Refactor digitador bot and add runtime monitoring
This commit is contained in:
erik
@@ -560,10 +560,29 @@ def post_event(
|
||||
metadata=body.metadata,
|
||||
timestamp=body.timestamp,
|
||||
)
|
||||
if body.index is not None and body.total is not None:
|
||||
overlay_keys = {
|
||||
"estado",
|
||||
"ultimo",
|
||||
"processed",
|
||||
"exitosos",
|
||||
"errores",
|
||||
"omitidos",
|
||||
"existentes",
|
||||
"reinicios",
|
||||
"radicado_actual",
|
||||
"elapsed_seconds",
|
||||
"speed_per_minute",
|
||||
}
|
||||
has_overlay = any(key in body.metadata for key in overlay_keys)
|
||||
if (body.index is not None and body.total is not None) or has_overlay:
|
||||
result = dict(job.result or {})
|
||||
result["processed"] = body.index
|
||||
result["total"] = body.total
|
||||
if body.index is not None and body.total is not None:
|
||||
result["total"] = body.total
|
||||
for key in overlay_keys:
|
||||
if key in body.metadata:
|
||||
result[key] = body.metadata[key]
|
||||
if "processed" not in body.metadata and body.index is not None:
|
||||
result["processed"] = body.index
|
||||
if body.nit:
|
||||
result["nit_actual"] = body.nit
|
||||
store.patch_job(digitador_job_id, result=result)
|
||||
|
||||
@@ -561,6 +561,9 @@ class JobStore:
|
||||
timeouts = 0
|
||||
existentes = 0
|
||||
uploaded = 0
|
||||
reinicios = 0
|
||||
elapsed_seconds = 0
|
||||
speed_per_minute = 0.0
|
||||
for c in children:
|
||||
r = dict(c.result or {})
|
||||
total += int(r.get("total") or 0)
|
||||
@@ -572,6 +575,9 @@ class JobStore:
|
||||
timeouts += int(r.get("timeouts") or r.get("timeout") or 0)
|
||||
existentes += int(r.get("existentes") or 0)
|
||||
uploaded += int(r.get("uploaded_count") or 0)
|
||||
reinicios += int(r.get("reinicios") or 0)
|
||||
elapsed_seconds = max(elapsed_seconds, int(r.get("elapsed_seconds") or 0))
|
||||
speed_per_minute += float(r.get("speed_per_minute") or 0)
|
||||
|
||||
processed = 0
|
||||
for c in children:
|
||||
@@ -605,6 +611,9 @@ class JobStore:
|
||||
"existentes": existentes,
|
||||
"processed": processed,
|
||||
"uploaded_count": uploaded,
|
||||
"reinicios": reinicios,
|
||||
"elapsed_seconds": elapsed_seconds,
|
||||
"speed_per_minute": round(speed_per_minute, 2),
|
||||
"reports_prefix": parent.reports_prefix,
|
||||
"shards": [
|
||||
{
|
||||
|
||||
@@ -0,0 +1,46 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import os
|
||||
from types import SimpleNamespace
|
||||
|
||||
os.environ["DATABASE_URL"] = "sqlite:///:memory:"
|
||||
|
||||
from api import main
|
||||
|
||||
|
||||
def test_post_event_persists_live_overlay_in_job_result(monkeypatch):
|
||||
patched: list[dict] = []
|
||||
job = SimpleNamespace(result={"total": 10})
|
||||
event = SimpleNamespace(to_dict=lambda: {"id": 1})
|
||||
fake_store = SimpleNamespace(
|
||||
get_job=lambda _job_id: job,
|
||||
add_event=lambda *_args, **_kwargs: event,
|
||||
patch_job=lambda _job_id, **kwargs: patched.append(kwargs) or job,
|
||||
)
|
||||
monkeypatch.setattr(main, "store", fake_store)
|
||||
|
||||
response = main.post_event(
|
||||
"job-1",
|
||||
main.EventBody(
|
||||
event_type="radicado_processing",
|
||||
message="Procesando",
|
||||
index=4,
|
||||
total=10,
|
||||
metadata={
|
||||
"estado": "Procesando",
|
||||
"processed": 3,
|
||||
"reinicios": 1,
|
||||
"radicado_actual": "RAD-4",
|
||||
"elapsed_seconds": 90,
|
||||
"speed_per_minute": 2.0,
|
||||
},
|
||||
),
|
||||
_="token",
|
||||
)
|
||||
|
||||
assert response == {"id": 1}
|
||||
result = patched[0]["result"]
|
||||
assert result["processed"] == 3
|
||||
assert result["total"] == 10
|
||||
assert result["radicado_actual"] == "RAD-4"
|
||||
assert result["reinicios"] == 1
|
||||
@@ -46,7 +46,7 @@ def test_distributed_parent_uses_agent_counters_and_live_progress(store):
|
||||
result={"total": 4, "processed": 2},
|
||||
status=JobStatus.RUNNING.value,
|
||||
)
|
||||
second = job_store.create_job(
|
||||
_second = job_store.create_job(
|
||||
"PC-02",
|
||||
"uploads/second/input.csv",
|
||||
parent_digitador_job_id=parent.id,
|
||||
@@ -72,6 +72,9 @@ def test_distributed_parent_uses_agent_counters_and_live_progress(store):
|
||||
"timeouts": 1,
|
||||
"existentes": 1,
|
||||
"uploaded_count": 3,
|
||||
"reinicios": 2,
|
||||
"elapsed_seconds": 90,
|
||||
"speed_per_minute": 1.5,
|
||||
},
|
||||
)
|
||||
live = job_store.refresh_parent_from_shards(parent.id)
|
||||
@@ -82,3 +85,6 @@ def test_distributed_parent_uses_agent_counters_and_live_progress(store):
|
||||
assert live.result["existentes"] == 1
|
||||
assert live.result["timeouts"] == 1
|
||||
assert live.result["uploaded_count"] == 3
|
||||
assert live.result["reinicios"] == 2
|
||||
assert live.result["elapsed_seconds"] == 90
|
||||
assert live.result["speed_per_minute"] == 1.5
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "ipsextractor-desktop",
|
||||
"version": "1.3.0",
|
||||
"version": "1.3.1",
|
||||
"private": true,
|
||||
"description": "Aplicación de escritorio Flujo IPS para automatizar procesos de cuentas médicas.",
|
||||
"main": "electron/main.cjs",
|
||||
|
||||
@@ -48,4 +48,35 @@ describe("App", () => {
|
||||
expect(screen.getByRole("heading", { name: /limpieza.*cop/i })).toBeInTheDocument();
|
||||
expect(screen.getByLabelText("2024")).toBeInTheDocument();
|
||||
});
|
||||
|
||||
it("offers the digitador package download inside an open job", async () => {
|
||||
const user = userEvent.setup();
|
||||
vi.spyOn(globalThis, "fetch").mockImplementation(async (input) => {
|
||||
const url = String(input);
|
||||
const body = url.endsWith("/jobs")
|
||||
? [{
|
||||
id: "job-1",
|
||||
status: "pending",
|
||||
title: "Trabajo prueba",
|
||||
metadata: {},
|
||||
created_at: "2026-08-30T12:00:00Z",
|
||||
updated_at: "2026-08-30T12:00:00Z",
|
||||
}]
|
||||
: [];
|
||||
return { ok: true, text: async () => JSON.stringify(body) } as Response;
|
||||
});
|
||||
render(<App />);
|
||||
|
||||
await user.click(await screen.findByText("Trabajo prueba"));
|
||||
const digitadorMenu = screen
|
||||
.getAllByRole("button", { name: /bot digitador/i })
|
||||
.find((button) => button.hasAttribute("aria-expanded"));
|
||||
expect(digitadorMenu).toBeDefined();
|
||||
await user.click(digitadorMenu!);
|
||||
|
||||
expect(screen.getByRole("button", { name: /descargar bot/i })).toHaveAttribute(
|
||||
"title",
|
||||
"Descargar el paquete del bot digitador",
|
||||
);
|
||||
});
|
||||
});
|
||||
|
||||
@@ -6,6 +6,7 @@ import {
|
||||
Download,
|
||||
Hash,
|
||||
Keyboard,
|
||||
Coins,
|
||||
Link2,
|
||||
Package,
|
||||
Settings,
|
||||
@@ -38,8 +39,10 @@ import "./modules/poblacion_apta/PoblacionAptaModule";
|
||||
import "./modules/downloads/DownloadsModule";
|
||||
import "./modules/digitador/DigitadorModule";
|
||||
|
||||
const BOT_PACKAGE_URL =
|
||||
const DOWNLOAD_BOT_PACKAGE_URL =
|
||||
"http://s3.builker.xyz/apps-versions/BotDescargaPYSCuentasMedicas/latest/bot_descarga.zip";
|
||||
const DIGITADOR_BOT_PACKAGE_URL =
|
||||
"http://s3.builker.xyz/apps-versions/BotDigitadorPYSCuentasMedicas/latest/bot_digitador.zip";
|
||||
|
||||
type Status = "idle" | "ok" | "error" | "busy";
|
||||
type View = "home" | "job" | "settings";
|
||||
@@ -100,8 +103,11 @@ function App() {
|
||||
const [busy, setBusy] = useState(false);
|
||||
const [releaseDate, setReleaseDate] = useState(APP_RELEASE_DATE);
|
||||
const [downloadsMenuOpen, setDownloadsMenuOpen] = useState(true);
|
||||
const [digitadorMenuOpen, setDigitadorMenuOpen] = useState(false);
|
||||
const [botDownloadBusy, setBotDownloadBusy] = useState(false);
|
||||
const [botDownloadError, setBotDownloadError] = useState<string | null>(null);
|
||||
const [digitadorDownloadBusy, setDigitadorDownloadBusy] = useState(false);
|
||||
const [digitadorDownloadError, setDigitadorDownloadError] = useState<string | null>(null);
|
||||
const update = useAutoUpdateCheck();
|
||||
|
||||
useEffect(() => {
|
||||
@@ -208,6 +214,7 @@ function App() {
|
||||
setActiveJob(job);
|
||||
setJobSection("downloads");
|
||||
setDownloadsMenuOpen(true);
|
||||
setDigitadorMenuOpen(false);
|
||||
setJobSectionContext({});
|
||||
setView("job");
|
||||
}
|
||||
@@ -216,6 +223,7 @@ function App() {
|
||||
setJobSection(section);
|
||||
setJobSectionContext(context);
|
||||
setDownloadsMenuOpen(section === "downloads");
|
||||
setDigitadorMenuOpen(section === "digitador");
|
||||
}
|
||||
|
||||
async function downloadBotPackage() {
|
||||
@@ -224,11 +232,11 @@ function App() {
|
||||
try {
|
||||
const api = window.desktopApi;
|
||||
if (typeof api?.downloadUrl === "function") {
|
||||
await api.downloadUrl(BOT_PACKAGE_URL);
|
||||
await api.downloadUrl(DOWNLOAD_BOT_PACKAGE_URL);
|
||||
return;
|
||||
}
|
||||
const a = document.createElement("a");
|
||||
a.href = BOT_PACKAGE_URL;
|
||||
a.href = DOWNLOAD_BOT_PACKAGE_URL;
|
||||
a.download = "bot_descarga.zip";
|
||||
a.rel = "noopener noreferrer";
|
||||
a.target = "_blank";
|
||||
@@ -242,6 +250,30 @@ function App() {
|
||||
}
|
||||
}
|
||||
|
||||
async function downloadDigitadorPackage() {
|
||||
setDigitadorDownloadBusy(true);
|
||||
setDigitadorDownloadError(null);
|
||||
try {
|
||||
const api = window.desktopApi;
|
||||
if (typeof api?.downloadUrl === "function") {
|
||||
await api.downloadUrl(DIGITADOR_BOT_PACKAGE_URL);
|
||||
return;
|
||||
}
|
||||
const a = document.createElement("a");
|
||||
a.href = DIGITADOR_BOT_PACKAGE_URL;
|
||||
a.download = "bot_digitador.zip";
|
||||
a.rel = "noopener noreferrer";
|
||||
a.target = "_blank";
|
||||
document.body.appendChild(a);
|
||||
a.click();
|
||||
a.remove();
|
||||
} catch (err) {
|
||||
setDigitadorDownloadError(err instanceof Error ? err.message : "No se pudo iniciar la descarga del bot digitador.");
|
||||
} finally {
|
||||
setDigitadorDownloadBusy(false);
|
||||
}
|
||||
}
|
||||
|
||||
function openSettings(section: SettingsSection = "connection") {
|
||||
setSettingsReturnView(view === "job" ? "job" : "home");
|
||||
setSettingsSection(section);
|
||||
@@ -405,13 +437,48 @@ function App() {
|
||||
<Users size={17} />
|
||||
<span>Población apta</span>
|
||||
</button>
|
||||
<div className="side-nav-group">
|
||||
<button
|
||||
type="button"
|
||||
className={jobSection === "digitador" ? "active" : ""}
|
||||
aria-expanded={digitadorMenuOpen}
|
||||
onClick={() => {
|
||||
if (jobSection === "digitador") {
|
||||
setDigitadorMenuOpen((open) => !open);
|
||||
} else {
|
||||
handleActivateSection("digitador");
|
||||
}
|
||||
}}
|
||||
>
|
||||
<Keyboard size={17} />
|
||||
<span>Bot digitador</span>
|
||||
<ChevronDown
|
||||
size={14}
|
||||
className={`side-nav-chevron${digitadorMenuOpen ? " side-nav-chevron--open" : ""}`}
|
||||
aria-hidden
|
||||
/>
|
||||
</button>
|
||||
{digitadorMenuOpen && (
|
||||
<div className="side-nav-sub" role="group" aria-label="Opciones del bot digitador">
|
||||
<button
|
||||
type="button"
|
||||
disabled={digitadorDownloadBusy}
|
||||
title={digitadorDownloadError || "Descargar el paquete del bot digitador"}
|
||||
onClick={() => void downloadDigitadorPackage()}
|
||||
>
|
||||
<Package size={15} />
|
||||
<span>{digitadorDownloadBusy ? "Descargando…" : "Descargar bot"}</span>
|
||||
</button>
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
<button
|
||||
type="button"
|
||||
className={jobSection === "digitador" ? "active" : ""}
|
||||
onClick={() => handleActivateSection("digitador")}
|
||||
className={jobSection === "costs" ? "active" : ""}
|
||||
onClick={() => handleActivateSection("costs")}
|
||||
>
|
||||
<Keyboard size={17} />
|
||||
<span>Bot digitador</span>
|
||||
<Coins size={17} />
|
||||
<span>Costos</span>
|
||||
</button>
|
||||
<button
|
||||
type="button"
|
||||
|
||||
@@ -0,0 +1,31 @@
|
||||
import { render, screen } from "@testing-library/react";
|
||||
import { DigitadorLiveMetrics } from "./DigitadorLiveMetrics";
|
||||
|
||||
describe("DigitadorLiveMetrics", () => {
|
||||
it("renders the remote equivalent of the local overlay", () => {
|
||||
render(
|
||||
<DigitadorLiveMetrics
|
||||
events={[{
|
||||
id: 1,
|
||||
job_id: "job-1",
|
||||
event_type: "radicado_processing",
|
||||
message: "Procesando",
|
||||
timestamp: "2026-08-30T12:00:00Z",
|
||||
metadata: {
|
||||
estado: "Procesando",
|
||||
radicado_actual: "RAD-123",
|
||||
reinicios: 2,
|
||||
speed_per_minute: 3.5,
|
||||
elapsed_seconds: 3661,
|
||||
ultimo: "Registro 3/10",
|
||||
},
|
||||
}]}
|
||||
/>,
|
||||
);
|
||||
|
||||
expect(screen.getByText("RAD-123")).toBeInTheDocument();
|
||||
expect(screen.getByText("3.5 rad/min")).toBeInTheDocument();
|
||||
expect(screen.getByText("01:01:01")).toBeInTheDocument();
|
||||
expect(screen.getByText("Registro 3/10")).toBeInTheDocument();
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,36 @@
|
||||
import type { DigitadorJobEvent } from "../../../shared/types";
|
||||
|
||||
type Props = { events: DigitadorJobEvent[] };
|
||||
|
||||
function value(metadata: Record<string, unknown>, key: string, fallback: string | number = "—") {
|
||||
const raw = metadata[key];
|
||||
return raw === undefined || raw === null || raw === "" ? String(fallback) : String(raw);
|
||||
}
|
||||
|
||||
function duration(secondsValue: unknown): string {
|
||||
const total = Math.max(0, Number(secondsValue) || 0);
|
||||
const hours = Math.floor(total / 3600);
|
||||
const minutes = Math.floor((total % 3600) / 60);
|
||||
const seconds = Math.floor(total % 60);
|
||||
return [hours, minutes, seconds].map((part) => String(part).padStart(2, "0")).join(":");
|
||||
}
|
||||
|
||||
export function DigitadorLiveMetrics({ events }: Props) {
|
||||
const latest = [...events].reverse().find((event) => {
|
||||
const metadata = event.metadata ?? {};
|
||||
return "estado" in metadata || "radicado_actual" in metadata;
|
||||
});
|
||||
if (!latest) return null;
|
||||
const metadata = latest.metadata ?? {};
|
||||
|
||||
return (
|
||||
<div className="digitador-live-metrics" aria-label="Estado en vivo del bot digitador">
|
||||
<div><span>Estado</span><strong>{value(metadata, "estado", latest.event_type)}</strong></div>
|
||||
<div><span>Radicado</span><strong title={value(metadata, "radicado_actual")}>{value(metadata, "radicado_actual")}</strong></div>
|
||||
<div><span>Reinicios</span><strong>{value(metadata, "reinicios", 0)}</strong></div>
|
||||
<div><span>Velocidad</span><strong>{value(metadata, "speed_per_minute", 0)} rad/min</strong></div>
|
||||
<div><span>Tiempo</span><strong>{duration(metadata.elapsed_seconds)}</strong></div>
|
||||
<div><span>Último evento</span><strong title={value(metadata, "ultimo", latest.message)}>{value(metadata, "ultimo", latest.message)}</strong></div>
|
||||
</div>
|
||||
);
|
||||
}
|
||||
@@ -17,6 +17,7 @@ import type {
|
||||
} from "../../../shared/types";
|
||||
import type { ApiClient } from "../../lib/apiClient";
|
||||
import { ParallelBotProgress } from "../../components/ParallelBotProgress";
|
||||
import { DigitadorLiveMetrics } from "./DigitadorLiveMetrics";
|
||||
|
||||
declare const desktopApi: import("../../../shared/types").DesktopBridge | undefined;
|
||||
|
||||
@@ -284,6 +285,8 @@ export function DigitadorResultsPanel({ client, job, tasks, onStartDigitador }:
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<DigitadorLiveMetrics events={events} />
|
||||
|
||||
<p className="wiz-hint" style={{ marginBottom: 12, flexShrink: 0 }}>
|
||||
{selected.shard_count && Number(selected.shard_count) > 1
|
||||
? `${selected.shard_count} bots (distribuido)`
|
||||
|
||||
@@ -15,6 +15,7 @@ import { useCallback, useEffect, useMemo, useRef, useState } from "react";
|
||||
import type { ApiClient } from "../../lib/apiClient";
|
||||
import { formatDateTime } from "../../lib/time";
|
||||
import { ParallelBotProgress } from "../../components/ParallelBotProgress";
|
||||
import { DigitadorLiveMetrics } from "./DigitadorLiveMetrics";
|
||||
import type {
|
||||
DigitadorJob,
|
||||
DigitadorJobEvent,
|
||||
@@ -627,6 +628,8 @@ export function DigitadorWizardModule({ client, job, initialSource = null }: Pro
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<DigitadorLiveMetrics events={events} />
|
||||
|
||||
<ParallelBotProgress shards={shards} unit="filas" />
|
||||
|
||||
<div className="wiz-section" style={{ marginTop: 16, width: "100%", maxWidth: 640 }}>
|
||||
|
||||
@@ -5635,3 +5635,65 @@ th {
|
||||
margin-left: 28px !important;
|
||||
margin-right: 28px !important;
|
||||
}
|
||||
|
||||
/* ── Costos de API por trabajo ── */
|
||||
.job-costs-view {
|
||||
box-sizing: border-box;
|
||||
height: 100%;
|
||||
overflow: auto;
|
||||
padding: 22px 28px 32px;
|
||||
}
|
||||
|
||||
.job-costs-toolbar { align-items: flex-start; display: flex; gap: 20px; justify-content: space-between; }
|
||||
.job-costs-toolbar h2 { font-size: 20px; margin: 0 0 4px; }
|
||||
.job-costs-toolbar p { color: var(--text-muted); font-size: 13px; margin: 0; }
|
||||
.job-costs-summary { display: grid; gap: 12px; grid-template-columns: repeat(4, minmax(0, 1fr)); margin: 20px 0; }
|
||||
.job-costs-summary > div { background: var(--surface, #fff); border: 1px solid var(--border); border-radius: 10px; display: flex; flex-direction: column; gap: 6px; padding: 14px 16px; }
|
||||
.job-costs-summary span { color: var(--text-muted); font-size: 12px; }
|
||||
.job-costs-summary strong { font-size: 20px; }
|
||||
.job-cost-warning { background: #fffbeb; border: 1px solid #fde68a; border-radius: 8px; color: #92400e; font-size: 12px; margin: -8px 0 16px; padding: 9px 12px; }
|
||||
.job-costs-list { display: flex; flex-direction: column; gap: 10px; }
|
||||
.job-cost-card { background: var(--surface, #fff); border: 1px solid var(--border); border-radius: 10px; overflow: hidden; }
|
||||
.job-cost-card > summary { align-items: center; cursor: pointer; display: grid; gap: 16px; grid-template-columns: minmax(220px, 1.5fr) auto auto minmax(100px, auto); list-style-position: inside; padding: 14px 16px; }
|
||||
.job-cost-process { display: inline-flex; flex-direction: column; min-width: 0; }
|
||||
.job-cost-process span { color: var(--text-muted); font-size: 12px; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
|
||||
.job-cost-api-badge { background: #f1f5f9; border-radius: 999px; color: #64748b; font-size: 11px; padding: 4px 8px; }
|
||||
.job-cost-api-badge.has-usage { background: #dcfce7; color: #15803d; }
|
||||
.job-cost-total { color: #0f766e; text-align: right; }
|
||||
.job-cost-detail { border-top: 1px solid var(--border); padding: 12px 16px 16px; }
|
||||
.job-cost-table { border-collapse: collapse; font-size: 12px; width: 100%; }
|
||||
.job-cost-table th, .job-cost-table td { border-bottom: 1px solid var(--border); padding: 9px 10px; text-align: right; }
|
||||
.job-cost-table th:first-child, .job-cost-table td:first-child { text-align: left; }
|
||||
.job-cost-table td small { color: var(--text-muted); display: block; font-size: 10px; margin-top: 2px; }
|
||||
.job-cost-calls { margin-top: 14px; }
|
||||
.job-cost-calls > summary { color: #2563eb; cursor: pointer; font-size: 12px; font-weight: 600; }
|
||||
.job-cost-calls-scroll { margin-top: 8px; max-height: 340px; overflow: auto; }
|
||||
.job-cost-no-api, .job-costs-empty { color: var(--text-muted); font-size: 13px; }
|
||||
.job-costs-empty { align-items: center; border: 1px dashed var(--border); border-radius: 10px; display: flex; flex-direction: column; padding: 40px; }
|
||||
@media (max-width: 900px) {
|
||||
.job-costs-summary { grid-template-columns: repeat(2, minmax(0, 1fr)); }
|
||||
.job-cost-card > summary { grid-template-columns: 1fr auto; }
|
||||
.job-cost-card > summary > span:not(.job-cost-api-badge) { display: none; }
|
||||
}
|
||||
|
||||
.digitador-live-metrics {
|
||||
display: grid;
|
||||
gap: 8px;
|
||||
grid-template-columns: repeat(3, minmax(0, 1fr));
|
||||
margin: 4px 0 16px;
|
||||
width: 100%;
|
||||
}
|
||||
|
||||
.digitador-live-metrics > div {
|
||||
background: var(--surface, #fff);
|
||||
border: 1px solid var(--border);
|
||||
border-radius: 8px;
|
||||
display: flex;
|
||||
flex-direction: column;
|
||||
gap: 3px;
|
||||
min-width: 0;
|
||||
padding: 9px 11px;
|
||||
}
|
||||
|
||||
.digitador-live-metrics span { color: var(--text-muted); font-size: 10px; text-transform: uppercase; }
|
||||
.digitador-live-metrics strong { font-size: 12px; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
|
||||
|
||||
@@ -5,4 +5,5 @@ export type JobSection =
|
||||
| "poblacion_apta"
|
||||
| "downloads"
|
||||
| "digitador"
|
||||
| "costs"
|
||||
| "results";
|
||||
|
||||
@@ -0,0 +1,50 @@
|
||||
import { render, screen } from "@testing-library/react";
|
||||
import type { Task } from "../../shared/types";
|
||||
import { JobCostsView } from "./JobCostsView";
|
||||
|
||||
function task(result: Task["result"]): Task {
|
||||
return {
|
||||
id: "task-1",
|
||||
type: "extraction",
|
||||
status: "completed",
|
||||
title: "Extracción facturas",
|
||||
request: {},
|
||||
result,
|
||||
error: null,
|
||||
owner_id: null,
|
||||
job_id: "job-1",
|
||||
created_at: "2026-08-30T12:00:00Z",
|
||||
updated_at: "2026-08-30T12:01:00Z",
|
||||
started_at: "2026-08-30T12:00:00Z",
|
||||
finished_at: "2026-08-30T12:01:00Z",
|
||||
};
|
||||
}
|
||||
|
||||
describe("JobCostsView", () => {
|
||||
it("shows real API usage totals for a process", () => {
|
||||
render(
|
||||
<JobCostsView
|
||||
refreshing={false}
|
||||
onRefresh={() => undefined}
|
||||
tasks={[task({
|
||||
api_usage: {
|
||||
summary: { requests: 2, total_tokens: 1250, total_cost_usd: 0.00125 },
|
||||
phases: {
|
||||
extract: { requests: 2, prompt_tokens: 1000, completion_tokens: 250, total_cost_usd: 0.00125, model: "gpt-5.6-luna" },
|
||||
},
|
||||
},
|
||||
})]}
|
||||
/>,
|
||||
);
|
||||
|
||||
expect(screen.getByText("Extracción facturas")).toBeInTheDocument();
|
||||
expect(screen.getByText("2 llamadas")).toBeInTheDocument();
|
||||
expect(screen.getAllByText("1.250").length).toBeGreaterThan(0);
|
||||
expect(screen.getAllByText("$0.00125").length).toBeGreaterThan(0);
|
||||
});
|
||||
|
||||
it("shows processes that did not consume a priced API", () => {
|
||||
render(<JobCostsView refreshing={false} onRefresh={() => undefined} tasks={[task({})]} />);
|
||||
expect(screen.getByText("Sin consumo API")).toBeInTheDocument();
|
||||
});
|
||||
});
|
||||
@@ -0,0 +1,190 @@
|
||||
import { Coins, RefreshCw } from "lucide-react";
|
||||
import type { Task } from "../../shared/types";
|
||||
|
||||
type UsageSummary = {
|
||||
requests?: number;
|
||||
prompt_tokens?: number;
|
||||
cached_tokens?: number;
|
||||
completion_tokens?: number;
|
||||
total_tokens?: number;
|
||||
input_usd?: number;
|
||||
cached_input_usd?: number;
|
||||
output_usd?: number;
|
||||
total_cost_usd?: number;
|
||||
unpriced_requests?: number;
|
||||
model?: string;
|
||||
};
|
||||
|
||||
type UsageDetail = UsageSummary & {
|
||||
timestamp?: string;
|
||||
phase?: string;
|
||||
model?: string;
|
||||
radicado?: string | null;
|
||||
filename?: string | null;
|
||||
cost?: { total_usd?: number };
|
||||
pricing_known?: boolean;
|
||||
};
|
||||
|
||||
type ApiUsage = {
|
||||
summary?: UsageSummary;
|
||||
phases?: Record<string, UsageSummary>;
|
||||
details?: UsageDetail[];
|
||||
};
|
||||
|
||||
interface Props {
|
||||
tasks: Task[];
|
||||
refreshing: boolean;
|
||||
onRefresh: () => void;
|
||||
}
|
||||
|
||||
const PROCESS_NAMES: Record<string, string> = {
|
||||
counting: "Conteo",
|
||||
extraction: "Extracción",
|
||||
cleaning: "Limpieza",
|
||||
poblacion_apta: "Población apta",
|
||||
downloads: "Bot de descarga",
|
||||
digitador: "Bot digitador",
|
||||
};
|
||||
|
||||
function usageOf(task: Task): ApiUsage {
|
||||
const value = task.result?.api_usage;
|
||||
return value && typeof value === "object" ? value as ApiUsage : {};
|
||||
}
|
||||
|
||||
function n(value: unknown): number {
|
||||
const parsed = Number(value ?? 0);
|
||||
return Number.isFinite(parsed) ? parsed : 0;
|
||||
}
|
||||
|
||||
function usd(value: unknown): string {
|
||||
return new Intl.NumberFormat("en-US", {
|
||||
style: "currency",
|
||||
currency: "USD",
|
||||
minimumFractionDigits: 4,
|
||||
maximumFractionDigits: 6,
|
||||
}).format(n(value));
|
||||
}
|
||||
|
||||
function integer(value: unknown): string {
|
||||
return new Intl.NumberFormat("es-CO").format(n(value));
|
||||
}
|
||||
|
||||
function phaseName(phase: string): string {
|
||||
if (phase === "extract") return "Extracción LLM";
|
||||
if (phase === "normalize") return "Normalización LLM";
|
||||
return phase;
|
||||
}
|
||||
|
||||
export function JobCostsView({ tasks, refreshing, onRefresh }: Props) {
|
||||
const ordered = [...tasks].sort((a, b) => new Date(b.created_at).getTime() - new Date(a.created_at).getTime());
|
||||
const summaries = ordered.map((task) => usageOf(task).summary ?? {});
|
||||
const totalCost = summaries.reduce((sum, item) => sum + n(item.total_cost_usd), 0);
|
||||
const totalTokens = summaries.reduce((sum, item) => sum + n(item.total_tokens), 0);
|
||||
const totalRequests = summaries.reduce((sum, item) => sum + n(item.requests), 0);
|
||||
const unpricedRequests = summaries.reduce((sum, item) => sum + n(item.unpriced_requests), 0);
|
||||
|
||||
return (
|
||||
<div className="job-costs-view">
|
||||
<div className="job-costs-toolbar">
|
||||
<div>
|
||||
<h2>Costos de API por proceso</h2>
|
||||
<p>Uso real reportado por el proveedor. Los valores históricos conservan la tarifa aplicada al procesar.</p>
|
||||
</div>
|
||||
<button type="button" className="btn btn-sm" onClick={onRefresh} disabled={refreshing}>
|
||||
<RefreshCw size={14} className={refreshing ? "spin" : ""} />
|
||||
Actualizar
|
||||
</button>
|
||||
</div>
|
||||
|
||||
<div className="job-costs-summary">
|
||||
<div><span>Costo total</span><strong>{usd(totalCost)}</strong></div>
|
||||
<div><span>Tokens</span><strong>{integer(totalTokens)}</strong></div>
|
||||
<div><span>Llamadas API</span><strong>{integer(totalRequests)}</strong></div>
|
||||
<div><span>Procesos</span><strong>{integer(tasks.length)}</strong></div>
|
||||
</div>
|
||||
|
||||
{unpricedRequests > 0 && (
|
||||
<div className="job-cost-warning">
|
||||
{integer(unpricedRequests)} llamadas tienen un modelo sin tarifa configurada; sus tokens se muestran, pero no se suman al costo.
|
||||
</div>
|
||||
)}
|
||||
|
||||
{ordered.length === 0 ? (
|
||||
<div className="job-costs-empty"><Coins size={28} /><p>Aún no hay procesos en este trabajo.</p></div>
|
||||
) : (
|
||||
<div className="job-costs-list">
|
||||
{ordered.map((task) => {
|
||||
const apiUsage = usageOf(task);
|
||||
const summary = apiUsage.summary ?? {};
|
||||
const phases = Object.entries(apiUsage.phases ?? {});
|
||||
const details = apiUsage.details ?? [];
|
||||
const hasUsage = n(summary.requests) > 0;
|
||||
return (
|
||||
<details className="job-cost-card" key={task.id}>
|
||||
<summary>
|
||||
<div className="job-cost-process">
|
||||
<strong>{PROCESS_NAMES[task.type] ?? task.type}</strong>
|
||||
<span>{task.title || `Proceso ${task.id.slice(0, 8)}`}</span>
|
||||
</div>
|
||||
<span className={`job-cost-api-badge${hasUsage ? " has-usage" : ""}`}>
|
||||
{hasUsage ? `${integer(summary.requests)} llamadas` : "Sin consumo API"}
|
||||
</span>
|
||||
<span>{integer(summary.total_tokens)} tokens</span>
|
||||
<strong className="job-cost-total">{usd(summary.total_cost_usd)}</strong>
|
||||
</summary>
|
||||
|
||||
<div className="job-cost-detail">
|
||||
{hasUsage ? (
|
||||
<>
|
||||
<table className="job-cost-table">
|
||||
<thead><tr><th>Fase / modelo</th><th>Llamadas</th><th>Entrada</th><th>En caché</th><th>Salida</th><th>Costo</th></tr></thead>
|
||||
<tbody>
|
||||
{phases.map(([phase, item]) => (
|
||||
<tr key={phase}>
|
||||
<td><strong>{phaseName(phase)}</strong><small>{item.model || "Modelo no informado"}</small></td>
|
||||
<td>{integer(item.requests)}</td>
|
||||
<td>{integer(item.prompt_tokens)}</td>
|
||||
<td>{integer(item.cached_tokens)}</td>
|
||||
<td>{integer(item.completion_tokens)}</td>
|
||||
<td>{usd(item.total_cost_usd)}</td>
|
||||
</tr>
|
||||
))}
|
||||
</tbody>
|
||||
</table>
|
||||
|
||||
{details.length > 0 && (
|
||||
<details className="job-cost-calls">
|
||||
<summary>Ver detalle de {details.length} llamadas</summary>
|
||||
<div className="job-cost-calls-scroll">
|
||||
<table className="job-cost-table">
|
||||
<thead><tr><th>Documento</th><th>Fase</th><th>Modelo</th><th>Entrada</th><th>Caché</th><th>Salida</th><th>Costo</th></tr></thead>
|
||||
<tbody>
|
||||
{details.map((item, index) => (
|
||||
<tr key={`${item.timestamp ?? "call"}-${index}`}>
|
||||
<td>{item.radicado || item.filename || "—"}<small>{item.radicado && item.filename ? item.filename : ""}</small></td>
|
||||
<td>{phaseName(item.phase || "API")}</td>
|
||||
<td>{item.model || "—"}</td>
|
||||
<td>{integer(item.prompt_tokens)}</td>
|
||||
<td>{integer(item.cached_tokens)}</td>
|
||||
<td>{integer(item.completion_tokens)}</td>
|
||||
<td>{item.pricing_known === false ? "Sin tarifa" : usd(item.cost?.total_usd)}</td>
|
||||
</tr>
|
||||
))}
|
||||
</tbody>
|
||||
</table>
|
||||
</div>
|
||||
</details>
|
||||
)}
|
||||
</>
|
||||
) : (
|
||||
<p className="job-cost-no-api">Este proceso no registró llamadas a una API LLM tarifada.</p>
|
||||
)}
|
||||
</div>
|
||||
</details>
|
||||
);
|
||||
})}
|
||||
</div>
|
||||
)}
|
||||
</div>
|
||||
);
|
||||
}
|
||||
@@ -11,6 +11,7 @@ import { JobExtractionWizard } from "../modules/extraction/JobExtractionWizard";
|
||||
import { DownloadsWizardModule } from "../modules/downloads/DownloadsWizardModule";
|
||||
import { DigitadorWizardModule } from "../modules/digitador/DigitadorWizardModule";
|
||||
import { JobResultsView } from "./JobResultsView";
|
||||
import { JobCostsView } from "./JobCostsView";
|
||||
|
||||
interface Props {
|
||||
client: ApiClient;
|
||||
@@ -30,6 +31,7 @@ export function JobView({ client, job, section, sectionContext, onActivateSectio
|
||||
const [extractionProgress, setExtractionProgress] = useState<BackgroundProgress | null>(null);
|
||||
const [cleaningProgress, setCleaningProgress] = useState<BackgroundProgress | null>(null);
|
||||
const [poblacionAptaProgress, setPoblacionAptaProgress] = useState<BackgroundProgress | null>(null);
|
||||
const [costsRefreshing, setCostsRefreshing] = useState(false);
|
||||
const pollingRef = useRef<ReturnType<typeof setInterval> | null>(null);
|
||||
|
||||
const backgroundRunning =
|
||||
@@ -45,6 +47,11 @@ export function JobView({ client, job, section, sectionContext, onActivateSectio
|
||||
} catch { /* silent */ }
|
||||
}, [client, job.id]);
|
||||
|
||||
const refreshCosts = useCallback(async () => {
|
||||
setCostsRefreshing(true);
|
||||
try { await loadTasks(); } finally { setCostsRefreshing(false); }
|
||||
}, [loadTasks]);
|
||||
|
||||
useEffect(() => {
|
||||
loadTasks();
|
||||
return () => { if (pollingRef.current) clearInterval(pollingRef.current); };
|
||||
@@ -129,6 +136,7 @@ export function JobView({ client, job, section, sectionContext, onActivateSectio
|
||||
{section === "poblacion_apta" && "Población apta"}
|
||||
{section === "downloads" && "Bot de descarga"}
|
||||
{section === "digitador" && "Bot digitador"}
|
||||
{section === "costs" && "Costos de API"}
|
||||
{section === "results" && "Resultados del trabajo"}
|
||||
</div>
|
||||
</header>
|
||||
@@ -236,6 +244,10 @@ export function JobView({ client, job, section, sectionContext, onActivateSectio
|
||||
onContinueToDigitador={(ctx) => onActivateSection("digitador", ctx)}
|
||||
/>
|
||||
</div>
|
||||
|
||||
<div className={`job-section-pane${section === "costs" ? "" : " job-section-pane--hidden"}`}>
|
||||
<JobCostsView tasks={tasks} refreshing={costsRefreshing} onRefresh={() => void refreshCosts()} />
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
);
|
||||
|
||||
@@ -8,6 +8,8 @@ from domain.normalizer import normalize_to_schema
|
||||
from domain.profiles import get_profile
|
||||
from domain.prompts import PromptBuilder
|
||||
from domain.service import InvoiceProcessingService
|
||||
from domain.config import settings
|
||||
from domain.usage_tracking import record_api_usage
|
||||
from infrastructure.database import ExtractionRow, Task, utcnow
|
||||
from infrastructure.serialization import flatten_extraction_result
|
||||
from sqlalchemy.orm import Session
|
||||
@@ -276,7 +278,7 @@ def process_desktop_ingest(
|
||||
else:
|
||||
prompt = instruction or PromptBuilder.build(selected_paths, {}, profile_name=prof.get("name", "invoice"))
|
||||
|
||||
data, _usage = service.extract_full_from_pdf(
|
||||
data, extraction_usage = service.extract_full_from_pdf(
|
||||
temp_path,
|
||||
instruction=prompt,
|
||||
tiff_page_range=tiff_page_range,
|
||||
@@ -292,8 +294,28 @@ def process_desktop_ingest(
|
||||
profile_name=prof.get("name", "invoice"),
|
||||
)
|
||||
normalized_obj = data
|
||||
normalizer_usage = None
|
||||
if isinstance(norm, dict):
|
||||
normalized_obj = norm.get("normalized") or data
|
||||
normalizer_usage = norm.get("_usage")
|
||||
|
||||
usage_result = record_api_usage(
|
||||
dict(task.result or {}),
|
||||
phase="extract",
|
||||
model=str(getattr(service.extractor, "model", model or "unknown")),
|
||||
usage=extraction_usage,
|
||||
radicado=radicado,
|
||||
filename=filename,
|
||||
)
|
||||
usage_result = record_api_usage(
|
||||
usage_result,
|
||||
phase="normalize",
|
||||
model=str(settings.api.normalizer_model),
|
||||
usage=normalizer_usage,
|
||||
radicado=radicado,
|
||||
filename=filename,
|
||||
)
|
||||
task.result = usage_result
|
||||
|
||||
return persist_desktop_ingest_result(
|
||||
session,
|
||||
|
||||
@@ -228,6 +228,28 @@ def serialize_task(task: Task) -> dict[str, Any]:
|
||||
|
||||
|
||||
def persist_extraction_request(request: ExtractionRequest) -> dict[str, Any]:
|
||||
from domain.usage_tracking import record_api_usage
|
||||
|
||||
usage_result: dict[str, Any] = {}
|
||||
|
||||
def capture_usage(
|
||||
phase: str,
|
||||
radicado: str,
|
||||
usage: dict[str, Any] | None,
|
||||
_cost: dict[str, Any] | None,
|
||||
) -> None:
|
||||
nonlocal usage_result
|
||||
model = request.model or settings.api.extractor_model
|
||||
if phase == "normalize":
|
||||
model = settings.api.normalizer_model
|
||||
usage_result = record_api_usage(
|
||||
usage_result,
|
||||
phase=phase,
|
||||
model=str(model),
|
||||
usage=usage,
|
||||
radicado=radicado,
|
||||
)
|
||||
|
||||
with session_scope() as session:
|
||||
batch = ExtractionBatch(
|
||||
status="running",
|
||||
@@ -252,6 +274,7 @@ def persist_extraction_request(request: ExtractionRequest) -> dict[str, Any]:
|
||||
structure_mode=request.structure_mode,
|
||||
retry_on_empty=request.retry_on_empty,
|
||||
radicado_filter=request.radicado_filter,
|
||||
on_usage=capture_usage,
|
||||
)
|
||||
|
||||
with session_scope() as session:
|
||||
@@ -279,6 +302,7 @@ def persist_extraction_request(request: ExtractionRequest) -> dict[str, Any]:
|
||||
session.flush()
|
||||
serialized = serialize_batch(batch)
|
||||
serialized["row_count"] = row_index
|
||||
serialized["api_usage"] = usage_result.get("api_usage", {})
|
||||
return serialized
|
||||
except Exception as exc:
|
||||
with session_scope() as session:
|
||||
@@ -312,6 +336,7 @@ def run_extraction_task(task_id: str) -> None:
|
||||
"extraction_batch_id": batch["id"],
|
||||
"row_count": batch.get("row_count", 0),
|
||||
"batch": batch,
|
||||
"api_usage": batch.get("api_usage", {}),
|
||||
}
|
||||
task.error = None
|
||||
task.finished_at = utcnow()
|
||||
@@ -620,6 +645,8 @@ async def _process_sync(
|
||||
from domain.profiles import get_profile
|
||||
from domain.prompts import PromptBuilder
|
||||
from domain.normalizer import normalize_to_schema
|
||||
from domain.config import settings as _settings
|
||||
from domain.usage_tracking import record_api_usage
|
||||
|
||||
_log = _logging.getLogger("extraction.ingest")
|
||||
|
||||
@@ -656,13 +683,13 @@ async def _process_sync(
|
||||
tiff_range = req.get("tiff_page_range")
|
||||
model_name = getattr(svc.extractor, "model", "?")
|
||||
_log.info("→ OpenAI [%s] radicado=%s filename=%s", model_name, radicado, effective_name)
|
||||
data, _usage = svc.extract_full_from_pdf(
|
||||
data, extraction_usage = svc.extract_full_from_pdf(
|
||||
temp_path,
|
||||
instruction=prompt,
|
||||
tiff_page_range=tiff_range,
|
||||
)
|
||||
tok_in = (_usage or {}).get("prompt_tokens", "?")
|
||||
tok_out = (_usage or {}).get("completion_tokens", "?")
|
||||
tok_in = (extraction_usage or {}).get("prompt_tokens", "?")
|
||||
tok_out = (extraction_usage or {}).get("completion_tokens", "?")
|
||||
_log.info("← OpenAI [%s] radicado=%s tokens=%s→%s", model_name, radicado, tok_in, tok_out)
|
||||
if isinstance(data, dict):
|
||||
data["radicado"] = radicado
|
||||
@@ -670,11 +697,13 @@ async def _process_sync(
|
||||
data = {"radicado": radicado, "data": data}
|
||||
norm = normalize_to_schema(data, paths=selected_paths, profile_name=prof.get("name", "invoice"))
|
||||
normalized_obj = data
|
||||
normalizer_usage = None
|
||||
if isinstance(norm, dict):
|
||||
normalized_obj = norm.get("normalized") or data
|
||||
return normalized_obj, selected_paths
|
||||
normalizer_usage = norm.get("_usage")
|
||||
return normalized_obj, selected_paths, extraction_usage, normalizer_usage, model_name
|
||||
|
||||
normalized_obj, selected_paths = await _asyncio.to_thread(_call_llm)
|
||||
normalized_obj, selected_paths, extraction_usage, normalizer_usage, model_name = await _asyncio.to_thread(_call_llm)
|
||||
|
||||
# ── Fase 2: persistir resultado (con FOR UPDATE para serializar) ───
|
||||
def _persist():
|
||||
@@ -682,6 +711,14 @@ async def _process_sync(
|
||||
task = session.get(Task, task_id, with_for_update=True)
|
||||
if task is None:
|
||||
return {"error": "task_not_found"}
|
||||
task.result = record_api_usage(
|
||||
dict(task.result or {}), phase="extract", model=str(model_name),
|
||||
usage=extraction_usage, radicado=radicado, filename=effective_name,
|
||||
)
|
||||
task.result = record_api_usage(
|
||||
dict(task.result or {}), phase="normalize", model=str(_settings.api.normalizer_model),
|
||||
usage=normalizer_usage, radicado=radicado, filename=effective_name,
|
||||
)
|
||||
payload = persist_desktop_ingest_result(
|
||||
session,
|
||||
task=task,
|
||||
|
||||
@@ -0,0 +1,90 @@
|
||||
"""Persistencia compacta del uso real de APIs LLM dentro del resultado de una tarea."""
|
||||
from __future__ import annotations
|
||||
|
||||
from datetime import datetime, timezone
|
||||
from typing import Any
|
||||
|
||||
from domain.costs import MODEL_PRICING, compute_cost_from_usage
|
||||
|
||||
|
||||
def _number(value: Any) -> float:
|
||||
try:
|
||||
return float(value or 0)
|
||||
except (TypeError, ValueError):
|
||||
return 0.0
|
||||
|
||||
|
||||
def _tokens(usage: dict[str, Any], key: str) -> int:
|
||||
try:
|
||||
return int(usage.get(key) or 0)
|
||||
except (TypeError, ValueError):
|
||||
return 0
|
||||
|
||||
|
||||
def record_api_usage(
|
||||
result: dict[str, Any],
|
||||
*,
|
||||
phase: str,
|
||||
model: str,
|
||||
usage: dict[str, Any] | None,
|
||||
radicado: str | None = None,
|
||||
filename: str | None = None,
|
||||
) -> dict[str, Any]:
|
||||
"""Agrega una llamada facturable a ``result.api_usage`` y devuelve una copia.
|
||||
|
||||
El proveedor es la fuente de los tokens. El costo se calcula con la tabla de
|
||||
precios configurada para el modelo y se conserva separado por fase y llamada.
|
||||
"""
|
||||
if not isinstance(usage, dict):
|
||||
return dict(result)
|
||||
|
||||
next_result = dict(result)
|
||||
ledger = dict(next_result.get("api_usage") or {})
|
||||
summary = dict(ledger.get("summary") or {})
|
||||
phases = dict(ledger.get("phases") or {})
|
||||
phase_summary = dict(phases.get(phase) or {})
|
||||
details = list(ledger.get("details") or [])
|
||||
|
||||
prompt_tokens = _tokens(usage, "prompt_tokens")
|
||||
completion_tokens = _tokens(usage, "completion_tokens")
|
||||
token_details = usage.get("prompt_tokens_details") or {}
|
||||
cached_tokens = _tokens(token_details, "cached_tokens") if isinstance(token_details, dict) else 0
|
||||
total_tokens = _tokens(usage, "total_tokens") or prompt_tokens + completion_tokens
|
||||
cost = compute_cost_from_usage(usage, phase=phase, model=model) or {}
|
||||
|
||||
values: dict[str, int | float] = {
|
||||
"requests": 1,
|
||||
"prompt_tokens": prompt_tokens,
|
||||
"cached_tokens": cached_tokens,
|
||||
"completion_tokens": completion_tokens,
|
||||
"total_tokens": total_tokens,
|
||||
"input_usd": _number(cost.get("input_usd")),
|
||||
"cached_input_usd": _number(cost.get("cached_input_usd")),
|
||||
"output_usd": _number(cost.get("output_usd")),
|
||||
"total_cost_usd": _number(cost.get("total_usd")),
|
||||
"unpriced_requests": 0 if model in MODEL_PRICING else 1,
|
||||
}
|
||||
for key, value in values.items():
|
||||
precision = 6 if key.endswith("usd") else None
|
||||
summary[key] = round(_number(summary.get(key)) + value, precision) if precision else int(_number(summary.get(key)) + value)
|
||||
phase_summary[key] = round(_number(phase_summary.get(key)) + value, precision) if precision else int(_number(phase_summary.get(key)) + value)
|
||||
|
||||
phase_summary["model"] = model
|
||||
phases[phase] = phase_summary
|
||||
details.append({
|
||||
"timestamp": datetime.now(timezone.utc).isoformat(),
|
||||
"phase": phase,
|
||||
"model": model,
|
||||
"radicado": radicado,
|
||||
"filename": filename,
|
||||
"prompt_tokens": prompt_tokens,
|
||||
"cached_tokens": cached_tokens,
|
||||
"completion_tokens": completion_tokens,
|
||||
"total_tokens": total_tokens,
|
||||
"cost": cost,
|
||||
"pricing_known": model in MODEL_PRICING,
|
||||
})
|
||||
|
||||
ledger.update({"summary": summary, "phases": phases, "details": details})
|
||||
next_result["api_usage"] = ledger
|
||||
return next_result
|
||||
@@ -0,0 +1,51 @@
|
||||
from domain.usage_tracking import record_api_usage
|
||||
|
||||
|
||||
def test_record_api_usage_aggregates_summary_phases_and_details():
|
||||
result = record_api_usage(
|
||||
{},
|
||||
phase="extract",
|
||||
model="gpt-5.6-luna",
|
||||
usage={
|
||||
"prompt_tokens": 1_000,
|
||||
"completion_tokens": 200,
|
||||
"total_tokens": 1_200,
|
||||
"prompt_tokens_details": {"cached_tokens": 100},
|
||||
},
|
||||
radicado="RAD-1",
|
||||
filename="factura.pdf",
|
||||
)
|
||||
result = record_api_usage(
|
||||
result,
|
||||
phase="extract",
|
||||
model="gpt-5.6-luna",
|
||||
usage={"prompt_tokens": 500, "completion_tokens": 50},
|
||||
radicado="RAD-2",
|
||||
)
|
||||
|
||||
ledger = result["api_usage"]
|
||||
assert ledger["summary"]["requests"] == 2
|
||||
assert ledger["summary"]["prompt_tokens"] == 1_500
|
||||
assert ledger["summary"]["completion_tokens"] == 250
|
||||
assert ledger["summary"]["total_tokens"] == 1_750
|
||||
assert ledger["summary"]["unpriced_requests"] == 0
|
||||
assert ledger["phases"]["extract"]["requests"] == 2
|
||||
assert ledger["details"][0]["filename"] == "factura.pdf"
|
||||
assert ledger["details"][0]["pricing_known"] is True
|
||||
|
||||
|
||||
def test_record_api_usage_marks_unknown_model_as_unpriced():
|
||||
result = record_api_usage(
|
||||
{},
|
||||
phase="extract",
|
||||
model="provider/model-without-price",
|
||||
usage={"prompt_tokens": 100, "completion_tokens": 10},
|
||||
)
|
||||
|
||||
assert result["api_usage"]["summary"]["unpriced_requests"] == 1
|
||||
assert result["api_usage"]["details"][0]["pricing_known"] is False
|
||||
|
||||
|
||||
def test_record_api_usage_ignores_missing_usage():
|
||||
original = {"processed": 1}
|
||||
assert record_api_usage(original, phase="extract", model="gpt-5.6-luna", usage=None) == original
|
||||
@@ -24,6 +24,8 @@ from domain.profiles import get_profile
|
||||
from domain.prompts import PromptBuilder
|
||||
from domain.normalizer import normalize_to_schema
|
||||
from domain.rate_limiter import get_rate_limiter
|
||||
from domain.config import settings
|
||||
from domain.usage_tracking import record_api_usage
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -131,8 +133,8 @@ async def process_document(
|
||||
await rate_limiter.acquire(estimated_tokens=estimated_tokens)
|
||||
|
||||
# Extracción + normalización (CPU + IO bound, en thread pool)
|
||||
def _extract_and_normalize() -> tuple[dict, dict]:
|
||||
data, _usage = service.extract_full_from_pdf(
|
||||
def _extract_and_normalize() -> tuple[dict, dict, dict | None, dict | None]:
|
||||
data, extraction_usage = service.extract_full_from_pdf(
|
||||
queue_path,
|
||||
instruction=prompt,
|
||||
tiff_page_range=tiff_range,
|
||||
@@ -149,11 +151,13 @@ async def process_document(
|
||||
profile_name=prof.get("name", "invoice"),
|
||||
)
|
||||
normalized_obj = data
|
||||
normalizer_usage = None
|
||||
if isinstance(norm, dict):
|
||||
normalized_obj = norm.get("normalized") or data
|
||||
return raw_data, normalized_obj
|
||||
normalizer_usage = norm.get("_usage")
|
||||
return raw_data, normalized_obj, extraction_usage, normalizer_usage
|
||||
|
||||
_raw, normalized_obj = await asyncio.to_thread(_extract_and_normalize)
|
||||
_raw, normalized_obj, extraction_usage, normalizer_usage = await asyncio.to_thread(_extract_and_normalize)
|
||||
|
||||
# Persistir resultado en BD
|
||||
def _persist() -> dict[str, Any]:
|
||||
@@ -164,6 +168,15 @@ async def process_document(
|
||||
return {"skipped": True, "reason": "task_not_found"}
|
||||
|
||||
result = dict(task.result or {})
|
||||
result = record_api_usage(
|
||||
result, phase="extract", model=str(effective_model), usage=extraction_usage,
|
||||
radicado=radicado, filename=filename,
|
||||
)
|
||||
result = record_api_usage(
|
||||
result, phase="normalize", model=str(settings.api.normalizer_model), usage=normalizer_usage,
|
||||
radicado=radicado, filename=filename,
|
||||
)
|
||||
task.result = result
|
||||
|
||||
if structure_mode == "radicados" and doc_types and doc_type:
|
||||
partials = dict(result.get("partials") or {})
|
||||
|
||||
Reference in New Issue
Block a user