feat(backend): add data loading scripts

This commit is contained in:
2025-04-26 17:39:10 +01:00
parent 2aee8e73cb
commit 9fb2d24833
2 changed files with 446 additions and 0 deletions

View File

@@ -0,0 +1,227 @@
#!/usr/bin/env python
"""Validate YAML content files for correctness."""
import sys
from pathlib import Path
from typing import Any
import yaml
def validate_categories(data_dir: Path) -> tuple[set[str], list[str]]:
"""Validate categories.yaml and return valid slugs."""
errors: list[str] = []
slugs: set[str] = set()
categories_file = data_dir / "categories" / "categories.yaml"
if not categories_file.exists():
errors.append(f"Missing file: {categories_file}")
return slugs, errors
with open(categories_file) as f:
data = yaml.safe_load(f)
if "categories" not in data:
errors.append("categories.yaml: missing 'categories' key")
return slugs, errors
for i, cat in enumerate(data["categories"]):
prefix = f"categories.yaml[{i}]"
if "name" not in cat:
errors.append(f"{prefix}: missing 'name'")
if "slug" not in cat:
errors.append(f"{prefix}: missing 'slug'")
else:
if cat["slug"] in slugs:
errors.append(f"{prefix}: duplicate slug '{cat['slug']}'")
slugs.add(cat["slug"])
return slugs, errors
def validate_patterns(data_dir: Path) -> tuple[set[str], list[str]]:
"""Validate patterns.yaml and return valid slugs."""
errors: list[str] = []
slugs: set[str] = set()
patterns_file = data_dir / "patterns" / "patterns.yaml"
if not patterns_file.exists():
errors.append(f"Missing file: {patterns_file}")
return slugs, errors
with open(patterns_file) as f:
data = yaml.safe_load(f)
if "patterns" not in data:
errors.append("patterns.yaml: missing 'patterns' key")
return slugs, errors
for i, pat in enumerate(data["patterns"]):
prefix = f"patterns.yaml[{i}]"
if "name" not in pat:
errors.append(f"{prefix}: missing 'name'")
if "slug" not in pat:
errors.append(f"{prefix}: missing 'slug'")
else:
if pat["slug"] in slugs:
errors.append(f"{prefix}: duplicate slug '{pat['slug']}'")
slugs.add(pat["slug"])
return slugs, errors
def validate_question(
question_file: Path,
valid_categories: set[str],
valid_patterns: set[str],
seen_slugs: set[str],
seen_leetcode_ids: set[int],
) -> list[str]:
"""Validate a single question file."""
errors: list[str] = []
filename = question_file.name
try:
with open(question_file) as f:
data: dict[str, Any] = yaml.safe_load(f)
except yaml.YAMLError as e:
errors.append(f"{filename}: invalid YAML - {e}")
return errors
# Required fields
required = ["title", "slug", "difficulty", "description"]
for field in required:
if field not in data:
errors.append(f"{filename}: missing required field '{field}'")
# Validate slug
if "slug" in data:
if data["slug"] in seen_slugs:
errors.append(f"{filename}: duplicate slug '{data['slug']}'")
seen_slugs.add(data["slug"])
# Slug should match filename
expected_filename = f"{data['slug']}.yaml"
if question_file.name != expected_filename:
errors.append(f"{filename}: filename should be '{expected_filename}'")
# Validate difficulty
if "difficulty" in data:
valid_difficulties = {"easy", "medium", "hard"}
if data["difficulty"] not in valid_difficulties:
errors.append(
f"{filename}: invalid difficulty '{data['difficulty']}' "
f"(must be one of {valid_difficulties})"
)
# Validate categories
for cat in data.get("categories", []):
if cat not in valid_categories:
errors.append(f"{filename}: unknown category '{cat}'")
# Validate patterns
for pat in data.get("patterns", []):
if pat not in valid_patterns:
errors.append(f"{filename}: unknown pattern '{pat}'")
# Validate leetcode_id uniqueness
if "leetcode_id" in data and data["leetcode_id"] is not None:
lid = data["leetcode_id"]
if lid in seen_leetcode_ids:
errors.append(f"{filename}: duplicate leetcode_id {lid}")
seen_leetcode_ids.add(lid)
# Validate explanation
if "explanation" in data:
exp = data["explanation"]
exp_required = ["approach", "intuition", "time_complexity", "space_complexity"]
for field in exp_required:
if field not in exp:
errors.append(f"{filename}: explanation missing '{field}'")
# Validate solutions
if "solutions" in data:
for i, sol in enumerate(data["solutions"]):
if "approach_name" not in sol:
errors.append(f"{filename}: solutions[{i}] missing 'approach_name'")
if "code" not in sol:
errors.append(f"{filename}: solutions[{i}] missing 'code'")
else:
errors.append(f"{filename}: missing 'solutions'")
return errors
def validate_questions(
data_dir: Path,
valid_categories: set[str],
valid_patterns: set[str],
) -> list[str]:
"""Validate all question files."""
errors: list[str] = []
questions_dir = data_dir / "questions"
if not questions_dir.exists():
errors.append(f"Missing directory: {questions_dir}")
return errors
seen_slugs: set[str] = set()
seen_leetcode_ids: set[int] = set()
question_files = list(questions_dir.glob("*.yaml"))
if not question_files:
errors.append("No question files found")
return errors
for question_file in sorted(question_files):
file_errors = validate_question(
question_file,
valid_categories,
valid_patterns,
seen_slugs,
seen_leetcode_ids,
)
errors.extend(file_errors)
return errors
def main() -> int:
"""Validate all content files."""
data_dir = Path(__file__).parent.parent / "data"
print(f"Validating content in {data_dir}...\n")
all_errors: list[str] = []
# Validate categories
valid_categories, cat_errors = validate_categories(data_dir)
all_errors.extend(cat_errors)
print(f"Categories: {len(valid_categories)} valid, {len(cat_errors)} errors")
# Validate patterns
valid_patterns, pat_errors = validate_patterns(data_dir)
all_errors.extend(pat_errors)
print(f"Patterns: {len(valid_patterns)} valid, {len(pat_errors)} errors")
# Validate questions
questions_dir = data_dir / "questions"
question_count = len(list(questions_dir.glob("*.yaml"))) if questions_dir.exists() else 0
question_errors = validate_questions(data_dir, valid_categories, valid_patterns)
all_errors.extend(question_errors)
print(f"Questions: {question_count} files, {len(question_errors)} errors")
print()
if all_errors:
print("Validation errors:")
for error in all_errors:
print(f" - {error}")
return 1
print("All content is valid!")
return 0
if __name__ == "__main__":
sys.exit(main())