feat(transit): 添加正文引用标记到书签超链接功能

- 新增 CITE_PATTERN 正则表达式匹配 [N] 引用格式
- 添加 base_dir 参数支持相对图片路径解析
- 实现书签创建和超链接替换功能
- 添加 link_body_citations 函数处理正文引用链接
- 在参考文献段落中添加书签标识
- 支持将 [N] 引用替换为指向参考文献的超链接

README.md

@@ -1,4 +1,5 @@
 使用方法
 ```shell
-py .\test.py .\毕业论文初稿.md
+# 需要 sample.docx 文件 且 该文件有 {{xxx}} 模板引擎的内容
+python .\test.py .\毕业论文初稿.md
 ```

transit/body.py

@@ -7,17 +7,25 @@ Markdown 正文 → Word 段落转换。
 
 import re
 from copy import deepcopy
+from pathlib import Path
+
 from docx import Document
 from docx.oxml.ns import qn
 
+from .images import make_image_paragraph, is_figure_caption, insert_image_paragraphs
+
 _PAT_HEADING = re.compile(r"^(#{1,6})\s+(.+)$", re.MULTILINE)
 
+# 匹配正文中的引用标记 [1] / [1,2,3]
+_CITE_PATTERN = re.compile(r"\[(\d+(?:[,，\s]*\d+)*)\]")
+
 
 def body_to_paragraphs(
     md_text: str,
     *,
     level_offset: int = 0,
     body_style: str = "Body Text Indent",
+    base_dir: str | Path | None = None,
 ) -> list[dict]:
     """将 Markdown 正文按标题和段落拆分为结构化列表。
 
@@ -29,21 +37,34 @@ def body_to_paragraphs(
         标题级别偏移量（正文从 ``##`` 开始时传 ``-1``，使其输出为 ``Heading 1``）。
     body_style : str
         正文段落的 Word 样式名。
+    base_dir : str | Path | None
+        Markdown 文件所在目录，用于解析图片相对路径。
     """
     paragraphs: list[dict] = []
     last_end = 0
 
+    def _add_block(block: str) -> None:
+        block = block.strip()
+        if not block:
+            return
+        # 图片段落
+        img = make_image_paragraph(block, base_dir)
+        if img:
+            paragraphs.append(img)
+            return
+        # 跳过紧跟在图片后的重复图标题
+        if paragraphs and paragraphs[-1].get("type") == "image" and is_figure_caption(block):
+            return
+        # 普通正文段落
+        paragraphs.append({"text": block, "level": 0, "style": body_style})
+
     for m in _PAT_HEADING.finditer(md_text):
         # 标题前的普通文本
         if m.start() > last_end:
             pre = md_text[last_end : m.start()].strip()
             if pre:
                 for block in re.split(r"\n\s*\n", pre):
-                    block = block.strip()
-                    if block:
-                        paragraphs.append(
-                            {"text": block, "level": 0, "style": body_style}
-                        )
+                    _add_block(block)
 
         level = len(m.group(1)) + level_offset
         heading_text = m.group(2).strip()
@@ -56,11 +77,7 @@ def body_to_paragraphs(
     tail = md_text[last_end:].strip()
     if tail:
         for block in re.split(r"\n\s*\n", tail):
-            block = block.strip()
-            if block:
-                paragraphs.append(
-                    {"text": block, "level": 0, "style": body_style}
-                )
+            _add_block(block)
 
     return paragraphs
 
@@ -93,7 +110,15 @@ def replace_placeholder(
 
             parent.remove(para._element)
 
+            # 为参考文献段落准备书签 ID
+            bm_id = _max_bookmark_id(doc) + 1
+
             for pd_data in reversed(paragraphs):
+                if pd_data.get("type") == "image":
+                    insert_image_paragraphs(
+                        doc, [pd_data], idx=idx, parent=parent
+                    )
+                else:
                     new_p = doc.add_paragraph(pd_data["text"])
                     style_name = pd_data["style"]
 
@@ -119,12 +144,123 @@ def replace_placeholder(
                         new_pPr.append(deepcopy(numPr))
 
                     parent.insert(idx, new_p._element)
+
+                    # 为参考文献条目添加书签
+                    ref_id = pd_data.get("ref_id")
+                    if ref_id is not None:
+                        _add_bookmark(new_p, f"ref-{ref_id}", bm_id)
+                        bm_id += 1
             break
 
     if not placeholder_found:
         print(f"警告：未找到占位符 '{placeholder}'，正文段落未注入。")
 
 
+def link_body_citations(doc: Document):
+    """将文档中正文段落的 ``[N]`` 引用替换为指向对应书签的超链接。"""
+    for para in doc.paragraphs:
+        # 跳过已有超链接的段落（如目录页）
+        if para._element.findall(qn("w:hyperlink")):
+            continue
+        _link_paragraph(para)
+
+
+def _max_bookmark_id(doc: Document) -> int:
+    """扫描文档返回最大书签 ID。"""
+    max_id = 0
+    for para in doc.paragraphs:
+        for bm in para._element.iter(qn("w:bookmarkStart")):
+            try:
+                max_id = max(max_id, int(bm.get(qn("w:id"))))
+            except (ValueError, TypeError):
+                pass
+    return max_id
+
+
+def _add_bookmark(paragraph, name: str, bm_id: int):
+    """为段落添加书签。"""
+    bm_start = paragraph._element.makeelement(qn("w:bookmarkStart"), {})
+    bm_start.set(qn("w:id"), str(bm_id))
+    bm_start.set(qn("w:name"), name)
+
+    bm_end = paragraph._element.makeelement(qn("w:bookmarkEnd"), {})
+    bm_end.set(qn("w:id"), str(bm_id))
+
+    pPr = paragraph._element.find(qn("w:pPr"))
+    if pPr is not None:
+        paragraph._element.insert(1, bm_start)
+    else:
+        paragraph._element.insert(0, bm_start)
+    paragraph._element.append(bm_end)
+
+
+def _link_paragraph(para):
+    """将单个段落中的 ``[N]`` 替换为 HYPERLINK 域。"""
+    runs = list(para._element.findall(qn("w:r")))
+    if not runs:
+        return
+
+    full_text = ""
+    for r in runs:
+        t = r.find(qn("w:t"))
+        if t is not None and t.text:
+            full_text += t.text
+
+    matches = list(_CITE_PATTERN.finditer(full_text))
+    if not matches:
+        return
+
+    first_rPr = runs[0].find(qn("w:rPr"))
+    for r in runs:
+        para._element.remove(r)
+
+    pos = 0
+    for m in matches:
+        before = full_text[pos : m.start()]
+        if before:
+            _add_run(para._element, before, first_rPr)
+
+        nums = re.findall(r"\d+", m.group(1))
+        _add_hlink(para._element, f"ref-{nums[0]}", m.group())
+
+        pos = m.end()
+
+    after = full_text[pos:]
+    if after:
+        _add_run(para._element, after, first_rPr)
+
+
+def _add_run(parent, text: str, rPr):
+    r = parent.makeelement(qn("w:r"), {})
+    if rPr is not None:
+        r.append(deepcopy(rPr))
+    t = r.makeelement(qn("w:t"), {})
+    t.text = text
+    r.append(t)
+    parent.append(r)
+
+
+def _add_hlink(parent, anchor: str, text: str):
+    hl = parent.makeelement(qn("w:hyperlink"), {})
+    hl.set(qn("w:anchor"), anchor)
+
+    r = parent.makeelement(qn("w:r"), {})
+    rPr = r.makeelement(qn("w:rPr"), {})
+    rStyle = rPr.makeelement(qn("w:rStyle"), {})
+    rStyle.set(qn("w:val"), "Hyperlink")
+    rPr.append(rStyle)
+    vertAlign = rPr.makeelement(qn("w:vertAlign"), {})
+    vertAlign.set(qn("w:val"), "superscript")
+    rPr.append(vertAlign)
+    r.append(rPr)
+
+    t = r.makeelement(qn("w:t"), {})
+    t.text = text
+    r.append(t)
+    hl.append(r)
+    parent.append(hl)
+
+
 def _apply_style(paragraph, doc, style_name: str) -> bool:
     """尝试给段落应用样式，成功返回 ``True``。"""
     try:

transit/images.py

@@ -0,0 +1,163 @@
+"""
+图片处理模块。
+
+将 Markdown 中的 ``<img>`` 标签解析为独立段落，
+并在 Word 文档中插入图片及居中图标题。
+"""
+
+import struct
+import re
+from pathlib import Path
+
+from docx import Document
+from docx.enum.text import WD_ALIGN_PARAGRAPH
+
+# 匹配 <img src="..." alt="...">
+_IMG_TAG = re.compile(
+    r'<img\s+([^>]*?)src=["\']([^"\']+)["\']([^>]*?)>', re.IGNORECASE
+)
+
+# 从标签属性块中提取名值对
+_ATTR = re.compile(r'(\w+)\s*=\s*["\']([^"\']*)["\']')
+
+# 匹配 **图X ...** 重复图标题（注入时跳过）
+_FIG_CAPTION = re.compile(
+    r'^\*\*(?:图|Table|Figure|Fig\.?)\s*\d+.*\*\*$', re.IGNORECASE
+)
+
+
+def make_image_paragraph(block: str, base_dir: str | Path | None = None) -> dict | None:
+    """若 *block* 包含 ``<img>`` 标签，返回图片段落字典；否则返回 ``None``。
+
+    Parameters
+    ----------
+    block : str
+        文本块。
+    base_dir : str | Path | None
+        Markdown 文件所在目录，用于解析图片相对路径。
+    """
+    m = _IMG_TAG.search(block)
+    if not m:
+        return None
+    attrs = dict(_ATTR.findall(block))
+    src = attrs.get("src", m.group(2))
+    if base_dir and not Path(src).is_absolute():
+        src = str(Path(base_dir) / src)
+    return {
+        "type": "image",
+        "src": src,
+        "alt": attrs.get("alt", ""),
+    }
+
+
+def is_figure_caption(block: str) -> bool:
+    """检查 *block* 是否为 ``**图X ...**`` 格式的重复图标题。"""
+    return bool(_FIG_CAPTION.match(block.strip()))
+
+
+def _get_image_dimensions(image_path: str) -> tuple[int, int] | None:
+    """读取图片文件头返回 ``(width_px, height_px)``，不支持格式返回 ``None``。
+
+    仅读取文件头，不依赖第三方库。支持 PNG / JPEG / GIF / BMP。
+    """
+    try:
+        with open(image_path, "rb") as f:
+            header = f.read(32)
+    except Exception:
+        return None
+
+    # PNG: 8-byte signature, then IHDR chunk
+    if header[:8] == b"\x89PNG\r\n\x1a\n":
+        w, h = struct.unpack_from(">II", header, 16)
+        return w, h
+
+    # JPEG: starts with FF D8, scan for SOF marker
+    if header[:2] == b"\xff\xd8":
+        pos = 2
+        while pos < len(header):
+            if header[pos] != 0xFF:
+                return None
+            marker = header[pos + 1]
+            if marker in (0xC0, 0xC1, 0xC2):
+                h, w = struct.unpack_from(">HH", header, pos + 5)
+                return w, h
+            seg_len = struct.unpack_from(">H", header, pos + 2)[0]
+            pos += 2 + seg_len
+        return None
+
+    # GIF: "GIF87a" or "GIF89a"
+    if header[:6] in (b"GIF87a", b"GIF89a"):
+        w, h = struct.unpack_from("<HH", header, 6)
+        return w, h
+
+    # BMP: "BM" signature
+    if header[:2] == b"BM":
+        w, h = struct.unpack_from("<ii", header, 18)
+        return w, abs(h)
+
+    return None
+
+
+def _get_native_emu(image_path: str) -> int | None:
+    """读取图片的原生宽度（EMU），失败返回 ``None``。
+
+    Word 默认以 72 DPI 渲染图片，1 px = 914400 / 72 = 12700 EMU。
+    """
+    dims = _get_image_dimensions(image_path)
+    if dims is None:
+        return None
+    w_px, _ = dims
+    return w_px * 12700
+
+
+def _constrain_width(image_path: str, page_text_width: int) -> int | None:
+    """返回图片宽度（EMU），超出页宽时缩至页宽。
+
+    Parameters
+    ----------
+    image_path : str
+        图片路径。
+    page_text_width : int
+        页面正文区宽度（EMU），来自 ``section.page_width - margins``。
+    """
+    native = _get_native_emu(image_path)
+    if native is None:
+        return None
+    return min(native, page_text_width)
+
+
+def insert_image_paragraphs(
+    doc: Document,
+    paragraphs: list[dict],
+    *,
+    idx: int,
+    parent,
+):
+    """在 *doc* 的指定位置插入图片段落序列。
+
+    每条图片段落生成两个 Word 段落：
+    1. 居中图片（超出页宽时自动缩放至页宽，否则保持原尺寸）
+    2. 居中图标题（从 ``alt`` 提取）
+
+    插入顺序保持 ``paragraphs`` 的原有顺序。
+    """
+    section = doc.sections[0]
+    page_text_width = section.page_width - section.left_margin - section.right_margin
+
+    for pd_data in reversed(paragraphs):
+        # 图标题（在 reversed 中先插入，最终位于图片下方）
+        cap_p = doc.add_paragraph(pd_data.get("alt", ""))
+        cap_p.alignment = WD_ALIGN_PARAGRAPH.CENTER
+        parent.insert(idx, cap_p._element)
+
+        # 图片段落（max-width 行为：超出页宽时压缩，否则原尺寸）
+        img_p = doc.add_paragraph()
+        img_p.alignment = WD_ALIGN_PARAGRAPH.CENTER
+        img_run = img_p.add_run()
+        try:
+            img_width = _constrain_width(pd_data["src"], page_text_width)
+            img_run.add_picture(pd_data["src"], width=img_width)
+        except Exception as exc:
+            print(f"警告：图片加载失败 {pd_data['src']} — {exc}")
+            img_run.add_text(f"[图片加载失败: {pd_data['src']}]")
+        parent.insert(idx, img_p._element)

transit/references.py

@@ -104,11 +104,15 @@ def references_to_paragraphs(
         ref = parse_reference_line(line)
         if ref:
             formatted = format_gb7714(ref)
+            ref_id = ref["number"]
         else:
             # 无法解析时，至少去掉 [N] 前缀
             fallback = re.sub(r"^\[\d+\]\s*", "", line)
             formatted = _normalize_period(fallback)
+            ref_id = None
 
-        paragraphs.append({"text": formatted, "level": 0, "style": ref_style})
+        paragraphs.append(
+            {"text": formatted, "level": 0, "style": ref_style, "ref_id": ref_id}
+        )
 
     return paragraphs

transit/renderer.py

@@ -11,7 +11,7 @@ from docx import Document
 
 from .config import load_config, ThesisConfig
 from .parser import parse_markdown
-from .body import body_to_paragraphs, replace_placeholder
+from .body import body_to_paragraphs, replace_placeholder, link_body_citations
 from .references import references_to_paragraphs
 
 
@@ -88,7 +88,12 @@ def generate_thesis(
     # 5. 解析正文为段落列表
     body_md = ctx.get("body_md", "")
     body_paragraphs = (
-        body_to_paragraphs(body_md, level_offset=config.level_offset, body_style=config.body_style)
+        body_to_paragraphs(
+            body_md,
+            level_offset=config.level_offset,
+            body_style=config.body_style,
+            base_dir=data_path.parent,
+        )
         if body_md else []
     )
 
@@ -114,6 +119,10 @@ def generate_thesis(
         final_doc, "__CONTEXT_PLACEHOLDER__", body_paragraphs,
         default_body_style=config.body_style,
     )
+
+    # 将正文中的 [N] 引用替换为超链接
+    link_body_citations(final_doc)
+
     replace_placeholder(
         final_doc, "__REFERENCE_PLACEHOLDER__", ref_paragraphs,
         default_body_style=config.reference_style,