@heripo/document-processor 0.1.12 → 0.1.13
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.cjs +9 -18
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.cts +2 -8
- package/dist/index.d.ts +2 -8
- package/dist/index.js +9 -18
- package/dist/index.js.map +1 -1
- package/package.json +6 -6
package/dist/index.d.cts
CHANGED
|
@@ -412,7 +412,8 @@ interface VisionLLMComponentOptions extends BaseLLMComponentOptions {
|
|
|
412
412
|
*/
|
|
413
413
|
interface ImageContent {
|
|
414
414
|
type: 'image';
|
|
415
|
-
image:
|
|
415
|
+
image: Uint8Array;
|
|
416
|
+
mediaType: string;
|
|
416
417
|
}
|
|
417
418
|
/**
|
|
418
419
|
* Abstract base class for vision-based LLM components
|
|
@@ -441,13 +442,6 @@ declare abstract class VisionLLMComponent extends BaseLLMComponent {
|
|
|
441
442
|
output: z.infer<TSchema>;
|
|
442
443
|
usage: ExtendedTokenUsage;
|
|
443
444
|
}>;
|
|
444
|
-
/**
|
|
445
|
-
* Load an image file and encode it as base64
|
|
446
|
-
*
|
|
447
|
-
* @param imagePath - Absolute path to the image file
|
|
448
|
-
* @returns Base64 encoded image string
|
|
449
|
-
*/
|
|
450
|
-
protected loadImageAsBase64(imagePath: string): string;
|
|
451
445
|
/**
|
|
452
446
|
* Build image content object for vision LLM messages
|
|
453
447
|
*
|
package/dist/index.d.ts
CHANGED
|
@@ -412,7 +412,8 @@ interface VisionLLMComponentOptions extends BaseLLMComponentOptions {
|
|
|
412
412
|
*/
|
|
413
413
|
interface ImageContent {
|
|
414
414
|
type: 'image';
|
|
415
|
-
image:
|
|
415
|
+
image: Uint8Array;
|
|
416
|
+
mediaType: string;
|
|
416
417
|
}
|
|
417
418
|
/**
|
|
418
419
|
* Abstract base class for vision-based LLM components
|
|
@@ -441,13 +442,6 @@ declare abstract class VisionLLMComponent extends BaseLLMComponent {
|
|
|
441
442
|
output: z.infer<TSchema>;
|
|
442
443
|
usage: ExtendedTokenUsage;
|
|
443
444
|
}>;
|
|
444
|
-
/**
|
|
445
|
-
* Load an image file and encode it as base64
|
|
446
|
-
*
|
|
447
|
-
* @param imagePath - Absolute path to the image file
|
|
448
|
-
* @returns Base64 encoded image string
|
|
449
|
-
*/
|
|
450
|
-
protected loadImageAsBase64(imagePath: string): string;
|
|
451
445
|
/**
|
|
452
446
|
* Build image content object for vision LLM messages
|
|
453
447
|
*
|
package/dist/index.js
CHANGED
|
@@ -2354,16 +2354,6 @@ var VisionLLMComponent = class extends BaseLLMComponent {
|
|
|
2354
2354
|
usage: result.usage
|
|
2355
2355
|
};
|
|
2356
2356
|
}
|
|
2357
|
-
/**
|
|
2358
|
-
* Load an image file and encode it as base64
|
|
2359
|
-
*
|
|
2360
|
-
* @param imagePath - Absolute path to the image file
|
|
2361
|
-
* @returns Base64 encoded image string
|
|
2362
|
-
*/
|
|
2363
|
-
loadImageAsBase64(imagePath) {
|
|
2364
|
-
const imageBuffer = fs.readFileSync(imagePath);
|
|
2365
|
-
return imageBuffer.toString("base64");
|
|
2366
|
-
}
|
|
2367
2357
|
/**
|
|
2368
2358
|
* Build image content object for vision LLM messages
|
|
2369
2359
|
*
|
|
@@ -2373,10 +2363,11 @@ var VisionLLMComponent = class extends BaseLLMComponent {
|
|
|
2373
2363
|
*/
|
|
2374
2364
|
buildImageContent(imagePath, mimeType = "image/png") {
|
|
2375
2365
|
const absolutePath = path.isAbsolute(imagePath) ? imagePath : path.resolve(this.outputPath, imagePath);
|
|
2376
|
-
const
|
|
2366
|
+
const imageData = new Uint8Array(fs.readFileSync(absolutePath));
|
|
2377
2367
|
return {
|
|
2378
2368
|
type: "image",
|
|
2379
|
-
image:
|
|
2369
|
+
image: imageData,
|
|
2370
|
+
mediaType: mimeType
|
|
2380
2371
|
};
|
|
2381
2372
|
}
|
|
2382
2373
|
};
|
|
@@ -2531,11 +2522,11 @@ var VisionTocExtractor = class extends VisionLLMComponent {
|
|
|
2531
2522
|
this.outputPath,
|
|
2532
2523
|
`pages/page_${pageNo - 1}.png`
|
|
2533
2524
|
);
|
|
2534
|
-
const
|
|
2535
|
-
const base64Image = imageBuffer.toString("base64");
|
|
2525
|
+
const imageData = new Uint8Array(fs2.readFileSync(imagePath));
|
|
2536
2526
|
imageContents.push({
|
|
2537
2527
|
type: "image",
|
|
2538
|
-
image:
|
|
2528
|
+
image: imageData,
|
|
2529
|
+
mediaType: "image/png"
|
|
2539
2530
|
});
|
|
2540
2531
|
}
|
|
2541
2532
|
return imageContents;
|
|
@@ -3113,12 +3104,12 @@ var PageRangeParser = class extends VisionLLMComponent {
|
|
|
3113
3104
|
this.outputPath,
|
|
3114
3105
|
`pages/page_${pageNo - 1}.png`
|
|
3115
3106
|
);
|
|
3116
|
-
const
|
|
3117
|
-
const base64Image = imageBuffer.toString("base64");
|
|
3107
|
+
const imageData = new Uint8Array(fs3.readFileSync(imagePath));
|
|
3118
3108
|
const mimeType = page.image.mimetype || "image/png";
|
|
3119
3109
|
imageContents.push({
|
|
3120
3110
|
type: "image",
|
|
3121
|
-
image:
|
|
3111
|
+
image: imageData,
|
|
3112
|
+
mediaType: mimeType
|
|
3122
3113
|
});
|
|
3123
3114
|
}
|
|
3124
3115
|
const schema = z4.object({
|