...

app/prompt_builder/prompt.py

@@ -220,7 +220,7 @@ I. ШАПКА. Lesson Plan Header — provide the core lesson metadata in a comp
 
 
 
-II **План урока** 
+II **План урока** (обычная таблица без классов)
 
 III **Teaching Reference Sheet** (style guide: this is a reference block and must fit a single vertical A4 compactly yet reader-friendly).
 Structure:

app/test_gradio_test.py

@@ -59,104 +59,92 @@ def upload_to_catbox(file_bytes: bytes, filename: str = "image.png", expire: str
 import re
 
 
+
+
 from docx import Document
 from bs4 import BeautifulSoup
+import tempfile
+import os
+from datetime import datetime
+
+from docx import Document
 from docx.shared import Pt
+from bs4 import BeautifulSoup
 import tempfile
 import os
 from datetime import datetime
 
 
 def generate_docx(html_content: str) -> str:
-    """Сохраняет HTML-контент в .docx с сохранением структуры"""
+    """Конвертирует HTML от LLM в DOCX с сохранением ключевой структуры"""
     doc = Document()
     soup = BeautifulSoup(html_content, 'html.parser')
 
-    # Обработка шапки
+    # Обработка шапки с метками
     header_table = soup.find('table', class_='header-table')
     if header_table:
-        table = doc.add_table(rows=0, cols=2)
+        table = doc.add_table(rows=0, cols=3)
         table.style = 'Table Grid'
 
         for row in header_table.find_all('tr'):
             cells = row.find_all('td')
-            if cells:
-                doc_row = table.add_row()
-                for i, cell in enumerate(cells[:2]):
-                    label = cell.find('span', class_='cell-label')
-                    value = cell.find('span', class_='cell-value')
-
-                    p = doc_row.cells[i].paragraphs[0]
-                    if label:
-                        run = p.add_run(label.get_text() + ": ")
-                        run.font.size = Pt(9)
-                        run.bold = True
-                    if value:
-                        run = p.add_run(value.get_text())
-                        run.font.size = Pt(11)
-
-    # Обработка основных блоков
+            doc_row = table.add_row()
+
+            for i, cell in enumerate(cells[:3]):  # Макс 3 колонки
+                content = " ".join([t.get_text() for t in cell.find_all(text=True)])
+                doc_row.cells[i].text = content.replace(" : ", ": ").strip()
+
+    # Обработка основной таблицы плана
+    lesson_table = soup.find('section', class_='output-block').find('table')
+    if lesson_table:
+        cols = max(len(row.find_all(['td', 'th'])) for row in lesson_table.find_all('tr'))
+        doc_table = doc.add_table(rows=0, cols=cols)
+        doc_table.style = 'Table Grid'
+
+        for row in lesson_table.find_all('tr'):
+            cells = row.find_all(['td', 'th'])[:cols]
+            if not cells:
+                continue
+
+            doc_row = doc_table.add_row()
+            for i, cell in enumerate(cells):
+                text = " ".join(cell.stripped_strings)
+                doc_cell = doc_row.cells[i]
+                doc_cell.text = text
+                if cell.find('strong'):
+                    doc_cell.paragraphs[0].runs[0].bold = True
+
+    # Обработка Teaching Reference Sheet
     for section in soup.find_all('section', class_='output-block'):
-        heading = section.find_previous(['h2', 'h3'])
-        if heading:
-            doc.add_heading(heading.get_text(), level=2)
-
-        for element in section.children:
-            # Обработка таблиц
-            if element.name == 'table':
-                rows = element.find_all('tr')
-                if not rows:
-                    continue
-
-                # Определяем количество колонок
-                cols = len(rows[0].find_all(['td', 'th']))
-                table = doc.add_table(rows=0, cols=cols)
-
-                for html_row in rows:
-                    cells = html_row.find_all(['td', 'th'])
-                    if len(cells) != cols:
-                        continue
-
-                    doc_row = table.add_row()
-                    for i, cell in enumerate(cells):
-                        text = cell.get_text().strip()
-                        doc_cell = doc_row.cells[i]
-                        doc_cell.text = text
-                        if cell.name == 'th':
-                            for paragraph in doc_cell.paragraphs:
-                                for run in paragraph.runs:
-                                    run.bold = True
-
-            # Обработка списков
-            elif element.name in ['ul', 'ol']:
-                list_type = 'ListBullet' if element.name == 'ul' else 'ListNumber'
-                for item in element.find_all('li'):
-                    p = doc.add_paragraph(style=list_type)
-                    p.add_run(item.get_text().strip())
-
-            # Обработка обычного текста
-            elif element.name == 'p':
-                text = element.get_text().strip()
-                if text:
-                    doc.add_paragraph(text)
-
-            # Обработка "сырого" текста без тегов
-            elif not element.name and element.strip():
-                doc.add_paragraph(element.strip())
-
-    # Сохранение документа
-    tmp_dir = tempfile.gettempdir()
-    filename = f"lesson_plan_{datetime.now().strftime('%Y%m%d%H%M%S')}.docx"
-    path = os.path.join(tmp_dir, filename)
-
-    try:
-        doc.save(path)
-        if not os.path.exists(path):
-            raise Exception("Ошибка создания файла")
-        return path
-    except Exception as e:
-        print(f"Ошибка: {str(e)}")
-        raise
+        if 'Teaching Reference Sheet' in section.get_text():
+            doc.add_heading('Teaching Reference Sheet', level=2)
+
+            # Двухколоночный макет через таблицу
+            cols_table = doc.add_table(rows=1, cols=2)
+            cols_table.autofit = True
+
+            # Левая колонка
+            left = cols_table.cell(0, 0)
+            for item in section.find_all(['h3', 'strong', 'em']):
+                if item.name == 'h3':
+                    left.add_paragraph(item.get_text(), style='Heading3')
+                else:
+                    p = left.add_paragraph()
+                    p.add_run(item.get_text()).bold = True
+
+            # Правая колонка
+            right = cols_table.cell(0, 1)
+            for item in section.find_all('h3'):
+                if 'Методические советы' in item.text:
+                    for sib in item.next_siblings:
+                        if sib.name == 'h3':
+                            break
+                        right.add_paragraph(sib.get_text())
+
+    # Сохранение
+    path = os.path.join(tempfile.gettempdir(), f"lesson_plan_{datetime.now().strftime('%Y%m%d%H%M%S')}.docx")
+    doc.save(path)
+    return path
 
 # названия textbooks для автозаполнения поля Учебник
 import json

requirements.txt

@@ -7,3 +7,4 @@ pillow
 bs4
 
 
+