Merge pull request #774 from dynamic-entropy/master

Add fetching locality instructions to helper scripts

helpers/README.md

@@ -44,4 +44,19 @@ This is to test HTTP REST API readiness for all our Tapes.
 9. Finally, set update_from_json to true for both the main and test _Tape RSEs.
 
 PS: Start with a small functional test transfer, the default dataset is a good example.
-If it works, then try with a larger dataset for stress testing.
+If it works, then try with a larger dataset for stress testing.
+
+
+---- 
+
+Adding scripts for retrieving file locality information on buffer and tape.
+
+```bash
+curl -X POST --cert $X509_USER_PROXY --key $X509_USER_PROXY --capath /etc/grid-security/certificates https://<hostname>:<port>/api/v1/tape/archiveinfo> \
+--data '{"paths":[
+<file_path1>,
+<file_path2>,
+]}' | jq
+```
+
+The document for rest api specification can be found [here](https://cernbox.cern.ch/pdf-viewer/public/vLhBpHDdaXJSqwW/WLCG%20Tape%20REST%20API%20reference%20document.pdf)

helpers/fetch_locality.py

@@ -0,0 +1,47 @@
+import requests
+import json
+from urllib.parse import urlparse
+import os
+
+
+def parse_apihost_path_map(urls):
+    parsed_urls = [urlparse(url) for url in urls]
+
+    host_files_map = {}
+    for url in parsed_urls:
+        hostname = 'https://' + url.netloc
+        if hostname not in host_files_map:
+            host_files_map[hostname] = []
+        host_files_map[hostname].append(url.path)
+
+    api_host_files_map = {}
+    for hostname, files in host_files_map.items():
+        response = requests.get(hostname + '/.well-known/wlcg-tape-rest-api',
+                                verify='/etc/grid-security/certificates',
+                                cert=(os.environ['X509_USER_PROXY'],
+                                      os.environ['X509_USER_PROXY']),
+                                timeout=180
+                                )
+        tape_host = response.json()["endpoints"][0]["uri"]
+        api_host_files_map[tape_host] = files
+    return api_host_files_map
+
+
+def get_locality(urls):
+    host_files_map = parse_apihost_path_map(urls)
+
+    for hostname, files in host_files_map.items():
+        response = requests.post(hostname + '/archiveinfo',
+                                 data=json.dumps({"paths": files}),
+                                 verify='/etc/grid-security/certificates',
+                                 cert=(os.environ['X509_USER_PROXY'],
+                                       os.environ['X509_USER_PROXY']),
+                                 headers={'Content-Type': 'application/json'},
+                                 timeout=180
+                                 )
+        print(response.text)
+
+
+if __name__ == "__main__":
+    urls = ["davs://<hostname>:<port><path>"]
+    get_locality(urls)