From b477aca7047cdb15cd16588a0976d0990d3a96a4 Mon Sep 17 00:00:00 2001
From: Manuel Souto Pico <m.soutopico@gmail.com>
Date: Wed, 4 Sep 2024 17:16:15 +0200
Subject: [PATCH] Patch to extract tu's note when there's one. Only the first
 one is extracted if there's more than one

---
 tmx2dataframe/tmx2dataframe.py | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/tmx2dataframe/tmx2dataframe.py b/tmx2dataframe/tmx2dataframe.py
index 11ea4cd..e263eae 100644
--- a/tmx2dataframe/tmx2dataframe.py
+++ b/tmx2dataframe/tmx2dataframe.py
@@ -10,6 +10,15 @@ def process_tuv(tuv):
     txt = seg.childNodes[0].data
     return lang, txt
 
+
+def extract_note(tu):
+    notes = tu.getElementsByTagName('note')
+    if len(notes) >= 1:
+        # if there's more than one note, only the first one will be extracted
+        # if all notes are to be extracted, they could be concatenated in the same value
+        return notes[0].childNodes[0].data
+
+
 def read(path):
 
     """Read function takes in a path to TMX translation file and outputs the metadata and a pandas dataframe.
@@ -51,6 +60,11 @@ def read(path):
                 'target_language': targetlang,
                 'target_sentence': targetsentence
             }
+
+            note = extract_note(tu)
+            if note:
+                item['note'] = note
+
             items.append(item)
 
     df = pd.DataFrame(items)