Merge pull request #4885 from kobotoolbox/asset-remove-kobocat-domain

Avoid storing KC domain in asset deployment data

kpi/deployment_backends/kobocat_backend.py

@@ -138,7 +138,7 @@ def connect(self, active=False):
         kc_server = settings.KOBOCAT_URL
         id_string = self.asset.uid
 
-        url = self.external_to_internal_url('{}/api/v1/forms'.format(kc_server))
+        url = self.normalize_internal_url('{}/api/v1/forms'.format(kc_server))
         xlsx_io = self.asset.to_xlsx_io(
             versioned=True, append={
                 'settings': {
@@ -162,12 +162,16 @@ def connect(self, active=False):
         files = {'xls_file': ('{}.xlsx'.format(id_string), xlsx_io)}
         json_response = self._kobocat_request(
             'POST', url, data=payload, files=files)
-        self.store_data({
-            'backend': 'kobocat',
-            'active': json_response['downloadable'],
-            'backend_response': json_response,
-            'version': self.asset.version_id,
-        })
+        # Store only path
+        json_response['url'] = urlparse(json_response['url']).path
+        self.store_data(
+            {
+                'backend': 'kobocat',
+                'active': json_response['downloadable'],
+                'backend_response': json_response,
+                'version': self.asset.version_id,
+            }
+        )
 
     @staticmethod
     def nlp_tracking_data(asset_ids, start_date=None):
@@ -240,7 +244,7 @@ def delete(self):
         """
         WARNING! Deletes all submitted data!
         """
-        url = self.external_to_internal_url(self.backend_response['url'])
+        url = self.normalize_internal_url(self.backend_response['url'])
         try:
             self._kobocat_request('DELETE', url)
         except KobocatDeploymentException as e:
@@ -455,22 +459,12 @@ def enketo_id(self):
         return enketo_id
 
     @staticmethod
-    def external_to_internal_url(url):
+    def normalize_internal_url(url: str) -> str:
         """
-        Replace the value of `settings.KOBOCAT_URL` with that of
-        `settings.KOBOCAT_INTERNAL_URL` when it appears at the beginning of
-        `url`
+        Normalize url to ensure KOBOCAT_INTERNAL_URL is used
         """
-        kobocat_url = settings.KOBOCAT_URL
-        # If old domain name is detected, use it for search&replace below
-        if settings.KOBOCAT_OLD_URL and settings.KOBOCAT_OLD_URL in url:
-            kobocat_url = settings.KOBOCAT_OLD_URL
-
-        return re.sub(
-            pattern='^{}'.format(re.escape(kobocat_url)),
-            repl=settings.KOBOCAT_INTERNAL_URL,
-            string=url
-        )
+        parsed_url = urlparse(url)
+        return f'{settings.KOBOCAT_INTERNAL_URL}{parsed_url.path}'
 
     def get_attachment(
         self,
@@ -836,7 +830,7 @@ def redeploy(self, active=None):
         """
         if active is None:
             active = self.active
-        url = self.external_to_internal_url(self.backend_response['url'])
+        url = self.normalize_internal_url(self.backend_response['url'])
         id_string = self.backend_response['id_string']
         xlsx_io = self.asset.to_xlsx_io(
             versioned=True, append={
@@ -860,6 +854,7 @@ def redeploy(self, active=None):
             'backend_response': json_response,
             'version': self.asset.version_id,
         })
+
         self.set_asset_uid()
 
     def remove_from_kc_only_flag(self,
@@ -916,7 +911,7 @@ def set_active(self, active):
         """
         # self.store_data is an alias for
         # self.asset._deployment_data.update(...)
-        url = self.external_to_internal_url(
+        url = self.normalize_internal_url(
             self.backend_response['url'])
         payload = {
             'downloadable': bool(active)
@@ -946,7 +941,7 @@ def set_asset_uid(self, force: bool = False) -> bool:
         if is_synchronized:
             return False
 
-        url = self.external_to_internal_url(self.backend_response['url'])
+        url = self.normalize_internal_url(self.backend_response['url'])
         payload = {
             'kpi_asset_uid': self.asset.uid
         }
@@ -991,7 +986,7 @@ def set_has_kpi_hooks(self):
         Store results in deployment data
         """
         has_active_hooks = self.asset.has_active_hooks
-        url = self.external_to_internal_url(
+        url = self.normalize_internal_url(
             self.backend_response['url'])
         payload = {
             'has_kpi_hooks': has_active_hooks,
@@ -1124,7 +1119,7 @@ def submission_url(self) -> str:
 
     def sync_media_files(self, file_type: str = AssetFile.FORM_MEDIA):
 
-        url = self.external_to_internal_url(self.backend_response['url'])
+        url = self.normalize_internal_url(self.backend_response['url'])
         response = self._kobocat_request('GET', url)
         kc_files = defaultdict(dict)
 
@@ -1400,7 +1395,7 @@ def __delete_kc_metadata(
         """
         # Delete file in KC
 
-        delete_url = self.external_to_internal_url(kc_file_['url'])
+        delete_url = self.normalize_internal_url(kc_file_['url'])
         self._kobocat_request('DELETE', url=delete_url, expect_formid=False)
 
         if file_ is None:
@@ -1661,7 +1656,6 @@ def __update_kc_metadata_hash(
         """
         server = settings.KOBOCAT_INTERNAL_URL
         metadata_detail_url = f'{server}/api/v1/metadata/{kc_metadata_id}'
-
         data = {'file_hash': file_.md5_hash}
         self._kobocat_request('PATCH',
                               url=metadata_detail_url,