correlate with eid

This commit is contained in:
Gergő Móricz 2025-04-15 23:06:13 -07:00
parent 512a2b1cd4
commit 80b507e64e
5 changed files with 15 additions and 4 deletions

View File

@ -22,6 +22,7 @@ type BatchExtractOptions = {
systemPrompt: string;
doc: Document;
useAgent: boolean;
extractId?: string;
};
/**
@ -44,7 +45,7 @@ export async function batchExtractPromise(options: BatchExtractOptions, logger:
smartScrapeCallCount: number;
otherCallCount: number;
}> {
const { multiEntitySchema, links, prompt, systemPrompt, doc, useAgent } = options;
const { multiEntitySchema, links, prompt, systemPrompt, doc, useAgent, extractId } = options;
const generationOptions: GenerateCompletionsOptions = {
@ -74,6 +75,7 @@ export async function batchExtractPromise(options: BatchExtractOptions, logger:
extractOptions: generationOptions,
urls: [doc.metadata.sourceURL || doc.metadata.url || ""],
useAgent,
extractId,
});
extractedDataArray = e;
warning = w;

View File

@ -14,7 +14,8 @@ export async function singleAnswerCompletion({
links,
prompt,
systemPrompt,
useAgent
useAgent,
extractId,
}: {
singleAnswerDocs: Document[];
rSchema: any;
@ -22,6 +23,7 @@ export async function singleAnswerCompletion({
prompt: string;
systemPrompt: string;
useAgent: boolean;
extractId?: string;
}): Promise<{
extract: any;
tokenUsage: TokenUsage;
@ -51,6 +53,7 @@ export async function singleAnswerCompletion({
extractOptions: generationOptions,
urls: singleAnswerDocs.map(doc => doc.metadata.url || doc.metadata.sourceURL || ""),
useAgent,
extractId,
});
const completion = {

View File

@ -430,7 +430,8 @@ export async function performExtraction(
prompt: request.prompt ?? "",
systemPrompt: request.systemPrompt ?? "",
doc,
useAgent: isAgentExtractModelValid(request.agent?.model)
useAgent: isAgentExtractModelValid(request.agent?.model),
extractId,
}, logger);
// Race between timeout and completion
@ -741,6 +742,7 @@ export async function performExtraction(
prompt: request.prompt ?? "",
systemPrompt: request.systemPrompt ?? "",
useAgent: isAgentExtractModelValid(request.agent?.model),
extractId,
});
costTracking.smartScrapeCost += singleAnswerSmartScrapeCost;
costTracking.smartScrapeCallCount += singleAnswerSmartScrapeCallCount;

View File

@ -184,10 +184,12 @@ export async function extractData({
extractOptions,
urls,
useAgent,
extractId,
}: {
extractOptions: GenerateCompletionsOptions;
urls: string[];
useAgent: boolean;
extractId?: string;
}): Promise<{
extractedDataArray: any[];
warning: any;
@ -273,7 +275,7 @@ export async function extractData({
let smartscrapeResults: SmartScrapeResult[];
if (isSingleUrl) {
smartscrapeResults = [
await smartScrape(urls[0], extract?.smartscrape_prompt),
await smartScrape(urls[0], extract?.smartscrape_prompt, extractId),
];
smartScrapeCost += smartscrapeResults[0].tokenUsage;
smartScrapeCallCount++;
@ -285,6 +287,7 @@ export async function extractData({
return await smartScrape(
urls[page.page_index],
page.smartscrape_prompt,
extractId,
);
}),
);

View File

@ -49,6 +49,7 @@ export async function smartScrape(
url: string,
prompt: string,
sessionId?: string,
extractId?: string,
): Promise<SmartScrapeResult> {
try {
logger.info("Initiating smart scrape request", { url, prompt, sessionId });