giorgio/read.markets
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

csv-parser: add CsvFormatTemplate model

Browse source 
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
Giorgio Gilestro 2026-05-27 11:51:01 +02:00
parent 08b4dddcdd
commit 3f1d2a1034
2 changed files with 66 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

35
app/models.py
View file

@ -426,3 +426,38 @@ class StripeEvent(Base):
UniqueConstraint("event_id", name="uq_stripe_events_event_id"),
Index("ix_stripe_events_type_received", "event_type", "received_at"),
)
class CsvFormatTemplate(Base):
"""Cached column-mapping for a single broker CSV format.
Populated on the first upload of a previously-unseen format via the
LLM-fallback parser. Subsequent uploads of the same format
(identified by ``fingerprint``, a sha256 of the normalised header
row) replay ``mapping`` deterministically with no LLM call.
The table holds the actual ``headers`` and one anonymous ``sample_row``
from the originating upload — there is no ``user_id`` column, no link
back to the uploader. The sample exists so the operator has concrete
material to look at when hand-writing future native parsers; the
system never auto-generates or modifies parser code from this data.
"""
__tablename__ = "csv_format_templates"
id: Mapped[int] = mapped_column(_PK, primary_key=True, autoincrement=True)
fingerprint: Mapped[str] = mapped_column(String(64), unique=True, nullable=False)
headers: Mapped[list] = mapped_column(JSON, nullable=False)
sample_row: Mapped[list] = mapped_column(JSON, nullable=False)
mapping: Mapped[dict] = mapped_column(JSON, nullable=False)
preamble_rows: Mapped[int] = mapped_column(Integer, nullable=False, default=0)
delimiter: Mapped[str] = mapped_column(String(1), nullable=False, default=",")
broker_label: Mapped[str | None] = mapped_column(String(128))
first_seen_at: Mapped[datetime] = mapped_column(
DateTime(timezone=True), nullable=False, default=utcnow,
)
use_count: Mapped[int] = mapped_column(Integer, nullable=False, default=1)
last_used_at: Mapped[datetime] = mapped_column(
DateTime(timezone=True), nullable=False, default=utcnow,
)
llm_model: Mapped[str | None] = mapped_column(String(64))
llm_cost_usd: Mapped[float | None] = mapped_column(Float)