46 lines
2.1 KiB
Python
46 lines
2.1 KiB
Python
"""Rerender cached PDF details; no remote requests or inferred dimensions."""
|
|
import base64
|
|
import hashlib
|
|
import math
|
|
|
|
|
|
def render_detail(job, source, page, box):
|
|
import pypdfium2 as pdfium
|
|
if type(page) is not int or page not in source.get('pages', []):
|
|
raise ValueError('只能放大已读取的 PDF 物理页')
|
|
if not isinstance(box, list) or len(box)!=4 or any(type(v) not in (int,float) or not math.isfinite(v) for v in box):
|
|
raise ValueError('box 必须为左上原点的四个 0–1 比例坐标')
|
|
x0,y0,x1,y1=box
|
|
if not (0<=x0<x1<=1 and 0<=y0<y1<=1) or min(x1-x0,y1-y0)<.04:
|
|
raise ValueError('放大区域超界或过小')
|
|
details=source.setdefault('details',[])
|
|
for detail in details:
|
|
if detail['page']==page and detail['box']==box: return detail
|
|
if len(details)>=12: raise ValueError('单份资料最多保留 12 个细节区域')
|
|
name=f"{source['id']}_p{page}_d{len(details)+1}.png"
|
|
pdf=job.path/f"{source['id']}.pdf"
|
|
if hashlib.sha256(pdf.read_bytes()).hexdigest()!=source['sha256']:
|
|
raise ValueError('参考 PDF 校验失败')
|
|
with pdfium.PdfDocument(pdf) as doc:
|
|
p=doc[page-1]
|
|
try:
|
|
w,h=p.get_size()
|
|
scale=min(6,2400/max(w*(x1-x0),h*(y1-y0)),6000/max(w,h))
|
|
bitmap=p.render(scale=scale,crop=(w*x0,h*(1-y1),w*(1-x1),h*y0))
|
|
try: bitmap.to_pil().save(job.path/name)
|
|
finally: bitmap.close()
|
|
finally: p.close()
|
|
result={'source_id':source['id'],'page':page,'box':box,'image':name,
|
|
'note':'由原始 PDF 重新渲染;box 为整页左上原点比例坐标,尺寸必须结合引出线和基准读取'}
|
|
details.append(result)
|
|
return result
|
|
|
|
|
|
def detail_content(job, details):
|
|
import json
|
|
content=[]
|
|
for d in details:
|
|
content.extend([{'type':'input_text','text':'原始图纸高清局部(不是生成模型):'+json.dumps(d,ensure_ascii=False)},
|
|
{'type':'input_image','detail':'high','image_url':'data:image/png;base64,'+base64.b64encode((job.path/d['image']).read_bytes()).decode()}])
|
|
return content
|