Bugfix + fixes for the links related to the Sensor Observation Service.

utils/fix_links.py

@@ -10,11 +10,14 @@ def translate_content_type( content_type ):
     # TODO: complete the following list!
     types = ['pdf', 'html', 'zip', 'xml', 'javascript', 'json', 'csv']
 
-    for type in types:
-        if type in content_type:
-            return type.upper()
+    for the_type in types:
+        if the_type in content_type:
+            output = the_type.upper()
+            break
         else:
-            return content_type
+            output = content_type
+
+    return output
 
 
 
@@ -27,7 +30,7 @@ def fix_links( links ):
 
         # the 'description' attribute ends, sometimes, with (OGC:WMS), (OGC:WCS), ..., which is redundant and, in some case, erroneus.
         if 'description' in link.keys():
-            fixed_links[k]['description'] = re.sub(r'\(OGC:WMS\)|\(OGC:WCS\)|\(OGC:WFS\)', '', link['description']).strip()
+            fixed_links[k]['description'] = re.sub(r'\(OGC:WMS\)|\(OGC:WCS\)|\(OGC:WFS\)|\(OGC:SOS\)', '', link['description']).strip()
 
         # # FIX links pointing to external websites
         # #logging.debug(link)
@@ -36,8 +39,7 @@ def fix_links( links ):
 
         # FIX links in which the declared protocol is as bizarre as "WWW:LINK-1.0-http--link"
         # The KML protocol needs also to be fixed.
-        if 'protocol' in link.keys() and any( [x in link['protocol'] for x in ['WWW', 'kml'] ] ):
-
+        if 'protocol' in link.keys() and any( [x in link['protocol'] for x in ['WWW', 'kml', 'html'] ] ):
             try:
                 # let's try getting the information from the Web Server...
                 resp = requests.head( link['url'], allow_redirects=True )
@@ -52,7 +54,7 @@ def fix_links( links ):
             except Exception as e:
                 logging.debug(e)
                 # ...otherwise, we make a guess on the basis of the information carried by the URL
-                known_formats = ['ecw', 'pdf', 'zip', 'kml', 'json', 'tif', 'tiff', 'csv']
+                known_formats = ['ecw', 'pdf', 'zip', 'kml', 'json', 'tif', 'tiff', 'csv', 'sos']
 
                 for known_format in known_formats:
                     if known_format in link['url'].lower():
@@ -86,6 +88,13 @@ def fix_links( links ):
                 del fixed_links[k]
             continue
 
+    # FIX SOS
+    for k, link in enumerate(fixed_links):
+
+        if link['protocol'] == 'SOS':
+            if '?' in link['url']:
+                fixed_links[k]['url'] = link['url'].split('?')[:-1]
+
 
     return fixed_links
 
@@ -141,6 +150,22 @@ if __name__ == '__main__':
                     'description': "Licence ouverte",
                     'unknown': "1",
                     'protocol': "WWW-LINK"
+                },
+                {
+                    'name': "Historique des disponibilités des stations Vélo'V(OGC:SOS)",
+                    'description': "Historique des disponibilités des stations Vélo'V(OGC:SOS)",
+                    'url': "https://download.data.grandlyon.com/sos/velov?request=getCapabilities&service=SOS",
+                    'protocol': "WWW:LINK-1.0-http--link",
+                    'content-type': "text/html",
+                    'unknown': "1"
+                },
+                {
+                    'name': "Démonstrateur WMS-Time",
+                    'description': "Visualisation cartographique temporelle (WMS-T) de l'historique des disponibilités des stations Vélo'V",
+                    'url': "http://demo.data.grandlyon.com/wmst/reseau_velov.html",
+                    'protocol': "text/html",
+                    'content-type': "text/html",
+                    'unknown': "1"
                 }
         ]