From b9ac270dae6bc36ef247ffda658aa9dc6c08d286 Mon Sep 17 00:00:00 2001
From: Jelte van Boheemen <j.vanboheemen@uu.nl>
Date: Wed, 15 May 2024 15:10:41 +0200
Subject: [PATCH] Remove manual override keys for uttid in ChatReader

Resolves #30
---
 corpus2alpino/readers/chat.py   | 13 ++-----------
 tests/example_chat_expected.txt |  4 ++--
 2 files changed, 4 insertions(+), 13 deletions(-)

diff --git a/corpus2alpino/readers/chat.py b/corpus2alpino/readers/chat.py
index 6fe5111..3331ca7 100644
--- a/corpus2alpino/readers/chat.py
+++ b/corpus2alpino/readers/chat.py
@@ -2,18 +2,16 @@
 """
 Module for reading CHAT cha files to parsable utterances.
 """
-from typing import cast, Dict, Iterable, List, Tuple
+from typing import cast, Dict, Iterable, List
 from chamd import ChatReader as ChatParser, ChatLine, ChatTier
 
-import os
-import re
 
 from corpus2alpino.abstracts import Reader
 from corpus2alpino.models import CollectedFile, Document, MetadataValue, Utterance
 
-MANUAL_IDS = ['xsid', 'xuid']
 UTTERANCE_NUMBER_ID = 'uttno'
 
+
 class ChatReader(Reader):
     """
     Class for converting a CHAT file to document.
@@ -31,13 +29,6 @@ def parse_utterances(self, chat_lines: List[ChatLine]):
         number = 0
         for line in chat_lines:
             number += 1  # start numbering utterances from 1
-            for id_override_key in MANUAL_IDS:
-                try:
-                    line.uttid = line.tiers[id_override_key].text
-                    line.metadata['uttid'].text = line.uttid
-                    break
-                except KeyError:
-                    pass
 
             yield Utterance(line.text,
                             str(line.uttid),
diff --git a/tests/example_chat_expected.txt b/tests/example_chat_expected.txt
index 9211a62..720ac96 100644
--- a/tests/example_chat_expected.txt
+++ b/tests/example_chat_expected.txt
@@ -26,11 +26,11 @@
 ##META text origutt = toen <ging zij &-eh haar &+ma ging zij> [//] wou zij de auto maken . 21792_23475
 ##META text parsefile = PRELAN_example_chat_u00000000002.xml
 ##META int uttendlineno = 10
-##META int uttid = 42
+##META int uttid = 2
 ##META int uttno = 2
 ##META int uttstartlineno = 9
 ##META text xsid = 42
-42|toen wou zij de auto maken .
+2|toen wou zij de auto maken .
 
 ##META text origutt = maar toen reed de auto er vandoor .
 ##META text parsefile = PRELAN_example_chat_u00000000003.xml