From 070962c63547b0ae5373c57164524ff95053373d Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 26 Oct 2021 12:25:31 -0700
Subject: ingest_html: update trafilatura TEI-XML output kwarg

---
 python/sandcrawler/ingest_html.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'python/sandcrawler')

diff --git a/python/sandcrawler/ingest_html.py b/python/sandcrawler/ingest_html.py
index 7e6e5e3..9c72dd5 100644
--- a/python/sandcrawler/ingest_html.py
+++ b/python/sandcrawler/ingest_html.py
@@ -21,7 +21,7 @@ TRAFILATURA_AGENT = f"trafilatura/{trafilatura.__version__}"
 def html_extract_body_teixml(doc: bytes) -> dict:
     try:
         tei_xml = trafilatura.extract(doc,
-            tei_output=True,
+            output_format='xmltei',
             include_comments=False,
             include_formatting=True,
         )
-- 
cgit v1.2.3