Merge pull request #2 from b97pla/feature/use_optional_path_parameter

if specified, use a supplied path to the pdc archive

Author: johandahlberg

.gitignore

@@ -2,3 +2,4 @@
 __pycache__
 *.swp
 *.swo
+.idea/

archive_verify/handlers.py

@@ -25,9 +25,11 @@ async def verify(request):
     host = body["host"]
     archive = body["archive"]
     description = body["description"]
-    
+    path = body.get("path")
+
     src_root = request.app["config"]["pdc_root_dir"].format(host)
-    archive_path = os.path.join(src_root, archive) 
+    # use a supplied path if available, otherwise construct it from the src_root and archive
+    archive_path = path or os.path.join(src_root, archive)
 
     redis_conn = Redis()
     q = Queue(connection=redis_conn)
@@ -38,7 +40,7 @@ async def verify(request):
     # config e.g. setups the queue to keep the job results indefinately, 
     # therefore they we will have to remove them ourselves afterwards. 
     job = q.enqueue_call(verify_archive, 
-                        args=(archive, host, description, request.app["config"]),
+                        args=(archive, archive_path, description, request.app["config"]),
                         timeout=request.app["config"]["job_timeout"],
                         result_ttl=request.app["config"]["job_result_ttl"],
                         ttl=request.app["config"]["job_ttl"])

archive_verify/workers.py

@@ -90,26 +90,24 @@ def compare_md5sum(archive_dir):
     else: 
         return True
 
-def verify_archive(archive, host, description, config):
+def verify_archive(archive, archive_path, description, config):
     """
     Our main worker function. This will be put into the RQ/Redis queue when the /verify endpoint gets called. 
     Downloads the specified archive from PDC and then verifies the MD5 sums. 
 
     :param archive: The name of the archive we shall download
-    :param host: The host the archive was previsouly uploaded from
+    :param archive_path: The path to the archive on PDC
     :param description: The unique description that was used when uploading the archive to PDC
     :param config: A dict containing the apps configuration
     :returns A JSON with the result that will be kept in the Redis queue
     """
     dest_root = config["verify_root_dir"]
     dsmc_log_dir = config["dsmc_log_dir"]
     whitelist = config["whitelisted_warnings"]
-    src_root = config["pdc_root_dir"].format(host) 
-    src = os.path.join(src_root, archive)
     job_id = rq.get_current_job().id
     dest = "{}_{}".format(os.path.join(dest_root, archive), job_id)
 
-    download_ok = download_from_pdc(src, description, dest, dsmc_log_dir, whitelist)
+    download_ok = download_from_pdc(archive_path, description, dest, dsmc_log_dir, whitelist)
 
     if not download_ok:
         return {"state": "error", "msg": "failed to properly download archive from pdc", "path": dest}