#19 checking for OGC Exceptions on/off in WWW:LINK Resource Responses

GeoHealthCheck/app.py

@@ -447,7 +447,7 @@ def add():
         return redirect(url_for('add', lang=g.current_lang))
 
     [title, success, response_time, message, start_time] = run_test_resource(
-        resource_type, url)
+        APP.config, resource_type, url)
 
     if not success:
         flash(message, 'danger')
@@ -502,7 +502,7 @@ def test(resource_identifier):
         return redirect(request.referrer)
 
     [title, success, response_time, message, start_time] = run_test_resource(
-        resource.resource_type, resource.url)
+        APP.config, resource.resource_type, resource.url)
 
     if message not in ['OK', None, 'None']:
         msg = gettext('ERROR')

GeoHealthCheck/config_main.py

@@ -39,6 +39,7 @@
 GHC_SELF_REGISTER = False
 GHC_NOTIFICATIONS = False
 GHC_NOTIFICATIONS_VERBOSITY = True
+GHC_WWW_LINK_EXCEPTION_CHECK = False
 GHC_ADMIN_EMAIL = 'you@example.com'
 GHC_SITE_TITLE = 'GeoHealthCheck Demonstration'
 GHC_SITE_URL = 'http://host'

GeoHealthCheck/healthcheck.py

@@ -47,7 +47,7 @@
 LOGGER = logging.getLogger(__name__)
 
 
-def run_test_resource(resource_type, url):
+def run_test_resource(config, resource_type, url):
     """tests a CSW service and provides run metrics"""
 
     if resource_type not in RESOURCE_TYPES.keys():
@@ -70,7 +70,7 @@ def run_test_resource(resource_type, url):
         elif resource_type == 'OGC:WFS':
             ows = WebFeatureService(url)
         elif resource_type == 'OGC:WCS':
-            ows = WebCoverageService(url)
+            ows = WebCoverageService(url, version='1.0.0')
         elif resource_type == 'OGC:WPS':
             ows = WebProcessingService(url)
         elif resource_type == 'OGC:CSW':
@@ -80,12 +80,36 @@ def run_test_resource(resource_type, url):
         elif resource_type in ['WWW:LINK', 'urn:geoss:waf']:
             ows = urlopen(url)
             if resource_type == 'WWW:LINK':
-                import re
-                try:
-                    title_re = re.compile("<title>(.+?)</title>")
-                    title = title_re.search(ows.read()).group(1)
-                except:
-                    title = url
+                content_type = ows.info().getheader('Content-Type')
+
+                # Check content if the response is not an image
+                if 'image/' not in content_type:
+                    content = ows.read()
+                    import re
+                    try:
+                        title_re = re.compile("<title>(.+?)</title>")
+                        title = title_re.search(content).group(1)
+                    except:
+                        title = url
+
+                    # Optional check for any OGC-Exceptions in Response
+                    if config and config['GHC_WWW_LINK_EXCEPTION_CHECK']:
+                        exception_text = None
+                        try:
+                            except_re = re.compile(
+                                "ServiceException>|ExceptionReport>")
+                            exception_text = except_re.search(content).group(0)
+                        except:
+                            # No Exception in Response text
+                            pass
+
+                        if exception_text:
+                            # Found OGC-Exception in Response text
+                            raise Exception(
+                                "Exception in response: %s" % exception_text)
+
+                    del content
+
             elif resource_type == 'urn:geoss:waf':
                 title = 'WAF %s %s' % (gettext('for'), urlparse(url).hostname)
         elif resource_type == 'FTP':
@@ -116,4 +140,5 @@ def run_test_resource(resource_type, url):
         print('Usage: %s <resource_type> <url>' % sys.argv[0])
         sys.exit(1)
 
-    print(run_test_resource(sys.argv[1], sys.argv[2]))
+    # TODO: need APP.config here, None for now
+    print(run_test_resource(None, sys.argv[1], sys.argv[2]))

GeoHealthCheck/models.py

@@ -262,7 +262,8 @@ def db_commit():
                     last_run_success = last_run.success
 
                 # Run test
-                run_to_add = run_test_resource(res.resource_type, res.url)
+                run_to_add = run_test_resource(
+                        APP.config, res.resource_type, res.url)
 
                 run1 = Run(res, run_to_add[1], run_to_add[2],
                            run_to_add[3], run_to_add[4])

docker/GeoHealthCheck/Dockerfile

@@ -13,6 +13,7 @@ ENV GHC_RUN_FREQUENCY 'hourly'
 ENV GHC_SELF_REGISTER False
 ENV GHC_NOTIFICATIONS False
 ENV GHC_NOTIFICATIONS_VERBOSITY True
+ENV GHC_WWW_LINK_EXCEPTION_CHECK False
 ENV GHC_ADMIN_EMAIL 'you@example.com'
 ENV GHC_SITE_TITLE 'GeoHealthCheck Demonstration'
 ENV GHC_SITE_URL 'http://localhost'

docker/GeoHealthCheck/config_site.py

@@ -41,6 +41,7 @@
 GHC_SELF_REGISTER = os.environ['GHC_SELF_REGISTER']
 GHC_NOTIFICATIONS = os.environ['GHC_NOTIFICATIONS']
 GHC_NOTIFICATIONS_VERBOSITY = os.environ['GHC_NOTIFICATIONS_VERBOSITY']
+GHC_WWW_LINK_EXCEPTION_CHECK = os.environ['GHC_WWW_LINK_EXCEPTION_CHECK']
 GHC_ADMIN_EMAIL = os.environ['GHC_ADMIN_EMAIL']
 GHC_SITE_TITLE = os.environ['GHC_SITE_TITLE']
 GHC_SITE_URL = os.environ['GHC_SITE_URL']

docs/index.rst

@@ -126,7 +126,8 @@ You can override these settings in ``instance/config_site.py``:
   (i.e. ``hourly``, ``daily``, ``monthly``)
 - **GHC_SELF_REGISTER**: allow registrations from users on the website
 - **GHC_NOTIFICATIONS**: turn on email notifications
-- **GHC_NOTIFICATIONS_VERBOSITY**: receive additional email notifications than just ``Failing`` and ``Fixed`` (default True)
+- **GHC_NOTIFICATIONS_VERBOSITY**: receive additional email notifications than just ``Failing`` and ``Fixed`` (default ``True``)
+- **GHC_WWW_LINK_EXCEPTION_CHECK**: turn on checking for OGC Exceptions in ``WWW:LINK`` Resource responses (default ``False``)
 - **GHC_ADMIN_EMAIL**: email address of administrator / contact
 - **GHC_SITE_TITLE**: title used for installation / deployment
 - **GHC_SITE_URL**: url of the installation / deployment