unpdf 0.1.1 → 0.2.1

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.
Files changed (12) hide show
  1. package/README.md +28 -1
  2. package/dist/index.node.cjs +34 -11
  3. package/dist/index.node.d.cts +3 -1
  4. package/dist/index.node.d.mts +3 -1
  5. package/dist/index.node.d.ts +3 -1
  6. package/dist/index.node.mjs +34 -12
  7. package/dist/index.web.cjs +4 -0
  8. package/dist/index.web.d.cts +3 -2
  9. package/dist/index.web.d.mts +3 -2
  10. package/dist/index.web.d.ts +3 -2
  11. package/dist/index.web.mjs +4 -1
  12. package/package.json +1 -1
package/README.md CHANGED
@@ -6,7 +6,7 @@ A collection of utilities to work with PDFs.
6
6
 
7
7
  **Why this package then?**
8
8
 
9
- - To circumvent build issues in serverless environments, where the `canvas` package is used by `PDF.js` not supported.
9
+ - To circumvent build issues in serverless environments, where the `canvas` package used by `PDF.js` is not supported.
10
10
  - WIP and more to come.
11
11
 
12
12
  ## Installation
@@ -36,6 +36,33 @@ const { totalPages, info, metadata, text } = await decodePDFText(
36
36
  )
37
37
  ```
38
38
 
39
+ ## Methods
40
+
41
+ ### `decodePDFText`
42
+
43
+ ```ts
44
+ interface PDFContent {
45
+ totalPages: number
46
+ info?: Record<string, any>
47
+ metadata?: any
48
+ text: string | string[]
49
+ }
50
+
51
+ function decodePDFText(
52
+ data: ArrayBuffer,
53
+ { mergePages }?: { mergePages?: boolean }
54
+ ): Promise<PDFContent>
55
+ ```
56
+
57
+ ### `getImagesFromPage`
58
+
59
+ ```ts
60
+ function getImagesFromPage(
61
+ data: ArrayBuffer,
62
+ pageNumber: number
63
+ ): Promise<ArrayBuffer[]>
64
+ ```
65
+
39
66
  ## License
40
67
 
41
68
  [MIT](./LICENSE) License © 2023-PRESENT [Johann Schopplich](https://github.com/johannschopplich)
package/dist/index.node.cjs CHANGED
@@ -1,6 +1,8 @@
1
1
  'use strict';
2
2
 
3
- async function decodePDFText(data, { mergePages = false } = {}) {
3
+ const pdfjsDist = require('pdfjs-dist');
4
+
5
+ async function getDocumentProxy(data) {
4
6
  const { getDocument } = await getPDFJSImports();
5
7
  const pdf = await getDocument({
6
8
  data,
@@ -8,6 +10,23 @@ async function decodePDFText(data, { mergePages = false } = {}) {
8
10
  useSystemFonts: true,
9
11
  isEvalSupported: false
10
12
  }).promise;
13
+ return pdf;
14
+ }
15
+ async function getPDFJSImports() {
16
+ try {
17
+ const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
18
+ const { getDocument, version } = mod;
19
+ return { getDocument, version };
20
+ } catch (e) {
21
+ console.error(e);
22
+ throw new Error(
23
+ "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
24
+ );
25
+ }
26
+ }
27
+
28
+ async function decodePDFText(data, { mergePages = false } = {}) {
29
+ const pdf = await getDocumentProxy(data);
11
30
  const meta = await pdf.getMetadata().catch(() => null);
12
31
  const texts = await Promise.all(
13
32
  Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
@@ -28,17 +47,21 @@ async function getPageText(pdf, pageNumber) {
28
47
  return;
29
48
  return items.map(({ str }) => str).join("\n");
30
49
  }
31
- async function getPDFJSImports() {
32
- try {
33
- const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
34
- const { getDocument, version } = mod;
35
- return { getDocument, version };
36
- } catch (e) {
37
- console.error(e);
38
- throw new Error(
39
- "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
40
- );
50
+
51
+ async function getImagesFromPage(data, pageNumber) {
52
+ const pdf = await getDocumentProxy(data);
53
+ const page = await pdf.getPage(pageNumber);
54
+ const operatorList = await page.getOperatorList();
55
+ const images = [];
56
+ for (const op of operatorList.fnArray) {
57
+ if (op !== pdfjsDist.OPS.paintImageXObject)
58
+ continue;
59
+ const image = await page.objs.get(operatorList.argsArray[op][0]);
60
+ if (image.data)
61
+ images.push(image.data.buffer);
41
62
  }
63
+ return images;
42
64
  }
43
65
 
44
66
  exports.decodePDFText = decodePDFText;
67
+ exports.getImagesFromPage = getImagesFromPage;
package/dist/index.node.d.cts CHANGED
@@ -9,4 +9,6 @@ declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
9
9
  mergePages?: boolean;
10
10
  }): Promise<PDFContent>;
11
11
 
12
- export { decodePDFText };
12
+ declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
13
+
14
+ export { decodePDFText, getImagesFromPage };
package/dist/index.node.d.mts CHANGED
@@ -9,4 +9,6 @@ declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
9
9
  mergePages?: boolean;
10
10
  }): Promise<PDFContent>;
11
11
 
12
- export { decodePDFText };
12
+ declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
13
+
14
+ export { decodePDFText, getImagesFromPage };
package/dist/index.node.d.ts CHANGED
@@ -9,4 +9,6 @@ declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
9
9
  mergePages?: boolean;
10
10
  }): Promise<PDFContent>;
11
11
 
12
- export { decodePDFText };
12
+ declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
13
+
14
+ export { decodePDFText, getImagesFromPage };
package/dist/index.node.mjs CHANGED
@@ -1,4 +1,6 @@
1
- async function decodePDFText(data, { mergePages = false } = {}) {
1
+ import { OPS } from 'pdfjs-dist';
2
+
3
+ async function getDocumentProxy(data) {
2
4
  const { getDocument } = await getPDFJSImports();
3
5
  const pdf = await getDocument({
4
6
  data,
@@ -6,6 +8,23 @@ async function decodePDFText(data, { mergePages = false } = {}) {
6
8
  useSystemFonts: true,
7
9
  isEvalSupported: false
8
10
  }).promise;
11
+ return pdf;
12
+ }
13
+ async function getPDFJSImports() {
14
+ try {
15
+ const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
16
+ const { getDocument, version } = mod;
17
+ return { getDocument, version };
18
+ } catch (e) {
19
+ console.error(e);
20
+ throw new Error(
21
+ "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
22
+ );
23
+ }
24
+ }
25
+
26
+ async function decodePDFText(data, { mergePages = false } = {}) {
27
+ const pdf = await getDocumentProxy(data);
9
28
  const meta = await pdf.getMetadata().catch(() => null);
10
29
  const texts = await Promise.all(
11
30
  Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
@@ -26,17 +45,20 @@ async function getPageText(pdf, pageNumber) {
26
45
  return;
27
46
  return items.map(({ str }) => str).join("\n");
28
47
  }
29
- async function getPDFJSImports() {
30
- try {
31
- const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
32
- const { getDocument, version } = mod;
33
- return { getDocument, version };
34
- } catch (e) {
35
- console.error(e);
36
- throw new Error(
37
- "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
38
- );
48
+
49
+ async function getImagesFromPage(data, pageNumber) {
50
+ const pdf = await getDocumentProxy(data);
51
+ const page = await pdf.getPage(pageNumber);
52
+ const operatorList = await page.getOperatorList();
53
+ const images = [];
54
+ for (const op of operatorList.fnArray) {
55
+ if (op !== OPS.paintImageXObject)
56
+ continue;
57
+ const image = await page.objs.get(operatorList.argsArray[op][0]);
58
+ if (image.data)
59
+ images.push(image.data.buffer);
39
60
  }
61
+ return images;
40
62
  }
41
63
 
42
- export { decodePDFText };
64
+ export { decodePDFText, getImagesFromPage };
package/dist/index.web.cjs CHANGED
@@ -3,5 +3,9 @@
3
3
  const decodePDFText = async () => {
4
4
  throw new Error("Not supported in browser context yet");
5
5
  };
6
+ const getImagesFromPage = async () => {
7
+ throw new Error("Not supported in browser context yet");
8
+ };
6
9
 
7
10
  exports.decodePDFText = decodePDFText;
11
+ exports.getImagesFromPage = getImagesFromPage;
package/dist/index.web.d.cts CHANGED
@@ -1,5 +1,6 @@
1
- import { decodePDFText as decodePDFText$1 } from './index.node.cjs';
1
+ import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.cjs';
2
2
 
3
3
  declare const decodePDFText: typeof decodePDFText$1;
4
+ declare const getImagesFromPage: typeof getImagesFromPage$1;
4
5
 
5
- export { decodePDFText };
6
+ export { decodePDFText, getImagesFromPage };
package/dist/index.web.d.mts CHANGED
@@ -1,5 +1,6 @@
1
- import { decodePDFText as decodePDFText$1 } from './index.node.mjs';
1
+ import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.mjs';
2
2
 
3
3
  declare const decodePDFText: typeof decodePDFText$1;
4
+ declare const getImagesFromPage: typeof getImagesFromPage$1;
4
5
 
5
- export { decodePDFText };
6
+ export { decodePDFText, getImagesFromPage };
package/dist/index.web.d.ts CHANGED
@@ -1,5 +1,6 @@
1
- import { decodePDFText as decodePDFText$1 } from './index.node.js';
1
+ import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.js';
2
2
 
3
3
  declare const decodePDFText: typeof decodePDFText$1;
4
+ declare const getImagesFromPage: typeof getImagesFromPage$1;
4
5
 
5
- export { decodePDFText };
6
+ export { decodePDFText, getImagesFromPage };
package/dist/index.web.mjs CHANGED
@@ -1,5 +1,8 @@
1
1
  const decodePDFText = async () => {
2
2
  throw new Error("Not supported in browser context yet");
3
3
  };
4
+ const getImagesFromPage = async () => {
5
+ throw new Error("Not supported in browser context yet");
6
+ };
4
7
 
5
- export { decodePDFText };
8
+ export { decodePDFText, getImagesFromPage };
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "unpdf",
3
3
  "type": "module",
4
- "version": "0.1.1",
4
+ "version": "0.2.1",
5
5
  "packageManager": "pnpm@8.6.12",
6
6
  "description": "Utilities to work with PDFs",
7
7
  "author": "Johann Schopplich <pkg@johannschopplich.com>",