iomgaa
61cace71c4
Unrecognizable bytes are rejected before any request. The result package is downloaded as a stream with a cumulative size cap, and error statuses are classified before the size check. Parsing runs in a worker thread so cancellation and call deadlines stay effective. parse_layout shares the first protocol stage without behavior change.
72 lines
2.3 KiB
Python
72 lines
2.3 KiB
Python
"""OCR 文档结果包的合成 fixtures(#34)。
|
|
|
|
全部**合成**: 保留 2026-10-21 实测的结构骨架(para_blocks 块类型、table/image 的嵌套
|
|
`blocks`、span 字段),一切识别文本替换为 `SYNTH-*` 占位——第三方论文内容许可不明、
|
|
医疗数据不入库(零业务假设 + P5),真实 ZIP 不提交。
|
|
"""
|
|
|
|
import io
|
|
import json
|
|
import zipfile
|
|
|
|
_BBOX = (10.0, 20.0, 110.0, 40.0)
|
|
|
|
|
|
def _span(content="SYNTH-TEXT", type_="text"):
|
|
return {"type": type_, "content": content, "bbox": list(_BBOX), "score": 1.0}
|
|
|
|
|
|
def _text_block(content="SYNTH-TEXT", *, type_="text", bbox=_BBOX):
|
|
return {
|
|
"type": type_,
|
|
"bbox": list(bbox),
|
|
"index": 0,
|
|
"lines": [{"bbox": list(bbox), "spans": [_span(content)]}],
|
|
}
|
|
|
|
|
|
def _body(type_, spans, bbox=_BBOX):
|
|
return {"type": type_, "bbox": list(bbox), "lines": [{"bbox": list(bbox), "spans": spans}]}
|
|
|
|
|
|
def _table_block(html="<table><tr><td>SYNTH-CELL</td></tr></table>", extra_subs=()):
|
|
body = _body("table_body", [{"type": "table", "html": html, "image_path": "synth.jpg"}])
|
|
return {"type": "table", "bbox": list(_BBOX), "index": 1, "blocks": [*extra_subs, body]}
|
|
|
|
|
|
def _image_block():
|
|
body = _body("image_body", [{"type": "image", "image_path": "synth.jpg"}])
|
|
return {"type": "image", "bbox": list(_BBOX), "index": 2, "blocks": [body]}
|
|
|
|
|
|
def _page(blocks=None, *, page_idx=0, page_size=(612.0, 792.0), **extra):
|
|
page = {"page_idx": page_idx, "page_size": list(page_size), "discarded_blocks": []}
|
|
if blocks is not None:
|
|
page["para_blocks"] = blocks
|
|
page.update(extra)
|
|
return page
|
|
|
|
|
|
def _package(
|
|
pages=None,
|
|
*,
|
|
md="# synth\n",
|
|
middle=None,
|
|
middle_name="doc_middle.json",
|
|
md_name=None,
|
|
extra=(),
|
|
):
|
|
"""组装结果包。`middle` 给原始 bytes 时原样写入;`md_name` 默认由 middle 名配对。"""
|
|
if middle is None:
|
|
middle = json.dumps({"pdf_info": pages if pages is not None else [_page([])]}).encode()
|
|
if md_name is None:
|
|
md_name = middle_name[: -len("_middle.json")] + ".md"
|
|
buf = io.BytesIO()
|
|
with zipfile.ZipFile(buf, "w") as zf:
|
|
zf.writestr(middle_name, middle)
|
|
if md is not None:
|
|
zf.writestr(md_name, md if isinstance(md, bytes) else md.encode())
|
|
for name, data in extra:
|
|
zf.writestr(name, data)
|
|
return buf.getvalue()
|