IMO_SL/segment_script/segment.py

# -----------------------------------------------------------------------------
# Author: Marina
# Date: 2024-11-15
# -----------------------------------------------------------------------------
"""Script to segment IMO shortlist md files using regex.
To run:
`python segment_script/segment.py`
To debug (or see covered use cases listed in fixtures/):
`pytest test_segment`
"""

from collections import defaultdict
import os
from pathlib import Path
import re
import pandas as pd
import json


section_re = re.compile(r"##\s+([A-Za-z]\w.*)")
problem_re = re.compile(
    r"^(?:##\s*)?((?:[AGNC]\s*\d+))\.*\s*(.*?)(?:\((.*?)\))?$", re.MULTILINE
)
solution_re = re.compile(
    r"^(?:##\s*)?(Solution(?: \d+)?\.)\s*(.*?)(?=(?:Solution|Comment|A\d+|G\d+|N\d+|C\d+|##|$))",
    re.MULTILINE | re.DOTALL,
)


def add_content(section, label, text_class, text, problems, solutions):
    text_str = " ".join(text).strip()
    if text_class == "problem":
        # print(f"ADD PROBLEM {section} {label} ")
        problems.append({"section": section, "label": label, "problem": text_str})
    elif text_class == "solution":
        # print(f"ADD SOLUTION {section} {label}")
        solutions.append({"label": label, "solution": text_str})


def parse(file: Path):
    content = file.read_text(encoding="utf-8")

    problems, solutions = [], []
    current_section, current_label, current_class = None, None, None
    current_lines = []
    for line in content.splitlines():
        if match := problem_re.match(line):
            label, text, country = match.groups()
            label = label.replace(" ", "")  # clean the label
            add_content(
                current_section,
                current_label,
                current_class,
                current_lines,
                problems,
                solutions,
            )
            current_class = "problem"
            current_label = label
            current_lines = [text]
        elif match := solution_re.match(line):
            label, text = match.groups()
            add_content(
                current_section,
                current_label,
                current_class,
                current_lines,
                problems,
                solutions,
            )
            current_class = "solution"
            current_lines = [text]
        elif match := section_re.match(line):
            add_content(
                current_section,
                current_label,
                current_class,
                current_lines,
                problems,
                solutions,
            )
            current_class = "section"
            (text,) = match.groups()
            current_section = text
        else:
            current_lines.append(line)
    add_content(
        current_section,
        current_label,
        current_class,
        current_lines,
        problems,
        solutions,
    )
    problems_df = pd.DataFrame(problems).drop_duplicates(subset=["label", "problem"])
    solutions_df = pd.DataFrame(solutions)
    return problems_df, solutions_df


def join(problems_df, solutions_df):
    pairs_df = problems_df.merge(solutions_df, on=["label"], how="left")
    return pairs_df


def add_metadata(pairs_df, year, resource_path):
    pairs_df.rename(
        columns={"section": "problem_type", "label": "problem_label"}, inplace=True
    )
    pairs_df["year"] = year
    pairs_df["tier"] = "T0"  # according to omnimath
    pairs_df["exam"] = ["IMO-SL"] * len(pairs_df)
    pairs_df["metadata"] = [{"resource_path": resource_path}] * len(pairs_df)
    return pairs_df[
        [
            "year",
            "tier",
            "problem_label",
            "problem_type",
            "exam",
            "problem",
            "solution",
            "metadata",
        ]
    ]


def write_pairs(file_path, pairs_df):
    pairs_df = pairs_df.replace({pd.NA: None, pd.NaT: None, float("nan"): None})
    pairs_dict = pairs_df.to_dict(orient="records")
    output_text = ""
    for pair in pairs_dict:
        output_text += json.dumps(pair, ensure_ascii=False) + "\n"
    file_path.write_text(output_text, encoding="utf-8")


if __name__ == "__main__":
    project_root = Path(__file__).parent.parent.parent
    compet_base_path = Path(__file__).resolve().parent.parent
    compet_md_path = compet_base_path / "md"
    seg_output_path = compet_base_path / "segmented"

    for md_file in compet_md_path.glob("**/*.md"):
        # en-compendium is segmented in segment_compendium.py
        if "compendium" not in md_file.name:
            year = re.search(r"(\d{4})", md_file.name).group(1)
            output_file = seg_output_path / md_file.relative_to(
                compet_md_path
            ).with_suffix(".jsonl")
            output_file.parent.mkdir(parents=True, exist_ok=True)

            print(md_file)
            problems, solutions = parse(md_file)
            pairs_df = join(problems, solutions)
            pairs_df = add_metadata(
                pairs_df, year, output_file.relative_to(project_root).as_posix()
            )
            print(pairs_df)
            write_pairs(output_file, pairs_df)