analyze_local.py

"""
analyze_local.py  --  Analyze a folder of local PDF files with LLM.

Reads every .pdf in the given directory, extracts full text, sends to LLM
for deep structured analysis, and writes per-paper Markdown files plus an
optional cross-paper synthesis report.

Usage:
    python analyze_local.py --dir path/to/pdfs
    python analyze_local.py --dir path/to/pdfs --lang zh --synthesis
    python analyze_local.py --dir path/to/pdfs --topic "CRISPR" --synthesis --pdf

Arguments:
    --dir    / -d   Path to folder containing PDF files (required)
    --topic  / -t   Topic label used in synthesis & index (default: folder name)
    --lang   / -l   Report language: en or zh              (default: en)
    --synthesis     Generate cross-paper synthesis report   (default: off)
    --pdf           Render Markdown output to PDF via pandoc (default: off)
    --model  / -m   LLM model                              (default: api_key.txt line 3)
    --env           Conda env name                         (default: DailyEssay)
"""

import argparse
import shutil
import subprocess
import sys
from datetime import datetime
from pathlib import Path

ENV_NAME = "DailyEssay"
HERE     = Path(__file__).parent
SCRIPT   = HERE / "analyze_local_runner.py"


def conda_exe() -> str:
    exe = shutil.which("conda")
    if not exe:
        print("[ERROR] conda not found.")
        sys.exit(1)
    return exe


def env_exists(conda: str, name: str) -> bool:
    result = subprocess.run([conda, "env", "list"], capture_output=True, text=True)
    return any(part == name
               for line in result.stdout.splitlines()
               for part in line.split())


def wizard() -> list:
    print("=" * 60)
    print("  Local PDF Analyzer -- Interactive Wizard")
    print("=" * 60)

    pdf_dir = input("Path to PDF folder: ").strip()
    if not pdf_dir:
        print("[ERROR] Folder path cannot be empty.")
        sys.exit(1)

    default_topic = Path(pdf_dir).name
    topic   = input(f"Topic label [default: {default_topic!r}]: ").strip() or default_topic
    lang    = input("Report language [en/zh, default: en]: ").strip().lower() or "en"
    if lang not in ("en", "zh"):
        lang = "en"
    do_syn  = input("Generate cross-paper synthesis? [y/N]: ").strip().lower() == "y"
    do_pdf  = input("Render output to PDF? [y/N]: ").strip().lower() == "y"

    args = ["--dir", pdf_dir, "--topic", topic, "--lang", lang]
    if do_syn:  args.append("--synthesis")
    if do_pdf:  args.append("--pdf")
    return args


def main():
    parser = argparse.ArgumentParser(add_help=False)
    parser.add_argument("--env", default=ENV_NAME)
    known, passthrough = parser.parse_known_args()

    conda = conda_exe()
    if not env_exists(conda, known.env):
        print(f"[ERROR] Conda env '{known.env}' not found.")
        sys.exit(1)

    extra_args = passthrough if passthrough else wizard()
    cmd = [conda, "run", "--no-capture-output", "-n", known.env,
           "python", str(SCRIPT)] + extra_args
    print(f"\n[Run] python analyze_local_runner.py {' '.join(extra_args)}\n")
    rc = subprocess.run(cmd).returncode
    print()
    if rc == 0:
        print("=" * 60)
        print("  Complete! Results saved in analysis_output/")
        print("=" * 60)
    else:
        print(f"[!] Script exited with code {rc}.")
    sys.exit(rc)


if __name__ == "__main__":
    main()