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
This commit is contained in:
123
transit/body.py
123
transit/body.py
@@ -7,6 +7,7 @@ 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
|
||||
@@ -15,12 +16,16 @@ from .images import make_image_paragraph, is_figure_caption, insert_image_paragr
|
||||
|
||||
_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。
|
||||
|
||||
@@ -32,6 +37,8 @@ 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
|
||||
@@ -41,7 +48,7 @@ def body_to_paragraphs(
|
||||
if not block:
|
||||
return
|
||||
# 图片段落
|
||||
img = make_image_paragraph(block)
|
||||
img = make_image_paragraph(block, base_dir)
|
||||
if img:
|
||||
paragraphs.append(img)
|
||||
return
|
||||
@@ -103,6 +110,9 @@ 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(
|
||||
@@ -134,12 +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:
|
||||
|
||||
@@ -7,6 +7,7 @@
|
||||
|
||||
import struct
|
||||
import re
|
||||
from pathlib import Path
|
||||
|
||||
from docx import Document
|
||||
from docx.enum.text import WD_ALIGN_PARAGRAPH
|
||||
@@ -25,15 +26,26 @@ _FIG_CAPTION = re.compile(
|
||||
)
|
||||
|
||||
|
||||
def make_image_paragraph(block: str) -> dict | None:
|
||||
"""若 *block* 包含 ``<img>`` 标签，返回图片段落字典；否则返回 ``None``。"""
|
||||
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": attrs.get("src", m.group(2)),
|
||||
"src": src,
|
||||
"alt": attrs.get("alt", ""),
|
||||
}
|
||||
|
||||
|
||||
@@ -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
|
||||
|
||||
@@ -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,
|
||||
|
||||
Reference in New Issue
Block a user