app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

print("Modeller yukleniyor...")
tr_to_en_tok = AutoTokenizer.from_pretrained("remox01/opus-mt-tc-big-tr-en")
tr_to_en_model = AutoModelForSeq2SeqLM.from_pretrained("remox01/opus-mt-tc-big-tr-en")
en_to_tr_tok = AutoTokenizer.from_pretrained("remox01/opus-mt-tc-big-en-tr")
en_to_tr_model = AutoModelForSeq2SeqLM.from_pretrained("remox01/opus-mt-tc-big-en-tr")

print("Zemberek yukleniyor...")
from zemberek import TurkishMorphology
morphology = TurkishMorphology.create_with_defaults()

print("spaCy yukleniyor...")
import spacy
nlp = spacy.load("en_core_web_sm")


def translate(text, direction):
    if not text.strip():
        return ""
    if direction == "TR -> EN":
        tokenizer, model = tr_to_en_tok, tr_to_en_model
    else:
        tokenizer, model = en_to_tr_tok, en_to_tr_model
    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True, max_length=128)
    outputs = model.generate(**inputs, max_length=128)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)


def zemberek_analyze(text):
    if not text.strip():
        return "Cumle bos."
    words = text.split()
    results = []
    for word in words:
        clean = word.strip(".,!?;:()")
        try:
            wa = morphology.analyze(clean)
            if wa.analysis_results:
                sa = wa.analysis_results[0]
                morphemes = sa.get_morphemes()
                suffix_list = [str(m) for m in morphemes[1:]] if len(morphemes) > 1 else []
                pos = morphemes[0].pos.name if morphemes else "Unknown"
                suffix_str = " + ".join(suffix_list) if suffix_list else "-"
                results.append(f"Kelime: {word}\n  Kok: {sa.get_stem()}\n  Ekler: {suffix_str}\n  POS: {pos}\n  Analiz: {sa.format_string()}")
            else:
                results.append(f"Kelime: {word}\n  Analiz bulunamadi")
        except Exception as e:
            results.append(f"Kelime: {word}\n  Hata: {str(e)}")
    return "\n\n".join(results)


def spacy_analyze(text):
    if not text.strip():
        return "Cumle bos."
    doc = nlp(text)
    results = []
    for token in doc:
        results.append(f"Kelime: {token.text}\n  Lemma: {token.lemma_}\n  POS: {token.pos_} ({token.tag_})\n  Dep: {token.dep_}\n  Shape: {token.shape_}\n  Stopword: {token.is_stop}")

    entities = [(ent.text, ent.label_) for ent in doc.ents]
    entity_str = "\n".join([f"  {text} -> {label}" for text, label in entities]) if entities else "  Varlik bulunamadi"

    return "\n\n".join(results) + "\n\nVARLIKLAR (NER):\n" + entity_str


def comprehensive_analysis(turkish, english):
    results = []
    if turkish.strip():
        results.append("=== TURKCE ZEMBEREK ANALIZI ===")
        results.append(zemberek_analyze(turkish))
        results.append("")
        results.append("=== TURKCE -> INGILIZCE (Helsinki-NLP) ===")
        results.append(translate(turkish, "TR -> EN"))
        results.append("")
    if english.strip():
        results.append("=== INGILIZCE SPACY ANALIZI ===")
        results.append(spacy_analyze(english))
        results.append("")
        results.append("=== INGILIZCE -> TURKCE (Helsinki-NLP) ===")
        results.append(translate(english, "EN -> TR"))
    return "\n".join(results) if results else "Lutfen bir cumle girin."


with gr.Blocks(title="XDE - Turkce-Ingilizce Dil Araci") as demo:
    gr.Markdown("# XDE - Turkce-Ingilizce Dil Araci")
    gr.Markdown("Helsinki-NLP, Zemberek ve spaCy ile dil analizi")

    with gr.Tab("Ceviri"):
        with gr.Row():
            direction = gr.Dropdown(["TR -> EN", "EN -> TR"], value="TR -> EN", label="Yon")
            input_text = gr.Textbox(label="Cumle", placeholder="Turkce veya Ingilizce cumle girin...")
        output_text = gr.Textbox(label="Ceviri")
        translate_btn = gr.Button("CEVIR")
        translate_btn.click(translate, inputs=[input_text, direction], outputs=output_text)

    with gr.Tab("Zemberek (Turkce)"):
        zemberek_input = gr.Textbox(label="Turkce Cumle", placeholder="Turkce cumle girin...")
        zemberek_output = gr.Textbox(label="Morfolojik Analiz", lines=15)
        zemberek_btn = gr.Button("ANALIZ ET")
        zemberek_btn.click(zemberek_analyze, inputs=zemberek_input, outputs=zemberek_output)

    with gr.Tab("spaCy (Ingilizce)"):
        spacy_input = gr.Textbox(label="Ingilizce Cumle", placeholder="Ingilizce cumle girin...")
        spacy_output = gr.Textbox(label="Dilbilgisi Analizi", lines=15)
        spacy_btn = gr.Button("ANALIZ ET")
        spacy_btn.click(spacy_analyze, inputs=spacy_input, outputs=spacy_output)

    with gr.Tab("Karsilastirmali Analiz"):
        with gr.Row():
            tr_input = gr.Textbox(label="Turkce Cumle", placeholder="Merhaba, nasilsiniz?")
            en_input = gr.Textbox(label="Ingilizce Cumle", placeholder="Hello, how are you?")
        comp_output = gr.Textbox(label="Sonuc", lines=20)
        comp_btn = gr.Button("KARSILASTIR")
        comp_btn.click(comprehensive_analysis, inputs=[tr_input, en_input], outputs=comp_output)

demo.launch()