@deepwhale-cn/dsh-doc-batch 0.0.0-stage → 0.1.4

package/lib/docx_tool.py

@@ -0,0 +1,651 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+dsh-docx-batch 的真实处理逻辑（python-docx）
+
+由 bin/cli.mjs 通过 spawnSync 调用：
+    python3 lib/docx_tool.py <action> '<json payload>'
+约定：
+  - stdout 只输出一行 JSON（最后一行），人类可读文案由 Node 侧负责
+  - stderr 输出异常栈（调试用）
+  - 退出码：0 成功 / 1 运行失败 / 2 参数或输入问题 / 3 缺依赖
+  - 只读输入文件，所有产物写进 --out 目录（默认 <输入目录>/_输出）
+"""
+
+import csv
+import json
+import os
+import re
+import sys
+import tempfile
+import traceback
+from pathlib import Path
+
+try:
+    from docx import Document
+    from docx.oxml.ns import qn
+    from docx.table import Table
+    from docx.text.paragraph import Paragraph
+except ImportError as exc:  # 缺依赖必须明确提示安装命令，不能静默
+    print(json.dumps({
+        "ok": False,
+        "code": 3,
+        "message": "缺少 Python 依赖：%s" % getattr(exc, "name", exc),
+        "hint": "请先安装：\n  python3 -m pip install python-docx\n"
+                "如果 python3 也没有：\n  macOS: brew install python3\n  Ubuntu: sudo apt install python3",
+    }, ensure_ascii=False))
+    sys.exit(3)
+
+DOCX_EXT = ".docx"
+SKIP_DIRS = {"_输出", "node_modules", ".git", "__pycache__", ".venv", "venv"}
+DEMO_HEADER = "机密 · 深鲸科技"
+DEMO_FOOTER = "第 1 页"
+
+
+# ── 错误 ──────────────────────────────────────────────────────────
+
+class ToolError(Exception):
+    def __init__(self, message, hint="", code=2):
+        super().__init__(message)
+        self.message = message
+        self.hint = hint
+        self.code = code
+
+
+def emit(obj):
+    print(json.dumps(obj, ensure_ascii=False))
+
+
+# ── 路径与文件收集 ────────────────────────────────────────────────
+
+def natural_key(text):
+    return [int(t) if t.isdigit() else t.lower() for t in re.split(r"(\d+)", text)]
+
+
+def resolve_out(payload, input_path):
+    """输出目录：--out 优先，否则 <输入目录>/_输出。"""
+    out = payload.get("out")
+    if out:
+        return Path(out).expanduser().resolve()
+    base = Path(input_path).expanduser().resolve()
+    if not base.is_dir():
+        base = base.parent
+    return base / "_输出"
+
+
+def _out_skip_set(out_dir):
+    try:
+        return {str(out_dir.resolve())}
+    except OSError:
+        return set()
+
+
+def collect_docx(input_path, out_dir=None):
+    """接受单个 .docx 或目录（递归），按文件名自然排序。"""
+    if not input_path:
+        raise ToolError("缺少 --input", "例：replace --input ./合同目录 --from 旧 --to 新")
+    root = Path(input_path).expanduser().resolve()
+    if not root.exists():
+        raise ToolError(
+            "输入不存在：%s" % input_path,
+            "请检查路径是否正确；传目录会递归处理其中的 .docx 文件。",
+        )
+    skip_out = _out_skip_set(out_dir) if out_dir else set()
+    if root.is_file():
+        if root.suffix.lower() != DOCX_EXT:
+            raise ToolError(
+                "输入不是 .docx 文件：%s" % root.name,
+                "本插件只处理 Word 2007+ 的 .docx；.doc 请先用 Word/WPS 另存为 .docx。",
+            )
+        return [root]
+    files = []
+    for dirpath, dirnames, filenames in os.walk(root):
+        dirnames[:] = sorted(
+            [d for d in dirnames if d not in SKIP_DIRS and not d.startswith("~$")],
+            key=natural_key,
+        )
+        if str(Path(dirpath).resolve()) in skip_out:
+            dirnames[:] = []
+            continue
+        for name in filenames:
+            if name.startswith("~$"):
+                continue  # Word 打开时的临时文件
+            p = Path(dirpath) / name
+            if p.suffix.lower() == DOCX_EXT:
+                files.append(p)
+    files.sort(key=lambda p: natural_key(str(p.relative_to(root))))
+    if not files:
+        raise ToolError(
+            "目录里没有 .docx 文件：%s" % root,
+            "确认目录选对了；本插件不处理 .doc（请先另存为 .docx）。",
+        )
+    return files
+
+
+def unique_path(directory, filename):
+    """同名文件不覆盖，追加 _1/_2 序号。"""
+    directory = Path(directory)
+    target = directory / filename
+    if not target.exists():
+        return target
+    stem, suffix = target.stem, target.suffix
+    n = 1
+    while (directory / ("%s_%d%s" % (stem, n, suffix))).exists():
+        n += 1
+    return directory / ("%s_%d%s" % (stem, n, suffix))
+
+
+# ── docx 遍历 ─────────────────────────────────────────────────────
+
+def iter_block_items(parent):
+    """按文档顺序遍历段落和表格。"""
+    body = parent.element.body
+    for child in body.iterchildren():
+        if child.tag == qn("w:p"):
+            yield Paragraph(child, parent)
+        elif child.tag == qn("w:tbl"):
+            yield Table(child, parent)
+
+
+def iter_cell_paragraphs(cell):
+    for p in cell.paragraphs:
+        yield p
+    for table in cell.tables:
+        for row in table.rows:
+            for c in row.cells:
+                yield from iter_cell_paragraphs(c)
+
+
+def iter_table_paragraphs(table):
+    for row in table.rows:
+        for cell in row.cells:
+            yield from iter_cell_paragraphs(cell)
+
+
+def all_editable_paragraphs(doc):
+    """正文 + 表格 + 页眉页脚，去重（链接到前一节的页眉是同一个 part）。"""
+    out, seen = [], set()
+
+    def add(p):
+        key = id(p._p)
+        if key not in seen:
+            seen.add(key)
+            out.append(p)
+
+    for p in doc.paragraphs:
+        add(p)
+    for t in doc.tables:
+        for p in iter_table_paragraphs(t):
+            add(p)
+    for section in doc.sections:
+        for part in (section.header, section.footer):
+            try:
+                for p in part.paragraphs:
+                    add(p)
+                for t in part.tables:
+                    for p in iter_table_paragraphs(t):
+                        add(p)
+            except Exception:
+                # 极少数残缺文档取不到页眉/页脚定义：跳过，不影响正文
+                continue
+    return out
+
+
+# ── 文本替换 ──────────────────────────────────────────────────────
+
+def replace_in_paragraph(paragraph, pairs):
+    """先做 run 内替换（保留格式），跨 run 的再做段落级重建。返回替换次数。"""
+    count = 0
+    if not paragraph.runs:
+        return 0
+    for old, new in pairs:
+        if not old:
+            continue
+        for run in paragraph.runs:
+            if old in run.text:
+                count += run.text.count(old)
+                run.text = run.text.replace(old, new)
+        if old in paragraph.text:
+            runs = paragraph.runs
+            full = "".join(r.text for r in runs)
+            if old in full:
+                count += full.count(old)
+                runs[0].text = full.replace(old, new)
+                for r in runs[1:]:
+                    r.text = ""
+    return count
+
+
+# ── 各动作 ────────────────────────────────────────────────────────
+
+def action_replace(payload):
+    files = collect_docx(payload.get("input"))
+    out = resolve_out(payload, payload.get("input"))
+    olds = payload.get("from") or []
+    news = payload.get("to") or []
+    if not olds:
+        raise ToolError("缺少 --from（要替换的旧文本）",
+                        "例：replace --input ./合同 --from 旧公司名 --to 新公司名")
+    if len(olds) != len(news):
+        raise ToolError(
+            "--from 与 --to 数量不一致（%d vs %d）" % (len(olds), len(news)),
+            "每一组 --from 都要配一个 --to，按顺序一一对应。",
+        )
+    pairs = list(zip(olds, news))
+    out.mkdir(parents=True, exist_ok=True)
+    results, total = [], 0
+    for src in files:
+        doc = Document(str(src))
+        n = 0
+        for p in all_editable_paragraphs(doc):
+            n += replace_in_paragraph(p, pairs)
+        target = unique_path(out, src.name)
+        doc.save(str(target))
+        results.append({
+            "input": str(src),
+            "input_name": src.name,
+            "output": str(target),
+            "output_name": target.name,
+            "replacements": n,
+        })
+        total += n
+    return {
+        "action": "replace",
+        "input": str(Path(payload["input"]).expanduser().resolve()),
+        "out": str(out),
+        "files": results,
+        "replacements": total,
+        "summary": "处理 %d 个文件，共替换 %d 处" % (len(files), total),
+        "artifacts": [r["output"] for r in results],
+    }
+
+
+def doc_body_text(doc):
+    lines = []
+    for block in iter_block_items(doc):
+        if isinstance(block, Paragraph):
+            lines.append(block.text)
+        else:
+            lines.append("")
+            lines.append("[表格]")
+            for row in block.rows:
+                lines.append("\t".join(c.text.replace("\n", " ") for c in row.cells))
+    return "\n".join(lines).strip() + "\n"
+
+
+def action_extract(payload):
+    files = collect_docx(payload.get("input"))
+    out = resolve_out(payload, payload.get("input"))
+    out.mkdir(parents=True, exist_ok=True)
+    results = []
+    for src in files:
+        text = doc_body_text(Document(str(src)))
+        target = unique_path(out, src.stem + ".txt")
+        target.write_text(text, encoding="utf-8")
+        results.append({
+            "input": str(src),
+            "input_name": src.name,
+            "output": str(target),
+            "output_name": target.name,
+            "chars": len(text),
+        })
+    return {
+        "action": "extract",
+        "input": str(Path(payload["input"]).expanduser().resolve()),
+        "out": str(out),
+        "files": results,
+        "summary": "提取 %d 个文件的正文文本" % len(files),
+        "artifacts": [r["output"] for r in results],
+    }
+
+
+def copy_run(dst_paragraph, src_run):
+    run = dst_paragraph.add_run(src_run.text)
+    try:
+        run.bold = src_run.bold
+        run.italic = src_run.italic
+        run.underline = src_run.underline
+        if src_run.font.size is not None:
+            run.font.size = src_run.font.size
+        if src_run.font.name:
+            run.font.name = src_run.font.name
+    except Exception:
+        pass  # 字体属性个别无法复制时不影响文本合并
+
+
+def copy_paragraph(dst_doc, src_paragraph):
+    p = dst_doc.add_paragraph()
+    try:
+        style_name = src_paragraph.style.name if src_paragraph.style else None
+        if style_name and style_name.lower() not in ("normal", "正文"):
+            p.style = style_name
+    except Exception:
+        pass
+    try:
+        if src_paragraph.alignment is not None:
+            p.alignment = src_paragraph.alignment
+    except Exception:
+        pass
+    for run in src_paragraph.runs:
+        copy_run(p, run)
+    return p
+
+
+def copy_table(dst_doc, src_table):
+    rows = len(src_table.rows)
+    cols = max((len(r.cells) for r in src_table.rows), default=0)
+    if rows == 0 or cols == 0:
+        return None
+    table = dst_doc.add_table(rows=rows, cols=cols)
+    for i, row in enumerate(src_table.rows):
+        for j, cell in enumerate(row.cells):
+            if j < cols:
+                table.cell(i, j).text = cell.text
+    try:
+        table.style = src_table.style.name
+    except Exception:
+        pass
+    return table
+
+
+def action_merge(payload):
+    files = collect_docx(payload.get("input"))
+    out = resolve_out(payload, payload.get("input"))
+    out.mkdir(parents=True, exist_ok=True)
+    merged = Document()
+    results = []
+    for src in files:
+        doc = Document(str(src))
+        paragraphs_before = len(merged.paragraphs)
+        for block in iter_block_items(doc):
+            if isinstance(block, Paragraph):
+                copy_paragraph(merged, block)
+            else:
+                copy_table(merged, block)
+        results.append({
+            "input": str(src),
+            "input_name": src.name,
+            "paragraphs": len(merged.paragraphs) - paragraphs_before,
+            "source_paragraphs": len(doc.paragraphs),
+        })
+    target = unique_path(out, "merged.docx")
+    merged.save(str(target))
+    return {
+        "action": "merge",
+        "input": str(Path(payload["input"]).expanduser().resolve()),
+        "out": str(out),
+        "files": results,
+        "output": str(target),
+        "output_name": target.name,
+        "paragraphs": len(merged.paragraphs),
+        "summary": "按文件名顺序合并 %d 个文件，共 %d 段" % (len(files), len(merged.paragraphs)),
+        "artifacts": [str(target)],
+    }
+
+
+def section_texts(doc):
+    sections = []
+    for idx, section in enumerate(doc.sections, 1):
+        item = {"section": idx, "header": [], "footer": []}
+        for label, part in (("header", section.header), ("footer", section.footer)):
+            try:
+                texts = [p.text.strip() for p in part.paragraphs if p.text.strip()]
+            except Exception:
+                texts = []
+            item[label] = texts
+        sections.append(item)
+    return sections
+
+
+def action_headers(payload):
+    files = collect_docx(payload.get("input"))
+    out = resolve_out(payload, payload.get("input"))
+    out.mkdir(parents=True, exist_ok=True)
+    results, lines = [], []
+    for src in files:
+        sections = section_texts(Document(str(src)))
+        hit = sum(1 for s in sections if s["header"] or s["footer"])
+        results.append({
+            "input": str(src),
+            "input_name": src.name,
+            "sections": len(sections),
+            "sections_with_text": hit,
+            "detail": sections,
+        })
+        lines.append("文件：%s（%d 节）" % (src.name, len(sections)))
+        for s in sections:
+            h = " / ".join(s["header"]) if s["header"] else "（空）"
+            f = " / ".join(s["footer"]) if s["footer"] else "（空）"
+            lines.append("  第 %d 节  页眉：%s" % (s["section"], h))
+            lines.append("          页脚：%s" % f)
+        lines.append("")
+    target = unique_path(out, "headers.txt")
+    target.write_text("\n".join(lines).strip() + "\n", encoding="utf-8")
+    return {
+        "action": "headers",
+        "input": str(Path(payload["input"]).expanduser().resolve()),
+        "out": str(out),
+        "files": results,
+        "output": str(target),
+        "output_name": target.name,
+        "summary": "列出 %d 个文件的页眉页脚" % len(files),
+        "artifacts": [str(target)],
+    }
+
+
+def inline_runs(paragraph):
+    parts = []
+    for run in paragraph.runs:
+        text = run.text
+        if not text:
+            continue
+        if run.bold and run.italic:
+            text = "***%s***" % text
+        elif run.bold:
+            text = "**%s**" % text
+        elif run.italic:
+            text = "*%s*" % text
+        parts.append(text)
+    text = "".join(parts).strip()
+    return text or paragraph.text.strip()
+
+
+def table_to_md(table):
+    rows = []
+    for row in table.rows:
+        cells = [" ".join(c.text.split()) for c in row.cells]
+        if any(cells):
+            rows.append(cells)
+    if not rows:
+        return ""
+    width = max(len(r) for r in rows)
+    rows = [r + [""] * (width - len(r)) for r in rows]
+
+    def esc(s):
+        return s.replace("|", "\\|")
+
+    out = ["| " + " | ".join(esc(c) for c in rows[0]) + " |",
+           "| " + " | ".join(["---"] * width) + " |"]
+    for r in rows[1:]:
+        out.append("| " + " | ".join(esc(c) for c in r) + " |")
+    return "\n".join(out)
+
+
+def paragraph_to_md(paragraph, state):
+    style = (paragraph.style.name if paragraph.style else "") or ""
+    style_low = style.lower()
+    text = inline_runs(paragraph)
+    m = re.match(r"^(heading|标题)\s*([1-6])$", style, re.I)
+    if m:
+        return "#" * int(m.group(2)) + " " + text
+    if style_low.startswith("title"):
+        return "# " + text
+    if "list bullet" in style_low or ("列表" in style and "项目" in style):
+        return "- " + text
+    if "list number" in style_low:
+        state["number"] += 1
+        return "%d. %s" % (state["number"], text)
+    if not text:
+        return ""
+    return text
+
+
+def doc_to_md(doc):
+    blocks = []
+    state = {"number": 0}
+    for block in iter_block_items(doc):
+        if isinstance(block, Paragraph):
+            md = paragraph_to_md(block, state)
+        else:
+            md = table_to_md(block)
+        if md.strip():
+            blocks.append(md)
+    return "\n\n".join(blocks).strip() + "\n"
+
+
+def action_to_md(payload):
+    files = collect_docx(payload.get("input"))
+    out = resolve_out(payload, payload.get("input"))
+    out.mkdir(parents=True, exist_ok=True)
+    results = []
+    for src in files:
+        md = doc_to_md(Document(str(src)))
+        target = unique_path(out, src.stem + ".md")
+        target.write_text(md, encoding="utf-8")
+        results.append({
+            "input": str(src),
+            "input_name": src.name,
+            "output": str(target),
+            "output_name": target.name,
+            "chars": len(md),
+        })
+    return {
+        "action": "to-md",
+        "input": str(Path(payload["input"]).expanduser().resolve()),
+        "out": str(out),
+        "files": results,
+        "summary": "转换 %d 个文件为 Markdown" % len(files),
+        "artifacts": [r["output"] for r in results],
+    }
+
+
+# ── demo ─────────────────────────────────────────────────────────
+
+def _make_demo_docx(path, heading, lines, table_rows=None, header=True):
+    doc = Document()
+    doc.add_heading(heading, level=1)
+    for line in lines:
+        doc.add_paragraph(line)
+    if table_rows:
+        table = doc.add_table(rows=len(table_rows), cols=len(table_rows[0]))
+        for i, row in enumerate(table_rows):
+            for j, value in enumerate(row):
+                table.cell(i, j).text = value
+    if header:
+        doc.sections[0].header.paragraphs[0].text = DEMO_HEADER
+        doc.sections[0].footer.paragraphs[0].text = DEMO_FOOTER
+    doc.save(str(path))
+
+
+def action_demo(payload):
+    root = Path(tempfile.mkdtemp(prefix="dsh-docx-batch-demo-"))
+    inputs = root / "样例"
+    inputs.mkdir(parents=True, exist_ok=True)
+    _make_demo_docx(
+        inputs / "合同A.docx",
+        "第一章 总则",
+        ["甲方：深鲸科技", "乙方：示例客户", "本合同由深鲸科技起草。"],
+        [["条款", "内容"], ["保密", "深鲸科技对数据保密"]],
+    )
+    _make_demo_docx(
+        inputs / "附件B.docx",
+        "附件 B",
+        ["附件内容第一段", "附件内容第二段"],
+    )
+    out = root / "输出"
+    steps, artifacts = [], []
+
+    def run(name, fn, extra):
+        p = {"input": str(inputs), "out": str(out / name)}
+        p.update(extra)
+        result = fn(p)
+        steps.append({
+            "step": name,
+            "action": result.get("action"),
+            "summary": result.get("summary", ""),
+            "out": result.get("out"),
+        })
+        artifacts.extend(result.get("artifacts", []))
+
+    run("替换", action_replace, {"from": ["深鲸科技"], "to": ["深海鲸"]})
+    replaced_dir = out / "替换"
+    # 提取/转 Markdown 用替换后的产物，顺带演示链路
+    p_extract = {"input": str(replaced_dir), "out": str(out / "文本")}
+    result = action_extract(p_extract)
+    steps.append({"step": "文本", "action": "extract", "summary": result["summary"], "out": result["out"]})
+    artifacts.extend(result["artifacts"])
+    p_md = {"input": str(replaced_dir), "out": str(out / "markdown")}
+    result = action_to_md(p_md)
+    steps.append({"step": "markdown", "action": "to-md", "summary": result["summary"], "out": result["out"]})
+    artifacts.extend(result["artifacts"])
+    run("合并", action_merge, {})
+    run("页眉页脚", action_headers, {})
+
+    return {
+        "action": "demo",
+        "demoRoot": str(root),
+        "steps": steps,
+        "summary": "演示完成：5 个动作全部跑通，产物 %d 个" % len(artifacts),
+        "artifacts": artifacts,
+    }
+
+
+ACTIONS = {
+    "replace": action_replace,
+    "extract": action_extract,
+    "merge": action_merge,
+    "headers": action_headers,
+    "to-md": action_to_md,
+    "demo": action_demo,
+}
+
+
+def main(argv):
+    if len(argv) < 3:
+        emit({"ok": False, "code": 2,
+              "message": "内部调用参数不足：需要 <action> 和 JSON payload",
+              "hint": "请通过 bin/cli.mjs 调用本脚本。"})
+        return 2
+    action, raw = argv[1], argv[2]
+    try:
+        payload = json.loads(raw) if raw else {}
+    except json.JSONDecodeError as exc:
+        emit({"ok": False, "code": 2, "message": "参数 JSON 解析失败：%s" % exc, "hint": ""})
+        return 2
+    fn = ACTIONS.get(action)
+    if not fn:
+        emit({"ok": False, "code": 2, "message": "未知动作：%s" % action,
+              "hint": "可用动作：%s" % " / ".join(k for k in ACTIONS if k != "demo")})
+        return 2
+    try:
+        result = fn(payload)
+        result.setdefault("ok", True)
+        emit(result)
+        return 0
+    except ToolError as exc:
+        emit({"ok": False, "code": exc.code, "message": exc.message, "hint": exc.hint})
+        return exc.code
+    except PermissionError as exc:
+        emit({"ok": False, "code": 1,
+              "message": "没有权限读写文件：%s" % exc,
+              "hint": "检查目标文件是否被 Word 占用、或输出目录是否有写权限。"})
+        return 1
+    except Exception as exc:  # noqa: BLE001 —— 兜底，保证永远输出 JSON
+        traceback.print_exc(file=sys.stderr)
+        emit({"ok": False, "code": 1, "message": "处理失败：%s" % exc,
+              "hint": "文件可能已损坏或被加密；可先用 Word 打开另存一次再试。"})
+        return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))

package/lib/index.js

@@ -0,0 +1,26 @@
+import { createPlugin, pkgRootFrom } from '@deepwhale-cn/dsh-guard-core/plugin';
+import { registerDocBatchApi } from './api.js';
+
+const base = createPlugin({
+  name: 'doc-batch',
+  pkgRoot: pkgRootFrom(import.meta.url),
+  skillFile: 'skill/doc-batch.md',
+  binFile: 'bin/cli.mjs',
+  description: '深鲸文档批量处理：Word / Excel / PDF 一次处理整目录（替换、提取、合并、拆分、压缩、水印、导出）',
+  whenToUse: '当用户要批量处理整目录的文档时使用：Word（替换/提取/合并/页眉页脚/转 Markdown）、Excel（合并/清洗/汇总/导出 CSV）、PDF（合并/拆分/压缩/水印/提取文本或图片）。典型说法：「把这批合同里的旧公司名全换掉」「这些表合并成一张」「把这些 PDF 按页拆开」。',
+});
+
+export const name = base.name;
+export const inject = [...base.inject, 'webServer'];
+
+export function apply(ctx) {
+  base.apply(ctx);
+  try {
+    if (ctx.webServer && typeof ctx.webServer.register === 'function') {
+      registerDocBatchApi(ctx);
+      ctx.logger?.info?.('[doc-batch] 已挂面板出口 /dsh-doc-batch/api');
+    }
+  } catch (e) {
+    ctx.logger?.warn?.('[doc-batch] 面板出口注册失败（CLI/skill 不受影响）: ' + (e?.message || e));
+  }
+}