Check survey files exported from Smartling

check_survey.py

from glob import glob
import csv
import os
import re
import sys

# Pass a path as an argument, the script will go through all .csv files included.
if len(sys.argv) < 2:
    sys.exit("Missing path argument")
path = os.path.abspath(sys.argv[1])

for file in glob(f"{path}/**/*.csv", recursive=True):
    errors = []
    with open(file) as f:
        reader = csv.reader(f, delimiter=',', quotechar='"')

        for row in reader:
            # Check if question references are preserved in the translation.
            # Example: ${q://QID68/ChoiceGroup/SelectedChoices}
            id = row[0]
            source = row[1]
            source_matches = re.findall(r"\$\{q://QID\d+[/a-z]+}", source, flags=re.IGNORECASE)
            translation = row[3]
            translation_matches = re.findall(r"\$\{q://QID\d+[/a-z]+}", translation, flags=re.IGNORECASE)

            if source_matches != translation_matches:
                errors.append(
                    "\n----"
                    f"\nERROR: question references don't match for string {id}"
                    f"\n\nSource matches: {source_matches}"
                    f"\nSource string: {source}"
                    f"\n\nTranslation matches: {translation_matches}"
                    f"\nTranslation: {translation}"
                )

            # Check numeric values
            try:
                num = int(source)
                if str(num) == source and translation != source:
                    errors.append(
                        "\n----"
                        f"\nWARNING: numeric value changed for string {id}"
                        f"\n\nSource string: {source}"
                        f"\n\nTranslation: {translation}"
                    )
            except ValueError:
                pass


            """
            # Check for untranslated strings
            if source == translation:
                if id.startswith(
                    (
                        "QID8_",
                        "QID9_",
                        "QID12_",
                        "QID18_",
                        "QID19_",
                        "QID12_",
                        "QID45_",
                        "QID87_",
                        "QID88_",
                    )
                ):
                    continue
                print("\n----\n")
                print("WARNING: translation is identical to source")
                print(f"ID: {id}")
                print(f"Source string: {source}")
                print(f"Translation: {translation}")
            """

    if errors:
        print(f"\nFile: {file}")
        print("\n".join(errors))