tabula-py вместо camelot и libreoffice, v1.30

2026-06-17 21:29:29 +04:00
parent aae6ea2a88
commit e613716178
3 changed files with 26 additions and 52 deletions
@@ -5,6 +5,6 @@ requests
 psycopg2-binary
 python-dotenv
 pdfplumber
-camelot-py[cv]
+tabula-py
 httpx
 h2
@@ -124,16 +124,24 @@ def _parse_doc(file_bytes: bytes) -> dict:
 def _parse_pdf(file_bytes: bytes) -> dict:
    result = {"elements": [], "errors": []}

-    # Попробовать libreoffice → docx (точные таблицы)
+    # Попробовать tabula (Java) — точнее таблицы
+    tables_from_tabula = False
    try:
-        docx_result = _pdf_via_libreoffice(file_bytes)
-        if docx_result["elements"]:
-            return docx_result
-        result["errors"].extend(docx_result.get("errors", []))
+        import tabula
+        tables = tabula.read_pdf(io.BytesIO(file_bytes), pages="all", multiple_tables=True)
+        for tbl in tables:
+            rows = [list(tbl.columns)] + tbl.values.tolist()
+            result["elements"].append({
+                "type": "table",
+                "rows": rows,
+            })
+        tables_from_tabula = len(tables) > 0
+    except ImportError:
+        result["errors"].append("tabula not installed, using pdfplumber")
    except Exception as e:
-        result["errors"].append(f"libreoffice pdf: {e}")
+        result["errors"].append(f"tabula error: {e}")

-    # Fallback: pdfplumber
+    # pdfplumber: текст + таблицы (если tabula не сработал)
    try:
        import pdfplumber
        with pdfplumber.open(io.BytesIO(file_bytes)) as pdf:
@@ -150,55 +158,21 @@ def _parse_pdf(file_bytes: bytes) -> dict:
                                "text": line,
                                "page": pn,
                            })
-                tables = page.extract_tables()
-                for tbl in tables:
-                    if tbl:
-                        result["elements"].append({
-                            "type": "table",
-                            "rows": tbl,
-                            "page": pn,
-                        })
+                if not tables_from_tabula:
+                    tables = page.extract_tables()
+                    for tbl in tables:
+                        if tbl:
+                            result["elements"].append({
+                                "type": "table",
+                                "rows": tbl,
+                                "page": pn,
+                            })
    except Exception as e:
        result["errors"].append(f"pdfplumber: {e}")

    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:
@@ -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;">v1.29</span></span>
+    <span class="title">Сверка договоров <span style="font-weight:400;color:var(--muted);font-size:12px;">v1.30</span></span>
  </div>

  <div class="content">