Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -47,9 +47,13 @@ def detect_language(text):
|
|
| 47 |
|
| 48 |
def summarize_text(text):
|
| 49 |
if not text or len(text) < 50:
|
| 50 |
-
return "⚠️ Недостаточно текста для суммаризации."
|
| 51 |
lang = detect_language(text)
|
| 52 |
model = summarizers.get(lang, summarizers["en"])
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
chunk_size = 2500
|
| 54 |
overlap = 200
|
| 55 |
summaries = []
|
|
@@ -60,7 +64,9 @@ def summarize_text(text):
|
|
| 60 |
summaries.append(result[0]['summary_text'])
|
| 61 |
except Exception as e:
|
| 62 |
summaries.append(f"[Ошибка в части {len(summaries)+1}: {str(e)}]")
|
| 63 |
-
|
|
|
|
|
|
|
| 64 |
|
| 65 |
def save_summary_as_docx(summary_text):
|
| 66 |
path = "summary.docx"
|
|
@@ -92,24 +98,26 @@ def save_summary_as_pdf(summary_text):
|
|
| 92 |
def summarize_file(file):
|
| 93 |
text = read_file(file)
|
| 94 |
if text.startswith("Ошибка"):
|
| 95 |
-
return text, None, None, None
|
| 96 |
-
summary = summarize_text(text)
|
| 97 |
txt_path = save_summary_as_txt(summary)
|
| 98 |
docx_path = save_summary_as_docx(summary)
|
| 99 |
pdf_path = save_summary_as_pdf(summary)
|
| 100 |
-
return summary, txt_path, docx_path, pdf_path
|
| 101 |
|
| 102 |
demo = gr.Interface(
|
| 103 |
fn=summarize_file,
|
| 104 |
inputs=gr.File(label="Загрузите документ (.pdf, .docx или .txt)"),
|
| 105 |
outputs=[
|
| 106 |
gr.Textbox(label="Краткое резюме"),
|
|
|
|
|
|
|
| 107 |
gr.File(label="📄 Скачать TXT"),
|
| 108 |
gr.File(label="📘 Скачать DOCX"),
|
| 109 |
gr.File(label="📕 Скачать PDF")
|
| 110 |
],
|
| 111 |
title="🧠 Eroha Summarizer",
|
| 112 |
-
description="
|
| 113 |
)
|
| 114 |
|
| 115 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|
|
|
|
| 47 |
|
| 48 |
def summarize_text(text):
|
| 49 |
if not text or len(text) < 50:
|
| 50 |
+
return "⚠️ Недостаточно текста для суммаризации.", "не определён", "нет модели"
|
| 51 |
lang = detect_language(text)
|
| 52 |
model = summarizers.get(lang, summarizers["en"])
|
| 53 |
+
|
| 54 |
+
lang_label = {"ru": "Русский", "en": "English", "kz": "Қазақ тілі"}.get(lang, "Unknown")
|
| 55 |
+
model_label = model.model.name_or_path if hasattr(model.model, "name_or_path") else "Custom"
|
| 56 |
+
|
| 57 |
chunk_size = 2500
|
| 58 |
overlap = 200
|
| 59 |
summaries = []
|
|
|
|
| 64 |
summaries.append(result[0]['summary_text'])
|
| 65 |
except Exception as e:
|
| 66 |
summaries.append(f"[Ошибка в части {len(summaries)+1}: {str(e)}]")
|
| 67 |
+
|
| 68 |
+
summary_text = "\n\n".join(summaries).strip()
|
| 69 |
+
return summary_text, lang_label, model_label
|
| 70 |
|
| 71 |
def save_summary_as_docx(summary_text):
|
| 72 |
path = "summary.docx"
|
|
|
|
| 98 |
def summarize_file(file):
|
| 99 |
text = read_file(file)
|
| 100 |
if text.startswith("Ошибка"):
|
| 101 |
+
return text, "ошибка", "ошибка", None, None, None
|
| 102 |
+
summary, lang_label, model_label = summarize_text(text)
|
| 103 |
txt_path = save_summary_as_txt(summary)
|
| 104 |
docx_path = save_summary_as_docx(summary)
|
| 105 |
pdf_path = save_summary_as_pdf(summary)
|
| 106 |
+
return summary, lang_label, model_label, txt_path, docx_path, pdf_path
|
| 107 |
|
| 108 |
demo = gr.Interface(
|
| 109 |
fn=summarize_file,
|
| 110 |
inputs=gr.File(label="Загрузите документ (.pdf, .docx или .txt)"),
|
| 111 |
outputs=[
|
| 112 |
gr.Textbox(label="Краткое резюме"),
|
| 113 |
+
gr.Textbox(label="🌍 Определённый язык"),
|
| 114 |
+
gr.Textbox(label="🧠 Используемая модель"),
|
| 115 |
gr.File(label="📄 Скачать TXT"),
|
| 116 |
gr.File(label="📘 Скачать DOCX"),
|
| 117 |
gr.File(label="📕 Скачать PDF")
|
| 118 |
],
|
| 119 |
title="🧠 Eroha Summarizer",
|
| 120 |
+
description="Определяет язык (RU / EN / KZ) и создаёт краткое резюме с выбором модели под язык."
|
| 121 |
)
|
| 122 |
|
| 123 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|