_PORT)
except gaierror:
print('Detected no active Internet connection.')
print('Overriding to use default IP and port configuration.')
app.run()
except socketerror:
print('Address {0}:{1} already in use!'.format(IPWBREPLAY_HOST,
IPWBREPLAY_PORT))
sys.exit()
def start(cdxjFilePath=INDEX_FILE):
hostPort = ipwbConfig.getIPWBReplayConfig()
if not hostPort:
ipwbConfig.setIPWBReplayConfig(IPWBREPLAY_IP, IPWBREPLAY_PORT)
hostPort = ipwbConfig.getIPWBReplayConfig()
ipwbConfig.firstRun()
ipwbConfig.setIPWBReplayIndexPath(cdxjFilePath)
app.cdxjFilePath = cdxjFilePath
app.run(host=IPWBREPLAY_IP, port=IPWBREPLAY_PORT)
def showTimeMap(urir, format):
urir = getCompleteURI(urir)
s = surt.surt(urir, path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbUtils.getIPWBReplayIndexPath()
cdxjLinesWithURIR = getCDXJLinesWithURIR(urir, indexPath)
tmContentType = ''
hostAndPort = ipwbUtils.getIPWBReplayConfig()
tgURI = 'http://{0}:{1}/timegate/{2}'.format(hostAndPort[0],
hostAndPort[1], urir)
tm = '' # Initialize for usage beyond below conditionals
if format == 'link':
tm = generateLinkTimeMapFromCDXJLines(cdxjLinesWithURIR, s,
request.url, tgURI)
tmContentType = 'application/link-format'
elif format == 'cdxj':
tm = generateCDXJTimeMapFromCDXJLines(cdxjLinesWithURIR, s,
request.url, tgURI)
tmContentType = 'application/cdxj+ors'
resp = Response(tm)
resp.headers['Content-Type'] = tmContentType
return resp
def getLinkHeaderAbbreviatedTimeMap(urir, pivotDatetime):
s = surt.surt(urir, path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbConfig.getIPWBReplayIndexPath()
cdxjLinesWithURIR = getCDXJLinesWithURIR(urir, indexPath)
hostAndPort = ipwbConfig.getIPWBReplayConfig()
tmURI = 'http://{0}:{1}/timemap/link/{2}'.format(
'localhost', # hostAndPort[0],
hostAndPort[1],
urir)
tm = generateLinkTimeMapFromCDXJLines(cdxjLinesWithURIR, s, tmURI)
# Fix base TM relation when viewing abbrev version in Link resp
tm = tm.replace('rel="self"', 'rel="timemap"')
# Only one memento in TimeMap
if 'rel="first last memento"' in tm:
return tm
tmLines = tm.split('\n')
for idx, line in enumerate(tmLines):
if len(re.findall('rel=.*memento"', line)) == 0:
continue # Not a memento
if pivotDatetime in line:
addBothNextAndPrev = False
if idx > 0 and idx < len(tmLines) - 1:
addBothNextAndPrev = True
if addBothNextAndPrev or idx == 0:
tmLines[idx + 1] = \
tmLines[idx + 1].replace('memento"', 'next memento"')
if addBothNextAndPrev or idx == len(tmLines) - 1:
tmLines[idx - 1] = \
tmLines[idx - 1].replace('memento"', 'prev memento"')
break
# Remove all mementos in abbrev TM that are not:
# first, last, prev, next, or pivot
for idx, line in enumerate(tmLines):
if len(re.findall('rel=.*memento"', line)) == 0:
continue # Not a memento
if pivotDatetime in line:
continue
if len(re.findall('rel=.*(next|prev|first|last)', line)) == 0:
tmLines[idx] = ''
tm = '\n'.join(tmLines)
return tm
return closest
def getCDXLines(surtURI):
with open('index.cdx', 'r') as cdxFile:
cdxlobj = []
bsResp = iter_exact(cdxFile, surtURI)
for cdxl in bsResp:
(suri, dttm, jobj) = cdxl.split(' ', 2)
if suri != surtURI:
break
cdxlobj.append((suri, dttm, jobj))
return cdxlobj
'''
if __name__ == "__main__":
hostPort = ipwbConfig.getIPWBReplayConfig()
if not hostPort:
ipwbConfig.setIPWBReplayConfig(IPWBREPLAY_IP, IPWBREPLAY_PORT)
hostPort = ipwbConfig.getIPWBReplayConfig()
# print hostPort
# sys.exit()
app.run(host=IPWBREPLAY_IP, port=IPWBREPLAY_PORT)
# Read in URI, convert to SURT
# surt(uriIn)
# Get SURTed URI lines in CDXJ
# Read CDXJ
# Do bin search to find relevant lines
# read IPFS hash from relevant lines (header, payload)