From ee797ddc0a1377423cfe1939634e6d019eecea9e Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Fri, 4 Oct 2019 14:09:59 -0700
Subject: grobid2json: language_code

---
 python/grobid2json.py         | 5 +++++
 python/tests/files/small.json | 3 ++-
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/python/grobid2json.py b/python/grobid2json.py
index e5af9d3..ae31486 100755
--- a/python/grobid2json.py
+++ b/python/grobid2json.py
@@ -28,6 +28,7 @@ import json
 import argparse
 import xml.etree.ElementTree as ET
 
+xml_ns = "http://www.w3.org/XML/1998/namespace"
 ns = "http://www.tei-c.org/ns/1.0"
 
 def all_authors(elem):
@@ -143,6 +144,10 @@ def teixml2json(content, encumbered=True):
         refs.append(ref)
     info['citations'] = refs
 
+    text = tei.find('.//{%s}text' % (ns))
+    print(text.attrib)
+    info['language_code'] = text.attrib['{%s}lang' % xml_ns]  # xml:lang
+
     if encumbered:
         el = tei.find('.//{%s}profileDesc/{%s}abstract' % (ns, ns))
         info['abstract'] = (el or None) and " ".join(el.itertext()).strip()
diff --git a/python/tests/files/small.json b/python/tests/files/small.json
index eb93d9e..35bf62d 100644
--- a/python/tests/files/small.json
+++ b/python/tests/files/small.json
@@ -56,5 +56,6 @@
   "annex": null,
   "fatcat_release": null,
   "grobid_timestamp": "2018-04-02T00:31+0000",
-  "grobid_version": "0.5.1-SNAPSHOT"
+  "grobid_version": "0.5.1-SNAPSHOT",
+  "language_code": "en"
 }
-- 
cgit v1.2.3