speech_data_recordings / check_short_sections.py
BaniGuru
Initial commit
911e360
Raw History Blame
3.88 kB
#!/usr/bin/env python3
"""Find transcript sections shorter than a duration threshold.
Searches JSON files recursively. Supports transcript files with a top-level
``sections`` array, including ``aligned_transcript.json`` and
``transcript_sections.json``.
Examples:
python find_short_sections.py /path/to/root
python find_short_sections.py /path/to/root --threshold 1.5
"""
from __future__ import annotations
import argparse
import json
from pathlib import Path
def get_sections(payload: object) -> list[dict]:
if isinstance(payload, dict):
sections = payload.get("sections", [])
elif isinstance(payload, list):
sections = payload
else:
return []
if not isinstance(sections, list):
return []
return [section for section in sections if isinstance(section, dict)]
def main() -> int:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument(
"root",
nargs="?",
type=Path,
default=Path.cwd(),
help="Folder to search recursively (default: current folder)",
)
parser.add_argument(
"--threshold",
type=float,
default=0.7,
metavar="SECONDS",
help="Report sections shorter than this many seconds (default: 2)",
)
args = parser.parse_args()
root = args.root.expanduser().resolve()
if not root.is_dir():
parser.error(f"root folder does not exist or is not a directory: {root}")
if args.threshold <= 0:
parser.error("--threshold must be greater than 0 seconds")
json_files = sorted(root.rglob("*.json"))
scanned_files = 0
examined_sections = 0
matches: list[tuple[Path, dict, float, float, float]] = []
errors: list[tuple[Path, str]] = []
for json_path in json_files:
if any(part.startswith(".") for part in json_path.relative_to(root).parts):
continue
try:
payload = json.loads(json_path.read_text(encoding="utf-8-sig"))
except (OSError, json.JSONDecodeError) as exc:
errors.append((json_path, str(exc)))
continue
sections = get_sections(payload)
if not sections:
continue
scanned_files += 1
for section in sections:
try:
start = float(section["start_time"])
end = float(section["end_time"])
except (KeyError, TypeError, ValueError):
continue
examined_sections += 1
duration = end - start
if duration < args.threshold:
matches.append((json_path, section, start, end, duration))
matches.sort(key=lambda item: (str(item[0]).casefold(), item[2], item[3]))
if matches:
for json_path, section, start, end, duration in matches:
try:
display_path = json_path.relative_to(root)
except ValueError:
display_path = json_path
transcript = " ".join(str(section.get("transcript", "")).split())
print(
f"{display_path}\n"
f" section_id: {section.get('section_id', '(no section_id)')}\n"
f" duration: {duration:.3f}s\n"
f" start: {start:.3f}\n"
f" end: {end:.3f}\n"
f" transcript: {transcript}\n"
)
else:
print(f"No sections shorter than {args.threshold:g} seconds found.")
print(f"Files with sections: {scanned_files}")
print(f"Sections checked: {examined_sections}")
print(f"Sections shorter than {args.threshold:g}s: {len(matches)}")
if errors:
print(f"Unreadable JSON files: {len(errors)}")
for path, error in errors:
print(f" {path.relative_to(root)}: {error}")
return 0
if __name__ == "__main__":
raise SystemExit(main())