whekin/household-bot
1
0
Fork 0
mirror of https://github.com/whekin/household-bot.git synced 2026-04-01 03:14:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(bot): improve calculated purchase confirmation flow

Browse Source
This commit is contained in:
whekin
2026-03-12 15:35:02 +04:00
parent 995725f121
commit 014d791bdc
8 changed files with 708 additions and 70 deletions
Download Patch File Download Diff File Expand all files Collapse all files

119
apps/bot/src/openai-purchase-interpreter.test.ts
View File

@@ -67,6 +67,8 @@ describe('createOpenAiPurchaseInterpreter', () => {
amountMinor: 100000n,
currency: 'GEL',
itemDescription: 'армянская золотая швабра',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 93,
parserMode: 'llm',
clarificationQuestion: null
@@ -104,6 +106,8 @@ describe('createOpenAiPurchaseInterpreter', () => {
amountMinor: 1000n,
currency: 'GEL',
itemDescription: 'сухари',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 88,
parserMode: 'llm',
clarificationQuestion: null
@@ -148,6 +152,8 @@ describe('createOpenAiPurchaseInterpreter', () => {
amountMinor: 5000n,
currency: 'GEL',
itemDescription: 'шампунь',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 92,
parserMode: 'llm',
clarificationQuestion: null
@@ -192,6 +198,8 @@ describe('createOpenAiPurchaseInterpreter', () => {
amountMinor: 4500n,
currency: 'GEL',
itemDescription: 'сосисоны',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 85,
parserMode: 'llm',
clarificationQuestion: null
@@ -201,7 +209,7 @@ describe('createOpenAiPurchaseInterpreter', () => {
}
})
test('corrects mis-scaled amountMinor when the source text contains a clear money amount', async () => {
test('keeps the llm provided amountMinor without local correction', async () => {
const interpreter = createOpenAiPurchaseInterpreter('test-key', 'gpt-4o-mini')
expect(interpreter).toBeDefined()
@@ -236,9 +244,11 @@ describe('createOpenAiPurchaseInterpreter', () => {
expect(result).toEqual<PurchaseInterpretation>({
decision: 'purchase',
amountMinor: 35000n,
amountMinor: 350n,
currency: 'GEL',
itemDescription: 'обои, 100 рулонов',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 86,
parserMode: 'llm',
clarificationQuestion: null
@@ -248,7 +258,7 @@ describe('createOpenAiPurchaseInterpreter', () => {
}
})
test('corrects mis-scaled amountMinor for simple clarification replies', async () => {
test('keeps llm provided amountMinor for clarification follow-ups without local correction', async () => {
const interpreter = createOpenAiPurchaseInterpreter('test-key', 'gpt-4o-mini')
expect(interpreter).toBeDefined()
@@ -283,9 +293,11 @@ describe('createOpenAiPurchaseInterpreter', () => {
expect(result).toEqual<PurchaseInterpretation>({
decision: 'purchase',
amountMinor: 35000n,
amountMinor: 350n,
currency: 'GEL',
itemDescription: 'Рулоны обоев',
amountSource: 'explicit',
calculationExplanation: null,
confidence: 89,
parserMode: 'llm',
clarificationQuestion: null
@@ -294,4 +306,103 @@ describe('createOpenAiPurchaseInterpreter', () => {
globalThis.fetch = originalFetch
}
})
test('preserves llm computed totals for quantity times unit-price purchases', async () => {
const interpreter = createOpenAiPurchaseInterpreter('test-key', 'gpt-5-mini')
expect(interpreter).toBeDefined()
const originalFetch = globalThis.fetch
globalThis.fetch = (async () =>
successfulResponse({
output: [
{
content: [
{
text: JSON.stringify({
decision: 'purchase',
amountMinor: '3000',
currency: 'GEL',
itemDescription: 'бутылки воды',
amountSource: 'calculated',
calculationExplanation: '5 × 6 лари = 30 лари',
confidence: 94,
clarificationQuestion: null
})
}
]
}
]
})) as unknown as typeof fetch
try {
const result = await interpreter!('Купил 5 бутылок воды, 6 лари за бутылку', {
defaultCurrency: 'GEL'
})
expect(result).toEqual<PurchaseInterpretation>({
decision: 'purchase',
amountMinor: 3000n,
currency: 'GEL',
itemDescription: 'бутылки воды',
amountSource: 'calculated',
calculationExplanation: '5 × 6 лари = 30 лари',
confidence: 94,
parserMode: 'llm',
clarificationQuestion: null
})
} finally {
globalThis.fetch = originalFetch
}
})
test('tells the llm to total per-item pricing and accept colloquial completed purchase phrasing', async () => {
const interpreter = createOpenAiPurchaseInterpreter('test-key', 'gpt-5-mini')
expect(interpreter).toBeDefined()
const originalFetch = globalThis.fetch
let requestBody: unknown = null
globalThis.fetch = (async (_url: unknown, init?: RequestInit) => {
requestBody = init?.body ? JSON.parse(String(init.body)) : null
return successfulResponse({
output: [
{
content: [
{
text: JSON.stringify({
decision: 'purchase',
amountMinor: '3000',
currency: 'GEL',
itemDescription: 'бутылки воды',
confidence: 94,
clarificationQuestion: null
})
}
]
}
]
})
}) as unknown as typeof fetch
try {
await interpreter!('Купил 5 бутылок воды, 6 лари за бутылку', {
defaultCurrency: 'GEL'
})
const systemMessage =
(
(requestBody as { input?: Array<{ role?: string; content?: string }> | null })?.input ??
[]
).find((entry) => entry.role === 'system')?.content ?? ''
expect(systemMessage).toContain(
'If the user gives quantity and per-item price, compute the total spend and return that total in amountMinor.'
)
expect(systemMessage).toContain(
'Treat colloquial completed-buy phrasing like "взял", "сходил и взял", or "сторговался до X" as a completed purchase when the message reports a real buy fact.'
)
} finally {
globalThis.fetch = originalFetch
}
})
})