From 64119bd84e6306695f80bca52de8cf9b7dbe8da2 Mon Sep 17 00:00:00 2001
From: Al Niessner <Al.Niessner@xxx.xxx>
Date: Tue, 1 Aug 2023 14:55:31 -0700
Subject: [PATCH] minor updates for running

---
 src/pds/registrysweepers/repairkit/__init__.py  | 9 ++++++---
 src/pds/registrysweepers/repairkit/allarrays.py | 6 ++++++
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/src/pds/registrysweepers/repairkit/__init__.py b/src/pds/registrysweepers/repairkit/__init__.py
index 9ce094d..1cd49ec 100644
--- a/src/pds/registrysweepers/repairkit/__init__.py
+++ b/src/pds/registrysweepers/repairkit/__init__.py
@@ -56,11 +56,14 @@ def run(base_url:str,
     host = Host(password, base_url, username, verify_host_certs)
     query = {"match_all":{}}
     for document in query_registry_db(host, query, {}):
+        id = document['_id']
+        src = document['_source']
         repairs = {}
-        for fieldname,data in document.items():
+        log.debug (f'working on document: {id}')
+        for fieldname,data in src.items():
             for regex,funcs in REPAIR_TOOLS.items():
-                if regex(filename):
-                    repairs.update(func(document, fieldname) for func in funcs)
+                if regex(fieldname):
+                    for func in funcs: repairs.update(func(src, fieldname))
         if repairs:
             log.info(f'Writing repairs to document: {id}')
             write_update_docs(host, {id:repairs})
diff --git a/src/pds/registrysweepers/repairkit/allarrays.py b/src/pds/registrysweepers/repairkit/allarrays.py
index cddc26a..cee304c 100644
--- a/src/pds/registrysweepers/repairkit/allarrays.py
+++ b/src/pds/registrysweepers/repairkit/allarrays.py
@@ -1,6 +1,12 @@
 '''change single strings to array of strings'''
 
+import logging
+
+log = logging.getLogger(__name__)
+
+
 def repair (document:{}, fieldname:str)->bool:
+    log.debug(f'checking {fieldname}')
     if isinstance (document[fieldname], str):
         log.info (f'found string for {fieldname} where it should be an array')
         return {fieldname:[document[fieldname]]}