CIMAC-CIDC/cidc-schemas

#!/usr/bin/env python3

import openpyxl
import csv
import os
import sys
import subprocess

from cidc_schemas.constants import TEMPLATE_DIR
from cidc_schemas.template import generate_analysis_template_schemas


def system(*args, **kwargs):
    kwargs.setdefault("stdout", subprocess.PIPE)
    proc = subprocess.Popen(args, **kwargs)
    out, err = proc.communicate()
    return out, err


def rtrim(seq):
    res = []
    iter = reversed(seq)
    for i in iter:
        if i:
            res.append(i)
            break
    res.extend(iter)
    return type(seq)(reversed(res))


def csv_from_excel(xlsx_dir, xslx_fname, csv_dir):
    csv_files = []

    wb = openpyxl.load_workbook(os.path.join(xlsx_dir, xslx_fname))

    for worksheet_name in wb.sheetnames:
        worksheet = wb[worksheet_name]

        csv_file = os.path.join(
            csv_dir, xslx_fname[:-5] + "__" + worksheet_name.replace(" ", "_") + ".csv"
        )
        with open(csv_file, "w") as fd:
            wr = csv.writer(fd, quoting=csv.QUOTE_ALL)

            for row in worksheet.iter_rows(values_only=True):
                trimmed_row = rtrim(row)
                if len(trimmed_row) == 1:  # only first cell with row type
                    continue  # skip empty rows
                wr.writerow(trimmed_row)

        csv_files.append(csv_file)
    return csv_files


def main():

    if not os.path.exists("./template_examples/csvs"):
        os.makedirs("./template_examples/csvs")

    for template_example in os.listdir("./template_examples/"):
        # remove any .csv in there, so nothing old will be left
        if template_example.endswith(".csv"):
            output, err = system(
                "rm", os.path.join("./template_examples/", template_example)
            )
            if output:
                print(output)
            if err:
                os.exit(err)
            output, err = system(
                "git", "rm", os.path.join("./template_examples/", template_example)
            )
            if output:
                print(output)
            if err:
                os.exit(err)

        # '~$___.xlsx' is something like lockfiles for Excel
        if not template_example.endswith(".xlsx") or template_example.startswith("~$"):
            continue

        csv_fnames = csv_from_excel(
            "./template_examples/", template_example, "./template_examples/csvs/"
        )
        for fname in csv_fnames:
            output, err = system("git", "add", fname)
            if output:
                print(output)
            if err:
                os.exit(err)

    generate_analysis_template_schemas()
    folder = os.path.join(TEMPLATE_DIR, "analyses")
    for fname in os.listdir(folder):
        if fname.endswith("_analysis_template.json"):
            output, err = system("git", "add", os.path.join(folder, fname))
            if output:
                print(output)
            if err:
                os.exit(err)


main()