...

app/test_gradio_test.py

@@ -58,60 +58,92 @@ def upload_to_catbox(file_bytes: bytes, filename: str = "image.png", expire: str
 
 import re
 
+import os, tempfile
+from datetime import datetime
+
 from bs4 import BeautifulSoup
 from docx import Document
 from docx.enum.section import WD_ORIENT
-from html_for_docx import HTML2DOCX  # класс из html‑for‑docx
-import tempfile, os
-from datetime import datetime
+from htmldocx import HtmlToDocx
 
 def generate_docx(html: str) -> str:
+    # 1. Парсим HTML
     soup = BeautifulSoup(html, "html.parser")
 
-    # Убираем все цвета, border, float; оставляем только font-size, padding
+    # 1.1 – очищаем все style, оставляем только font-size (em/px→pt), padding, vertical-align
     for tag in soup.find_all(True):
-        style = tag.get("style","")
-        safe=[]
-        for part in style.split(";"):
-            if part.strip().startswith(("font-size","padding","vertical-align")):
-                # конвертируем em/px→pt
-                prop,val=part.split(":")
-                if val.endswith("em"):
-                    val=f"{int(float(val[:-2])*12)}pt"
-                elif val.endswith("px"):
-                    val=f"{round(int(val[:-2])*72/96)}pt"
-                safe.append(f"{prop}:{val}")
-        if safe: tag["style"]=";".join(safe)
-        else: tag.attrs.pop("style",None)
-
-    # Преобразуем два <div> в section.output-block → таблицу 1×2
-    for section in soup.select("section.output-block"):
-        divs=section.find_all("div",recursive=False)
-        if len(divs)==2:
-            table=soup.new_tag("table", style="width:100%;border-collapse:collapse")
-            tr=soup.new_tag("tr")
+        style = tag.get("style", "")
+        if style:
+            parts = []
+            for part in style.split(";"):
+                if ":" not in part:
+                    continue
+                prop, val = part.split(":", 1)
+                prop, val = prop.strip(), val.strip()
+                if prop == "font-size":
+                    # em→pt, px→pt
+                    if val.endswith("em"):
+                        pt = int(float(val[:-2]) * 12)
+                        parts.append(f"font-size:{pt}pt")
+                    elif val.endswith("px"):
+                        pt = round(int(val[:-2]) * 72/96)
+                        parts.append(f"font-size:{pt}pt")
+                elif prop in ("padding", "vertical-align", "margin-bottom"):
+                    parts.append(f"{prop}:{val}")
+            if parts:
+                tag["style"] = ";".join(parts)
+            else:
+                tag.attrs.pop("style", None)
+
+    # 1.2 – шапка: cell-label и cell-value
+    for span in soup.select("span.cell-label"):
+        span["style"] = "display:block;font-size:9pt;color:#555;margin-bottom:0.2em"
+    for span in soup.select("span.cell-value"):
+        span["style"] = "display:block;font-size:12pt"
+
+    # 1.3 – заголовки h4 → p.bold
+    for h4 in soup.find_all("h4"):
+        p = soup.new_tag("p")
+        p.string = h4.get_text()
+        p["style"] = "font-size:12pt;font-weight:bold;margin-bottom:0.2em"
+        h4.replace_with(p)
+
+    # 1.4 – Teaching Reference Sheet (только секция без header‑block)
+    for section in soup.find_all("section", class_="output-block"):
+        classes = section.get("class", [])
+        if "header-block" in classes or section.find("table"):
+            continue
+        divs = section.find_all("div", recursive=False)
+        if len(divs) == 2:
+            table = soup.new_tag("table")
+            table["style"] = "width:100%;border-collapse:collapse"
+            tr = soup.new_tag("tr")
             for div in divs:
-                td=soup.new_tag("td", style="vertical-align:top;padding:0.5em;font-size:11pt;")
-                td.extend(div.contents)
+                td = soup.new_tag("td")
+                td["style"] = "vertical-align:top;padding:0.5em;font-size:11pt"
+                for ch in div.contents:
+                    td.append(ch)
                 tr.append(td)
             table.append(tr)
-            section.clear(); section.append(table)
+            section.clear()
+            section.append(table)
 
-    clean_html=str(soup)
+    clean_html = str(soup)
 
-    # Создаём документ и ландшафт
-    doc=Document()
-    sec=doc.sections[-1]
-    sec.orientation=WD_ORIENT.LANDSCAPE
-    sec.page_width,sec.page_height=sec.page_height,sec.page_width
+    # 2. Создаём DOCX и делаем landscape
+    doc = Document()
+    sec = doc.sections[-1]
+    sec.orientation = WD_ORIENT.LANDSCAPE                # альбом :contentReference[oaicite:4]{index=4}
+    sec.page_width, sec.page_height = sec.page_height, sec.page_width
 
-    # Конвертируем через html-for-docx, задаём видимые границы
-    converter=HTML2DOCX()
-    converter.table_style="TableGrid"
-    converter.add_html_to_document(clean_html,doc)
+    # 3. Конвертируем HTML → DOCX с видимыми границами
+    parser = HtmlToDocx()
+    parser.table_style = "TableGrid"                    # TableGrid для всех таблиц :contentReference[oaicite:5]{index=5}
+    parser.add_html_to_document(clean_html, doc)
 
-    fn=f"lesson_{datetime.now():%Y%m%d%H%M%S}.docx"
-    path=os.path.join(tempfile.gettempdir(),fn)
+    # 4. Сохраняем во временный файл
+    fname = f"lesson_plan_{datetime.now():%Y%m%d%H%M%S}.docx"
+    path = os.path.join(tempfile.gettempdir(), fname)
     doc.save(path)
     return path