Tanner Davis commited on
Commit
2cd20ef
·
1 Parent(s): 840cb29

remove references to storage for now

Browse files
Files changed (1) hide show
  1. src/papertrail/pipeline.py +16 -16
src/papertrail/pipeline.py CHANGED
@@ -23,14 +23,14 @@ from papertrail.events import Done, Event, Failed, Stage, Token
23
  from papertrail.inference.client import BasePaperTrailClient, get_client
24
  from papertrail.preprocess import load_document
25
  from papertrail.schema import EXTRACTION_INSTRUCTION, ProcessedExpense
26
- from papertrail.storage import Repository
27
 
28
 
29
  def process_document(
30
  path: str | Path,
31
  *,
32
  client: BasePaperTrailClient | None = None,
33
- repo: Repository | None = None,
34
  instruction: str = EXTRACTION_INSTRUCTION,
35
  doc_type: str = "upload",
36
  ) -> ProcessedExpense:
@@ -44,12 +44,12 @@ def process_document(
44
 
45
  result = ProcessedExpense(expense=expense, eligibility=verdict, readiness=readiness)
46
 
47
- if repo is not None:
48
- doc_id = repo.add_document(
49
- Path(path).name, path, doc_type, raw_model_output=expense.model_dump_json()
50
- )
51
- result.document_id = doc_id
52
- result.expense_id = repo.add_expense(doc_id, expense, verdict, readiness)
53
 
54
  return result
55
 
@@ -58,7 +58,7 @@ def process_document_stream(
58
  path: str | Path,
59
  *,
60
  client: BasePaperTrailClient | None = None,
61
- repo: Repository | None = None,
62
  instruction: str = EXTRACTION_INSTRUCTION,
63
  doc_type: str = "upload",
64
  ) -> Iterator[Event]:
@@ -89,13 +89,13 @@ def process_document_stream(
89
  yield Stage("scoring", "scoring audit readiness…")
90
  readiness = score(expense)
91
  result = ProcessedExpense(expense=expense, eligibility=verdict, readiness=readiness)
92
- if repo is not None:
93
- yield Stage("storage", "saving record…")
94
- doc_id = repo.add_document(
95
- Path(path).name, path, doc_type, raw_model_output=expense.model_dump_json()
96
- )
97
- result.document_id = doc_id
98
- result.expense_id = repo.add_expense(doc_id, expense, verdict, readiness)
99
  yield Done(result)
100
  except Exception as exc: # surface as a terminal event, don't crash the stream
101
  yield Failed(f"{type(exc).__name__}: {exc}")
 
23
  from papertrail.inference.client import BasePaperTrailClient, get_client
24
  from papertrail.preprocess import load_document
25
  from papertrail.schema import EXTRACTION_INSTRUCTION, ProcessedExpense
26
+ # from papertrail.storage import Repository
27
 
28
 
29
  def process_document(
30
  path: str | Path,
31
  *,
32
  client: BasePaperTrailClient | None = None,
33
+ # repo: Repository | None = None,
34
  instruction: str = EXTRACTION_INSTRUCTION,
35
  doc_type: str = "upload",
36
  ) -> ProcessedExpense:
 
44
 
45
  result = ProcessedExpense(expense=expense, eligibility=verdict, readiness=readiness)
46
 
47
+ # if repo is not None:
48
+ # doc_id = repo.add_document(
49
+ # Path(path).name, path, doc_type, raw_model_output=expense.model_dump_json()
50
+ # )
51
+ # result.document_id = doc_id
52
+ # result.expense_id = repo.add_expense(doc_id, expense, verdict, readiness)
53
 
54
  return result
55
 
 
58
  path: str | Path,
59
  *,
60
  client: BasePaperTrailClient | None = None,
61
+ # repo: Repository | None = None,
62
  instruction: str = EXTRACTION_INSTRUCTION,
63
  doc_type: str = "upload",
64
  ) -> Iterator[Event]:
 
89
  yield Stage("scoring", "scoring audit readiness…")
90
  readiness = score(expense)
91
  result = ProcessedExpense(expense=expense, eligibility=verdict, readiness=readiness)
92
+ # if repo is not None:
93
+ # yield Stage("storage", "saving record…")
94
+ # doc_id = repo.add_document(
95
+ # Path(path).name, path, doc_type, raw_model_output=expense.model_dump_json()
96
+ # )
97
+ # result.document_id = doc_id
98
+ # result.expense_id = repo.add_expense(doc_id, expense, verdict, readiness)
99
  yield Done(result)
100
  except Exception as exc: # surface as a terminal event, don't crash the stream
101
  yield Failed(f"{type(exc).__name__}: {exc}")