Merge pull request #375 from tomkralidis/issue-374

Sanitize URL for services (#374)

owslib/csw.py

@@ -68,7 +68,7 @@ def __init__(self, url, lang='en-US', version='2.0.2', timeout=10, skip_caps=Fal
 
         """
 
-        self.url = url
+        self.url = util.clean_ows_url(url)
         self.lang = lang
         self.version = version
         self.timeout = timeout

owslib/sos.py

@@ -14,6 +14,7 @@
 from __future__ import (absolute_import, division, print_function)
 
 from .swe.observation import sos100, sos200
+from .util import clean_ows_url
 
 
 def SensorObservationService(url,
@@ -30,11 +31,14 @@ def SensorObservationService(url,
     :param password: password for the username
     :return: a version specific SensorObservationService object
     """
+
+    clean_url = clean_ows_url(url)
+
     if version in ['1.0', '1.0.0']:
         return sos100.SensorObservationService_1_0_0.__new__(
-            sos100.SensorObservationService_1_0_0, url, version,
+            sos100.SensorObservationService_1_0_0, clean_url, version,
             xml, username, password)
     elif version in ['2.0', '2.0.0']:
         return sos200.SensorObservationService_2_0_0.__new__(
-            sos200.SensorObservationService_2_0_0, url, version,
+            sos200.SensorObservationService_2_0_0, clean_url, version,
             xml, username, password)

owslib/util.py

@@ -15,11 +15,7 @@
 import pytz
 from owslib.etree import etree, ParseError
 from owslib.namespaces import Namespaces
-try:                    # Python 3
-    from urllib.parse import urlsplit, urlencode
-except ImportError:     # Python 2
-    from urlparse import urlsplit
-    from urllib import urlencode
+from six.moves.urllib.parse import urlsplit, urlencode, urlparse, parse_qs, urlunparse
 
 try:
     from StringIO import StringIO  # Python 2
@@ -573,6 +569,35 @@ def strip_bom(raw_text):
     return raw_text
 
 
+def clean_ows_url(url):
+    """
+    clean an OWS URL of basic service elements
+
+    source: https://stackoverflow.com/a/11640565
+    """
+
+    filtered_kvp = {}
+    basic_service_elements = ('service', 'version', 'request')
+
+    parsed = urlparse(url)
+    qd = parse_qs(parsed.query, keep_blank_values=True)
+
+    for key, value in qd.items():
+        if key.lower() not in basic_service_elements:
+            filtered_kvp[key] = value
+
+    newurl = urlunparse([
+        parsed.scheme,
+        parsed.netloc,
+        parsed.path,
+        parsed.params,
+        urlencode(filtered_kvp, doseq=True),
+        parsed.fragment
+    ])
+
+    return newurl
+
+
 def bind_url(url):
     """binds an HTTP GET query string endpiont"""
     if url.find('?') == -1: # like http://host/wms

owslib/wcs.py

@@ -17,7 +17,7 @@
 
 from . import etree
 from .coverage import wcs100, wcs110, wcs111, wcsBase
-from owslib.util import openURL
+from owslib.util import clean_ows_url, openURL
 
 
 def WebCoverageService(url, version=None, xml=None, cookies=None, timeout=30):
@@ -33,9 +33,11 @@ def WebCoverageService(url, version=None, xml=None, cookies=None, timeout=30):
         version = capabilities.get('version')
         del capabilities
 
+    clean_url = clean_ows_url(url)
+
     if version == '1.0.0':
-        return wcs100.WebCoverageService_1_0_0.__new__(wcs100.WebCoverageService_1_0_0, url, xml, cookies)
+        return wcs100.WebCoverageService_1_0_0.__new__(wcs100.WebCoverageService_1_0_0, clean_url, xml, cookies)
     elif version == '1.1.0':
-        return wcs110.WebCoverageService_1_1_0.__new__(wcs110.WebCoverageService_1_1_0,url, xml, cookies)
+        return wcs110.WebCoverageService_1_1_0.__new__(wcs110.WebCoverageService_1_1_0, clean_url, xml, cookies)
     elif version == '1.1.1':
-        return wcs111.WebCoverageService_1_1_1.__new__(wcs111.WebCoverageService_1_1_1,url, xml, cookies)
+        return wcs111.WebCoverageService_1_1_1.__new__(wcs111.WebCoverageService_1_1_1, clean_url, xml, cookies)

owslib/wfs.py

@@ -16,6 +16,7 @@
 from __future__ import (absolute_import, division, print_function)
 
 from .feature import wfs100, wfs110, wfs200
+from .util import clean_ows_url
 
 
 def WebFeatureService(url, version='1.0.0', xml=None, parse_remote_metadata=False,
@@ -33,18 +34,21 @@ def WebFeatureService(url, version='1.0.0', xml=None, parse_remote_metadata=Fals
     @param password: service authentication password
     @return: initialized WebFeatureService_2_0_0 object
     '''
+
+    clean_url = clean_ows_url(url)
+
     if version in  ['1.0', '1.0.0']:
-        return wfs100.WebFeatureService_1_0_0(url, version, xml, parse_remote_metadata,
+        return wfs100.WebFeatureService_1_0_0(clean_url, version, xml, parse_remote_metadata,
                                               timeout=timeout,
                                               username=username,
                                               password=password)
     elif version in  ['1.1', '1.1.0']:
-        return wfs110.WebFeatureService_1_1_0(url, version, xml, parse_remote_metadata,
+        return wfs110.WebFeatureService_1_1_0(clean_url, version, xml, parse_remote_metadata,
                                               timeout=timeout,
                                               username=username,
                                               password=password)
     elif version in ['2.0', '2.0.0']:
-        return wfs200.WebFeatureService_2_0_0(url,  version, xml, parse_remote_metadata,
+        return wfs200.WebFeatureService_2_0_0(clean_url, version, xml, parse_remote_metadata,
                                               timeout=timeout,
                                               username=username,
                                               password=password)

owslib/wms.py

@@ -17,6 +17,7 @@
 
 from __future__ import (absolute_import, division, print_function)
 from .map import wms111, wms130
+from .util import clean_ows_url
 
 
 def WebMapService(url,
@@ -39,13 +40,16 @@ def WebMapService(url,
     @param timeout: time (in seconds) after which requests should timeout
     @return: initialized WebFeatureService_2_0_0 object
     '''
+
+    clean_url = clean_ows_url(url)
+
     if version in ['1.1.1']:
-        return wms111.WebMapService_1_1_1(url, version=version, xml=xml,
+        return wms111.WebMapService_1_1_1(clean_url, version=version, xml=xml,
                                           parse_remote_metadata=parse_remote_metadata,
                                           username=username, password=password,
                                           timeout=timeout, headers=headers)
     elif version in ['1.3.0']:
-        return wms130.WebMapService_1_3_0(url, version=version, xml=xml,
+        return wms130.WebMapService_1_3_0(clean_url, version=version, xml=xml,
                                           parse_remote_metadata=parse_remote_metadata,
                                           username=username, password=password,
                                           timeout=timeout, headers=headers)

owslib/wmts.py

@@ -42,7 +42,7 @@
     from urllib import urlencode
     from urlparse import urlparse, urlunparse, parse_qs, ParseResult
 from .etree import etree
-from .util import openURL, testXMLValue, getXMLInteger
+from .util import clean_ows_url, openURL, testXMLValue, getXMLInteger
 from .fgdc import Metadata
 from .iso import MD_Metadata
 from .ows import ServiceProvider, ServiceIdentification, OperationsMetadata
@@ -159,7 +159,7 @@ def __init__(self, url, version='1.0.0', xml=None, username=None,
             requests.
 
         """
-        self.url = url
+        self.url = clean_ows_url(url)
         self.username = username
         self.password = password
         self.version = version

owslib/wps.py

@@ -89,7 +89,7 @@
 from owslib.etree import etree
 from owslib.ows import DEFAULT_OWS_NAMESPACE, ServiceIdentification, ServiceProvider, OperationsMetadata, BoundingBox
 from time import sleep
-from owslib.util import (testXMLValue, build_get_url, dump, getTypedValue,
+from owslib.util import (testXMLValue, build_get_url, clean_ows_url, dump, getTypedValue,
                          getNamespace, element_to_string, nspath, openURL, nspath_eval, log)
 from xml.dom.minidom import parseString
 from owslib.namespaces import Namespaces
@@ -188,7 +188,7 @@ def __init__(self, url, version=WPS_DEFAULT_VERSION, username=None, password=Non
         """
 
         # fields passed in from object initializer
-        self.url = url
+        self.url = clean_ows_url(url)
         self.username = username
         self.password = password
         self.version = version

tests/doctests/util.txt

@@ -0,0 +1,24 @@
+
+Imports
+
+    >>> from __future__ import (absolute_import, division, print_function)
+    >>> from owslib.util import clean_ows_url
+
+Tests
+
+    >>> clean_ows_url('http//example.org/wms')
+    'http//example.org/wms'
+    >>> clean_ows_url('http//example.org/wms?service=WMS')
+    'http//example.org/wms'
+    >>> clean_ows_url('http//example.org/wms?SERVICE=WMS')
+    'http//example.org/wms'
+    >>> clean_ows_url('http//example.org/wms?SeRvIcE=WMS')
+    'http//example.org/wms'
+    >>> clean_ows_url('http//example.org/wms?SeRvIcE=WMS&version=1.3.0&request=GetCapabilities')
+    'http//example.org/wms'
+    >>> clean_ows_url('http//example.org/wms?foo=bar&SeRvIcE=WMS&version=1.3.0&request=GetCapabilities')
+    'http//example.org/wms?foo=bar'
+    >>> clean_ows_url('http://example.org/wms?map=/path/to/foo.map&SERVICE=WMS&version=1.3.0&request=GetCapabilities')
+    'http://example.org/wms?map=%2Fpath%2Fto%2Ffoo.map'
+    >>> clean_ows_url('http://example.org/wms?map=/path/to/foo.map&foo=bar&&SERVICE=WMS&version=1.3.0&request=GetCapabilities')
+    'http://example.org/wms?map=%2Fpath%2Fto%2Ffoo.map&foo=bar'