From 5ccdf446f0603467ab8f8833110efbe197bc0f0f Mon Sep 17 00:00:00 2001
From: Corentin Thomasset <corentin.thomasset74@gmail.com>
Date: Thu, 13 Nov 2025 18:41:22 +0100
Subject: [PATCH] feat(extractors): add logger support to text extraction
 functions (#627)

---
 .changeset/tame-jobs-jump.md                  |  5 ++++
 .../modules/documents/documents.services.ts   |  2 +-
 packages/lecture/src/extractors.usecases.ts   | 28 +++++++++++++------
 packages/lecture/src/types.ts                 |  7 +++++
 4 files changed, 32 insertions(+), 10 deletions(-)
 create mode 100644 .changeset/tame-jobs-jump.md

diff --git a/.changeset/tame-jobs-jump.md b/.changeset/tame-jobs-jump.md
new file mode 100644
index 0000000..8dc555e
--- /dev/null
+++ b/.changeset/tame-jobs-jump.md
@@ -0,0 +1,5 @@
+---
+"@papra/lecture": patch
+---
+
+Added logger parameter
diff --git a/apps/papra-server/src/modules/documents/documents.services.ts b/apps/papra-server/src/modules/documents/documents.services.ts
index b2748dc..727ad48 100644
--- a/apps/papra-server/src/modules/documents/documents.services.ts
+++ b/apps/papra-server/src/modules/documents/documents.services.ts
@@ -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');
diff --git a/packages/lecture/src/extractors.usecases.ts b/packages/lecture/src/extractors.usecases.ts
index 8f180ea..28db39f 100644
--- a/packages/lecture/src/extractors.usecases.ts
+++ b/packages/lecture/src/extractors.usecases.ts
@@ -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 });
 }
diff --git a/packages/lecture/src/types.ts b/packages/lecture/src/types.ts
index 03ac6cb..d44056f 100644
--- a/packages/lecture/src/types.ts
+++ b/packages/lecture/src/types.ts
@@ -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;
+};