Move data cleaning scripts to clean module
This commit is contained in:
parent
87be49f30f
commit
3b78a84782
7 changed files with 5 additions and 5 deletions
65
code/clean/packages.py
Normal file
65
code/clean/packages.py
Normal file
|
|
@ -0,0 +1,65 @@
|
|||
import csv
|
||||
import json
|
||||
from pathlib import Path
|
||||
from typing import cast
|
||||
|
||||
|
||||
def packages_csv(input_dir: Path, output_dir: Path) -> None:
|
||||
output_file = output_dir / "packages.csv"
|
||||
with output_file.open("w") as fw:
|
||||
writer = csv.writer(fw)
|
||||
writer.writerow(["date", "package", "version", "count"])
|
||||
|
||||
for j in input_dir.glob("*.json"):
|
||||
with open(j) as fr:
|
||||
date = j.stem
|
||||
data: dict[str, object] = {}
|
||||
try:
|
||||
data = json.load(fr)
|
||||
except json.JSONDecodeError:
|
||||
print(f"WARN: Could not decode JSON data for file {j}")
|
||||
continue
|
||||
|
||||
if "Versions" not in data or not isinstance(data["Versions"], dict):
|
||||
print(
|
||||
f"WARN: No correct json structure containing 'Versions' field in file {j}"
|
||||
)
|
||||
continue
|
||||
|
||||
data_versions = cast(dict[str, dict[str, int]], data["Versions"])
|
||||
for package_name, package_versions in data_versions.items():
|
||||
if not isinstance(package_versions, dict):
|
||||
print(
|
||||
f"WARN: No correct json version structure containing versions in the Version field in file {j}"
|
||||
)
|
||||
continue
|
||||
for version, count in package_versions.items():
|
||||
p_name = package_name
|
||||
p_version = version
|
||||
v_count = count
|
||||
p_date = date
|
||||
writer.writerow([p_date, p_name, p_version, v_count])
|
||||
|
||||
|
||||
def ensure_dirs(input_dir: Path, output_dir: Path):
|
||||
if not input_dir.is_dir():
|
||||
raise ValueError
|
||||
output_dir.mkdir(exist_ok=True, parents=True)
|
||||
|
||||
|
||||
def main(input: str, output: str) -> None:
|
||||
input_dir = Path(input)
|
||||
output_dir = Path(output)
|
||||
ensure_dirs(input_dir, output_dir)
|
||||
packages_csv(input_dir, output_dir)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
import sys
|
||||
|
||||
if not len(sys.argv) == 3:
|
||||
print("Please provide exactly one input directory and one output directory.")
|
||||
sys.exit(1)
|
||||
inp = sys.argv[1]
|
||||
out = sys.argv[2]
|
||||
main(inp, out)
|
||||
Loading…
Add table
Add a link
Reference in a new issue