feat:增加对DOTX，DOTM，DOT，XML的去水印功能

backend/app.py

@@ -9,8 +9,13 @@ from werkzeug.utils import secure_filename
 
 from config import ALLOWED_EXTENSIONS, MAX_CONTENT_LENGTH
 from handlers.doc_handler import remove_watermark as remove_doc_watermark
+from handlers.docm_handler import remove_watermark as remove_docm_watermark
 from handlers.docx_handler import remove_watermark as remove_docx_watermark
+from handlers.dot_handler import remove_watermark as remove_dot_watermark
+from handlers.dotm_handler import remove_watermark as remove_dotm_watermark
+from handlers.dotx_handler import remove_watermark as remove_dotx_watermark
 from handlers.pdf_handler import remove_watermark as remove_pdf_watermark
+from handlers.xml_handler import remove_watermark as remove_xml_watermark
 from utils.file_utils import (
     cleanup_dir,
     is_allowed_file,
@@ -44,7 +49,7 @@ def remove_watermark():
 
         original_filename = file.filename
         if not is_allowed_file(original_filename, ALLOWED_EXTENSIONS):
-            return jsonify({"message": "仅支持 doc、docx、pdf 格式"}), 400
+            return jsonify({"message": "仅支持 doc、docx、docm、dotx、dotm、dot、xml、pdf 格式"}), 400
 
         temp_dir = make_temp_dir()
         ext = original_filename.rsplit(".", 1)[1].lower()
@@ -59,6 +64,16 @@ def remove_watermark():
             remove_docx_watermark(input_path, output_path)
         elif ext == "doc":
             remove_doc_watermark(input_path, output_path)
+        elif ext == "docm":
+            remove_docm_watermark(input_path, output_path)
+        elif ext == "dotx":
+            remove_dotx_watermark(input_path, output_path)
+        elif ext == "dotm":
+            remove_dotm_watermark(input_path, output_path)
+        elif ext == "dot":
+            remove_dot_watermark(input_path, output_path)
+        elif ext == "xml":
+            remove_xml_watermark(input_path, output_path)
         elif ext == "pdf":
             remove_pdf_watermark(input_path, output_path)
         else:

backend/config.py

@@ -4,7 +4,7 @@ from pathlib import Path
 BASE_DIR = Path(__file__).resolve().parent
 
 # 允许格式（扩展名大小写不敏感）
-ALLOWED_EXTENSIONS = {"doc", "docx", "pdf"}
+ALLOWED_EXTENSIONS = {"doc", "docx", "docm", "pdf", "dotx", "dotm", "dot", "xml"}
 
 # 上传文件大小限制（字节）
 MAX_CONTENT_LENGTH = 20 * 1024 * 1024  # 20 MB
@@ -12,6 +12,11 @@ MAX_CONTENT_LENGTH = 20 * 1024 * 1024  # 20 MB
 # LibreOffice 转换超时（秒）
 SOFFICE_TIMEOUT = 120
 
+# UNO 直改去水印（doc/dot/xml，需系统 python3 的 uno 模块）
+UNO_SCRIPT_REL = "scripts/lo_uno_remove.py"
+UNO_CONNECT_TIMEOUT = 30
+SYSTEM_PYTHON = "/usr/bin/python3"
+
 # 常见水印关键词（用于启发式识别）
 WATERMARK_KEYWORDS = {
     "机密", "秘密", "绝密", "内部", "样稿", "样张", "草稿", "草案",

backend/handlers/doc_handler.py

+import logging
 import shutil
 import subprocess
 from pathlib import Path
@@ -5,6 +6,17 @@ from typing import List
 
 from config import SOFFICE_TIMEOUT
 from handlers.docx_handler import remove_watermark as remove_docx_watermark
+from utils.lo_uno import (
+    UnoProcessError,
+    UnoTimeoutError,
+    UnoUnavailableError,
+    uno_remove_watermark,
+)
+
+logger = logging.getLogger(__name__)
+
+# UNO 直改另存 DOC 所用的 FilterName
+FILTER_NAME = "MS Word 97"
 
 
 def _run_soffice(args: List[str], timeout: int = SOFFICE_TIMEOUT) -> None:
@@ -27,7 +39,7 @@ def _run_soffice(args: List[str], timeout: int = SOFFICE_TIMEOUT) -> None:
         raise RuntimeError(f"LibreOffice 转换失败: {err}") from exc
 
 
-def remove_watermark(input_path: Path, output_path: Path) -> None:
+def _remove_watermark_via_conversion(input_path: Path, output_path: Path) -> None:
     """DOC 文件：先转 DOCX，去水印，再转回 DOC。"""
     temp_dir = input_path.parent
     docx_path = temp_dir / (input_path.stem + ".docx")
@@ -59,3 +71,20 @@ def remove_watermark(input_path: Path, output_path: Path) -> None:
         raise RuntimeError("DOCX 转换回 DOC 失败")
 
     shutil.copy2(generated_doc, output_path)
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """优先 UNO 直改（不经过 docx 转换），失败回退两步转换方案。"""
+    try:
+        removed = uno_remove_watermark(input_path, output_path, FILTER_NAME)
+        logger.info("UNO 直改去水印完成: removed=%d (%s)", removed, input_path.name)
+        return
+    except UnoUnavailableError as exc:
+        logger.warning("UNO 不可用，回退转换方案: %s", exc)
+    except UnoTimeoutError as exc:
+        logger.warning("UNO 处理超时，回退转换方案: %s", exc)
+    except UnoProcessError as exc:
+        logger.warning("UNO 处理失败，回退转换方案: %s", exc)
+    except Exception:
+        logger.warning("UNO 直改发生未知异常，回退转换方案", exc_info=True)
+    _remove_watermark_via_conversion(input_path, output_path)

backend/handlers/docm_handler.py

+from pathlib import Path
+
+from handlers.docx_handler import remove_watermark as remove_docx_watermark
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """去除 DOCM（启用宏的 Word 文档）文件中的文字和图片水印。
+
+    DOCM 与 DOCX 同为 OOXML ZIP 结构，水印处理逻辑完全一致；
+    宏文件（vbaProject.bin）不受影响，会随原样重新打包保留。
+    """
+    remove_docx_watermark(input_path, output_path)

backend/handlers/docx_handler.py

@@ -90,24 +90,50 @@ def remove_watermark(input_path: Path, output_path: Path) -> None:
 
                 # 3) 删除独立的图片水印（仅出现在 header/footer 中且大面积覆盖的通常难以自动识别，
                 #    这里删除看起来“孤立”的图片：即所在段落除图片外无其他可见文本且图片没有正文字号）
+                def _paragraph_visible_texts(paragraph) -> list:
+                    texts = paragraph.findall(".//w:t", NSMAP)
+                    # 文本框（txbxContent）内的文字不算段落可见文本，
+                    # 否则文本框水印会让段落被误判为“有内容”而跳过删除
+                    tb_texts = set(paragraph.findall(".//w:txbxContent//w:t", NSMAP))
+                    return [t for t in texts if t not in tb_texts]
+
+                def _remove_paragraph(paragraph) -> bool:
+                    parent = None
+                    for parent_candidate in root.iter():
+                        if paragraph in list(parent_candidate):
+                            parent = parent_candidate
+                            break
+                    if parent is not None:
+                        parent.remove(paragraph)
+                        return True
+                    return False
+
                 for pict in root.findall(".//w:pict", NSMAP):
-                    # 判断其所在段落是否已无文本
                     parent_p = None
                     for parent_candidate in root.iter():
                         if pict in list(parent_candidate):
                             parent_p = parent_candidate
                             break
                     if parent_p is not None:
-                        texts = parent_p.findall(".//w:t", NSMAP)
-                        if all(not (t.text or "").strip() for t in texts):
+                        if all(not (t.text or "").strip() for t in _paragraph_visible_texts(parent_p)):
                             removed_rels |= gather_image_rels_in_element(pict, NSMAP)
-                            parent_p_parent = None
-                            for ppc in root.iter():
-                                if parent_p in list(ppc):
-                                    parent_p_parent = ppc
-                                    break
-                            if parent_p_parent is not None:
-                                parent_p_parent.remove(parent_p)
+                            if _remove_paragraph(parent_p):
+                                changed = True
+
+                # 3b) 删除 DrawingML 文本框水印（LibreOffice 转换 VML 水印的产物），
+                #     即段落中没有可见文本、只包含文本框的情况
+                for drawing in root.findall(".//w:drawing", NSMAP):
+                    if not drawing.findall(".//w:txbxContent", NSMAP):
+                        continue
+                    parent_p = None
+                    for parent_candidate in root.iter():
+                        if drawing in list(parent_candidate):
+                            parent_p = parent_candidate
+                            break
+                    if parent_p is not None:
+                        if all(not (t.text or "").strip() for t in _paragraph_visible_texts(parent_p)):
+                            removed_rels |= gather_image_rels_in_element(drawing, NSMAP)
+                            if _remove_paragraph(parent_p):
                                 changed = True
 
                 if changed:

backend/handlers/dot_handler.py

+import logging
+import shutil
+from pathlib import Path
+
+from handlers.doc_handler import _run_soffice
+from handlers.docx_handler import remove_watermark as remove_docx_watermark
+from utils.lo_uno import (
+    UnoProcessError,
+    UnoTimeoutError,
+    UnoUnavailableError,
+    uno_remove_watermark,
+)
+
+logger = logging.getLogger(__name__)
+
+# UNO 直改另存 DOT 所用的 FilterName
+FILTER_NAME = "MS Word 97 Vorlage"
+
+
+def _remove_watermark_via_conversion(input_path: Path, output_path: Path) -> None:
+    """DOT 文件：先转 DOCX，去水印，再转回 DOT。"""
+    temp_dir = input_path.parent
+    docx_path = temp_dir / (input_path.stem + ".docx")
+    cleaned_docx = temp_dir / (input_path.stem + "_cleaned.docx")
+
+    # 1) DOT -> DOCX
+    _run_soffice([
+        "--convert-to", "docx",
+        "--outdir", str(temp_dir),
+        str(input_path),
+    ])
+
+    if not docx_path.exists():
+        raise RuntimeError("DOT 转换为 DOCX 失败")
+
+    # 2) 去水印
+    remove_docx_watermark(docx_path, cleaned_docx)
+
+    # 3) DOCX -> DOT
+    _run_soffice([
+        "--convert-to", "dot",
+        "--outdir", str(temp_dir),
+        str(cleaned_docx),
+    ])
+
+    generated_dot = temp_dir / (cleaned_docx.stem + ".dot")
+    if not generated_dot.exists():
+        raise RuntimeError("DOCX 转换回 DOT 失败")
+
+    shutil.copy2(generated_dot, output_path)
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """优先 UNO 直改（不经过 docx 转换），失败回退两步转换方案。"""
+    try:
+        removed = uno_remove_watermark(input_path, output_path, FILTER_NAME)
+        logger.info("UNO 直改去水印完成: removed=%d (%s)", removed, input_path.name)
+        return
+    except UnoUnavailableError as exc:
+        logger.warning("UNO 不可用，回退转换方案: %s", exc)
+    except UnoTimeoutError as exc:
+        logger.warning("UNO 处理超时，回退转换方案: %s", exc)
+    except UnoProcessError as exc:
+        logger.warning("UNO 处理失败，回退转换方案: %s", exc)
+    except Exception:
+        logger.warning("UNO 直改发生未知异常，回退转换方案", exc_info=True)
+    _remove_watermark_via_conversion(input_path, output_path)

backend/handlers/dotm_handler.py

+from pathlib import Path
+
+from handlers.docx_handler import remove_watermark as remove_docx_watermark
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """去除 DOTM（启用宏的 Word 模板）文件中的文字和图片水印。
+
+    DOTM 与 DOCX 同为 OOXML ZIP 结构，水印处理逻辑完全一致；
+    宏文件（vbaProject.bin）不受影响，会随原样重新打包保留。
+    """
+    remove_docx_watermark(input_path, output_path)

backend/handlers/dotx_handler.py

+from pathlib import Path
+
+from handlers.docx_handler import remove_watermark as remove_docx_watermark
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """去除 DOTX（Word 模板）文件中的文字和图片水印。
+
+    DOTX 与 DOCX 同为 OOXML ZIP 结构，水印处理逻辑完全一致。
+    """
+    remove_docx_watermark(input_path, output_path)

backend/handlers/xml_handler.py

+import logging
+import shutil
+from pathlib import Path
+
+from handlers.doc_handler import _run_soffice
+from handlers.docx_handler import remove_watermark as remove_docx_watermark
+from utils.lo_uno import (
+    UnoProcessError,
+    UnoTimeoutError,
+    UnoUnavailableError,
+    uno_remove_watermark,
+)
+
+logger = logging.getLogger(__name__)
+
+# LibreOffice 输出 Word 2003 XML 所用的过滤器（--convert-to 需要带 xml: 前缀）
+WORD2003_XML_FILTER = "xml:MS Word 2003 XML"
+
+# UNO 直改另存 Word 2003 XML 所用的 FilterName（storeToURL 不带前缀）
+FILTER_NAME = "MS Word 2003 XML"
+
+
+def _remove_watermark_via_conversion(input_path: Path, output_path: Path) -> None:
+    """Word 2003 XML 文件：先转 DOCX，去水印，再转回 XML。"""
+    temp_dir = input_path.parent
+    docx_path = temp_dir / (input_path.stem + ".docx")
+    cleaned_docx = temp_dir / (input_path.stem + "_cleaned.docx")
+
+    # 1) XML -> DOCX
+    _run_soffice([
+        "--convert-to", "docx",
+        "--outdir", str(temp_dir),
+        str(input_path),
+    ])
+
+    if not docx_path.exists():
+        raise RuntimeError("XML 转换为 DOCX 失败")
+
+    # 2) 去水印
+    remove_docx_watermark(docx_path, cleaned_docx)
+
+    # 3) DOCX -> Word 2003 XML
+    _run_soffice([
+        "--convert-to", WORD2003_XML_FILTER,
+        "--outdir", str(temp_dir),
+        str(cleaned_docx),
+    ])
+
+    generated_xml = temp_dir / (cleaned_docx.stem + ".xml")
+    if not generated_xml.exists():
+        raise RuntimeError("DOCX 转换回 XML 失败")
+
+    shutil.copy2(generated_xml, output_path)
+
+
+def remove_watermark(input_path: Path, output_path: Path) -> None:
+    """优先 UNO 直改（不经过 docx 转换），失败回退两步转换方案。"""
+    try:
+        removed = uno_remove_watermark(input_path, output_path, FILTER_NAME)
+        logger.info("UNO 直改去水印完成: removed=%d (%s)", removed, input_path.name)
+        return
+    except UnoUnavailableError as exc:
+        logger.warning("UNO 不可用，回退转换方案: %s", exc)
+    except UnoTimeoutError as exc:
+        logger.warning("UNO 处理超时，回退转换方案: %s", exc)
+    except UnoProcessError as exc:
+        logger.warning("UNO 处理失败，回退转换方案: %s", exc)
+    except Exception:
+        logger.warning("UNO 直改发生未知异常，回退转换方案", exc_info=True)
+    _remove_watermark_via_conversion(input_path, output_path)

backend/scripts/lo_uno_remove.py

+#!/usr/bin/env python3
+"""LibreOffice UNO 直改去水印脚本（须由系统 python3 执行，venv 无 uno）。
+
+用 headless soffice listener 直接加载原格式文档（doc/dot/xml），删除 DrawPage 上的
+水印 shape，再以原格式另存，避免"转 docx 再转回"的两次格式转换。
+
+水印识别规则（先收集后删除）：
+  1. shape Name 以 "PowerPlusWaterMarkObject" 开头（Word 文字水印的固定命名）→ 命中
+  2. CustomShape 的 CustomShapeGeometry -> "TextPath" -> "Text" 提取文字，
+     与 --keywords 传入的关键词子串匹配（大小写不敏感）→ 命中
+  3. TextEmbeddedObject（OLE 对象）一律跳过
+
+限制：doc 中的图片水印不做识别（命名模式未知，易误删页眉图片），此类文件输出等价原文件。
+
+退出码：
+  0  成功（stdout 末行 RESULT: removed=N）
+  2  uno 模块或 soffice 命令缺失
+  3  listener 启动/连接失败
+  4  文档加载失败
+  5  处理/保存失败
+"""
+
+import argparse
+import os
+import shutil
+import socket
+import subprocess
+import sys
+import tempfile
+import time
+from pathlib import Path
+
+EXIT_OK = 0
+EXIT_MISSING = 2
+EXIT_LISTENER = 3
+EXIT_LOAD = 4
+EXIT_PROCESS = 5
+
+CONNECT_RETRY_INTERVAL = 0.3
+
+try:
+    import uno
+    from com.sun.star.beans import PropertyValue
+    from com.sun.star.connection import NoConnectException
+
+    UNO_OK = True
+except ImportError:
+    UNO_OK = False
+
+
+def pick_free_port() -> int:
+    with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as sock:
+        sock.bind(("", 0))
+        return sock.getsockname()[1]
+
+
+def start_listener(port: int, profile_dir: Path) -> subprocess.Popen:
+    cmd = [
+        "soffice",
+        "--headless", "--invisible", "--nologo", "--norestore", "--nolockcheck",
+        f"--accept=socket,host=127.0.0.1,port={port};urp;",
+        f"-env:UserInstallation=file://{profile_dir}",
+    ]
+    return subprocess.Popen(
+        cmd,
+        stdout=subprocess.DEVNULL,
+        stderr=subprocess.DEVNULL,
+        start_new_session=True,
+    )
+
+
+def connect(port: int, deadline: float):
+    local_ctx = uno.getComponentContext()
+    resolver = local_ctx.ServiceManager.createInstanceWithContext(
+        "com.sun.star.bridge.UnoUrlResolver", local_ctx
+    )
+    url = f"uno:socket,host=127.0.0.1,port={port};urp;StarOffice.ComponentContext"
+    while time.monotonic() < deadline:
+        try:
+            return resolver.resolve(url)
+        except NoConnectException:
+            time.sleep(CONNECT_RETRY_INTERVAL)
+    raise NoConnectException(f"连接 127.0.0.1:{port} 超时", None)
+
+
+def make_prop(name: str, value) -> "PropertyValue":
+    prop = PropertyValue()
+    prop.Name = name
+    prop.Value = value
+    return prop
+
+
+def get_custom_shape_type(shape) -> str:
+    try:
+        geom = shape.getPropertyValue("CustomShapeGeometry")
+    except Exception:
+        return ""
+    for pv in geom:
+        if pv.Name == "Type" and isinstance(pv.Value, str):
+            return pv.Value
+    return ""
+
+
+def extract_text_path_text(shape) -> str:
+    """从 CustomShapeGeometry 的 TextPath 中提取文字。"""
+    try:
+        geom = shape.getPropertyValue("CustomShapeGeometry")
+    except Exception:
+        return ""
+    text_path = None
+    for pv in geom:
+        if pv.Name == "TextPath":
+            text_path = pv.Value
+            break
+    if not text_path:
+        return ""
+    for pv in text_path:
+        if pv.Name == "Text" and isinstance(pv.Value, str):
+            return pv.Value
+    return ""
+
+
+def is_watermark_shape(shape, keywords) -> bool:
+    try:
+        if shape.supportsService("com.sun.star.text.TextEmbeddedObject"):
+            return False
+    except Exception:
+        pass
+
+    name = ""
+    try:
+        name = shape.getPropertyValue("Name") or ""
+    except Exception:
+        pass
+    if name.lower().startswith("powerpluswatermarkobject"):
+        return True
+
+    try:
+        if shape.supportsService("com.sun.star.drawing.CustomShape"):
+            # fontwork 系（Word 文字水印即 fontwork-plain-text）：文字可能在
+            # TextPath geometry 中，也可能在 shape 正文（Word 2003 XML 导入如此）
+            if get_custom_shape_type(shape).startswith("fontwork"):
+                parts = [extract_text_path_text(shape)]
+                try:
+                    parts.append(shape.getString() or "")
+                except Exception:
+                    pass
+                text = " ".join(parts).lower()
+                if text and any(k in text for k in keywords):
+                    return True
+    except Exception:
+        pass
+    return False
+
+
+def stop_listener(proc: subprocess.Popen, kill_after: float = 3.0) -> None:
+    if proc.poll() is not None:
+        return
+    proc.terminate()
+    try:
+        proc.wait(timeout=kill_after)
+    except subprocess.TimeoutExpired:
+        proc.kill()
+        try:
+            proc.wait(timeout=5)
+        except subprocess.TimeoutExpired:
+            pass
+
+
+def remove_lock_files(*paths: Path) -> None:
+    for p in paths:
+        lock = p.parent / (".~lock." + p.name + "#")
+        try:
+            lock.unlink()
+        except OSError:
+            pass
+
+
+def run(args) -> int:
+    input_path = Path(args.input).resolve()
+    output_path = Path(args.output).resolve()
+    keywords = [k.strip().lower() for k in args.keywords.split(",") if k.strip()]
+
+    profile_dir = Path(tempfile.mkdtemp(prefix=f"lo_uno_{os.getpid()}_"))
+    proc = None
+    desktop = None
+    removed = 0
+    try:
+        # 启动 listener 并连接；失败则清理后换端口重试一次
+        ctx = None
+        last_exc = None
+        for attempt in range(2):
+            if attempt > 0:
+                shutil.rmtree(profile_dir, ignore_errors=True)
+                profile_dir = Path(tempfile.mkdtemp(prefix=f"lo_uno_{os.getpid()}_"))
+            port = pick_free_port()
+            proc = start_listener(port, profile_dir)
+            try:
+                connect_timeout = min(float(args.connect_timeout), float(args.timeout))
+                ctx = connect(port, time.monotonic() + connect_timeout)
+                break
+            except Exception as exc:
+                last_exc = exc
+                stop_listener(proc)
+                proc = None
+        if ctx is None:
+            print(f"ERROR: UNO listener 启动/连接失败: {last_exc}", file=sys.stderr)
+            return EXIT_LISTENER
+
+        smgr = ctx.ServiceManager
+        desktop = smgr.createInstanceWithContext("com.sun.star.frame.Desktop", ctx)
+
+        in_url = uno.systemPathToFileUrl(str(input_path))
+        out_url = uno.systemPathToFileUrl(str(output_path))
+        try:
+            doc = desktop.loadComponentFromURL(
+                in_url, "_blank", 0,
+                (make_prop("Hidden", True), make_prop("ReadOnly", False)),
+            )
+        except Exception as exc:
+            print(f"ERROR: 文档加载失败: {exc}", file=sys.stderr)
+            return EXIT_LOAD
+        if doc is None:
+            print("ERROR: 文档加载失败（loadComponentFromURL 返回 None）", file=sys.stderr)
+            return EXIT_LOAD
+
+        try:
+            draw_page = doc.getDrawPage()
+            to_remove = []
+            for i in range(draw_page.getCount()):
+                shape = draw_page.getByIndex(i)
+                if is_watermark_shape(shape, keywords):
+                    to_remove.append(shape)
+            for shape in to_remove:
+                draw_page.remove(shape)
+                removed += 1
+            doc.storeToURL(
+                out_url,
+                (make_prop("FilterName", args.filter), make_prop("Overwrite", True)),
+            )
+        except Exception as exc:
+            print(f"ERROR: 处理/保存失败: {exc}", file=sys.stderr)
+            return EXIT_PROCESS
+        finally:
+            try:
+                doc.close(False)
+            except Exception:
+                pass
+    finally:
+        if desktop is not None:
+            try:
+                desktop.terminate()
+            except Exception:
+                pass
+        if proc is not None:
+            stop_listener(proc)
+        shutil.rmtree(profile_dir, ignore_errors=True)
+        remove_lock_files(input_path, output_path)
+
+    print(f"RESULT: removed={removed}")
+    return EXIT_OK
+
+
+def main(argv=None) -> int:
+    parser = argparse.ArgumentParser(description="LibreOffice UNO 直改去水印")
+    parser.add_argument("--input", required=True, help="输入文件路径")
+    parser.add_argument("--output", required=True, help="输出文件路径")
+    parser.add_argument("--filter", required=True, help="另存 FilterName，如 MS Word 97")
+    parser.add_argument("--keywords", default="", help="水印关键词，逗号分隔")
+    parser.add_argument("--timeout", type=float, default=120, help="处理超时（秒）")
+    parser.add_argument("--connect-timeout", type=float, default=30, help="listener 连接超时（秒）")
+    args = parser.parse_args(argv)
+
+    if not UNO_OK:
+        print("ERROR: python3-uno 模块不可用", file=sys.stderr)
+        return EXIT_MISSING
+    if shutil.which("soffice") is None:
+        print("ERROR: 未找到 soffice 命令", file=sys.stderr)
+        return EXIT_MISSING
+
+    return run(args)
+
+
+if __name__ == "__main__":
+    sys.exit(main())

backend/utils/lo_uno.py

+"""LibreOffice UNO 直改去水印的 venv 侧编排（本模块永不 import uno）。
+
+通过子进程调用系统 python3 执行 backend/scripts/lo_uno_remove.py：
+UNO 脚本自持 listener 生命周期，本模块只负责编排（预检、超时、进程组清理、结果解析）。
+"""
+
+import os
+import re
+import signal
+import subprocess
+from pathlib import Path
+
+from config import (
+    BASE_DIR,
+    SOFFICE_TIMEOUT,
+    SYSTEM_PYTHON,
+    UNO_CONNECT_TIMEOUT,
+    UNO_SCRIPT_REL,
+    WATERMARK_KEYWORDS,
+)
+
+UNO_SCRIPT = BASE_DIR / UNO_SCRIPT_REL
+
+_RESULT_RE = re.compile(r"^RESULT:\s*removed=(\d+)\s*$", re.MULTILINE)
+
+_uno_checked = False
+
+
+class UnoUnavailableError(RuntimeError):
+    """UNO 环境不可用（uno 模块/soffice 缺失，或 listener 无法启动连接）。"""
+
+
+class UnoTimeoutError(RuntimeError):
+    """UNO 处理超时。"""
+
+
+class UnoProcessError(RuntimeError):
+    """UNO 加载/处理/保存失败。"""
+
+
+def _check_uno_available() -> None:
+    global _uno_checked
+    if _uno_checked:
+        return
+    try:
+        subprocess.run(
+            [SYSTEM_PYTHON, "-c", "import uno"],
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
+            timeout=15,
+            check=True,
+        )
+    except FileNotFoundError as exc:
+        raise UnoUnavailableError(f"系统 python3 不存在: {SYSTEM_PYTHON}") from exc
+    except subprocess.TimeoutExpired as exc:
+        raise UnoUnavailableError("系统 python3 import uno 预检超时") from exc
+    except subprocess.CalledProcessError as exc:
+        raise UnoUnavailableError("系统 python3 缺少 uno 模块（未安装 python3-uno）") from exc
+    _uno_checked = True
+
+
+def _kill_process_group(proc: subprocess.Popen) -> None:
+    try:
+        os.killpg(os.getpgid(proc.pid), signal.SIGKILL)
+    except (ProcessLookupError, PermissionError):
+        try:
+            proc.kill()
+        except Exception:
+            pass
+
+
+def uno_remove_watermark(
+    input_path: Path,
+    output_path: Path,
+    filter_name: str,
+    timeout: int = SOFFICE_TIMEOUT,
+) -> int:
+    """UNO 直改去水印，返回删除的 shape 数。失败按三类异常抛出。"""
+    _check_uno_available()
+
+    cmd = [
+        SYSTEM_PYTHON,
+        str(UNO_SCRIPT),
+        "--input", str(input_path),
+        "--output", str(output_path),
+        "--filter", filter_name,
+        "--keywords", ",".join(WATERMARK_KEYWORDS),
+        "--timeout", str(timeout),
+        "--connect-timeout", str(UNO_CONNECT_TIMEOUT),
+    ]
+
+    proc = subprocess.Popen(
+        cmd,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        start_new_session=True,
+    )
+    try:
+        out, err = proc.communicate(timeout=timeout + 30)
+    except subprocess.TimeoutExpired as exc:
+        _kill_process_group(proc)
+        proc.communicate()
+        raise UnoTimeoutError(f"UNO 处理超时（{timeout + 30}s）: {input_path.name}") from exc
+
+    stdout = out.decode("utf-8", errors="ignore")
+    stderr = err.decode("utf-8", errors="ignore")
+
+    if proc.returncode == 0:
+        match = _RESULT_RE.search(stdout)
+        if not match:
+            raise UnoProcessError(f"UNO 脚本输出异常: {stdout!r}")
+        return int(match.group(1))
+
+    message = stderr.strip() or stdout.strip()
+    if proc.returncode in (2, 3):
+        raise UnoUnavailableError(f"UNO 不可用（退出码 {proc.returncode}）: {message}")
+    if proc.returncode in (4, 5):
+        raise UnoProcessError(f"UNO 处理失败（退出码 {proc.returncode}）: {message}")
+    raise UnoProcessError(f"UNO 脚本异常退出（码 {proc.returncode}）: {message}")

frontend/src/App.vue

@@ -2,7 +2,7 @@
   <div class="app">
     <header class="header">
       <h1>文档去水印工具</h1>
-      <p class="subtitle">全自动识别并删除 DOC、DOCX、PDF 文件中的文字与图片水印</p>
+      <p class="subtitle">全自动识别并删除 DOC、DOCX、DOCM、DOTX、DOTM、DOT、XML、PDF 文件中的文字与图片水印</p>
     </header>
     <main class="main">
       <FileUploader />
@@ -13,13 +13,14 @@
           <li>下载文件名将自动加上下划线前缀，例如 <code>报告.docx</code> → <code>_报告.docx</code>。</li>
           <li>服务器不会保留您的原始文件与结果文件，处理完成后立即删除。</li>
           <li>对于复杂 PDF 水印或自定义图片水印，去除效果可能有限。</li>
-          <li>DOC 文件会经过 LibreOffice 转换为 DOCX 处理后再转回 DOC，排版可能存在细微变化。</li>
+          <li>DOC 文件通过 LibreOffice UNO 直接识别并删除水印，保持原格式保存，不经过格式转换；特殊情况下会自动回退到转换方案，排版可能存在细微变化。</li>
+          <li>DOCM 文件直接按 DOCX 结构处理，文档中的宏会被完整保留。</li>
+          <li>DOTX / DOTM 模板文件与 DOCX 结构相同，处理逻辑一致（DOTM 的宏同样保留）。</li>
+          <li>DOT 与 XML 文件与 DOC 相同，优先直接原格式去除水印，失败时回退转换方案。</li>
+          <li>DOC / DOT / XML 中的图片水印暂不支持识别，仅处理文字水印。</li>
         </ul>
       </div>
     </main>
-    <footer class="footer">
-      <p>本地运行 · 数据不留存</p>
-    </footer>
   </div>
 </template>
 

frontend/src/components/FileUploader.vue

@@ -12,13 +12,13 @@
       <input
         ref="fileInput"
         type="file"
-        accept=".doc,.docx,.pdf"
+        accept=".doc,.docx,.docm,.dotx,.dotm,.dot,.xml,.pdf"
         class="hidden-input"
         @change="handleFileChange"
       />
       <p v-if="!selectedFile">点击或拖拽上传文件</p>
       <p v-else class="file-name">已选择：{{ selectedFile.name }}</p>
-      <p class="hint">支持格式：doc、docx、pdf（最大 20MB）</p>
+      <p class="hint">支持格式：doc、docx、docm、dotx、dotm、dot、xml、pdf（最大 20MB）</p>
     </div>
 
     <div class="actions">
@@ -52,7 +52,7 @@
 import { ref } from 'vue'
 import { removeWatermark } from '../api/watermark.js'
 
-const ALLOWED_EXTENSIONS = ['doc', 'docx', 'pdf']
+const ALLOWED_EXTENSIONS = ['doc', 'docx', 'docm', 'dotx', 'dotm', 'dot', 'xml', 'pdf']
 const MAX_SIZE = 20 * 1024 * 1024
 
 const fileInput = ref(null)
@@ -75,7 +75,7 @@ function setMessage(text, type = 'info') {
 function handleFile(file) {
   if (!file) return
   if (!isAllowed(file.name)) {
-    setMessage('仅支持 doc、docx、pdf 格式', 'error')
+    setMessage('仅支持 doc、docx、docm、dotx、dotm、dot、xml、pdf 格式', 'error')
     selectedFile.value = null
     return
   }