first commit
This commit is contained in:
384
mcp_docx.py
Normal file
384
mcp_docx.py
Normal file
@@ -0,0 +1,384 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
docx_editor.py — 保留原格式替换文本 + 修改字体颜色 + 替换图片
|
||||
|
||||
用法:
|
||||
# 列出文档中所有图片
|
||||
python3 docx_editor.py input.docx --list-images
|
||||
|
||||
# 文本替换 + 颜色
|
||||
python3 docx_editor.py input.docx output.docx \
|
||||
--replace "原文" "新文" \
|
||||
--color "关键词" "FF0000"
|
||||
|
||||
# 图片替换（按文档中出现的顺序，从1开始）
|
||||
python3 docx_editor.py input.docx output.docx \
|
||||
--image 1 new_chart.png \
|
||||
--image 2 new_photo.jpg
|
||||
|
||||
# 同时替换文字和图片
|
||||
python3 docx_editor.py input.docx output.docx \
|
||||
--replace "旧标题" "新标题" \
|
||||
--image 1 new_image.png \
|
||||
--color "重点" "FF0000"
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import os
|
||||
import tempfile
|
||||
import zipfile
|
||||
from lxml import etree
|
||||
from PIL import Image
|
||||
|
||||
W = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main'
|
||||
WD = 'http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing'
|
||||
A = 'http://schemas.openxmlformats.org/drawingml/2006/main'
|
||||
R = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships'
|
||||
REL_TYPE_IMAGE = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/image'
|
||||
|
||||
EXT_TO_MIME = {
|
||||
'png': 'image/png', 'jpg': 'image/jpeg', 'jpeg': 'image/jpeg',
|
||||
'gif': 'image/gif', 'bmp': 'image/bmp', 'tiff': 'image/tiff',
|
||||
'webp': 'image/webp',
|
||||
}
|
||||
|
||||
|
||||
def unpack(docx_path, out_dir):
|
||||
"""使用 zipfile 直接解包 .docx 到临时目录，替代外部 unpack.py 脚本。"""
|
||||
with zipfile.ZipFile(docx_path, 'r') as zf:
|
||||
zf.extractall(out_dir)
|
||||
|
||||
|
||||
def pack(unpacked_dir, output_docx, original_docx):
|
||||
"""
|
||||
使用 zipfile 将修改后的目录重新打包为 .docx。
|
||||
|
||||
original_docx 参数目前保留只是为了兼容原函数签名，没有实际使用。
|
||||
"""
|
||||
# 确保输出目录存在
|
||||
out_dir = os.path.dirname(os.path.abspath(output_docx))
|
||||
if out_dir and not os.path.exists(out_dir):
|
||||
os.makedirs(out_dir, exist_ok=True)
|
||||
|
||||
# 将解包目录中的所有文件打成 ZIP（保持相对路径结构）
|
||||
with zipfile.ZipFile(output_docx, 'w', compression=zipfile.ZIP_DEFLATED) as zf:
|
||||
for root, _, files in os.walk(unpacked_dir):
|
||||
for fname in files:
|
||||
abs_path = os.path.join(root, fname)
|
||||
# docx 内部使用 / 作为路径分隔符
|
||||
arcname = os.path.relpath(abs_path, unpacked_dir).replace(os.sep, '/')
|
||||
zf.write(abs_path, arcname)
|
||||
|
||||
|
||||
def build_image_index(unpacked_dir):
|
||||
"""返回按文档顺序排列的图片列表"""
|
||||
word_dir = os.path.join(unpacked_dir, 'word')
|
||||
doc_xml = os.path.join(word_dir, 'document.xml')
|
||||
rels_xml = os.path.join(word_dir, '_rels', 'document.xml.rels')
|
||||
|
||||
rels_root = etree.parse(rels_xml).getroot()
|
||||
rid_to_media = {}
|
||||
for rel in rels_root:
|
||||
if rel.get('Type', '') == REL_TYPE_IMAGE:
|
||||
rid_to_media[rel.get('Id')] = rel.get('Target')
|
||||
|
||||
doc_root = etree.parse(doc_xml).getroot()
|
||||
results = []
|
||||
for blip in doc_root.iter(f'{{{A}}}blip'):
|
||||
rid = blip.get(f'{{{R}}}embed')
|
||||
if not rid or rid not in rid_to_media:
|
||||
continue
|
||||
media_rel = rid_to_media[rid]
|
||||
media_abs = os.path.join(word_dir, media_rel.replace('/', os.sep))
|
||||
ext = os.path.splitext(media_rel)[1].lstrip('.').lower()
|
||||
|
||||
inline = blip
|
||||
while inline is not None and inline.tag not in (f'{{{WD}}}inline', f'{{{WD}}}anchor'):
|
||||
inline = inline.getparent()
|
||||
w_cm = h_cm = None
|
||||
docpr_name = ''
|
||||
if inline is not None:
|
||||
ext_el = inline.find(f'{{{WD}}}extent')
|
||||
if ext_el is not None:
|
||||
w_cm = round(int(ext_el.get('cx', 0)) / 360000, 2)
|
||||
h_cm = round(int(ext_el.get('cy', 0)) / 360000, 2)
|
||||
dp = inline.find(f'{{{WD}}}docPr')
|
||||
if dp is not None:
|
||||
docpr_name = dp.get('name', '')
|
||||
|
||||
results.append({
|
||||
'index': len(results) + 1, 'rid': rid,
|
||||
'media_file': media_rel, 'abs_path': media_abs,
|
||||
'ext': ext, 'docpr_name': docpr_name,
|
||||
'width_cm': w_cm, 'height_cm': h_cm,
|
||||
})
|
||||
return results
|
||||
|
||||
|
||||
def list_images(docx_path):
|
||||
imgs = get_images_info(docx_path)
|
||||
if not imgs:
|
||||
print("文档中没有找到图片。")
|
||||
return
|
||||
print(f"共找到 {len(imgs)} 张图片：\n")
|
||||
print(f" {'#':<4} {'文件名':<20} {'尺寸':<18} Word内部名称")
|
||||
print(" " + "-" * 62)
|
||||
for img in imgs:
|
||||
size = f"{img['width_cm']}×{img['height_cm']}cm" if img['width_cm'] else "未知"
|
||||
print(f" {img['index']:<4} {os.path.basename(img['media_file']):<20} {size:<18} {img['docpr_name']}")
|
||||
|
||||
|
||||
def get_images_info(docx_path):
|
||||
"""
|
||||
返回给定 DOCX 文件中所有图片的结构化信息列表。
|
||||
|
||||
该函数专门为其他模块（例如 MCP 服务器）复用而设计，
|
||||
行为等价于原来的 list_images 内部逻辑，但不做任何打印。
|
||||
"""
|
||||
with tempfile.TemporaryDirectory() as tmpdir:
|
||||
unpack(docx_path, tmpdir)
|
||||
return build_image_index(tmpdir)
|
||||
|
||||
|
||||
def replace_image(unpacked_dir, index, new_image_path):
|
||||
"""替换第 index 张图片（1-based）"""
|
||||
imgs = build_image_index(unpacked_dir)
|
||||
if index < 1 or index > len(imgs):
|
||||
raise ValueError(f"图片序号 {index} 超出范围（共 {len(imgs)} 张）")
|
||||
|
||||
info = imgs[index - 1]
|
||||
old_abs = info['abs_path']
|
||||
old_ext = info['ext']
|
||||
new_ext = os.path.splitext(new_image_path)[1].lstrip('.').lower()
|
||||
if new_ext == 'jpg':
|
||||
new_ext = 'jpeg'
|
||||
|
||||
print(f" 图片#{index} {os.path.basename(info['media_file'])}({old_ext.upper()})"
|
||||
f" ← {os.path.basename(new_image_path)}({new_ext.upper()})")
|
||||
|
||||
if old_ext == new_ext:
|
||||
# ── 同格式：直接覆盖 ──────────────────────────────
|
||||
import shutil
|
||||
shutil.copy2(new_image_path, old_abs)
|
||||
|
||||
else:
|
||||
# ── 不同格式：Pillow 转换 + 更新 rels + ContentTypes
|
||||
new_abs = os.path.splitext(old_abs)[0] + '.' + new_ext
|
||||
img = Image.open(new_image_path)
|
||||
fmt = {'jpeg': 'JPEG', 'png': 'PNG', 'gif': 'GIF',
|
||||
'bmp': 'BMP', 'tiff': 'TIFF', 'webp': 'WEBP'}.get(new_ext, new_ext.upper())
|
||||
if fmt == 'JPEG' and img.mode in ('RGBA', 'P'):
|
||||
img = img.convert('RGB')
|
||||
img.save(new_abs, format=fmt)
|
||||
if os.path.abspath(new_abs) != os.path.abspath(old_abs):
|
||||
os.remove(old_abs)
|
||||
|
||||
# 更新 rels
|
||||
old_media = info['media_file']
|
||||
new_media = os.path.splitext(old_media)[0] + '.' + new_ext
|
||||
word_dir = os.path.join(unpacked_dir, 'word')
|
||||
rels_path = os.path.join(word_dir, '_rels', 'document.xml.rels')
|
||||
rels_tree = etree.parse(rels_path)
|
||||
for rel in rels_tree.getroot():
|
||||
if rel.get('Id') == info['rid']:
|
||||
rel.set('Target', new_media)
|
||||
break
|
||||
rels_tree.write(rels_path, xml_declaration=True, encoding='UTF-8', standalone=True)
|
||||
|
||||
# 更新 ContentTypes
|
||||
ct_path = os.path.join(unpacked_dir, '[Content_Types].xml')
|
||||
ct_tree = etree.parse(ct_path)
|
||||
ct_root = ct_tree.getroot()
|
||||
existing = {el.get('Extension', '') for el in ct_root}
|
||||
if new_ext not in existing:
|
||||
etree.SubElement(ct_root, 'Default', Extension=new_ext,
|
||||
ContentType=EXT_TO_MIME.get(new_ext, f'image/{new_ext}'))
|
||||
ct_tree.write(ct_path, xml_declaration=True, encoding='UTF-8', standalone=True)
|
||||
print(f" 格式转换 {old_ext}→{new_ext}，rels 和 ContentTypes 已更新")
|
||||
|
||||
|
||||
def paragraph_replace(para_el, replacements):
|
||||
"""在 <w:t> 层面替换文本，完全不碰图片和格式"""
|
||||
for t_el in para_el.iter(f'{{{W}}}t'):
|
||||
if not t_el.text:
|
||||
continue
|
||||
new_text = t_el.text
|
||||
for old, new in replacements:
|
||||
new_text = new_text.replace(old, new)
|
||||
if new_text != t_el.text:
|
||||
t_el.text = new_text
|
||||
if new_text and (new_text[0] == ' ' or new_text[-1] == ' '):
|
||||
t_el.set('{http://www.w3.org/XML/1998/namespace}space', 'preserve')
|
||||
|
||||
|
||||
def ensure_rpr(run_el):
|
||||
rpr = run_el.find(f'{{{W}}}rPr')
|
||||
if rpr is None:
|
||||
rpr = etree.Element(f'{{{W}}}rPr')
|
||||
run_el.insert(0, rpr)
|
||||
return rpr
|
||||
|
||||
def set_color_on_rpr(rpr_el, hex_color):
|
||||
c = rpr_el.find(f'{{{W}}}color')
|
||||
if c is None:
|
||||
c = etree.SubElement(rpr_el, f'{{{W}}}color')
|
||||
c.set(f'{{{W}}}val', hex_color.lstrip('#'))
|
||||
|
||||
def apply_color_to_keyword(doc_el, keyword, hex_color):
|
||||
"""
|
||||
只给匹配到的关键字本身着色，而不是整个 run。
|
||||
|
||||
做法：在有关键字的 run 上，把文本拆成多段 run：
|
||||
[前缀][关键字][后缀]，只有“关键字”这个 run 设置颜色。
|
||||
"""
|
||||
# 先 list 一下，避免在遍历时修改树结构导致问题
|
||||
runs = list(doc_el.iter(f'{{{W}}}r'))
|
||||
for run in runs:
|
||||
t_nodes = list(run.findall(f'{{{W}}}t'))
|
||||
if not t_nodes:
|
||||
continue
|
||||
full_text = ''.join(t.text or '' for t in t_nodes)
|
||||
if keyword not in full_text:
|
||||
continue
|
||||
|
||||
parent = run.getparent()
|
||||
if parent is None:
|
||||
continue
|
||||
insert_pos = parent.index(run)
|
||||
|
||||
# 原 run 的 rPr 复制给新 run
|
||||
orig_rpr = run.find(f'{{{W}}}rPr')
|
||||
if orig_rpr is not None:
|
||||
rpr_bytes = etree.tostring(orig_rpr)
|
||||
else:
|
||||
rpr_bytes = None
|
||||
|
||||
def make_run(text, colored):
|
||||
new_r = etree.Element(f'{{{W}}}r')
|
||||
if rpr_bytes is not None:
|
||||
new_r.append(etree.fromstring(rpr_bytes))
|
||||
t_el = etree.SubElement(new_r, f'{{{W}}}t')
|
||||
t_el.text = text
|
||||
if text and (text[0] == ' ' or text[-1] == ' '):
|
||||
t_el.set('{http://www.w3.org/XML/1998/namespace}space', 'preserve')
|
||||
if colored:
|
||||
set_color_on_rpr(ensure_rpr(new_r), hex_color)
|
||||
return new_r
|
||||
|
||||
segments = []
|
||||
s = full_text
|
||||
start = 0
|
||||
klen = len(keyword)
|
||||
while True:
|
||||
idx = s.find(keyword, start)
|
||||
if idx == -1:
|
||||
if start < len(s):
|
||||
segments.append((s[start:], False))
|
||||
break
|
||||
if idx > start:
|
||||
segments.append((s[start:idx], False))
|
||||
segments.append((keyword, True))
|
||||
start = idx + klen
|
||||
|
||||
# 用新 run 替换原 run
|
||||
parent.remove(run)
|
||||
for offset, (seg_text, colored) in enumerate(segments):
|
||||
if seg_text:
|
||||
parent.insert(insert_pos + offset, make_run(seg_text, colored))
|
||||
|
||||
def process(input_docx, output_docx, replacements, image_replacements,
|
||||
color_keywords):
|
||||
with tempfile.TemporaryDirectory() as tmpdir:
|
||||
print(f"📂 解包 {input_docx} ...")
|
||||
unpack(input_docx, tmpdir)
|
||||
|
||||
doc_xml_path = os.path.join(tmpdir, 'word', 'document.xml')
|
||||
|
||||
if image_replacements:
|
||||
print(f"🖼️ 替换 {len(image_replacements)} 张图片...")
|
||||
for idx, new_img in image_replacements:
|
||||
replace_image(tmpdir, idx, new_img)
|
||||
|
||||
tree = etree.parse(doc_xml_path)
|
||||
root = tree.getroot()
|
||||
|
||||
if replacements:
|
||||
print(f"✏️ 替换 {len(replacements)} 条文本...")
|
||||
for para in root.iter(f'{{{W}}}p'):
|
||||
paragraph_replace(para, replacements)
|
||||
|
||||
# 根据 span 解析出的关键字上色
|
||||
for keyword, color in color_keywords:
|
||||
print(f"🎨 关键词「{keyword}」→ #{color}")
|
||||
apply_color_to_keyword(root, keyword, color)
|
||||
|
||||
tree.write(doc_xml_path, xml_declaration=True, encoding='UTF-8', standalone=True)
|
||||
print(f"📦 打包 → {output_docx} ...")
|
||||
pack(tmpdir, output_docx, input_docx)
|
||||
print(f"✅ 完成！输出: {output_docx}")
|
||||
|
||||
|
||||
def _parse_span_replacement(new_text):
|
||||
"""
|
||||
解析 NEW 文本中的 span 标签，用于决定颜色。
|
||||
|
||||
约定格式（不区分大小写）：
|
||||
<span color="FF0000">待补充</span>
|
||||
<span color="#FF0000">待补充</span>
|
||||
|
||||
返回: (纯文本, [(keyword, hex_color), ...])
|
||||
"""
|
||||
import re
|
||||
|
||||
span_pattern = re.compile(
|
||||
r'<span\s+[^>]*?color=["\']?(#?[0-9a-fA-F]{6})["\']?[^>]*>(.*?)</span>',
|
||||
re.IGNORECASE | re.DOTALL,
|
||||
)
|
||||
|
||||
color_keywords = []
|
||||
|
||||
def _repl(m):
|
||||
hex_color = m.group(1).lstrip('#')
|
||||
keyword = m.group(2)
|
||||
color_keywords.append((keyword, hex_color))
|
||||
return keyword
|
||||
|
||||
plain_text = span_pattern.sub(_repl, new_text)
|
||||
return plain_text, color_keywords
|
||||
|
||||
|
||||
def main():
|
||||
parser = argparse.ArgumentParser(description='DOCX 格式保留：替换文本/图片/颜色')
|
||||
parser.add_argument('input', help='输入 .docx')
|
||||
parser.add_argument('output', nargs='?', help='输出 .docx')
|
||||
parser.add_argument('--list-images', action='store_true', help='列出所有图片')
|
||||
parser.add_argument('--replace', nargs=2, metavar=('OLD', 'NEW'),
|
||||
action='append', default=[])
|
||||
parser.add_argument('--image', nargs=2, metavar=('INDEX', 'FILE'),
|
||||
action='append', default=[], help='图片替换')
|
||||
args = parser.parse_args()
|
||||
|
||||
if args.list_images:
|
||||
list_images(args.input)
|
||||
return
|
||||
if not args.output:
|
||||
parser.error("需要指定输出文件")
|
||||
|
||||
# 处理 span 颜色：把 NEW 中的 <span color="...">文字</span> 抽出来
|
||||
replacements = []
|
||||
color_keywords = []
|
||||
for old, new_raw in args.replace:
|
||||
new_plain, spans = _parse_span_replacement(new_raw)
|
||||
replacements.append((old, new_plain))
|
||||
color_keywords.extend(spans)
|
||||
|
||||
process(
|
||||
input_docx = args.input,
|
||||
output_docx = args.output,
|
||||
replacements = replacements,
|
||||
image_replacements= [(int(i), f) for i, f in args.image],
|
||||
color_keywords = color_keywords,
|
||||
)
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
Reference in New Issue
Block a user