def set(self, url, post_data, data, headers):
if post_data:
url_hash = hashlib.sha1(url + '?' + post_data).hexdigest()
else:
url_hash = hashlib.sha1(url).hexdigest()
domain = ".".join(urlparse.urlparse(url)[1].split('.')[-2:])
conn = self.connect()
c = conn.cursor()
# Insert a row of data
if not post_data: post_data=""
only_headers = 0
if data == -1:
only_headers = 1
data = ""
created = time.mktime(time.localtime())
content_type = headers.get('content-type', '').split(';')[0].strip()
if content_type in COMPRESS_TYPES:
compressed = 1
data = zlib.compress(data)
else:
compressed = 0
data = sqlite3.Binary(data)
#fixme: this looks wrong
try:
_headers = json.dumps(headers)
except:
for h in headers:
headers[h] = headers[h].decode(detect_encoding(headers[h]))
_headers = json.dumps(headers)
t = (url_hash, domain, url, post_data, _headers, created,
data, only_headers, compressed)
c.execute(u"""INSERT OR REPLACE INTO cache values (?, ?, ?, ?, ?, ?, ?, ?, ?)""", t)
# Save (commit) the changes and clean up
conn.commit()
c.close()
conn.close()
result = store.get(url, data, headers, timeout)
if not result:
try:
url_headers, result = net.read_url(url, data, headers, return_headers=True)
except urllib2.HTTPError, e:
e.headers['Status'] = "%s" % e.code
url_headers = dict(e.headers)
result = e.read()
if url_headers.get('content-encoding', None) == 'gzip':
result = gzip.GzipFile(fileobj=StringIO.StringIO(result)).read()
if not valid or valid(result, url_headers):
store.set(url, post_data=data, data=result, headers=url_headers)
else:
raise InvalidResult(result, url_headers)
if unicode:
encoding = detect_encoding(result)
if not encoding:
encoding = 'latin-1'
result = result.decode(encoding)
return result
def save_url(url, filename, overwrite=False):
if not os.path.exists(filename) or overwrite:
dirname = os.path.dirname(filename)
if not os.path.exists(dirname):
os.makedirs(dirname)
data = read_url(url)
f = open(filename, 'w')
f.write(data)
f.close()
