Implement census analysis: read, count, summarise, write and CLI

Closes #11
Closes #12
Closes #13
Closes #14
Closes #15

Task: MIL-002#4
Task: MIL-002#5
Task: MIL-002#6
Task: MIL-002#7
Task: MIL-002#8

src/squirrel/__main__.py

@@ -0,0 +1,10 @@
+##
+# @file __main__.py
+# @brief Allows `python -m squirrel` to run the command line.
+
+"""Allows `python -m squirrel` to run the command line."""
+
+from squirrel.cli import main
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/squirrel/census.py

@@ -0,0 +1,63 @@
+##
+# @file census.py
+# @brief Pure steps of the census analysis: read, count, summarise, write.
+
+"""Pure steps of the census analysis: read, count, summarise, write."""
+
+import collections
+import csv
+import pathlib
+from collections.abc import Iterable, Mapping, Sequence
+
+from squirrel.constants import (
+    FILE_ENCODING,
+    FUR_COLOR_COLUMN,
+    FUR_COLORS,
+    SUMMARY_COLUMNS,
+)
+
+
+##
+# @brief Read a CSV file into one dictionary per row, keyed by the header names.
+# @param path Location of the CSV file.
+# @return The rows of the file, in file order.
+def read_csv(path: pathlib.Path) -> list[dict[str, str]]:
+    """Read a CSV file into one dictionary per row, keyed by the header names."""
+    with path.open(encoding=FILE_ENCODING, newline="") as file:
+        return list(csv.DictReader(file))
+
+
+##
+# @brief Count the rows per fur colour; other or blank colours are counted in none.
+# @param rows Census rows, each holding the fur colour column.
+# @param colors Fur colours to count, in the order of the result.
+# @return Count per colour of `colors`, 0 for a colour that does not occur.
+def count_by_fur_color(
+    rows: Iterable[Mapping[str, str]],
+    colors: Sequence[str] = FUR_COLORS,
+) -> dict[str, int]:
+    """Count the rows per fur colour; other or blank colours are counted in none."""
+    seen = collections.Counter(row.get(FUR_COLOR_COLUMN) for row in rows)
+    return {color: seen[color] for color in colors}
+
+
+##
+# @brief Turn the counts into summary rows, keeping the order of the counts.
+# @param counts Count per fur colour.
+# @return One (colour, count) pair per colour.
+def build_summary(counts: Mapping[str, int]) -> list[tuple[str, int]]:
+    """Turn the counts into summary rows, keeping the order of the counts."""
+    return list(counts.items())
+
+
+##
+# @brief Write the summary as a CSV file: the header row, then one row per colour.
+# @param summary Rows to write, as built by `build_summary`.
+# @param path Location of the CSV file; its folder must exist.
+# @return None.
+def write_csv(summary: Sequence[tuple[str, int]], path: pathlib.Path) -> None:
+    """Write the summary as a CSV file: the header row, then one row per colour."""
+    with path.open("w", encoding=FILE_ENCODING, newline="") as file:
+        writer = csv.writer(file)
+        writer.writerow(SUMMARY_COLUMNS)
+        writer.writerows(summary)

src/squirrel/cli.py

@@ -0,0 +1,67 @@
+##
+# @file cli.py
+# @brief Command line entry point: count the squirrels and write the summary.
+
+"""Command line entry point: count the squirrels and write the summary."""
+
+import argparse
+import pathlib
+import sys
+from collections.abc import Sequence
+
+from squirrel import census
+from squirrel.constants import (
+    DEFAULT_INPUT_PATH,
+    DEFAULT_OUTPUT_PATH,
+    EXIT_INPUT_MISSING,
+)
+
+
+##
+# @brief Build the argument parser of the command line.
+# @return The parser with the optional `input` and `output` positionals.
+def _build_parser() -> argparse.ArgumentParser:
+    """Build the argument parser of the command line."""
+    parser = argparse.ArgumentParser(
+        prog="squirrel",
+        description="Count the census squirrels by primary fur colour.",
+    )
+    parser.add_argument(
+        "input",
+        nargs="?",
+        type=pathlib.Path,
+        default=DEFAULT_INPUT_PATH,
+        help="census CSV file (default: %(default)s)",
+    )
+    parser.add_argument(
+        "output",
+        nargs="?",
+        type=pathlib.Path,
+        default=DEFAULT_OUTPUT_PATH,
+        help="summary CSV file to write (default: %(default)s)",
+    )
+    return parser
+
+
+##
+# @brief Run the census analysis from the command line.
+# @param argv Arguments without the program name; `sys.argv` is used when None.
+# @return 0 on success, 2 when the census file does not exist.
+def main(argv: Sequence[str] | None = None) -> int:
+    """Run the census analysis from the command line."""
+    args = _build_parser().parse_args(argv)
+    input_path: pathlib.Path = args.input
+    output_path: pathlib.Path = args.output
+
+    if not input_path.is_file():
+        print(f"Census file not found: {input_path}", file=sys.stderr)
+        return EXIT_INPUT_MISSING
+
+    counts = census.count_by_fur_color(census.read_csv(input_path))
+    summary = census.build_summary(counts)
+    census.write_csv(summary, output_path)
+
+    # print is the intended output here: the counts are shown to the participant.
+    for color, count in summary:
+        print(f"{color}: {count}")
+    return 0

src/squirrel/constants.py

@@ -23,5 +23,8 @@ FUR_COLOR_COLUMN: str = "Primary Fur Color"
 ## @brief Fur colours that are counted, in the order they appear in the summary.
 FUR_COLORS: tuple[str, ...] = ("Gray", "Cinnamon", "Black")
 
+## @brief Exit status of the command line when the census file does not exist.
+EXIT_INPUT_MISSING: int = 2
+
 ## @brief Header of the summary file: the colour column, then the count column.
 SUMMARY_COLUMNS: tuple[str, str] = ("Fur Color", "Count")