Upload knessetCorpus.py

knessetCorpus.py

@@ -357,8 +357,7 @@ class KnessetCorpus(datasets.GeneratorBasedBuilder):
 
     def _generate_examples(self, data_files):
         """
-        Generate examples for both streaming and non-streaming modes.
-        Handles `.bz2` and `.zip` files based on mode.
+        Generate examples from files, supporting both streaming and non-streaming modes.
         """
         if "all_features_sentences" in self.config.name:
             id_field_name = "sentence_id"
@@ -373,54 +372,31 @@ class KnessetCorpus(datasets.GeneratorBasedBuilder):
 
         for filepath in data_files:
             try:
-                if filepath.endswith(".bz2"):
-                    # Stream `.bz2` file
-                    with bz2.open(filepath, "rt", encoding="utf-8") as f:
-                        for line_number, line in enumerate(f):
-                            yield from self._process_line(line, id_field_name, line_number, filepath)
-                elif filepath.endswith(".zip"):
-                    # Stream `.zip` file
-                    with zipfile.ZipFile(filepath, "r") as archive:
-                        for filename in archive.namelist():
-                            with archive.open(filename) as f:
-                                for line_number, line in enumerate(io.TextIOWrapper(f, encoding="utf-8")):
-                                    yield from self._process_line(line, id_field_name, line_number, filename)
-                else:
-                    # Stream `.jsonl` or other uncompressed files
-                    with open(filepath, "rt", encoding="utf-8") as f:
-                        for line_number, line in enumerate(f):
-                            yield from self._process_line(line, id_field_name, line_number, filepath)
+                # Hugging Face automatically handles streaming
+                for line_number, row in enumerate(filepath):
+                    try:
+                        row = json.loads(row)
+                    except json.JSONDecodeError as e:
+                        print(f"Failed to decode JSON at line {line_number} in file {filepath}: {e}. Skipping.")
+                        continue
+
+                    # Extract the ID field
+                    id_ = row.get(id_field_name, f"unknown_{line_number}")
+                    if not id_:
+                        print(f"Key '{id_field_name}' not found in row at line {line_number}. Skipping row.")
+                        continue
+
+                    # Prepare the sample dictionary
+                    sample = {feature: row.get(feature, None) for feature in self._info().features}
+
+                    # Handle nested structures (e.g., `protocol_sentences`)
+                    if "protocol_sentences" in row:
+                        protocol_sentences = row.get("protocol_sentences", [])
+                        for sentence in protocol_sentences:
+                            sentence_id = sentence.get("sentence_id", None)
+                            if sentence_id:
+                                yield sentence_id, sentence
+                    else:
+                        yield id_, sample
             except Exception as e:
                 print(f"Error processing file '{filepath}': {e}. Skipping.")
-
-    def _process_line(self, line, id_field_name, line_number, filepath):
-        """
-        Process a single line of input, handling JSON decoding and nested structures.
-        """
-        try:
-            row = json.loads(line)
-        except json.JSONDecodeError as e:
-            print(f"Failed to decode JSON at line {line_number} in file {filepath}: {e}. Skipping.")
-            return
-
-        # Extract the ID field
-        id_ = row.get(id_field_name, f"unknown_{line_number}")
-        if not id_:
-            print(f"Key '{id_field_name}' not found in row at line {line_number}. Skipping row.")
-            return
-
-        # Prepare the sample dictionary
-        sample = {feature: row.get(feature, None) for feature in self._info().features}
-
-        # Handle nested structures (e.g., `protocol_sentences`)
-        if "protocol_sentences" in row:
-            protocol_sentences = row.get("protocol_sentences", [])
-            for sentence in protocol_sentences:
-                sentence_id = sentence.get("sentence_id", None)
-                if sentence_id:
-                    yield sentence_id, sentence
-        else:
-            try:
-                yield id_, sample
-            except Exception as e:
-                print(f"Failed to yield sample at line {line_number} in file {filepath}. Error: {e}. Sample: {sample}.")