Update multiconer_v2.py

multiconer_v2.py

@@ -287,52 +287,38 @@ class MultiCoNER2(datasets.GeneratorBasedBuilder):
         logger.info("⏳ Generating examples from = %s", filepath)
         
 
-        content = open(filepath, 'r', encoding="utf-8").read().strip()
-        lines = content.split('\n')
-
-        items = [(4, {'id': '5fcb44ad-c1ef-4925-be27-008a76667b3b', 'tokens': ['রাস্তায়', 'রঙিন', 'লাইট', 'এবং', 'বাদশাহী', 'মসজিদ', 'এর', 'দৃশ্য', 'রয়েছে', 'এবং', 'এটি', 'রোশনাই', 'গেট', 'এর', 'নিকটে', 'রয়েছে'], 'ner_tags': ['O', 'O', 'O', 'O', 'B-Facility', 'I-Facility', 'O', 'O', 'O', 'O', 'O', 'B-OtherPROD', 'I-OtherPROD', 'O', 'O', 'O']}),
-            (5, {'id': '44334ba4-0ebb-4001-83a7-525473eba921', 'tokens': ['ফাইল', ':', 'একে', '৭৪', 'বোল্ট।', 'জেপিজি', '|', 'একে-৭৪', 'বোল্ট', 'এবং', 'ফায়ারিং', 'পিন।'], 'ner_tags': ['O', 'O', 'O', 'O', 'O', 'O', 'O', 'B-OtherPROD', 'O', 'O', 'O', 'O']}),
-            (6, {'id': '8c87e2f6-cacc-43a8-bb0f-40aa3817af48', 'tokens': ['যখন', 'নিমজ্জিত', 'হয়', 'প্রতিটি', 'প্রোপেলার', 'একটি', 'বৈদ্যুতিক', 'মোটর', 'দ্বারা', 'চালিত', 'হয়েছিল।'], 'ner_tags': ['O', 'O', 'O', 'O', 'O', 'O', 'B-OtherPROD', 'I-OtherPROD', 'O', 'O', 'O']}),
-            (7, {'id': '624ef8af-c8d6-4f28-b946-53344d721b6e', 'tokens': ['বেশ', 'কয়েকটি', 'অ্যাসকোমাইসেট', 'ছত্রাকের', 'বংশাণুসমগ্র', 'পুরোপুরি', 'ক্রমযুক্ত', 'হয়েছে।'], 'ner_tags': ['O', 'O', 'O', 'O', 'B-OtherPROD', 'O', 'O', 'O']})]
-
-        items.append((6, {'id': 'asdas624ef8af-c8d6-4f28-b946-53344d721b6e', 'tokens': lines[:5], 'ner_tags': lines[5:10]} ))
-        for it in items:
-            yield it
-            
-        guid = -1
-        s_id = None
-        tokens = []
-        ner_tags = []
-
-        for line in lines:
-            if line.startswith("# id"):
-                s_id = line.split('\t')[0].split(' ')[-1].strip()
-                guid += 1
-                tokens = []
-                ner_tags = []            
-            elif len(line.strip()) < 3:
-                if s_id and len(tokens) >= 1 and len(tokens) == len(ner_tags):
-                    yield (guid, {
-                        "id": str(guid),
-                        "sample_id": str(s_id),
-                        "tokens": tokens,
-                        "ner_tags": ner_tags,
-                    })
-                    s_id = None
+        with open(filepath) as f:                    
+            guid = -1
+            s_id = None
+            tokens = []
+            ner_tags = []
+    
+            for line in f:
+                if line.strip().startswith("# id"):
+                    s_id = line.split('\t')[0].split(' ')[-1].strip()
+                    guid += 1
                     tokens = []
                     ner_tags = []
-                else:
-                    continue
-            else:
-                # Separator is " _ _ "
-                splits = line.split()
-                if len(splits) > 2:
+                elif '_ _' in line:
+                    # Separator is " _ _ "
+                    splits = line.split("_ _")
                     tokens.append(splits[0].strip())
-                    ner_tags.append(splits[-1].strip())
-        if s_id:
-            yield (guid, {
-                "id": str(guid),
-                "sample_id": str(s_id),
-                "tokens": tokens,
-                "ner_tags": ner_tags,
-            })
+                    ner_tags.append(splits[1].strip())
+                elif len(line.strip()) == 0:
+                    if s_id and len(tokens) >= 1 and len(tokens) == len(ner_tags):
+                        yield guid, {
+                            "id": s_id,
+                            "tokens": tokens,
+                            "ner_tags": ner_tags,
+                        }
+                        s_id = None
+                        tokens = []
+                        ner_tags = []
+                    else:
+                        continue
+            if s_id and len(tokens) >= 1 and len(tokens) == len(ner_tags):
+                yield guid, {
+                    "id": s_id,
+                    "tokens": tokens,
+                    "ner_tags": ner_tags,
+                }