def store_text_cap(url, title, link_guid):
bid, tdata, success = instapaper_capture(url, title)
if success:
asset = Asset.objects.get(link__guid=link_guid)
asset.instapaper_timestamp = datetime.datetime.now()
h = smhasher.murmur3_x86_128(tdata)
asset.instapaper_hash = h
asset.instapaper_id = bid
asset.save()
file_path = GENERATED_ASSETS_STORAGE + '/' + asset.base_storage_path
if not os.path.exists(file_path):
os.makedirs(file_path)
f = open(file_path + '/instapaper_cap.html', 'w')
f.write(tdata)
os.fsync(f)
f.close
if os.path.exists(file_path + '/instapaper_cap.html'):
asset.text_capture = 'instapaper_cap.html'
asset.save()
else:
logger.info("Text (instapaper) capture failed for %s" % target_url)
asset.text_capture = 'failed'
asset.save()
else:
# Must have received something other than an HTTP 200 from Instapaper, or no response object at all
logger.info("Text (instapaper) capture failed for %s" % target_url)
asset = Asset.objects.get(link__guid=link_guid)
asset.text_capture = 'failed'
asset.save()
def store_text_cap(url, title, link_guid):
asset = Asset.objects.get(link__guid=link_guid)
bid, tdata = instapaper_capture(url, title)
asset.instapaper_timestamp = datetime.datetime.now()
h = smhasher.murmur3_x86_128(tdata)
asset.instapaper_hash = h
asset.instapaper_id = bid
asset.save()
file_path = GENERATED_ASSETS_STORAGE + '/' + asset.base_storage_path
if not os.path.exists(file_path):
os.makedirs(file_path)
f = open(file_path + '/instapaper_cap.html', 'w')
f.write(tdata)
os.fsync(f)
f.close
if os.path.exists(file_path + '/instapaper_cap.html'):
asset.text_capture = 'instapaper_cap.html'
asset.save()
else:
logger.info("Text (instapaper) capture failed for %s" % target_url)
asset.text_capture = 'failed'
asset.save()
def store_text_cap(link_guid, target_url, base_storage_path, title):
# basic setup
asset_query = get_asset_query(link_guid)
storage_path = get_storage_path(base_storage_path)
create_storage_dir(storage_path)
bid, tdata, success = instapaper_capture(target_url, title)
if success:
with open(storage_path + '/instapaper_cap.html', 'wb') as f:
f.write(tdata)
if os.path.exists(storage_path + '/instapaper_cap.html'):
text_capture = 'instapaper_cap.html'
else:
logger.info("Text (instapaper) capture failed for %s" % target_url)
text_capture = 'failed'
asset_query.update(
text_capture=text_capture,
instapaper_timestamp=datetime.datetime.now(),
instapaper_hash=smhasher.murmur3_x86_128(tdata),
instapaper_id=bid
)
else:
# Must have received something other than an HTTP 200 from Instapaper, or no response object at all
logger.info("Text (instapaper) capture failed for %s" % target_url)
asset_query.update(text_capture='failed')
def render(self, context):
asset = self.asset.resolve(context)
linky = self.linky.resolve(context)
id, cap = instapaper_capture(linky.submitted_url, linky.submitted_title)
# Only say it's changed if we are sure it has changed.
if asset.instapaper_hash is None or str(smhasher.murmur3_x86_128(cap)) == asset.instapaper_hash:
return ''
else:
return ' (changed)'
def hash(self):
""" The full hash of the file """
hash_ = self.size
blocks = 0
with open(self.path, "rb") as f:
data = f.read(BIG_BLOCKSIZE)
while data:
next_hash = smhasher.murmur3_x86_128(data, GenericFileWrapper.seed)
hash_ ^= next_hash << blocks
data = f.read(BIG_BLOCKSIZE)
blocks = blocks + 1 if blocks < 16 else 0
return hash_
def hash(self):
""" The full hash of the file """
hash_ = self.size
blocks = 0
with open(self.path, 'rb') as f:
data = f.read(BIG_BLOCKSIZE)
while data:
next_hash = smhasher.murmur3_x86_128(data, GenericFileWrapper.seed)
hash_ ^= (next_hash << blocks)
data = f.read(BIG_BLOCKSIZE)
blocks = blocks + 1 if blocks < 16 else 0
return hash_
def mhash(data, n=32):
"""
Returns an N-bit hash value of provided data (string)
generated with the Murmur3 hashing.
n - is a number of bits in a final mask [0..128].
None means no masking.
"""
if not isinstance(data, basestring):
try: data = str(data)
except: data = unicode(data)
if isinstance(data, unicode):
data = data.encode('utf-8')
value = murmur3_x86_128(data)
if n is None:
return value
else:
mask = (1L << n) - 1
return value & mask
def first_block_hash(self):
with open(self.path, "rb") as f:
data = f.read(SMALL_BLOCKSIZE)
return smhasher.murmur3_x86_128(data, GenericFileWrapper.seed)
def first_block_hash(self):
with open(self.path, 'rb') as f:
data = f.read(SMALL_BLOCKSIZE)
return smhasher.murmur3_x86_128(data, GenericFileWrapper.seed)
