Github/stackdump
1
0
Fork 0
mirror of https://github.com/djohnlewis/stackdump synced 2025-04-05 01:03:27 +00:00
Code Issues Projects Releases Wiki Activity

Skips valid XML character scrubbing if configured for faster performance.

Browse Source
This commit is contained in:
Samuel Lai 2013-11-28 14:01:00 +11:00
parent 644269dd5d
commit c0766de8d4
2 changed files with 12 additions and 3 deletions
python
packages
pysolr.py
src/stackdump/commands
import_site.py

11
python/packages/pysolr.py

@ -246,6 +246,11 @@ class Solr(object):
Optionally accepts ``timeout`` for wait seconds until giving up on a Optionally accepts ``timeout`` for wait seconds until giving up on a
request. Default is ``60`` seconds. request. Default is ``60`` seconds.
Optionally accepts ``assume_clean`` to skip cleaning request of invalid XML
characters. This offers a slight performance improvement, but only set this
to ``True`` if you know your request is clean (e.g. coming from other XML
data). Bad things will happen otherwise. Default is ``False``.
Usage:: Usage::
solr = pysolr.Solr('http://localhost:8983/solr') solr = pysolr.Solr('http://localhost:8983/solr')
@ -253,10 +258,11 @@ class Solr(object):
solr = pysolr.Solr('http://localhost:8983/solr', timeout=10) solr = pysolr.Solr('http://localhost:8983/solr', timeout=10)
""" """
def __init__(self, url, decoder=None, timeout=60): def __init__(self, url, decoder=None, timeout=60, assume_clean=False):
self.decoder = decoder or json.JSONDecoder() self.decoder = decoder or json.JSONDecoder()
self.url = url self.url = url
self.timeout = timeout self.timeout = timeout
self.assume_clean = assume_clean
self.log = self._get_log() self.log = self._get_log()
self.session = requests.Session() self.session = requests.Session()
self.session.stream = False self.session.stream = False
@ -506,6 +512,9 @@ class Solr(object):
value = "{0}".format(value) value = "{0}".format(value)
if self.assume_clean:
return value
else:
return clean_xml_string(value) return clean_xml_string(value)
def _to_python(self, value): def _to_python(self, value):

2
python/src/stackdump/commands/import_site.py

@ -600,7 +600,7 @@ def import_site(xml_root, site_name, dump_date, site_desc, site_key,
# connect to solr # connect to solr
print('Connecting to solr...') print('Connecting to solr...')
solr = Solr(settings.SOLR_URL) solr = Solr(settings.SOLR_URL, assume_clean=True)
# pysolr doesn't try to connect until a request is made, so we'll make a ping request # pysolr doesn't try to connect until a request is made, so we'll make a ping request
try: try:
solr._send_request('GET', 'admin/ping') solr._send_request('GET', 'admin/ping')