Update app.py

app.py

@@ -3,107 +3,85 @@ from transformers import pipeline
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 from langdetect import detect
-import re
-import datetime
-import hashlib
-import io
+import re, datetime, hashlib, io, json
 
-# ================== Утилиты ==================
-
-def clean_text(text: str):
-    text = text.replace("\n", " ").replace("\r", " ")
-    text = re.sub(r"\s+", " ", text)
+# === Утилиты ===
+def clean_text(text):
+    text = re.sub(r"\s+", " ", text.replace("\n", " ").replace("\r", " "))
     text = re.sub(r"[^\w\s.,!?%\-–:;()\"'’«»]", "", text)
     return text.strip()
 
-def detect_language(text: str):
+def detect_language(text):
     try:
         lang = detect(text)
     except:
         lang = "en"
-
-    kazakh_letters = "қңәөүһіұ"
-    if any(ch in text.lower() for ch in kazakh_letters):
+    if any(ch in text.lower() for ch in "қңәөүһіұ"):
         lang = "kk"
     return lang
 
-def generate_slug(title: str):
+def generate_slug(title):
     slug = re.sub(r"[^a-zA-Zа-яА-Я0-9]+", "-", title.lower()).strip("-")
-    slug_hash = hashlib.md5(title.encode()).hexdigest()[:6]
-    return f"/news/{slug}-{slug_hash}"
-
-# ================== Модели ==================
-
-summarizers = {}
-analyzers = {}
-
-def get_summarizer(lang: str):
-    if lang == "ru":
-        model_name = "IlyaGusev/mbart_ru_sum_gazeta"
-    elif lang == "kk":
-        model_name = "facebook/mbart-large-50-many-to-many-mmt"
-    else:
-        model_name = "facebook/bart-large-cnn"
-    if model_name not in summarizers:
-        summarizers[model_name] = pipeline("summarization", model=model_name)
-    return summarizers[model_name]
-
-def get_sentiment_analyzer(lang: str):
-    if lang in ["ru", "kk"]:
-        model_name = "nlptown/bert-base-multilingual-uncased-sentiment"
-    else:
-        model_name = "cardiffnlp/twitter-roberta-base-sentiment"
-    if model_name not in analyzers:
-        analyzers[model_name] = pipeline("sentiment-analysis", model=model_name)
-    return analyzers[model_name]
-
-# ================== Контент ==================
-
-def extract_keywords(text: str, top_n: int = 7):
+    return f"/news/{slug}-{hashlib.md5(title.encode()).hexdigest()[:6]}"
+
+# === Модели ===
+summarizers, analyzers = {}, {}
+
+def get_summarizer(lang):
+    model = {
+        "ru": "IlyaGusev/mbart_ru_sum_gazeta",
+        "kk": "facebook/mbart-large-50-many-to-many-mmt",
+    }.get(lang, "facebook/bart-large-cnn")
+    if model not in summarizers:
+        summarizers[model] = pipeline("summarization", model=model)
+    return summarizers[model]
+
+def get_sentiment_analyzer(lang):
+    model = (
+        "nlptown/bert-base-multilingual-uncased-sentiment"
+        if lang in ["ru", "kk"]
+        else "cardiffnlp/twitter-roberta-base-sentiment"
+    )
+    if model not in analyzers:
+        analyzers[model] = pipeline("sentiment-analysis", model=model)
+    return analyzers[model]
+
+# === Логика ===
+def extract_keywords(text, top_n=7):
     words = re.findall(r"\b\w{5,}\b", text.lower())
     freq = {}
     for w in words:
         freq[w] = freq.get(w, 0) + 1
-    keywords = sorted(freq, key=freq.get, reverse=True)[:top_n]
-    return ", ".join(keywords)
+    return ", ".join(sorted(freq, key=freq.get, reverse=True)[:top_n])
 
-def detect_topic(text: str):
+def detect_topic(text):
     topics = {
         "Экономика": ["рынок", "компания", "инвестиция", "қаржы", "сату"],
         "Технологии": ["ai", "робот", "интернет", "жасанды интеллект"],
         "Саясат": ["үкімет", "закон", "президент", "выборы"],
         "Ғылым": ["зерттеу", "ғалым", "эксперимент"],
-        "Спорт": ["матч", "команда", "спорт"]
+        "Спорт": ["матч", "команда", "спорт"],
     }
-    text_lower = text.lower()
+    t = text.lower()
     for topic, words in topics.items():
-        if any(w in text_lower for w in words):
+        if any(w in t for w in words):
             return topic
     return "Жалпы тақырып / Общая тема"
 
-# ================== Основная логика ==================
-
-def summarize_text(text: str):
+def summarize_text(text):
     if not text.strip():
         return "⚠️ Введите текст для анализа.", None
-
     text = clean_text(text)
     lang = detect_language(text)
 
-    summarizer = get_summarizer(lang)
-    sentiment_model = get_sentiment_analyzer(lang)
+    summ = get_summarizer(lang)
+    sent_model = get_sentiment_analyzer(lang)
 
     words = len(text.split())
-    if words < 80:
-        max_len, min_len = 70, 20
-    elif words < 300:
-        max_len, min_len = 140, 40
-    else:
-        max_len, min_len = 220, 60
-
-    summary = summarizer(text, max_length=max_len, min_length=min_len, do_sample=False)[0]["summary_text"]
+    max_len, min_len = (70, 20) if words < 80 else (140, 40) if words < 300 else (220, 60)
 
-    sentiment = sentiment_model(summary)[0]["label"].lower()
+    summary = summ(text, max_length=max_len, min_length=min_len, do_sample=False)[0]["summary_text"]
+    sentiment = sent_model(summary)[0]["label"].lower()
     if "5" in sentiment or "pos" in sentiment:
         sentiment = "😊 Позитивті / Позитивное"
     elif "1" in sentiment or "neg" in sentiment:
@@ -114,49 +92,47 @@ def summarize_text(text: str):
     topic = detect_topic(text)
     keywords = extract_keywords(text)
     title = summary.split(".")[0][:80].strip()
-    meta_description = summary[:160].strip()
+    meta_desc = summary[:160].strip()
     slug = generate_slug(title)
     date_now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M")
 
-    score = 0
-    score += 1 if len(keywords.split(",")) >= 5 else 0
-    score += 1 if len(meta_description) >= 100 else 0
-    score += 1 if len(title) > 20 else 0
-    seo_status = "✅ Оптимально для публикации" if score >= 2 else "⚠️ Недостаточно данных для SEO"
-
-    lang_name = {
-        "kk": "Қазақ (Kazakh)",
-        "ru": "Русский (Russian)",
-        "en": "Ағылшын (English)"
-    }.get(lang, "Multilingual")
-
-    output = f"# 🧠 Eroha Summarizer PRO++++ v2.4 Publisher Edition\n"
-    output += f"## 🌍 Language: {lang_name}\n"
-    output += f"### 📅 Date: {date_now}\n"
-    output += f"### 📌 Topic: {topic}\n"
-    output += f"### 💬 Sentiment: {sentiment}\n\n"
-    output += "---\n\n"
-    output += f"📄 **Summary:**\n{summary}\n\n"
-    output += "---\n\n"
-    output += f"## 🧭 SEO Optimization\n"
-    output += f"**📰 Title:** {title}\n\n"
+    # === JSON-LD Schema.org ===
+    schema_data = {
+        "@context": "https://schema.org",
+        "@type": "NewsArticle",
+        "headline": title,
+        "datePublished": datetime.datetime.now().isoformat(),
+        "articleSection": topic,
+        "keywords": keywords.split(", "),
+        "description": meta_desc,
+        "inLanguage": lang,
+        "mainEntityOfPage": f"https://eroha.ai{slug}",
+        "author": {"@type": "Organization", "name": "Eroha Intelligence Suite"},
+        "publisher": {
+            "@type": "Organization",
+            "name": "Eroha Intelligence Suite",
+            "logo": {"@type": "ImageObject", "url": "https://eroha.ai/logo.png"},
+        },
+    }
+    json_ld = json.dumps(schema_data, indent=2, ensure_ascii=False)
+
+    output = f"# 🧠 Eroha Summarizer PRO++++ v2.5 Press-Optimized Edition\n"
+    output += f"## 🌍 Language: {lang.upper()}\n### 📅 Date: {date_now}\n"
+    output += f"### 📌 Topic: {topic}\n### 💬 Sentiment: {sentiment}\n\n"
+    output += f"---\n\n📄 **Summary:**\n{summary}\n\n---\n\n"
+    output += f"## 🧭 SEO Optimization\n**📰 Title:** {title}\n\n"
     output += f"**🔑 Keywords:** {keywords}\n\n"
-    output += f"**📄 Meta Description:** {meta_description}\n\n"
+    output += f"**📄 Meta Description:** {meta_desc}\n\n"
     output += f"**🔗 Slug:** `{slug}`\n\n"
-    output += f"**📊 SEO Score:** {seo_status}\n\n"
-    output += "---\n\n"
-    output += f"🔖 **Tags:** #Eroha #AI #SEO #Publisher #Kazakhstan #Press #News\n"
-
-    # Создание Markdown-файла
-    filename = f"Eroha_Summary_{datetime.datetime.now().strftime('%Y-%m-%d_%H-%M')}.md"
-    md_bytes = io.BytesIO(output.encode('utf-8'))
-    md_bytes.name = filename
+    output += "---\n\n### 🧱 JSON-LD (Schema.org)\n```json\n{json_ld}\n```\n\n"
+    output += "---\n\n🔖 **Tags:** #Eroha #AI #SEO #Press #Kazakhstan #News\n"
 
-    return output, md_bytes
+    md_file = io.BytesIO(output.encode("utf-8"))
+    md_file.name = f"Eroha_Summary_{datetime.datetime.now().strftime('%Y-%m-%d_%H-%M')}.md"
+    return output, md_file
 
-# ================== API и UI ==================
-
-app = FastAPI(title="Eroha Summarizer PRO++++ v2.4 Publisher Edition")
+# === API + UI ===
+app = FastAPI(title="Eroha Summarizer PRO++++ v2.5")
 app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"], allow_headers=["*"])
 
 @app.post("/api/summarize")
@@ -165,25 +141,15 @@ async def summarize_api(data: dict):
     summary, _ = summarize_text(text)
     return {"summary": summary}
 
-# Gradio интерфейс
-with gr.Blocks(title="Eroha Summarizer PRO++++ v2.4 Publisher Edition") as iface:
-    gr.Markdown("# 🧠 Eroha Summarizer PRO++++ v2.4 Publisher Edition")
-    gr.Markdown("AI-инструмент для суммаризации, анализа, SEO и экспорта Markdown (с поддержкой казахского 🇰🇿)")
-
-    input_box = gr.Textbox(lines=8, label="Введите текст / Мәтінді енгізіңіз")
-    summarize_btn = gr.Button("🚀 Анализ и SEO-суммаризация")
-    clear_btn = gr.Button("🧹 Очистить")
-    copy_btn = gr.Button("📋 Копировать результат")
-    download_btn = gr.File(label="💾 Скачать результат в Markdown")
-
-    output_box = gr.Markdown(label="Результат / Result")
-
-    def process_input(text):
-        summary, md_file = summarize_text(text)
-        return summary, md_file
-
-    summarize_btn.click(process_input, inputs=input_box, outputs=[output_box, download_btn])
-    clear_btn.click(lambda: "", None, input_box)
-    copy_btn.click(lambda t: t, input_box, input_box)
+with gr.Blocks(title="Eroha Summarizer PRO++++ v2.5 Press-Optimized Edition") as iface:
+    gr.Markdown("# 🧠 Eroha Summarizer PRO++++ v2.5 Press-Optimized Edition")
+    gr.Markdown("AI-инструмент для суммаризации, SEO и экспорта с JSON-LD микроразметкой (NewsArticle)")
+    text_in = gr.Textbox(lines=8, label="Введите текст / Мәтінді енгізіңіз")
+    btn = gr.Button("🚀 Анализ и экспорт SEO-структуры")
+    clear = gr.Button("🧹 Очистить")
+    file_out = gr.File(label="💾 Скачать Markdown")
+    text_out = gr.Markdown(label="Результат")
+    btn.click(lambda t: summarize_text(t), inputs=text_in, outputs=[text_out, file_out])
+    clear.click(lambda: "", None, text_in)
 
 iface.launch(server_name="0.0.0.0", server_port=7860)