gctools/pdf/pdf_merge.py

import config
from PyPDF2 import PdfMerger
from pathlib import Path
import re
import json


def pdf_merge(base_dir: str, report_name, config=None):
    if config is None:
        config = {"_0": []}
    report_prefix = re.sub(r"\[([^\]]+)\]", "", report_name)
    if report_prefix[-2:] == "__":
        report_prefix = report_prefix[:-1]
    for suffix, entries in config.items():
        target = f"{base_dir}/{report_prefix}{suffix}.pdf"
        if Path(target).exists():
            Path(target).unlink()
        pdfs = [f for f in Path(base_dir).glob(report_prefix + "*.pdf")]
        pdfs.sort(key=lambda f: f.stat().st_mtime)
        if len(entries) > 0:
            pdfs = [f for e in entries for f in pdfs if e + ".pdf" in f]
        if len(pdfs) > 0:
            with PdfMerger() as merger:
                for pdf in pdfs:
                    merger.append(pdf)
                merger.write(target)


def get_reports(reports_file, folder):
    with open(reports_file, "r") as frh:
        reports = json.load(frh)
    return [r["name"] for r in reports if r["path"].startswith(folder)]


def merge_reports_in_folder(
    cfg: config.Config, folder="Team Content/ReportOutput", config=None
):
    reports = get_reports(cfg.cognos11.reports_file, folder)
    print(folder)
    for r in reports:
        print(r)
        pdf_merge(cfg.cognos11.reportoutput_dir, r, config)


def get_config_json(config, config_dir):
    if config == "":
        return None
    if config.startswith("{"):
        return json.loads(config)
    config_file = f"{config_dir}/merge/{config}.json"
    if not Path(config_file).exists():
        return None
    with open(config_file, "r") as frh:
        return json.load(frh)


if __name__ == "__main__":
    merge_reports_in_folder(config.Config())