mirror/papra
1
0
Fork 0
mirror of https://github.com/papra-hq/papra.git synced 2025-12-30 08:59:39 -06:00
Code Issues Packages Projects Releases 80 Wiki Activity

feat(extractors): add logger support to text extraction functions (#627)

Browse Source
This commit is contained in:
Corentin Thomasset
2025-11-13 18:41:22 +01:00
committed by GitHub
parent 11ad13058e
commit 5ccdf446f0
4 changed files with 32 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
.changeset/tame-jobs-jump.md Normal file
View File

@@ -0,0 +1,5 @@
---
"@papra/lecture": patch
---
Added logger parameter

2
apps/papra-server/src/modules/documents/documents.services.ts
View File

@@ -18,7 +18,7 @@ export async function extractDocumentText({
ocrLanguages?: string[];
logger?: Logger;
}) {
const { textContent, error, extractorName, extractorType } = await extractTextFromFile({ file, config: { tesseract: { languages: ocrLanguages } } });
const { textContent, error, extractorName, extractorType } = await extractTextFromFile({ file, config: { tesseract: { languages: ocrLanguages } }, logger });
if (error) {
logger.error({ error, extractorName, extractorType }, 'Error while extracting text from document');

28
packages/lecture/src/extractors.usecases.ts
View File

@@ -1,8 +1,8 @@
import type { PartialExtractorConfig } from './types';
import type { Logger, PartialExtractorConfig } from './types';
import { parseConfig } from './config';
import { getExtractor } from './extractors.registry';
export async function extractText({ arrayBuffer, mimeType, config: rawConfig }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: PartialExtractorConfig }): Promise<{
export async function extractText({ arrayBuffer, mimeType, config: rawConfig, logger }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: PartialExtractorConfig; logger?: Logger }): Promise<{
extractorName: string | undefined;
extractorType: string | undefined;
textContent: string | undefined;
@@ -13,6 +13,8 @@ export async function extractText({ arrayBuffer, mimeType, config: rawConfig }:
const { extractor } = getExtractor({ mimeType });
if (!extractor) {
logger?.warn({ mimeType }, 'No extractor found');
return {
extractorName: undefined,
extractorType: undefined,
@@ -21,19 +23,27 @@ export async function extractText({ arrayBuffer, mimeType, config: rawConfig }:
};
}
const extractorName = extractor.name;
try {
logger?.debug({ extractorName, mimeType }, 'Starting extraction');
const startTime = Date.now();
const { content, subExtractorsUsed } = await extractor.extract({ arrayBuffer, config });
const duration = Date.now() - startTime;
const extractorType = [extractorName, ...subExtractorsUsed ?? []].join(':');
logger?.info({ extractorName, extractorType, mimeType, durationMs: duration }, 'Extraction completed');
return {
extractorName: extractor.name,
extractorType: [extractor.name, ...subExtractorsUsed ?? []].join(':'),
extractorName,
extractorType,
textContent: content,
subExtractorsUsed,
};
} catch (error) {
return {
error,
extractorName: extractor.name,
extractorName,
extractorType: undefined,
textContent: undefined,
subExtractorsUsed: [],
@@ -41,13 +51,13 @@ export async function extractText({ arrayBuffer, mimeType, config: rawConfig }:
}
}
export async function extractTextFromBlob({ blob, config }: { blob: Blob; config?: PartialExtractorConfig }) {
export async function extractTextFromBlob({ blob, ...rest }: { blob: Blob; config?: PartialExtractorConfig; logger?: Logger }) {
const arrayBuffer = await blob.arrayBuffer();
const mimeType = blob.type;
return extractText({ arrayBuffer, mimeType, config });
return extractText({ arrayBuffer, mimeType, ...rest });
}
export async function extractTextFromFile({ file, config }: { file: File; config?: PartialExtractorConfig }) {
return extractTextFromBlob({ blob: file, config });
export async function extractTextFromFile({ file, ...rest }: { file: File; config?: PartialExtractorConfig; logger?: Logger }) {
return extractTextFromBlob({ blob: file, ...rest });
}

7
packages/lecture/src/types.ts
View File

@@ -9,3 +9,10 @@ export type ExtractorConfig = {
};
export type PartialExtractorConfig = undefined | DeepPartial<ExtractorConfig>;
export type Logger = {
debug: (...args: [data: Record<string, unknown>, message: string] | [message: string]) => void;
info: (...args: [data: Record<string, unknown>, message: string] | [message: string]) => void;
warn: (...args: [data: Record<string, unknown>, message: string] | [message: string]) => void;
error: (...args: [data: Record<string, unknown>, message: string] | [message: string]) => void;
};