.

minipileoflaw.py

@@ -94,21 +94,21 @@ class MiniPileOfLaw(datasets.GeneratorBasedBuilder):
                 name=datasets.Split.VALIDATION, gen_kwargs={"filepaths": validation_downloaded_files}
             ),
         ]
-
-
-
+    
     def _generate_examples(self, filepaths):
-        """This function returns the examples from CSV files, handling escape characters."""
+        """This function returns the examples from JSONL files, with each line being a separate JSON object."""
         for filepath in filepaths:
             try:
-                # Reading the CSV file with the appropriate escape character
-                df = pd.read_csv(filepath, escapechar='\\')
-                for _, row in df.iterrows():
-                    yield {
-                        "text": row.get("text", ""),
-                        "created_timestamp": row.get("created_timestamp", ""),
-                        "downloaded_timestamp": row.get("downloaded_timestamp", ""),
-                        "url": row.get("url", "")
-                    }
+                with open(filepath, 'r', encoding='utf-8') as file:
+                    for line in file:
+                        # Parse each line as a separate JSON object
+                        item = json.loads(line.strip())
+                        yield {
+                            "text": item.get("text", ""),
+                            "created_timestamp": item.get("created_timestamp", ""),
+                            "downloaded_timestamp": item.get("downloaded_timestamp", ""),
+                            "url": item.get("url", "")
+                        }
             except Exception as e:
                 print(f"Error reading file {filepath}: {e}")
+