46 lines
2.0 KiB
Python
46 lines
2.0 KiB
Python
"""Shared orchestration for the contract reconstruction engine.
|
|
|
|
This module deliberately accepts the batch's persistence-backed extractor and
|
|
dossier builder as dependencies. Both the browser workflow and the JSON
|
|
integration boundary therefore execute the same business flow in-process,
|
|
while tests can replace either dependency without constructing HTTP calls or
|
|
making network requests.
|
|
"""
|
|
|
|
|
|
def analyze_batch(batch_id, documents, extractor, dossier_builder,
|
|
progress_callback=None, llm_client=None):
|
|
"""Extract document passports, assemble dossiers, and shape the API result.
|
|
|
|
`extractor` owns LLM interaction and persistence of each extraction;
|
|
`dossier_builder` consumes those saved passports and persists the resulting
|
|
graph. Keeping the calls in this order is an invariant: grouping must never
|
|
observe a partially extracted batch. The callback remains optional because
|
|
small batches run synchronously, while the existing large-batch route uses
|
|
it to update persistent progress.
|
|
"""
|
|
extraction = extractor(
|
|
batch_id,
|
|
llm_client=llm_client,
|
|
progress_callback=progress_callback,
|
|
documents=documents,
|
|
)
|
|
dossier_result = dossier_builder(batch_id) if extraction.get("ok") else {"ok": False}
|
|
classes = extraction.get("classes", {})
|
|
math_checks_passed = sum(
|
|
1 for result in extraction.get("results", [])
|
|
if (result.get("extraction") or {}).get("math_validation", {}).get("status") == "VERIFIED"
|
|
)
|
|
return {
|
|
"ok": bool(extraction.get("ok")) and bool(dossier_result.get("ok")),
|
|
"v3": True,
|
|
"done": extraction.get("done", 0),
|
|
"total": len(documents),
|
|
"failed": extraction.get("failed", 0),
|
|
"classes": classes,
|
|
"ignored_accounting": classes.get("ignored_accounting", 0),
|
|
"math_checks_passed": math_checks_passed,
|
|
"dossier_count": len(dossier_result.get("dossiers", [])),
|
|
"ambiguous": dossier_result.get("ambiguous", 0),
|
|
"unlinked": dossier_result.get("unlinked", []),
|
|
} |