PDF → libreoffice → DOCX таблицы, fallback pdfplumber, v1.29

site/parser.py

@@ -123,21 +123,22 @@ def _parse_doc(file_bytes: bytes) -> dict:
 
 def _parse_pdf(file_bytes: bytes) -> dict:
     result = {"elements": [], "errors": []}
-    import tempfile, os as _os
 
-    # Сохранить PDF во временный файл (camelot требует путь)
+    # Попробовать libreoffice → docx (точные таблицы)
-    tmp_path = None
     try:
-        with tempfile.NamedTemporaryFile(suffix=".pdf", delete=False) as tmp:
+        docx_result = _pdf_via_libreoffice(file_bytes)
-            tmp.write(file_bytes)
+        if docx_result["elements"]:
-            tmp_path = tmp.name
+            return docx_result
+        result["errors"].extend(docx_result.get("errors", []))
+    except Exception as e:
+        result["errors"].append(f"libreoffice pdf: {e}")
 
+    # Fallback: pdfplumber
+    try:
         import pdfplumber
-
         with pdfplumber.open(io.BytesIO(file_bytes)) as pdf:
             for page in pdf.pages:
                 pn = page.page_number
-                # Текст страницы
                 text = page.extract_text()
                 if text:
                     for line in text.split("\n"):
@@ -149,35 +150,6 @@ def _parse_pdf(file_bytes: bytes) -> dict:
                                 "text": line,
                                 "page": pn,
                             })
-
-        # Таблицы — camelot (точнее), иначе pdfplumber
-        tables_found = False
-        try:
-            import camelot
-            # Сначала lattice (таблицы с рамками)
-            for flavor in ("lattice", "stream"):
-                try:
-                    ctables = camelot.read_pdf(tmp_path, pages="all", flavor=flavor)
-                    for ct in ctables:
-                        rows = [list(ct.df.columns)] + ct.df.values.tolist()
-                        result["elements"].append({
-                            "type": "table",
-                            "rows": rows,
-                            "page": ct.page,
-                        })
-                        tables_found = True
-                except Exception:
-                    pass
-        except ImportError:
-            result["errors"].append("camelot not installed, using pdfplumber")
-        except Exception as e:
-            result["errors"].append(f"camelot error: {e}")
-
-        # Fallback: pdfplumber таблицы если camelot ничего не нашёл
-        if not tables_found:
-            with pdfplumber.open(io.BytesIO(file_bytes)) as pdf:
-                for page in pdf.pages:
-                    pn = page.page_number
                 tables = page.extract_tables()
                 for tbl in tables:
                     if tbl:
@@ -186,16 +158,47 @@ def _parse_pdf(file_bytes: bytes) -> dict:
                             "rows": tbl,
                             "page": pn,
                         })
-
     except Exception as e:
-        result["errors"].append(f"pdf parse error: {e}")
+        result["errors"].append(f"pdfplumber: {e}")
-    finally:
-        if tmp_path and _os.path.exists(tmp_path):
-            _os.unlink(tmp_path)
 
     return result
 
 
+def _pdf_via_libreoffice(file_bytes: bytes) -> dict:
+    """Конвертировать PDF → DOCX через libreoffice, затем python-docx."""
+    result = {"elements": [], "errors": []}
+    tmp_pdf = None
+    tmp_dir = None
+    try:
+        with tempfile.NamedTemporaryFile(suffix=".pdf", delete=False) as tmp:
+            tmp.write(file_bytes)
+            tmp_pdf = tmp.name
+
+        tmp_dir = tempfile.mkdtemp()
+        subprocess.run(
+            ["libreoffice", "--headless", "--convert-to", "docx", "--outdir", tmp_dir, tmp_pdf],
+            timeout=60, capture_output=True,
+        )
+        docx_files = [f for f in os.listdir(tmp_dir) if f.endswith(".docx")]
+        if docx_files:
+            with open(os.path.join(tmp_dir, docx_files[0]), "rb") as f:
+                result = _parse_docx(f.read())
+        else:
+            result["errors"].append("libreoffice PDF→DOCX: no output")
+    except FileNotFoundError:
+        result["errors"].append("libreoffice not installed")
+    except Exception as e:
+        result["errors"].append(f"libreoffice PDF→DOCX: {e}")
+    finally:
+        if tmp_pdf and os.path.exists(tmp_pdf):
+            os.unlink(tmp_pdf)
+        if tmp_dir and os.path.exists(tmp_dir):
+            for f in os.listdir(tmp_dir):
+                os.unlink(os.path.join(tmp_dir, f))
+            os.rmdir(tmp_dir)
+    return result
+
+
 # ── ZIP ──
 
 def _parse_zip(file_bytes: bytes) -> dict:

site/templates/upload.html

@@ -59,7 +59,7 @@
 <body>
   <div class="topbar">
     <img src="{{ url_for('static', filename='nubes-logo.svg') }}" alt="Nubes">
-    <span class="title">Сверка договоров <span style="font-weight:400;color:var(--muted);font-size:12px;">v1.28</span></span>
+    <span class="title">Сверка договоров <span style="font-weight:400;color:var(--muted);font-size:12px;">v1.29</span></span>
   </div>
 
   <div class="content">