From 8f964b9b48572ac71f27ba64207816dfd3a6dc36 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 3 Nov 2020 22:40:57 -0800
Subject: small fixes from local testing for XML ingest

---
 python/sandcrawler/ingest.py  |  2 +-
 python/sandcrawler/persist.py | 11 ++++++++---
 python/sandcrawler_worker.py  |  2 --
 3 files changed, 9 insertions(+), 6 deletions(-)

(limited to 'python')

diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py
index 363485e..2e227bf 100644
--- a/python/sandcrawler/ingest.py
+++ b/python/sandcrawler/ingest.py
@@ -336,7 +336,7 @@ class IngestFileWorker(SandcrawlerWorker):
         )
 
     def want(self, request: dict) -> bool:
-        if not request.get('ingest_type') in ('file', 'pdf'):
+        if not request.get('ingest_type') in ('file', 'pdf', 'xml'):
             return False
         return True
 
diff --git a/python/sandcrawler/persist.py b/python/sandcrawler/persist.py
index 81cf664..c225d5a 100644
--- a/python/sandcrawler/persist.py
+++ b/python/sandcrawler/persist.py
@@ -20,7 +20,7 @@ grobid
 """
 
 import os
-from typing import Optional
+from typing import Optional, AnyStr
 import xml.etree.ElementTree
 
 from sandcrawler.workers import SandcrawlerWorker
@@ -518,12 +518,17 @@ class GenericPersistDocWorker(SandcrawlerWorker):
         self.s3_folder = kwargs.get('s3_folder', "unknown")
         self.doc_key = "unknown"
 
-    def process(self, record: dict, key: Optional[str] = None) -> None:
+    def process(self, record: dict, raw_key: Optional[AnyStr] = None) -> None:
 
         if record.get('status') != 'success' or not record.get(self.doc_key):
             return
 
-        assert key is not None and len(key) == 40 and isinstance(key, str)
+        assert raw_key is not None
+        if isinstance(raw_key, bytes):
+            key = raw_key.decode('utf-8')
+        elif isinstance(raw_key, str):
+            key = raw_key
+        assert len(key) == 40
         if 'sha1hex' in record:
             assert key == record['sha1hex']
 
diff --git a/python/sandcrawler_worker.py b/python/sandcrawler_worker.py
index 24dbdd0..3681d7f 100755
--- a/python/sandcrawler_worker.py
+++ b/python/sandcrawler_worker.py
@@ -163,7 +163,6 @@ def run_persist_xml_doc(args: argparse.Namespace) -> None:
         consume_topic=consume_topic,
         group="persist-xml-doc",
         push_batches=False,
-        raw_records=True,
         batch_size=25,
     )
     pusher.run()
@@ -182,7 +181,6 @@ def run_persist_html_teixml(args: argparse.Namespace) -> None:
         consume_topic=consume_topic,
         group="persist-html-teixml",
         push_batches=False,
-        raw_records=True,
         batch_size=25,
     )
     pusher.run()
-- 
cgit v1.2.3