CLDR-json data extraction script

data/de/lexicon.txt

@@ -0,0 +1,19 @@
+Januar: noun masculine nominative singular
+Februar: noun masculine nominative singular
+März: noun masculine nominative singular
+April: noun masculine nominative singular
+Mai: noun masculine nominative singular
+Juni: noun masculine nominative singular
+Juli: noun masculine nominative singular
+August: noun masculine nominative singular
+September: noun masculine nominative singular
+Oktober: noun masculine nominative singular
+November: noun masculine nominative singular
+Dezember: noun masculine nominative singular
+Sonntag: noun masculine nominative singular
+Montag: noun masculine nominative singular
+Dienstag: noun masculine nominative singular
+Mittwoch: noun masculine nominative singular
+Donnerstag: noun masculine nominative singular
+Freitag: noun masculine nominative singular
+Samstag: noun masculine nominative singular

data/en/lexicon.txt

@@ -0,0 +1,19 @@
+January: noun singular
+February: noun singular
+March: noun singular
+April: noun singular
+May: noun singular
+June: noun singular
+July: noun singular
+August: noun singular
+September: noun singular
+October: noun singular
+November: noun singular
+December: noun singular
+Sunday: noun singular
+Monday: noun singular
+Tuesday: noun singular
+Wednesday: noun singular
+Thursday: noun singular
+Friday: noun singular
+Saturday: noun singular

data/es/lexicon.txt

@@ -0,0 +1,19 @@
+enero: noun masculine nominative singular inanimate
+febrero: noun masculine nominative singular inanimate
+marzo: noun masculine nominative singular inanimate
+abril: noun masculine nominative singular inanimate
+mayo: noun masculine nominative singular inanimate
+junio: noun masculine nominative singular inanimate
+julio: noun masculine nominative singular inanimate
+agosto: noun masculine nominative singular inanimate
+septiembre: noun masculine nominative singular inanimate
+octubre: noun masculine nominative singular inanimate
+noviembre: noun masculine nominative singular inanimate
+diciembre: noun masculine nominative singular inanimate
+domingo: noun masculine nominative singular inanimate
+lunes: noun masculine nominative singular inanimate
+martes: noun masculine nominative singular inanimate
+miércoles: noun masculine nominative singular inanimate
+jueves: noun masculine nominative singular inanimate
+viernes: noun masculine nominative singular inanimate
+sábado: noun masculine nominative singular inanimate

data/fr/lexicon.txt

@@ -0,0 +1,19 @@
+janvier: noun masculine singular
+février: noun masculine singular
+mars: noun masculine singular
+avril: noun masculine singular
+mai: noun masculine singular
+juin: noun masculine singular
+juillet: noun masculine singular
+août: noun masculine singular
+septembre: noun masculine singular
+octobre: noun masculine singular
+novembre: noun masculine singular
+décembre: noun masculine singular
+dimanche: noun masculine singular
+lundi: noun masculine singular
+mardi: noun masculine singular
+mercredi: noun masculine singular
+jeudi: noun masculine singular
+vendredi: noun masculine singular
+samedi: noun masculine singular

data/sr/lexicon.txt

@@ -0,0 +1,19 @@
+јануар: noun masculine nominative singular inanimate
+фебруар: noun masculine nominative singular inanimate
+март: noun masculine nominative singular inanimate
+април: noun masculine nominative singular inanimate
+мај: noun masculine nominative singular inanimate
+јун: noun masculine nominative singular inanimate
+јул: noun masculine nominative singular inanimate
+август: noun masculine nominative singular inanimate
+септембар: noun masculine nominative singular inanimate
+октобар: noun masculine nominative singular inanimate
+новембар: noun masculine nominative singular inanimate
+децембар: noun masculine nominative singular inanimate
+недеља: noun feminine nominative singular inanimate
+понедељак: noun masculine nominative singular inanimate
+уторак: noun masculine nominative singular inanimate
+среда: noun feminine nominative singular inanimate
+четвртак: noun masculine nominative singular inanimate
+петак: noun masculine nominative singular inanimate
+субота: noun feminine nominative singular inanimate

data/tools/extract_cldr_data.py

@@ -0,0 +1,84 @@
+"""
+Extracts data from CLDR-JSON repository, e.g. nouns like month or day names.
+Script either creates a new inflection file, or appends data to existing one.
+
+Part of Speech, Gender, Case, Number. Animacy
+
+Run script from data folder.
+
+Before running the script clone cldr-json repository:
+
+gh repo clone unicode-org/cldr-json
+
+and install jsonpath-ng package:
+
+pip install jsonpath-ng
+"""
+
+import argparse
+import json
+import os
+
+from jsonpath_ng import jsonpath, parse
+
+
+def load_json(filename):
+    """Loads JSON data from the specified file.
+
+    Args:
+        filename: The name of the JSON file.
+
+    Returns:
+        The parsed JSON data.
+    """
+
+    try:
+        with open(filename, 'r', encoding='utf-8') as file:
+            return json.load(file)
+    except FileNotFoundError:
+        print(f"Error: File '{filename}' not found.")
+        return None
+
+
+def write_to_lexicon(output_file, language, json_data):
+    """Extracts specified data from cldr-json file
+       and writes it to the lexicon file.
+
+    Args:
+        output_file: name of the lexicon.
+        language: cldr-json file language.
+        json_data: parsed cldr-json data.
+    """
+    MONTH_NAMES_EXPRESSION = parse('main..dates.calendars.gregorian.months.format.wide.*')
+    DAY_NAMES_EXPRESSION = parse('main..dates.calendars.gregorian.days.format.wide.*')
+    EXPRESSIONS = [MONTH_NAMES_EXPRESSION, DAY_NAMES_EXPRESSION]
+
+    results = []
+    for expression in EXPRESSIONS:
+        match = expression.find(json_data)
+        for m in match:
+            results.append(m.value + ': noun masculine nominative singular inanimate\n')
+
+    full_filename = os.path.join(language, output_file)
+    try:
+        os.makedirs(os.path.dirname(full_filename), exist_ok=True)
+        with open(full_filename, 'a', encoding='utf-8') as file:
+            file.writelines(results)
+    except FileNotFoundError:
+        print(f"Error: file '{output_file}' can't be created.")
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description='Load and process CLDR-JSON files for given languages.')
+    parser.add_argument('--cldr_root', help='The path to CLDR-JSON data.', default='../../cldr-json/cldr-json/cldr-dates-full/main')
+    parser.add_argument('--input_file', help='Data file to read from, e.g. ca-gregorian.json.', default='ca-gregorian.json')
+    parser.add_argument('--output_file', help='Data file to create/append to, e.g. lexicon.txt.', default='lexicon.txt')
+    parser.add_argument('--language_list', nargs='+', default=['sr', 'en', 'de', 'es', 'fr'])
+    args = parser.parse_args()
+
+    for language in args.language_list:
+        full_filename = os.path.join(args.cldr_root, language, args.input_file)
+        data = load_json(full_filename)
+
+        if data:
+            write_to_lexicon(args.output_file, language, data)