add: multiple language support

README.md

@@ -25,7 +25,7 @@ Check out the configuration reference at <https://huggingface.co/docs/hub/spaces
 
 ## Introduction
 
-I love reading manga, and I can't wait for the next chapter of my favorite manga to be released. However, the newest chapters are usually in Japanese, and they are translated to English after some time. I want to read the newest chapters as soon as possible, so I decided to build a manga translator that can translate Japanese manga to English.
+I love reading manga, and I can't wait for the next chapter of my favorite manga to be released. However, the newest chapters are usually in Japanese, and they are translated to English after some time. I want to read the newest chapters as soon as possible, so I decided to build a manga translator that can translate Japanese manga to English. (**NEW** I've added support for multiple languages, which include Arabic, Greek, Cyrillic and other non Latin languages.)
 
 ## GitHub Project
 
@@ -37,7 +37,6 @@ I want to translate the text in the manga images from Japanese to English. I wil
 
 ![Manga Translator](./assets/MangaTranslator.png)
 
-
 ### Data Collection
 
 This [dataset](https://universe.roboflow.com/speechbubbledetection-y9yz3/bubble-detection-gbjon/dataset/2#) contains over 8500 images of manga pages together with their annotations from Roboflow. I will use this dataset to train `Yolov8` to detect the speech bubbles in the manga images. To use this dataset with Yolov8, I will need to convert the annotations to the YOLO format, which is a text file containing the class label and the bounding box coordinates of the object in the image.

app.py

@@ -2,45 +2,58 @@ import numpy as np
 from PIL import Image
 import gradio as gr
 
+from utils.langs import languages
 from main import predict
 
-def process_image(image):
-	if image is not None:
-		if not isinstance(image, np.ndarray):
-			image = np.array(Image.open(image))
-			print(image)
+language_choices = [(name.title(), code) for name, code in languages.items()]
 
-		translated_image = predict(image)
-		return translated_image
-	return None
+
+def process_image(image, target_lang):
+    if image is not None:
+        if not isinstance(image, np.ndarray):
+            image = np.array(Image.open(image))
+
+        translated_image = predict(image, target_lang=target_lang)
+        return translated_image
+    return None
 
 
 with gr.Blocks() as demo:
-	gr.Markdown(
-		"""
-		<div style="display: flex; align-items: center; flex-direction: row; justify-content: center; margin-bottom: 20px; text-align: center;">
-			<a href="https://github.com/Detopall/manga-translator" target="_blank" rel="noopener noreferrer" style="text-decoration: none;">
-				<h1 style="display: inline; margin-left: 10px; text-decoration: underline;">Manga Translator</h1>
-			</a>
-		</div>
-		"""
-	)
-
-	with gr.Row():
-		with gr.Column(scale=1):
-			image_input = gr.Image()
-			submit_button = gr.Button("Translate")
-		with gr.Column(scale=1):
-			image_output = gr.Image()
-
-	submit_button.click(process_image, inputs=image_input, outputs=image_output)
-
-	examples = gr.Examples(examples=[
-		["./examples/ex1.jpg"],
-		["./examples/ex2.jpg"],
-		["./examples/ex3.jpg"],
-		["./examples/ex4.jpg"],
-	], inputs=image_input)
+    gr.Markdown(
+        """
+        <div style="display: flex; align-items: center; flex-direction: row; justify-content: center; margin-bottom: 20px; text-align: center;">
+            <a href="https://github.com/Detopall/manga-translator" target="_blank" rel="noopener noreferrer" style="text-decoration: none;">
+                <h1 style="display: inline; margin-left: 10px; text-decoration: underline;">Manga Translator</h1>
+            </a>
+        </div>
+        """
+    )
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            image_input = gr.Image()
+            language_dropdown = gr.Dropdown(
+                choices=language_choices,
+                label="Target Language",
+                value="en-GB",
+            )
+            submit_button = gr.Button("Translate")
+        with gr.Column(scale=1):
+            image_output = gr.Image()
+
+    submit_button.click(
+        process_image, inputs=[image_input, language_dropdown], outputs=image_output
+    )
+
+    examples = gr.Examples(
+        examples=[
+            ["./examples/ex1.jpg"],
+            ["./examples/ex2.jpg"],
+            ["./examples/ex3.jpg"],
+            ["./examples/ex4.jpg"],
+        ],
+        inputs=image_input,
+    )
 
 if __name__ == "__main__":
-	demo.launch()
+    demo.launch()

fonts/NotoNaskhArabic-Regular.ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46060349550000637973beb3f7cf8e4ae5f6d76e8319f0f9da50c0c1f987d260
+size 327956

fonts/NotoSans-Regular.ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe8c022f48d8dd29f17b744d16f9346f4357e16f7d4f7be58b000ae7c291b614
+size 629024

fonts/NotoSansCyrillic-Regular.ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2c786babb3fd1603d174e6108cc222cede9b0968540a16a7b34ac454467d5e
+size 555264

fonts/NotoSansGreek-Regular.ttf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac2c786babb3fd1603d174e6108cc222cede9b0968540a16a7b34ac454467d5e
+size 555264

main.py

@@ -1,5 +1,6 @@
 import io
 import base64
+from typing import Dict, Any
 
 import numpy as np
 from PIL import Image
@@ -14,40 +15,42 @@ from utils.write_text_on_image import add_text
 MODEL_PATH = "./model_creation/runs/detect/train5/weights/best.pt"
 object_detection_model = YOLO(MODEL_PATH)
 
-def extract_text_from_regions(image: np.ndarray, results: list):
 
-	for result in results:
-		x1, y1, x2, y2, _, _ = result
-		detected_image = image[int(y1):int(y2), int(x1):int(x2)]
-		if detected_image.shape[-1] == 4:
-			detected_image = detected_image[:, :, :3]
-		im = Image.fromarray(np.uint8(detected_image * 255))
-		text = get_text_from_image(im)
+def extract_text_from_regions(
+    image: np.ndarray, target_lang: str, results: list
+) -> Dict[str, Any]:
 
-		processed_image, cont = process_contour(detected_image)
-		translated_text = translate_manga(text, source_lang="auto", target_lang="en")
-		add_text(processed_image, translated_text, cont)
+    for result in results:
+        x1, y1, x2, y2, _, _ = result
+        detected_image = image[int(y1) : int(y2), int(x1) : int(x2)]
+        if detected_image.shape[-1] == 4:
+            detected_image = detected_image[:, :, :3]
+        im = Image.fromarray(np.uint8(detected_image * 255))
+        text = get_text_from_image(im)
 
+        processed_image, cont = process_contour(detected_image)
+        translated_text = translate_manga(
+            text, target_lang=target_lang, source_lang="ja-JP"
+        )
+        if translated_text is None:
+            translated_text = "Translation failed"
 
-def convert_image_to_base64(image: Image.Image) -> str:
-	buff = io.BytesIO()
-	image.save(buff, format="PNG")
-	return base64.b64encode(buff.getvalue()).decode("utf-8")
+        add_text(processed_image, translated_text, cont)
 
 
-def predict(image: np.ndarray):
+def predict(image: np.ndarray, target_lang: str):
 
-	image = Image.fromarray(image)
-	image.save("image.png")
+    image = Image.fromarray(image)
+    image.save("image.png")
 
-	try:
-		np_image = np.array(image)
+    try:
+        np_image = np.array(image)
 
-		results = predict_bounding_boxes(object_detection_model, "image.png")
-		extract_text_from_regions(np_image, results)
+        results = predict_bounding_boxes(object_detection_model, "image.png")
+        extract_text_from_regions(np_image, target_lang, results)
 
-		return np_image
+        return np_image
 
-	except Exception as e:
-		print(f"Error: {str(e)}")
-		return None
+    except Exception as e:
+        print(f"Error: {str(e)}")
+        return None

requirements.txt

@@ -3,3 +3,5 @@ ultralytics==8.3.78
 manga-ocr==0.1.14
 deep-translator==1.11.4
 torch==2.6.0
+python-bidi==0.6.6
+arabic-reshaper==3.0.0

utils/langs.py

@@ -0,0 +1,325 @@
+languages = {
+    "acehnese": "ace-ID",
+    "afrikaans": "af-ZA",
+    "akan": "ak-GH",
+    "albanian": "sq-AL",
+    "amharic": "am-ET",
+    "antigua and barbuda creole english": "aig-AG",
+    "arabic": "ar-SA",
+    "arabic egyptian": "ar-EG",
+    "aragonese": "an-ES",
+    "armenian": "hy-AM",
+    "assamese": "as-IN",
+    "asturian": "ast-ES",
+    "austrian german": "de-AT",
+    "awadhi": "awa-IN",
+    "ayacucho quechua": "quy-PE",
+    "azerbaijani": "az-AZ",
+    "bahamas creole english": "bah-BS",
+    "bajan": "bjs-BB",
+    "balinese": "ban-ID",
+    "balkan gipsy": "rm-RO",
+    "bambara": "bm-ML",
+    "banjar": "bjn-ID",
+    "bashkir": "ba-RU",
+    "basque": "eu-ES",
+    "belarusian": "be-BY",
+    "belgian french": "fr-BE",
+    "bemba": "bem-ZM",
+    "bengali": "bn-IN",
+    "bhojpuri": "bho-IN",
+    "bihari": "bh-IN",
+    "bislama": "bi-VU",
+    "borana": "gax-KE",
+    "bosnian": "bs-BA",
+    "bosnian (cyrillic)": "bs-Cyrl-BA",
+    "breton": "br-FR",
+    "buginese": "bug-ID",
+    "bulgarian": "bg-BG",
+    "burmese": "my-MM",
+    "catalan": "ca-ES",
+    "catalan valencian": "cav-ES",
+    "cebuano": "ceb-PH",
+    "central atlas tamazight": "tzm-MA",
+    "central aymara": "ayr-BO",
+    "central kanuri (latin script)": "knc-NG",
+    "chadian arabic": "shu-TD",
+    "chamorro": "ch-GU",
+    "cherokee": "chr-US",
+    "chhattisgarhi": "hne-IN",
+    "chinese simplified": "zh-CN",
+    "chinese trad. (hong kong)": "zh-HK",
+    "chinese traditional": "zh-TW",
+    "chinese traditional macau": "zh-MO",
+    "chittagonian": "ctg-BD",
+    "chokwe": "cjk-AO",
+    "classical greek": "grc-GR",
+    "comorian ngazidja": "zdj-KM",
+    "coptic": "cop-EG",
+    "crimean tatar": "crh-RU",
+    "crioulo upper guinea": "pov-GW",
+    "croatian": "hr-HR",
+    "czech": "cs-CZ",
+    "danish": "da-DK",
+    "dari": "prs-AF",
+    "dimli": "diq-TR",
+    "dutch": "nl-NL",
+    "dyula": "dyu-CI",
+    "dzongkha": "dz-BT",
+    "eastern yiddish": "ydd-US",
+    "emakhuwa": "vmw-MZ",
+    "english": "en-GB",
+    "english australia": "en-AU",
+    "english canada": "en-CA",
+    "english india": "en-IN",
+    "english ireland": "en-IE",
+    "english new zealand": "en-NZ",
+    "english singapore": "en-SG",
+    "english south africa": "en-ZA",
+    "english us": "en-US",
+    "esperanto": "eo-EU",
+    "estonian": "et-EE",
+    "ewe": "ee-GH",
+    "fanagalo": "fn-FNG",
+    "faroese": "fo-FO",
+    "fijian": "fj-FJ",
+    "filipino": "fil-PH",
+    "finnish": "fi-FI",
+    "flemish": "nl-BE",
+    "fon": "fon-BJ",
+    "french": "fr-FR",
+    "french canada": "fr-CA",
+    "french swiss": "fr-CH",
+    "friulian": "fur-IT",
+    "fula": "ff-FUL",
+    "galician": "gl-ES",
+    "gamargu": "mfi-NG",
+    "garo": "grt-IN",
+    "georgian": "ka-GE",
+    "german": "de-DE",
+    "gilbertese": "gil-KI",
+    "glavda": "glw-NG",
+    "greek": "el-GR",
+    "grenadian creole english": "gcl-GD",
+    "guarani": "gn-PY",
+    "gujarati": "gu-IN",
+    "guyanese creole english": "gyn-GY",
+    "haitian creole french": "ht-HT",
+    "halh mongolian": "khk-MN",
+    "hausa": "ha-NE",
+    "hawaiian": "haw-US",
+    "hebrew": "he-IL",
+    "higi": "hig-NG",
+    "hiligaynon": "hil-PH",
+    "hill mari": "mrj-RU",
+    "hindi": "hi-IN",
+    "hmong": "hmn-CN",
+    "hungarian": "hu-HU",
+    "icelandic": "is-IS",
+    "igbo ibo": "ibo-NG",
+    "igbo ig": "ig-NG",
+    "ilocano": "ilo-PH",
+    "indonesian": "id-ID",
+    "inuktitut greenlandic": "kl-GL",
+    "irish gaelic": "ga-IE",
+    "italian": "it-IT",
+    "italian swiss": "it-CH",
+    "jamaican creole english": "jam-JM",
+    "japanese": "ja-JP",
+    "javanese": "jv-ID",
+    "jingpho": "kac-MM",
+    "k'iche": "quc-GT",
+    "kabiyè": "kbp-TG",
+    "kabuverdianu": "kea-CV",
+    "kabylian": "kab-DZ",
+    "kalenjin": "kln-KE",
+    "kamba": "kam-KE",
+    "kannada": "kn-IN",
+    "kanuri": "kr-KAU",
+    "karen": "kar-MM",
+    "kashmiri (devanagari script)": "ks-IN",
+    "kashmiri (arabic script)": "kas-IN",
+    "kazakh": "kk-KZ",
+    "khasi": "kha-IN",
+    "khmer": "km-KH",
+    "kikuyu kik": "kik-KE",
+    "kikuyu ki": "ki-KE",
+    "kimbundu": "kmb-AO",
+    "kinyarwanda": "rw-RW",
+    "kirundi": "rn-BI",
+    "kisii": "guz-KE",
+    "kongo": "kg-CG",
+    "konkani": "kok-IN",
+    "korean": "ko-KR",
+    "northern kurdish": "kmr-TR",
+    "kurdish sorani": "ckb-IQ",
+    "kyrgyz": "ky-KG",
+    "lao": "lo-LA",
+    "latgalian": "ltg-LV",
+    "latin": "la-XN",
+    "latvian": "lv-LV",
+    "ligurian": "lij-IT",
+    "limburgish": "li-NL",
+    "lingala": "ln-LIN",
+    "lithuanian": "lt-LT",
+    "lombard": "lmo-IT",
+    "luba-kasai": "lua-CD",
+    "luganda": "lg-UG",
+    "luhya": "luy-KE",
+    "luo": "luo-KE",
+    "luxembourgish": "lb-LU",
+    "maa": "mas-KE",
+    "macedonian": "mk-MK",
+    "magahi": "mag-IN",
+    "maithili": "mai-IN",
+    "malagasy": "mg-MG",
+    "malay": "ms-MY",
+    "malayalam": "ml-IN",
+    "maldivian": "dv-MV",
+    "maltese": "mt-MT",
+    "mandara": "mfi-CM",
+    "manipuri": "mni-IN",
+    "manx gaelic": "gv-IM",
+    "maori": "mi-NZ",
+    "marathi": "mr-IN",
+    "margi": "mrt-NG",
+    "mari": "mhr-RU",
+    "marshallese": "mh-MH",
+    "mende": "men-SL",
+    "meru": "mer-KE",
+    "mijikenda": "nyf-KE",
+    "minangkabau": "min-ID",
+    "mizo": "lus-IN",
+    "mongolian": "mn-MN",
+    "montenegrin": "sr-ME",
+    "morisyen": "mfe-MU",
+    "moroccan arabic": "ar-MA",
+    "mossi": "mos-BF",
+    "ndau": "ndc-MZ",
+    "ndebele": "nr-ZA",
+    "nepali": "ne-NP",
+    "nigerian fulfulde": "fuv-NG",
+    "niuean": "niu-NU",
+    "north azerbaijani": "azj-AZ",
+    "sesotho": "nso-ZA",
+    "northern uzbek": "uzn-UZ",
+    "norwegian bokmål": "nb-NO",
+    "norwegian nynorsk": "nn-NO",
+    "nuer": "nus-SS",
+    "nyanja": "ny-MW",
+    "occitan": "oc-FR",
+    "occitan aran": "oc-ES",
+    "odia": "or-IN",
+    "oriya": "ory-IN",
+    "urdu": "ur-PK",
+    "palauan": "pau-PW",
+    "pali": "pi-IN",
+    "pangasinan": "pag-PH",
+    "papiamentu": "pap-CW",
+    "pashto": "ps-PK",
+    "persian": "fa-IR",
+    "pijin": "pis-SB",
+    "plateau malagasy": "plt-MG",
+    "polish": "pl-PL",
+    "portuguese": "pt-PT",
+    "portuguese brazil": "pt-BR",
+    "potawatomi": "pot-US",
+    "punjabi": "pa-IN",
+    "punjabi (pakistan)": "pnb-PK",
+    "quechua": "qu-PE",
+    "rohingya": "rhg-MM",
+    "rohingyalish": "rhl-MM",
+    "romanian": "ro-RO",
+    "romansh": "roh-CH",
+    "rundi": "run-BI",
+    "russian": "ru-RU",
+    "saint lucian creole french": "acf-LC",
+    "samoan": "sm-WS",
+    "sango": "sg-CF",
+    "sanskrit": "sa-IN",
+    "santali": "sat-IN",
+    "sardinian": "sc-IT",
+    "scots gaelic": "gd-GB",
+    "sena": "seh-ZW",
+    "serbian cyrillic": "sr-Cyrl-RS",
+    "serbian latin": "sr-Latn-RS",
+    "seselwa creole french": "crs-SC",
+    "setswana (south africa)": "tn-ZA",
+    "shan": "shn-MM",
+    "shona": "sn-ZW",
+    "sicilian": "scn-IT",
+    "silesian": "szl-PL",
+    "sindhi snd": "snd-PK",
+    "sindhi sd": "sd-PK",
+    "sinhala": "si-LK",
+    "slovak": "sk-SK",
+    "slovenian": "sl-SI",
+    "somali": "so-SO",
+    "sotho southern": "st-LS",
+    "south azerbaijani": "azb-AZ",
+    "southern pashto": "pbt-PK",
+    "southwestern dinka": "dik-SS",
+    "spanish": "es-ES",
+    "spanish argentina": "es-AR",
+    "spanish colombia": "es-CO",
+    "spanish latin america": "es-419",
+    "spanish mexico": "es-MX",
+    "spanish united states": "es-US",
+    "sranan tongo": "srn-SR",
+    "standard latvian": "lvs-LV",
+    "standard malay": "zsm-MY",
+    "sundanese": "su-ID",
+    "swahili": "sw-KE",
+    "swati": "ss-SZ",
+    "swedish": "sv-SE",
+    "swiss german": "de-CH",
+    "syriac (aramaic)": "syc-TR",
+    "tagalog": "tl-PH",
+    "tahitian": "ty-PF",
+    "tajik": "tg-TJ",
+    "tamashek (tuareg)": "tmh-DZ",
+    "tamasheq": "taq-ML",
+    "tamil india": "ta-IN",
+    "tamil sri lanka": "ta-LK",
+    "taroko": "trv-TW",
+    "tatar": "tt-RU",
+    "telugu": "te-IN",
+    "tetum": "tet-TL",
+    "thai": "th-TH",
+    "tibetan": "bo-CN",
+    "tigrinya": "ti-ET",
+    "tok pisin": "tpi-PG",
+    "tokelauan": "tkl-TK",
+    "tongan": "to-TO",
+    "tosk albanian": "als-AL",
+    "tsonga": "ts-ZA",
+    "tswa": "tsc-MZ",
+    "tswana": "tn-BW",
+    "tumbuka": "tum-MW",
+    "turkish": "tr-TR",
+    "turkmen": "tk-TM",
+    "tuvaluan": "tvl-TV",
+    "twi": "tw-GH",
+    "udmurt": "udm-RU",
+    "ukrainian": "uk-UA",
+    "uma": "ppk-ID",
+    "umbundu": "umb-AO",
+    "uyghur uig": "uig-CN",
+    "uyghur ug": "ug-CN",
+    "uzbek": "uz-UZ",
+    "venetian": "vec-IT",
+    "vietnamese": "vi-VN",
+    "vincentian creole english": "svc-VC",
+    "virgin islands creole english": "vic-US",
+    "wallisian": "wls-WF",
+    "waray (philippines)": "war-PH",
+    "welsh": "cy-GB",
+    "west central oromo": "gaz-ET",
+    "western persian": "pes-IR",
+    "wolof": "wo-SN",
+    "xhosa": "xh-ZA",
+    "yiddish": "yi-YD",
+    "yoruba": "yo-NG",
+    "zulu": "zu-ZA",
+}
+

utils/translate_manga.py

@@ -2,10 +2,10 @@
 This module is used to translate manga from one language to another.
 """
 
-from deep_translator import GoogleTranslator
+from deep_translator import MyMemoryTranslator
 
 
-def translate_manga(text: str, source_lang: str = "ja", target_lang: str = "en") -> str:
+def translate_manga(text: str, target_lang: str, source_lang: str = "ja-JP") -> str:
     """
     Translate manga from one language to another.
     """
@@ -13,9 +13,13 @@ def translate_manga(text: str, source_lang: str = "ja", target_lang: str = "en")
     if source_lang == target_lang:
         return text
 
-    translated_text = GoogleTranslator(
-        source=source_lang, target=target_lang).translate(text)
+    if text == "．．．":
+        return text
+
+    translated_text = MyMemoryTranslator(
+        source=source_lang, target=target_lang
+    ).translate(text)
     print("Original text:", text)
     print("Translated text:", translated_text)
 
-    return translated_text
+    return translated_text if translated_text != "．．．" else text

utils/write_text_on_image.py

@@ -1,59 +1,95 @@
 """
 This module contains a function to add text to an image with a bounding box.
 """
+
+import unicodedata
 import textwrap
 from PIL import Image, ImageDraw, ImageFont
 import numpy as np
 import cv2
 
+import arabic_reshaper
+from bidi.algorithm import get_display
+
+
+def detect_script(text):
+    """
+    Detect the script of the text
+    """
+    scripts = set()
+    for char in text:
+        if char.isalpha():
+            name = unicodedata.name(char, "")
+            if "LATIN" in name:
+                scripts.add("Latin")
+            elif "ARABIC" in name:
+                scripts.add("Arabic")
+            elif "CYRILLIC" in name:
+                scripts.add("Cyrillic")
+            elif "GREEK" in name:
+                scripts.add("Greek")
+            elif "HEBREW" in name:
+                scripts.add("Hebrew")
+            elif "DEVANAGARI" in name:
+                scripts.add("Devanagari")
+    if not scripts:
+        return "Latin"
+    return list(scripts)[0]
+
+
+def get_font_path(script):
+    if script == "Latin":
+        return "./fonts/NotoSans-Regular.ttf"
+    elif script == "Arabic":
+        return "./fonts/NotoNaskhArabic-Regular.ttf"
+    elif script == "Cyrillic":
+        return "./fonts/NotoSansCyrillic-Regular.ttf"
+    elif script == "Greek":
+        return "./fonts/NotoSansGreek-Regular.ttf"
+    else:
+        return "./fonts/NotoSans-Regular.ttf"
 
-def add_text(image: np.ndarray, text: str, contour: np.ndarray):
-	"""
-	Add text to an image with a bounding box.
-	"""
-
-	font_path = "./fonts/fonts_animeace_i.ttf"
-	pil_image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
-	draw = ImageDraw.Draw(pil_image)
-
-	x, y, w, h = cv2.boundingRect(contour)
-
-	line_height = 16
-	font_size = 14
-	wrapping_ratio = 0.075
-
-	wrapped_text = textwrap.fill(text, width=int(w * wrapping_ratio),
-									break_long_words=True)
-
-	font = ImageFont.truetype(font_path, size=font_size)
-
-	lines = wrapped_text.split('\n')
-	total_text_height = (len(lines)) * line_height
-
-	while total_text_height > h:
-		line_height -= 2
-		font_size -= 2
-		wrapping_ratio += 0.025
 
-		wrapped_text = textwrap.fill(text, width=int(w * wrapping_ratio),
-										break_long_words=True)
-
-		font = ImageFont.truetype(font_path, size=font_size)
-
-		lines = wrapped_text.split('\n')
-		total_text_height = (len(lines)) * line_height
-
-	# Vertical centering
-	text_y = y + (h - total_text_height) // 2
-
-	for line in lines:
-		text_length = draw.textlength(line, font=font)
-
-		# Horizontal centering
-		text_x = x + (w - text_length) // 2
-
-		draw.text((text_x, text_y), line, font=font, fill=(0, 0, 0))
-
-		text_y += line_height
-
-	image[:, :, :] = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR)
+def add_text(image: np.ndarray, text: str, contour: np.ndarray):
+    script = detect_script(text)
+    font_path = get_font_path(script)
+    if script == "Arabic":
+        reshaped_text = arabic_reshaper.reshape(text)
+        text = get_display(reshaped_text)
+    pil_image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
+    draw = ImageDraw.Draw(pil_image)
+
+    x, y, w, h = cv2.boundingRect(contour)
+
+    line_height = 16
+    font_size = 14
+    wrapping_ratio = 0.075
+
+    wrap_width = max(1, int(w * wrapping_ratio))
+    wrapped_text = textwrap.fill(text, width=wrap_width, break_long_words=True)
+
+    font = ImageFont.truetype(font_path, size=font_size)
+
+    lines = wrapped_text.split("\n")
+    total_text_height = (len(lines)) * line_height
+
+    while total_text_height > h:
+        line_height -= 2
+        font_size -= 2
+        wrapping_ratio += 0.025
+        wrap_width = max(1, int(w * wrapping_ratio))
+        wrapped_text = textwrap.fill(text, width=wrap_width, break_long_words=True)
+        font = ImageFont.truetype(font_path, size=font_size)
+        lines = wrapped_text.split("\n")
+        total_text_height = (len(lines)) * line_height
+
+    # Vertical centering
+    text_y = y + (h - total_text_height) // 2
+
+    for line in lines:
+        text_length = draw.textlength(line, font=font)
+        text_x = x + (w - text_length) // 2
+        draw.text((text_x, text_y), line, font=font, fill=(0, 0, 0))
+        text_y += line_height
+
+    image[:, :, :] = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR)