Migrate to gitea

scripts/printline.py

@@ -0,0 +1,38 @@
+import json
+import pathlib
+from datetime import datetime
+
+
+INPUT_FILE_NAME = "fr_raw-wiktextract-data.jsonl" 
+SCRIPT_DIR = pathlib.Path(__file__).parent
+ROOT_DIR = SCRIPT_DIR.parent
+INPUT_FILE = ROOT_DIR / "raw_data" / INPUT_FILE_NAME
+
+
+
+# --- Configuration ---
+START_LINE = 99  # 1-based index (first line is 1)
+NUM_LINES = 99  # Number of lines/objects to write
+
+
+def extract_lines_to_file(file_path, start_line, num_lines):
+    # Generate timestamp filename
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    output_file = file_path.parent / f"{timestamp}.json"
+    
+    with open(file_path, 'r', encoding='utf-8') as infile:
+        with open(output_file, 'w', encoding='utf-8') as outfile:
+            for i, line in enumerate(infile, start=1):
+                if i >= start_line and i < start_line + num_lines:
+                    try:
+                        element = json.loads(line)
+                        outfile.write(json.dumps(element, indent=2, ensure_ascii=False))
+                        outfile.write('\n')
+                    except json.JSONDecodeError:
+                        outfile.write(f"Error: Line {i} is not valid JSON.\n")
+    
+    print(f"Output written to: {output_file}")
+
+
+if __name__ == "__main__":
+    extract_lines_to_file(INPUT_FILE, START_LINE, NUM_LINES)