synthocr 2.1.0__py3-none-any.whl

synthocr/kurdish_characters.json

@@ -0,0 +1,442 @@
+{
+    "0": {
+        "char": "0",
+        "count": 2,
+        "codepoint": "U+0030"
+    },
+    "1": {
+        "char": "1",
+        "count": 2,
+        "codepoint": "U+0031"
+    },
+    "2": {
+        "char": "2",
+        "count": 2,
+        "codepoint": "U+0032"
+    },
+    "3": {
+        "char": "3",
+        "count": 2,
+        "codepoint": "U+0033"
+    },
+    "4": {
+        "char": "4",
+        "count": 2,
+        "codepoint": "U+0034"
+    },
+    "5": {
+        "char": "5",
+        "count": 2,
+        "codepoint": "U+0035"
+    },
+    "6": {
+        "char": "6",
+        "count": 2,
+        "codepoint": "U+0036"
+    },
+    "7": {
+        "char": "7",
+        "count": 2,
+        "codepoint": "U+0037"
+    },
+    "8": {
+        "char": "8",
+        "count": 2,
+        "codepoint": "U+0038"
+    },
+    "9": {
+        "char": "9",
+        "count": 2,
+        "codepoint": "U+0039"
+    },
+    " ": {
+        "char": " ",
+        "count": 1,
+        "codepoint": "U+0020"
+    },
+    "ە": {
+        "char": "ە",
+        "count": 1,
+        "codepoint": "U+06D5"
+    },
+    "ا": {
+        "char": "ا",
+        "count": 1,
+        "codepoint": "U+0627"
+    },
+    "ی": {
+        "char": "ی",
+        "count": 1,
+        "codepoint": "U+06CC"
+    },
+    "و": {
+        "char": "و",
+        "count": 1,
+        "codepoint": "U+0648"
+    },
+    "ن": {
+        "char": "ن",
+        "count": 1,
+        "codepoint": "U+0646"
+    },
+    "ر": {
+        "char": "ر",
+        "count": 1,
+        "codepoint": "U+0631"
+    },
+    "د": {
+        "char": "د",
+        "count": 1,
+        "codepoint": "U+062F"
+    },
+    "ه": {
+        "char": "ه",
+        "count": 1,
+        "codepoint": "U+0647"
+    },
+    "ت": {
+        "char": "ت",
+        "count": 1,
+        "codepoint": "U+062A"
+    },
+    "ێ": {
+        "char": "ێ",
+        "count": 1,
+        "codepoint": "U+06CE"
+    },
+    "ب": {
+        "char": "ب",
+        "count": 1,
+        "codepoint": "U+0628"
+    },
+    "ک": {
+        "char": "ک",
+        "count": 1,
+        "codepoint": "U+06A9"
+    },
+    "م": {
+        "char": "م",
+        "count": 1,
+        "codepoint": "U+0645"
+    },
+    "‌": {
+        "char": "‌",
+        "count": 1,
+        "codepoint": "U+200C"
+    },
+    "س": {
+        "char": "س",
+        "count": 1,
+        "codepoint": "U+0633"
+    },
+    "ل": {
+        "char": "ل",
+        "count": 1,
+        "codepoint": "U+0644"
+    },
+    "ۆ": {
+        "char": "ۆ",
+        "count": 1,
+        "codepoint": "U+06C6"
+    },
+    "ئ": {
+        "char": "ئ",
+        "count": 1,
+        "codepoint": "U+0626"
+    },
+    "ش": {
+        "char": "ش",
+        "count": 1,
+        "codepoint": "U+0634"
+    },
+    "گ": {
+        "char": "گ",
+        "count": 1,
+        "codepoint": "U+06AF"
+    },
+    "ز": {
+        "char": "ز",
+        "count": 1,
+        "codepoint": "U+0632"
+    },
+    "پ": {
+        "char": "پ",
+        "count": 1,
+        "codepoint": "U+067E"
+    },
+    "خ": {
+        "char": "خ",
+        "count": 1,
+        "codepoint": "U+062E"
+    },
+    "ڵ": {
+        "char": "ڵ",
+        "count": 1,
+        "codepoint": "U+06B5"
+    },
+    "ڕ": {
+        "char": "ڕ",
+        "count": 1,
+        "codepoint": "U+0695"
+    },
+    "ژ": {
+        "char": "ژ",
+        "count": 1,
+        "codepoint": "U+0698"
+    },
+    "چ": {
+        "char": "چ",
+        "count": 1,
+        "codepoint": "U+0686"
+    },
+    "ج": {
+        "char": "ج",
+        "count": 1,
+        "codepoint": "U+062C"
+    },
+    "ڤ": {
+        "char": "ڤ",
+        "count": 1,
+        "codepoint": "U+06A4"
+    },
+    "ف": {
+        "char": "ف",
+        "count": 1,
+        "codepoint": "U+0641"
+    },
+    "ق": {
+        "char": "ق",
+        "count": 1,
+        "codepoint": "U+0642"
+    },
+    "ح": {
+        "char": "ح",
+        "count": 1,
+        "codepoint": "U+062D"
+    },
+    "ع": {
+        "char": "ع",
+        "count": 1,
+        "codepoint": "U+0639"
+    },
+    "ھ": {
+        "char": "ھ",
+        "count": 1,
+        "codepoint": "U+06BE"
+    },
+    "غ": {
+        "char": "غ",
+        "count": 1,
+        "codepoint": "U+063A"
+    },
+    "ص": {
+        "char": "ص",
+        "count": 1,
+        "codepoint": "U+0635"
+    },
+    "ط": {
+        "char": "ط",
+        "count": 1,
+        "codepoint": "U+0637"
+    },
+    "ظ": {
+        "char": "ظ",
+        "count": 1,
+        "codepoint": "U+0638"
+    },
+    "ض": {
+        "char": "ض",
+        "count": 1,
+        "codepoint": "U+0636"
+    },
+    "ذ": {
+        "char": "ذ",
+        "count": 1,
+        "codepoint": "U+0630"
+    },
+    "ث": {
+        "char": "ث",
+        "count": 1,
+        "codepoint": "U+062B"
+    },
+    "؛": {
+        "char": "؛",
+        "count": 1,
+        "codepoint": "U+061B"
+    },
+    ".": {
+        "char": ".",
+        "count": 1,
+        "codepoint": "U+002E"
+    },
+    "\\": {
+        "char": "\\",
+        "count": 1,
+        "codepoint": "U+005C"
+    },
+    "/": {
+        "char": "/",
+        "count": 1,
+        "codepoint": "U+002F"
+    },
+    ">": {
+        "char": ">",
+        "count": 1,
+        "codepoint": "U+003E"
+    },
+    "<": {
+        "char": "<",
+        "count": 1,
+        "codepoint": "U+003C"
+    },
+    "(": {
+        "char": "(",
+        "count": 1,
+        "codepoint": "U+0028"
+    },
+    ")": {
+        "char": ")",
+        "count": 1,
+        "codepoint": "U+0029"
+    },
+    "–": {
+        "char": "–",
+        "count": 1,
+        "codepoint": "U+2013"
+    },
+    "«": {
+        "char": "«",
+        "count": 1,
+        "codepoint": "U+00AB"
+    },
+    "»": {
+        "char": "»",
+        "count": 1,
+        "codepoint": "U+00BB"
+    },
+    "%": {
+        "char": "%",
+        "count": 1,
+        "codepoint": "U+0025"
+    },
+    "•": {
+        "char": "•",
+        "count": 1,
+        "codepoint": "U+2022"
+    },
+    ":": {
+        "char": ":",
+        "count": 1,
+        "codepoint": "U+003A"
+    },
+    "،": {
+        "char": "،",
+        "count": 1,
+        "codepoint": "U+060C"
+    },
+    "-": {
+        "char": "-",
+        "count": 1,
+        "codepoint": "U+002D"
+    },
+    "+": {
+        "char": "+",
+        "count": 1,
+        "codepoint": "U+002B"
+    },
+    "=": {
+        "char": "=",
+        "count": 1,
+        "codepoint": "U+003D"
+    },
+    "@": {
+        "char": "@",
+        "count": 1,
+        "codepoint": "U+0040"
+    },
+    "\"": {
+        "char": "\"",
+        "count": 1,
+        "codepoint": "U+0022"
+    },
+    "~": {
+        "char": "~",
+        "count": 1,
+        "codepoint": "U+007E"
+    },
+    "#": {
+        "char": "#",
+        "count": 1,
+        "codepoint": "U+0023"
+    },
+    "؟": {
+        "char": "؟",
+        "count": 1,
+        "codepoint": "U+061F"
+    },
+    "!": {
+        "char": "!",
+        "count": 1,
+        "codepoint": "U+0021"
+    },
+    "ـ": {
+        "char": "ـ",
+        "count": 1,
+        "codepoint": "U+0640"
+    },
+    "'": {
+        "char": "'",
+        "count": 1,
+        "codepoint": "U+0027"
+    },
+    "٠": {
+        "char": "٠",
+        "count": 0,
+        "codepoint": "U+0660"
+    },
+    "١": {
+        "char": "١",
+        "count": 0,
+        "codepoint": "U+0661"
+    },
+    "٢": {
+        "char": "٢",
+        "count": 0,
+        "codepoint": "U+0662"
+    },
+    "٣": {
+        "char": "٣",
+        "count": 0,
+        "codepoint": "U+0663"
+    },
+    "٤": {
+        "char": "٤",
+        "count": 0,
+        "codepoint": "U+0664"
+    },
+    "٥": {
+        "char": "٥",
+        "count": 0,
+        "codepoint": "U+0665"
+    },
+    "٦": {
+        "char": "٦",
+        "count": 0,
+        "codepoint": "U+0666"
+    },
+    "٧": {
+        "char": "٧",
+        "count": 0,
+        "codepoint": "U+0667"
+    },
+    "٨": {
+        "char": "٨",
+        "count": 0,
+        "codepoint": "U+0668"
+    },
+    "٩": {
+        "char": "٩",
+        "count": 0,
+        "codepoint": "U+0669"
+    }
+}

synthocr/kurdish_spelling.json

@@ -0,0 +1,45 @@
+{
+  "version": 1,
+  "scope": "Common Sorani and Latin Kurmanji spelling patterns, plus labelled Unicode rendering probes. Authored examples, not a dictionary or a claim of every regional spelling.",
+  "references": [
+    "https://unicode.org/cldr/charts/49/summary/ckb.html",
+    "https://unicode.org/cldr/charts/49/summary/ku.html",
+    "https://www.unicode.org/versions/Unicode16.0.0/core-spec/chapter-9/"
+  ],
+  "groups": [
+    {"id":"ckb-vowel-initial", "language":"ckb", "feature":"vowel_initial", "kind":"words", "texts":["ئاو", "ئاگر", "ئازادی", "ئامێر", "ئێوارە", "ئێمە", "ئۆتۆمبێل", "ئومێد"]},
+    {"id":"ckb-yeh-vowels", "language":"ckb", "feature":"yeh_vowels", "kind":"words", "texts":["شیر", "شێر", "ژیان", "پێنج", "هێواش", "بیر", "خوێندن", "گوێ"]},
+    {"id":"ckb-reh", "language":"ckb", "feature":"reh_rreh", "kind":"words", "texts":["ڕۆژ", "ڕێگا", "باران", "دەرگا", "ڕووبار", "هەزار", "شار", "ڕاست"]},
+    {"id":"ckb-lam", "language":"ckb", "feature":"lam_llam", "kind":"words", "texts":["گوڵ", "گەل", "دڵ", "ماڵ", "ساڵ", "ئاڵا", "هەڵبژاردن", "منداڵ"]},
+    {"id":"ckb-waw", "language":"ckb", "feature":"waw_oe_double_waw", "kind":"words", "texts":["دوو", "نوور", "ڕوون", "سوور", "نووسین", "بوو", "سوێند", "گۆران"]},
+    {"id":"ckb-unwritten-vowel", "language":"ckb", "feature":"unwritten_short_vowel", "kind":"words", "texts":["گرتن", "بردن", "کردن"]},
+    {"id":"ckb-lam-alef", "language":"ckb", "feature":"lam_alef_context", "kind":"words", "texts":["لاو", "لاوان", "لە", "سڵاو"]},
+    {"id":"ckb-compounds", "language":"ckb", "feature":"compounds", "kind":"words", "texts":["دەستنووس", "کتێبخانە", "نیشتمان", "پێشکەش"]},
+    {"id":"ckb-heh", "language":"ckb", "feature":"heh_ae", "kind":"words", "texts":["هەور", "هەولێر", "هاوڕێ", "بەهار", "نەورۆز", "هەوا", "هێمن", "هەموو"]},
+    {"id":"ckb-consonants", "language":"ckb", "feature":"consonant_distinctions", "kind":"words", "texts":["باخچە", "چاو", "جوان", "ژن", "زانست", "ڤیدیۆ", "ڤاکسین", "خۆش", "غەریب", "حەوت", "عێراق", "قەڵا", "کورد", "گوند"]},
+    {"id":"ckb-ezafe", "language":"ckb", "feature":"ezafe", "kind":"words", "texts":["کتێبی من", "ماڵی ئێمە", "گوڵی سوور", "ناوی شار"]},
+    {"id":"ckb-definite", "language":"ckb", "feature":"definiteness", "kind":"words", "texts":["کتێبێک", "کتێبەکە", "ماڵێک", "ماڵەکە", "ژنێک", "ژنەکە", "منداڵێک", "منداڵەکە"]},
+    {"id":"ckb-plural", "language":"ckb", "feature":"plurals", "kind":"words", "texts":["کتێبەکان", "کچەکان", "پیاوەکان", "منداڵەکان", "مامۆستاکان", "خوێندکارەکان"]},
+    {"id":"ckb-possessive", "language":"ckb", "feature":"possessive_suffixes", "kind":"words", "texts":["کتێبەکەم", "ماڵەکەت", "ناوم", "ناوت", "قوتابخانەکەمان"]},
+    {"id":"ckb-verbs", "language":"ckb", "feature":"verb_prefixes", "kind":"words", "texts":["دەنووسم", "ناڕۆم", "نەهات", "بڕۆ", "بخوێنەوە", "بنووسە"]},
+    {"id":"ckb-sentences", "language":"ckb", "feature":"word_boundaries", "kind":"words", "texts":["سڵاو جیهان", "ئەمە کتێبێکی نوێیە.", "من کوردی دەنووسم.", "ئێمە لە کتێبخانەدا دەخوێنینەوە."]},
+    {"id":"ckb-heh-encodings", "language":"ckb", "feature":"heh_encoding_variants", "kind":"encoding_probe", "texts":["ه ھ ە ه‌", "هەور ھەور", "کەکه‌ کەکە", "بەب به‌ب"]},
+    {"id":"ckb-keyboard-variants", "language":"ckb", "feature":"keyboard_variants", "kind":"encoding_probe", "texts":["ک ك", "ی ي ى", "کی كي کى", "ڕ ر ڵ ل"]},
+    {"id":"ckb-join-controls", "language":"ckb", "feature":"join_controls", "kind":"encoding_probe", "texts":["بب ب‌ب ب‍ب", "سڵاو سڵا‌و", "هه ه‌ه ه‍ه", "لا ل‌ا ل‍ا"]},
+    {"id":"ckb-marks", "language":"ckb", "feature":"combining_marks", "kind":"mark_probe", "texts":["بَ بِ بُ بّ بْ", "بَّ بِّ بُّ", "ئ ئ", "ئێمە ئێمە"]},
+    {"id":"ckb-digits", "language":"ckb", "feature":"digit_sets", "kind":"format_probe", "texts":["0123456789", "٠١٢٣٤٥٦٧٨٩", "۰۱۲۳۴۵۶۷۸۹", "ژمارە 123 ١٢٣ ۱۲۳"]},
+    {"id":"ckb-formatted-numbers", "language":"ckb", "feature":"formatted_numbers", "kind":"format_probe", "texts":["١٬٢٣٤٫٥٦ دینار", "۱۲٬۳۴۵٫۶۷ دینار", "+123 −٤٥٦ ۷۸۹٪", "٢٠٢٦/١٠/٢٥ ١٤:٣٠", "ژمارە ١٢٣، نرخ ٤٥٦ دینار"]},
+    {"id":"ckb-punctuation", "language":"ckb", "feature":"punctuation", "kind":"format_probe", "texts":["«سڵاو»، چۆنی؟", "هەولێر؛ سلێمانی؛ دهۆک.", "(سڵاو) [کوردی]", "سڵاو… جیهان!", "سڵاو — جیهان – کوردستان"]},
+    {"id":"ckb-mixed", "language":"ckb", "feature":"mixed_direction", "kind":"format_probe", "texts":["سڵاو SynthOCR 2026", "ID-2026 — هەولێر", "کۆد (ABC-123): سڵاو", "بەستەر https://example.org/ckb", "ئیمەیڵ user@example.org", "سڵاو ⁦ABC-123⁩ جیهان", "سڵاو ‏١٢٣‏ جیهان"]},
+    {"id":"kmr-accented", "language":"kmr", "feature":"accented_letters", "kind":"words", "texts":["çav", "êvar", "îro", "şev", "şêr", "sûr", "dûr", "bîr", "nû", "jiyan", "ziman", "xwendin", "xwarin", "qelem", "welat"]},
+    {"id":"kmr-capitals", "language":"kmr", "feature":"capitalization", "kind":"words", "texts":["ÇAV", "ÊVAR", "ÎRO", "ŞEV", "ŞÊR", "SÛR", "NÛ", "Çav", "Êvar", "Îro", "Şev"]},
+    {"id":"kmr-ezafe", "language":"kmr", "feature":"ezafe", "kind":"words", "texts":["mala me", "kitêba min", "dilê te", "xwendekarên me"]},
+    {"id":"kmr-plural", "language":"kmr", "feature":"plurals_and_oblique", "kind":"words", "texts":["zarok", "zarokan", "kitêb", "kitêban", "jin", "jinan", "mêvan", "mêvanan"]},
+    {"id":"kmr-verbs", "language":"kmr", "feature":"verb_prefixes", "kind":"words", "texts":["ez dixwînim", "ew dinivîse", "ez naçim", "ew nayê", "nivîsîn", "dîtin"]},
+    {"id":"kmr-sentences", "language":"kmr", "feature":"word_boundaries", "kind":"words", "texts":["Ez bi kurdî dinivîsim.", "Em li vir in.", "Tu çawa yî?", "Ev kitêbeke nû ye."]},
+    {"id":"kmr-canonical-accents", "language":"kmr", "feature":"canonical_accents", "kind":"encoding_probe", "texts":["ç ç Ç Ç", "ş ş Ş Ş", "ê ê Ê Ê", "î î Î Î", "û û Û Û"]},
+    {"id":"kmr-keyboard-variants", "language":"kmr", "feature":"keyboard_variants", "kind":"encoding_probe", "texts":["i I î Î ı İ", "ş ș Ş Ș", "ç c Ç C", "ê e î i û u"]},
+    {"id":"kmr-punctuation", "language":"kmr", "feature":"punctuation", "kind":"format_probe", "texts":["«Silav», tu çawa yî?", "Silav… baş î?", "(kurdî) [Kurmancî]", "kurdî — Kurmancî – Kurdistan", "'kurdî' “kurdî”"]},
+    {"id":"kmr-mixed", "language":"kmr", "feature":"mixed_direction", "kind":"format_probe", "texts":["Kurdî 123 / ١٢٣ / ۱۲۳", "Kurdî: سڵاو جیهان", "Kurdî (ABC-123) 2026-10-25", "Kurdî https://example.org/kmr", "Kurdî ⁧سڵاو ١٢٣⁩"]}
+  ]
+}

synthocr/lexicons.json

@@ -0,0 +1,82 @@
+{
+  "en": {
+    "subjects": ["The students", "The teachers", "The readers", "The visitors", "The researchers", "The children", "The writers", "The neighbors"],
+    "verbs": ["read books", "write notes", "discuss new ideas", "work together", "ask questions", "share stories"],
+    "places": ["in the library", "at the school", "in the garden", "at the museum", "in the city", "near the river"],
+    "prose": "{subject} {verb} {place}.",
+    "items": ["Book", "Notebook", "Pencil", "Tea", "Rice", "Bread", "Coffee", "Paper", "Milk", "Apple", "Soap", "Bag"],
+    "streets": ["Library Road", "Garden Street", "River Avenue", "School Lane", "Market Road", "Museum Street"],
+    "cities": ["Erbil", "Sulaymaniyah", "Duhok", "Baghdad", "Basra", "Kirkuk"],
+    "total": "Total", "invoice": "Invoice", "unit": "Unit", "quantity": "Quantity"
+  },
+  "ckb": {
+    "subjects": ["خوێندکارەکان", "مامۆستاکان", "خوێنەران", "گەشتیاران", "نووسەران", "منداڵەکان", "هاوڕێکان", "هاوسێکان"],
+    "verbs": ["کتێب دەخوێننەوە", "تێبینی دەنووسن", "گفتوگۆ دەکەن", "کار دەکەن", "پرسیار دەکەن", "چیرۆک دەگێڕنەوە"],
+    "places": ["لە کتێبخانەدا", "لە قوتابخانەدا", "لە باخچەدا", "لە مۆزەخانەدا", "لە شاردا", "لە نزیک ڕووبارەکەدا"],
+    "prose": "{subject} {place} {verb}.",
+    "items": ["کتێب", "دەفتەر", "پێنووس", "چا", "برنج", "نان", "قاوە", "کاغەز", "شیر", "سێو", "سابوون", "جانتا"],
+    "streets": ["شەقامی کتێبخانە", "شەقامی باخچە", "شەقامی ڕووبار", "شەقامی قوتابخانە", "شەقامی بازاڕ", "شەقامی مۆزەخانە"],
+    "cities": ["هەولێر", "سلێمانی", "دهۆک", "بەغدا", "بەسرە", "کەرکووک"],
+    "total": "کۆی گشتی", "invoice": "پسووڵە", "unit": "نرخی یەکە", "quantity": "ژمارە"
+  },
+  "ar": {
+    "subjects": ["الطلاب", "المعلمون", "القراء", "الزوار", "الباحثون", "الأطفال", "الكتاب", "الجيران"],
+    "verbs": ["يقرؤون الكتب", "يكتبون الملاحظات", "يناقشون أفكارًا جديدة", "يعملون معًا", "يطرحون الأسئلة", "يتبادلون القصص"],
+    "places": ["في المكتبة", "في المدرسة", "في الحديقة", "في المتحف", "في المدينة", "بالقرب من النهر"],
+    "prose": "{subject} {verb} {place}.",
+    "items": ["كتاب", "دفتر", "قلم", "شاي", "أرز", "خبز", "قهوة", "ورق", "حليب", "تفاح", "صابون", "حقيبة"],
+    "streets": ["شارع المكتبة", "شارع الحديقة", "شارع النهر", "شارع المدرسة", "شارع السوق", "شارع المتحف"],
+    "cities": ["أربيل", "السليمانية", "دهوك", "بغداد", "البصرة", "كركوك"],
+    "total": "المجموع", "invoice": "فاتورة", "unit": "سعر الوحدة", "quantity": "الكمية"
+  },
+  "fa": {
+    "subjects": ["دانشجویان", "معلمان", "خوانندگان", "بازدیدکنندگان", "پژوهشگران", "کودکان", "نویسندگان", "همسایه‌ها"],
+    "verbs": ["کتاب می‌خوانند", "یادداشت می‌نویسند", "گفت‌وگو می‌کنند", "با هم کار می‌کنند", "پرسش می‌کنند", "داستان تعریف می‌کنند"],
+    "places": ["در کتابخانه", "در مدرسه", "در باغ", "در موزه", "در شهر", "کنار رودخانه"],
+    "prose": "{subject} {place} {verb}.",
+    "items": ["کتاب", "دفتر", "مداد", "چای", "برنج", "نان", "قهوه", "کاغذ", "شیر", "سیب", "صابون", "کیف"],
+    "streets": ["خیابان کتابخانه", "خیابان باغ", "خیابان رودخانه", "خیابان مدرسه", "خیابان بازار", "خیابان موزه"],
+    "cities": ["تهران", "شیراز", "تبریز", "اصفهان", "رشت", "کرمانشاه"],
+    "total": "مجموع", "invoice": "صورت‌حساب", "unit": "قیمت واحد", "quantity": "تعداد"
+  },
+  "de": {
+    "subjects": ["Die Studierenden", "Die Lehrkräfte", "Die Leser", "Die Besucher", "Die Forschenden", "Die Kinder", "Die Autoren", "Die Nachbarn"],
+    "verbs": ["lesen Bücher", "schreiben Notizen", "diskutieren neue Ideen", "arbeiten zusammen", "stellen Fragen", "erzählen Geschichten"],
+    "places": ["in der Bibliothek", "in der Schule", "im Garten", "im Museum", "in der Stadt", "am Fluss"],
+    "prose": "{subject} {verb} {place}.",
+    "items": ["Buch", "Heft", "Bleistift", "Tee", "Reis", "Brot", "Kaffee", "Papier", "Milch", "Apfel", "Seife", "Tasche"],
+    "streets": ["Bibliotheksstraße", "Gartenstraße", "Flussweg", "Schulstraße", "Marktstraße", "Museumsstraße"],
+    "cities": ["Berlin", "München", "Köln", "Hamburg", "Bremen", "Dresden"],
+    "total": "Gesamt", "invoice": "Rechnung", "unit": "Einzelpreis", "quantity": "Menge"
+  },
+  "fr": {
+    "subjects": ["Les étudiants", "Les enseignants", "Les lecteurs", "Les visiteurs", "Les chercheurs", "Les enfants", "Les auteurs", "Les voisins"],
+    "verbs": ["lisent des livres", "écrivent des notes", "discutent de nouvelles idées", "travaillent ensemble", "posent des questions", "racontent des histoires"],
+    "places": ["à la bibliothèque", "à l’école", "dans le jardin", "au musée", "en ville", "près de la rivière"],
+    "prose": "{subject} {verb} {place}.",
+    "items": ["Livre", "Cahier", "Crayon", "Thé", "Riz", "Pain", "Café", "Papier", "Lait", "Pomme", "Savon", "Sac"],
+    "streets": ["Rue de la Bibliothèque", "Rue du Jardin", "Rue de la Rivière", "Rue de l’École", "Rue du Marché", "Rue du Musée"],
+    "cities": ["Paris", "Lyon", "Nice", "Lille", "Nantes", "Bordeaux"],
+    "total": "Total", "invoice": "Facture", "unit": "Prix unitaire", "quantity": "Quantité"
+  },
+  "es": {
+    "subjects": ["Los estudiantes", "Los profesores", "Los lectores", "Los visitantes", "Los investigadores", "Los niños", "Los autores", "Los vecinos"],
+    "verbs": ["leen libros", "escriben notas", "comentan nuevas ideas", "trabajan juntos", "hacen preguntas", "cuentan historias"],
+    "places": ["en la biblioteca", "en la escuela", "en el jardín", "en el museo", "en la ciudad", "cerca del río"],
+    "prose": "{subject} {verb} {place}.",
+    "items": ["Libro", "Cuaderno", "Lápiz", "Té", "Arroz", "Pan", "Café", "Papel", "Leche", "Manzana", "Jabón", "Bolsa"],
+    "streets": ["Calle de la Biblioteca", "Calle del Jardín", "Calle del Río", "Calle de la Escuela", "Calle del Mercado", "Calle del Museo"],
+    "cities": ["Madrid", "Sevilla", "Valencia", "Málaga", "Bilbao", "Toledo"],
+    "total": "Total", "invoice": "Factura", "unit": "Precio unitario", "quantity": "Cantidad"
+  },
+  "tr": {
+    "subjects": ["Öğrenciler", "Öğretmenler", "Okurlar", "Ziyaretçiler", "Araştırmacılar", "Çocuklar", "Yazarlar", "Komşular"],
+    "verbs": ["kitap okuyor", "not yazıyor", "yeni fikirleri tartışıyor", "birlikte çalışıyor", "soru soruyor", "hikâye anlatıyor"],
+    "places": ["kütüphanede", "okulda", "bahçede", "müzede", "şehirde", "nehir kenarında"],
+    "prose": "{subject} {place} {verb}.",
+    "items": ["Kitap", "Defter", "Kalem", "Çay", "Pirinç", "Ekmek", "Kahve", "Kâğıt", "Süt", "Elma", "Sabun", "Çanta"],
+    "streets": ["Kütüphane Caddesi", "Bahçe Sokak", "Nehir Caddesi", "Okul Sokak", "Pazar Caddesi", "Müze Sokak"],
+    "cities": ["Ankara", "İstanbul", "İzmir", "Bursa", "Konya", "Van"],
+    "total": "Toplam", "invoice": "Fatura", "unit": "Birim fiyat", "quantity": "Miktar"
+  }
+}