From 1a63c31b71e72ef4a25a4cf331db4f70e064057e Mon Sep 17 00:00:00 2001
From: Tessa Walsh <tessa@bitarchivist.net>
Date: Wed, 26 Apr 2023 17:36:48 -0400
Subject: [PATCH] backend: errors endpoint: Parse JSON-l errors before
 returning (#799)

---
 backend/btrixcloud/crawls.py | 13 ++++++++-----
 backend/btrixcloud/utils.py  | 18 ++++++++++++++++++
 2 files changed, 26 insertions(+), 5 deletions(-)

diff --git a/backend/btrixcloud/crawls.py b/backend/btrixcloud/crawls.py
index 2c0a5913..ec82475b 100644
--- a/backend/btrixcloud/crawls.py
+++ b/backend/btrixcloud/crawls.py
@@ -24,7 +24,7 @@ from .orgs import Organization, MAX_CRAWL_SCALE
 from .pagination import DEFAULT_PAGE_SIZE, paginated_format
 from .storages import get_presigned_url, delete_crawl_file_object, get_wacz_logs
 from .users import User
-from .utils import dt_now, ts_now, get_redis_crawl_stats
+from .utils import dt_now, ts_now, get_redis_crawl_stats, parse_jsonl_error_messages
 
 
 CRAWL_STATES = (
@@ -784,12 +784,14 @@ class CrawlOps:
         try:
             redis = await self.get_redis(crawl_id)
             errors = await redis.lrange(f"{crawl_id}:e", skip, page_size)
-            total = len(errors)
         except exceptions.ConnectionError:
             # pylint: disable=raise-missing-from
             raise HTTPException(status_code=503, detail="redis_connection_error")
 
-        return errors, total
+        parsed_errors = parse_jsonl_error_messages(errors)
+        total = len(parsed_errors)
+
+        return parsed_errors, total
 
     async def get_redis(self, crawl_id):
         """get redis url for crawl id"""
@@ -1240,8 +1242,9 @@ def init_crawls_api(app, mdb, users, crawl_manager, crawl_config_ops, orgs, user
             skip = (page - 1) * pageSize
             upper_bound = skip + pageSize - 1
             errors = crawl.errors[skip:upper_bound]
-            total = len(errors)
-            return paginated_format(errors, total, page, pageSize)
+            parsed_errors = parse_jsonl_error_messages(errors)
+            total = len(parsed_errors)
+            return paginated_format(parsed_errors, total, page, pageSize)
 
         errors, total = await ops.get_errors_from_redis(crawl_id, pageSize, page)
         return paginated_format(errors, total, page, pageSize)
diff --git a/backend/btrixcloud/utils.py b/backend/btrixcloud/utils.py
index c82f48c5..10cff21e 100644
--- a/backend/btrixcloud/utils.py
+++ b/backend/btrixcloud/utils.py
@@ -2,6 +2,7 @@
 
 import os
 import asyncio
+import json
 import sys
 import signal
 import atexit
@@ -83,3 +84,20 @@ def register_exit_handler():
         sys.exit(1)
 
     loop.add_signal_handler(signal.SIGTERM, exit_handler)
+
+
+def parse_jsonl_error_messages(errors):
+    """parse json-l error strings from redis/db into json"""
+    parsed_errors = []
+    for error_line in errors:
+        if not error_line:
+            continue
+        try:
+            result = json.loads(error_line)
+            parsed_errors.append(result)
+        except json.JSONDecodeError as err:
+            print(
+                f"Error decoding json-l error line: {error_line}. Error: {err}",
+                flush=True,
+            )
+    return parsed_errors