unpdf 0.10.1 → 0.11.0

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 (59) hide show
  1. package/README.md +9 -7
  2. package/dist/index.cjs +6 -12
  3. package/dist/index.d.cts +32 -13
  4. package/dist/index.d.mts +32 -13
  5. package/dist/index.d.ts +32 -13
  6. package/dist/index.mjs +6 -12
  7. package/dist/pdfjs.d.mts +1 -0
  8. package/dist/pdfjs.d.ts +1 -0
  9. package/dist/pdfjs.mjs +1 -1
  10. package/dist/types/src/display/annotation_layer.d.ts +10 -3
  11. package/dist/types/src/display/annotation_storage.d.ts +1 -0
  12. package/dist/types/src/display/api.d.ts +50 -22
  13. package/dist/types/src/display/base_factory.d.ts +3 -1
  14. package/dist/types/src/display/canvas.d.ts +2 -0
  15. package/dist/types/src/display/display_utils.d.ts +4 -3
  16. package/dist/types/src/display/draw_layer.d.ts +12 -14
  17. package/dist/types/src/display/editor/alt_text.d.ts +22 -0
  18. package/dist/types/src/display/editor/annotation_editor_layer.d.ts +24 -2
  19. package/dist/types/src/display/editor/color_picker.d.ts +19 -0
  20. package/dist/types/src/display/editor/editor.d.ts +75 -15
  21. package/dist/types/src/display/editor/freetext.d.ts +6 -1
  22. package/dist/types/src/display/editor/highlight.d.ts +88 -0
  23. package/dist/types/src/display/editor/ink.d.ts +4 -4
  24. package/dist/types/src/display/editor/outliner.d.ts +53 -9
  25. package/dist/types/src/display/editor/stamp.d.ts +1 -1
  26. package/dist/types/src/display/editor/toolbar.d.ts +9 -1
  27. package/dist/types/src/display/editor/tools.d.ts +39 -4
  28. package/dist/types/src/display/fetch_stream.d.ts +3 -4
  29. package/dist/types/src/display/font_loader.d.ts +1 -5
  30. package/dist/types/src/display/network.d.ts +2 -3
  31. package/dist/types/src/display/node_stream.d.ts +4 -5
  32. package/dist/types/src/display/node_utils.d.ts +6 -5
  33. package/dist/types/src/display/optional_content_config.d.ts +6 -1
  34. package/dist/types/src/display/text_layer.d.ts +40 -108
  35. package/dist/types/src/display/transport_stream.d.ts +3 -7
  36. package/dist/types/src/display/worker_options.d.ts +20 -22
  37. package/dist/types/src/pdf.d.ts +3 -3
  38. package/dist/types/src/shared/util.d.ts +21 -22
  39. package/dist/types/web/annotation_editor_layer_builder.d.ts +10 -13
  40. package/dist/types/web/annotation_layer_builder.d.ts +7 -4
  41. package/dist/types/web/app_options.d.ts +1 -2
  42. package/dist/types/web/draw_layer_builder.d.ts +23 -0
  43. package/dist/types/web/genericl10n.d.ts +5 -3
  44. package/dist/types/web/interfaces.d.ts +8 -5
  45. package/dist/types/web/l10n.d.ts +3 -2
  46. package/dist/types/web/pdf_find_controller.d.ts +1 -2
  47. package/dist/types/web/pdf_history.d.ts +1 -53
  48. package/dist/types/web/pdf_link_service.d.ts +4 -107
  49. package/dist/types/web/pdf_page_view.d.ts +4 -15
  50. package/dist/types/web/pdf_scripting_manager.d.ts +6 -10
  51. package/dist/types/web/pdf_thumbnail_view.d.ts +1 -13
  52. package/dist/types/web/pdf_thumbnail_viewer.d.ts +0 -16
  53. package/dist/types/web/pdf_viewer.component.d.ts +1 -2
  54. package/dist/types/web/pdf_viewer.d.ts +45 -29
  55. package/dist/types/web/text_accessibility.d.ts +1 -1
  56. package/dist/types/web/text_highlighter.d.ts +1 -1
  57. package/dist/types/web/text_layer_builder.d.ts +17 -29
  58. package/package.json +20 -19
  59. package/dist/types/web/l10n_utils.d.ts +0 -9
package/README.md CHANGED
@@ -17,7 +17,7 @@ This library is also intended as a modern alternative to the unmaintained but st
17
17
  ## PDF.js Compatibility
18
18
 
19
19
  > [!NOTE]
20
- > This package is currently using PDF.js v4.0.189.
20
+ > This package is currently using PDF.js v4.3.136.
21
21
 
22
22
  ## Installation
23
23
 
@@ -50,7 +50,7 @@ const buffer = await fetch(
50
50
  const buffer = await readFile("./dummy.pdf");
51
51
 
52
52
  // Load PDF from buffer
53
- const pdf = await getDocumentProxy(new Uint8Array(pdf));
53
+ const pdf = await getDocumentProxy(new Uint8Array(buffer));
54
54
  // Extract text from PDF
55
55
  const { totalPages, text } = await extractText(pdf, { mergePages: true });
56
56
  ```
@@ -91,7 +91,7 @@ Generally speaking, you don't need to worry about the PDF.js build. `unpdf` ship
91
91
  // if you need another PDF.js build
92
92
  import { configureUnPDF } from "unpdf";
93
93
 
94
- configureUnPDF({
94
+ await configureUnPDF({
95
95
  // Use the official PDF.js build (make sure to install it first)
96
96
  pdfjs: () => import("pdfjs-dist"),
97
97
  });
@@ -138,7 +138,9 @@ function getResolvedPDFJS(): Promise<PDFJS>;
138
138
  ### `getMeta`
139
139
 
140
140
  ```ts
141
- function getMeta(data: BinaryData | PDFDocumentProxy): Promise<{
141
+ function getMeta(
142
+ data: DocumentInitParameters["data"] | PDFDocumentProxy,
143
+ ): Promise<{
142
144
  info: Record<string, any>;
143
145
  metadata: Record<string, any>;
144
146
  }>;
@@ -150,7 +152,7 @@ Extracts all text from a PDF. If `mergePages` is set to `true`, the text of all
150
152
 
151
153
  ```ts
152
154
  function extractText(
153
- data: BinaryData | PDFDocumentProxy,
155
+ data: DocumentInitParameters["data"] | PDFDocumentProxy,
154
156
  { mergePages }?: { mergePages?: boolean },
155
157
  ): Promise<{
156
158
  totalPages: number;
@@ -175,7 +177,7 @@ In order to use this method, you have to meet the following requirements:
175
177
  ```ts
176
178
  import { configureUnPDF, renderPageAsImage } from "unpdf";
177
179
 
178
- configureUnPDF({
180
+ await configureUnPDF({
179
181
  // Use the official PDF.js build
180
182
  pdfjs: () => import("pdfjs-dist"),
181
183
  });
@@ -194,7 +196,7 @@ await writeFile("dummy-page-1.png", Buffer.from(result));
194
196
 
195
197
  ```ts
196
198
  declare function renderPageAsImage(
197
- data: BinaryData,
199
+ data: DocumentInitParameters["data"],
198
200
  pageNumber: number,
199
201
  options?: {
200
202
  canvas?: () => Promise<typeof import("canvas")>;
package/dist/index.cjs CHANGED
@@ -69,20 +69,15 @@ async function extractText$1(data, options = {}) {
69
69
  const texts = await Promise.all(
70
70
  Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
71
71
  );
72
- const filteredTexts = texts.filter(Boolean);
73
72
  return {
74
73
  totalPages: pdf.numPages,
75
- text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
74
+ text: mergePages ? texts.join("\n").replace(/\s+/g, " ") : texts
76
75
  };
77
76
  }
78
- async function getPageText(pdf, pageNumber) {
79
- const page = await pdf.getPage(pageNumber);
77
+ async function getPageText(document, pageNumber) {
78
+ const page = await document.getPage(pageNumber);
80
79
  const content = await page.getTextContent();
81
- const items = content.items;
82
- if (items.length === 0) {
83
- return;
84
- }
85
- return items.map(({ str }) => str).join("\n");
80
+ return content.items.filter((item) => item.str != null).map((item) => item.str + (item.hasEOL ? "\n" : "")).join("");
86
81
  }
87
82
 
88
83
  async function extractImages$1(data, pageNumber) {
@@ -109,7 +104,7 @@ async function extractImages$1(data, pageNumber) {
109
104
  }
110
105
  async function renderPageAsImage$1(data, pageNumber, options = {}) {
111
106
  const canvasFactory = await createIsomorphicCanvasFactory(options.canvas);
112
- const pdf = await getDocumentProxy(data, { canvasFactory });
107
+ const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data, { canvasFactory });
113
108
  const page = await pdf.getPage(pageNumber);
114
109
  if (pageNumber < 1 || pageNumber > pdf.numPages) {
115
110
  throw new Error(
@@ -187,7 +182,6 @@ async function configureUnPDF(options) {
187
182
  }
188
183
  }
189
184
 
190
- const defineUnPDFConfig = configureUnPDF;
191
185
  const getMeta = async (...args) => {
192
186
  await resolvePDFJSImports();
193
187
  return await getMeta$1(...args);
@@ -206,7 +200,7 @@ const renderPageAsImage = async (...args) => {
206
200
  };
207
201
 
208
202
  exports.configureUnPDF = configureUnPDF;
209
- exports.defineUnPDFConfig = defineUnPDFConfig;
203
+ exports.createIsomorphicCanvasFactory = createIsomorphicCanvasFactory;
210
204
  exports.extractImages = extractImages;
211
205
  exports.extractText = extractText;
212
206
  exports.getDocumentProxy = getDocumentProxy;
package/dist/index.d.cts CHANGED
@@ -1,27 +1,49 @@
1
- import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from './types/src/display/api';
1
+ import { DocumentInitParameters, PDFDocumentProxy } from './types/src/display/api';
2
2
  import * as canvas from 'canvas';
3
+ import { Canvas } from 'canvas';
3
4
  import * as _PDFJS from './types/src/pdf';
4
5
 
5
- declare function getMeta$1(data: BinaryData | PDFDocumentProxy): Promise<{
6
+ declare function getMeta$1(data: DocumentInitParameters["data"] | PDFDocumentProxy): Promise<{
6
7
  info: Record<string, any>;
7
8
  metadata: Record<string, any>;
8
9
  }>;
9
10
 
10
- declare function extractText$1(data: BinaryData | PDFDocumentProxy, options?: {
11
- mergePages?: boolean;
11
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options?: {
12
+ mergePages?: false;
12
13
  }): Promise<{
13
14
  totalPages: number;
14
- text: string | string[];
15
+ text: string[];
16
+ }>;
17
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options: {
18
+ mergePages: true;
19
+ }): Promise<{
20
+ totalPages: number;
21
+ text: string;
15
22
  }>;
16
23
 
17
- declare function extractImages$1(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
18
- declare function renderPageAsImage$1(data: BinaryData, pageNumber: number, options?: {
24
+ declare function extractImages$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
25
+ declare function renderPageAsImage$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number, options?: {
19
26
  canvas?: () => Promise<typeof canvas>;
20
27
  /** @default 1 */
21
28
  scale?: number;
22
29
  width?: number;
23
30
  height?: number;
24
31
  }): Promise<ArrayBuffer>;
32
+ declare function createIsomorphicCanvasFactory(canvas?: () => Promise<typeof canvas>): Promise<{
33
+ _createCanvas(width: number, height: number): Canvas | HTMLCanvasElement;
34
+ create(width: number, height: number): {
35
+ canvas: Canvas | HTMLCanvasElement;
36
+ context: CanvasRenderingContext2D;
37
+ };
38
+ reset(ctx: {
39
+ canvas?: Canvas;
40
+ context?: CanvasRenderingContext2D;
41
+ }, width: number, height: number): void;
42
+ destroy(ctx: {
43
+ canvas?: Canvas;
44
+ context?: CanvasRenderingContext2D;
45
+ }): void;
46
+ }>;
25
47
 
26
48
  type PDFJS = typeof _PDFJS;
27
49
  interface UnPDFConfiguration {
@@ -37,8 +59,6 @@ interface UnPDFConfiguration {
37
59
  pdfjs?: () => Promise<any>;
38
60
  }
39
61
 
40
- declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
41
-
42
62
  /**
43
63
  * Returns a PDFDocumentProxy instance from a given binary data.
44
64
  *
@@ -46,18 +66,17 @@ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
46
66
  * - `isEvalSupported: false`
47
67
  * - `useSystemFonts: true`
48
68
  */
49
- declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
69
+ declare function getDocumentProxy(data: DocumentInitParameters["data"], options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
50
70
  declare function getResolvedPDFJS(): Promise<PDFJS>;
51
71
  declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<any>, { force }?: {
52
72
  force?: boolean | undefined;
53
73
  }): Promise<void>;
54
74
 
55
- /** @deprecated Use `configureUnPDF` instead */
56
- declare const defineUnPDFConfig: typeof configureUnPDF;
75
+ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
57
76
 
58
77
  declare const getMeta: typeof getMeta$1;
59
78
  declare const extractText: typeof extractText$1;
60
79
  declare const extractImages: typeof extractImages$1;
61
80
  declare const renderPageAsImage: typeof renderPageAsImage$1;
62
81
 
63
- export { configureUnPDF, defineUnPDFConfig, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
82
+ export { configureUnPDF, createIsomorphicCanvasFactory, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
package/dist/index.d.mts CHANGED
@@ -1,27 +1,49 @@
1
- import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from './types/src/display/api';
1
+ import { DocumentInitParameters, PDFDocumentProxy } from './types/src/display/api';
2
2
  import * as canvas from 'canvas';
3
+ import { Canvas } from 'canvas';
3
4
  import * as _PDFJS from './types/src/pdf';
4
5
 
5
- declare function getMeta$1(data: BinaryData | PDFDocumentProxy): Promise<{
6
+ declare function getMeta$1(data: DocumentInitParameters["data"] | PDFDocumentProxy): Promise<{
6
7
  info: Record<string, any>;
7
8
  metadata: Record<string, any>;
8
9
  }>;
9
10
 
10
- declare function extractText$1(data: BinaryData | PDFDocumentProxy, options?: {
11
- mergePages?: boolean;
11
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options?: {
12
+ mergePages?: false;
12
13
  }): Promise<{
13
14
  totalPages: number;
14
- text: string | string[];
15
+ text: string[];
16
+ }>;
17
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options: {
18
+ mergePages: true;
19
+ }): Promise<{
20
+ totalPages: number;
21
+ text: string;
15
22
  }>;
16
23
 
17
- declare function extractImages$1(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
18
- declare function renderPageAsImage$1(data: BinaryData, pageNumber: number, options?: {
24
+ declare function extractImages$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
25
+ declare function renderPageAsImage$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number, options?: {
19
26
  canvas?: () => Promise<typeof canvas>;
20
27
  /** @default 1 */
21
28
  scale?: number;
22
29
  width?: number;
23
30
  height?: number;
24
31
  }): Promise<ArrayBuffer>;
32
+ declare function createIsomorphicCanvasFactory(canvas?: () => Promise<typeof canvas>): Promise<{
33
+ _createCanvas(width: number, height: number): Canvas | HTMLCanvasElement;
34
+ create(width: number, height: number): {
35
+ canvas: Canvas | HTMLCanvasElement;
36
+ context: CanvasRenderingContext2D;
37
+ };
38
+ reset(ctx: {
39
+ canvas?: Canvas;
40
+ context?: CanvasRenderingContext2D;
41
+ }, width: number, height: number): void;
42
+ destroy(ctx: {
43
+ canvas?: Canvas;
44
+ context?: CanvasRenderingContext2D;
45
+ }): void;
46
+ }>;
25
47
 
26
48
  type PDFJS = typeof _PDFJS;
27
49
  interface UnPDFConfiguration {
@@ -37,8 +59,6 @@ interface UnPDFConfiguration {
37
59
  pdfjs?: () => Promise<any>;
38
60
  }
39
61
 
40
- declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
41
-
42
62
  /**
43
63
  * Returns a PDFDocumentProxy instance from a given binary data.
44
64
  *
@@ -46,18 +66,17 @@ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
46
66
  * - `isEvalSupported: false`
47
67
  * - `useSystemFonts: true`
48
68
  */
49
- declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
69
+ declare function getDocumentProxy(data: DocumentInitParameters["data"], options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
50
70
  declare function getResolvedPDFJS(): Promise<PDFJS>;
51
71
  declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<any>, { force }?: {
52
72
  force?: boolean | undefined;
53
73
  }): Promise<void>;
54
74
 
55
- /** @deprecated Use `configureUnPDF` instead */
56
- declare const defineUnPDFConfig: typeof configureUnPDF;
75
+ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
57
76
 
58
77
  declare const getMeta: typeof getMeta$1;
59
78
  declare const extractText: typeof extractText$1;
60
79
  declare const extractImages: typeof extractImages$1;
61
80
  declare const renderPageAsImage: typeof renderPageAsImage$1;
62
81
 
63
- export { configureUnPDF, defineUnPDFConfig, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
82
+ export { configureUnPDF, createIsomorphicCanvasFactory, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
package/dist/index.d.ts CHANGED
@@ -1,27 +1,49 @@
1
- import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from './types/src/display/api';
1
+ import { DocumentInitParameters, PDFDocumentProxy } from './types/src/display/api';
2
2
  import * as canvas from 'canvas';
3
+ import { Canvas } from 'canvas';
3
4
  import * as _PDFJS from './types/src/pdf';
4
5
 
5
- declare function getMeta$1(data: BinaryData | PDFDocumentProxy): Promise<{
6
+ declare function getMeta$1(data: DocumentInitParameters["data"] | PDFDocumentProxy): Promise<{
6
7
  info: Record<string, any>;
7
8
  metadata: Record<string, any>;
8
9
  }>;
9
10
 
10
- declare function extractText$1(data: BinaryData | PDFDocumentProxy, options?: {
11
- mergePages?: boolean;
11
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options?: {
12
+ mergePages?: false;
12
13
  }): Promise<{
13
14
  totalPages: number;
14
- text: string | string[];
15
+ text: string[];
16
+ }>;
17
+ declare function extractText$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, options: {
18
+ mergePages: true;
19
+ }): Promise<{
20
+ totalPages: number;
21
+ text: string;
15
22
  }>;
16
23
 
17
- declare function extractImages$1(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
18
- declare function renderPageAsImage$1(data: BinaryData, pageNumber: number, options?: {
24
+ declare function extractImages$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number): Promise<Uint8ClampedArray[]>;
25
+ declare function renderPageAsImage$1(data: DocumentInitParameters["data"] | PDFDocumentProxy, pageNumber: number, options?: {
19
26
  canvas?: () => Promise<typeof canvas>;
20
27
  /** @default 1 */
21
28
  scale?: number;
22
29
  width?: number;
23
30
  height?: number;
24
31
  }): Promise<ArrayBuffer>;
32
+ declare function createIsomorphicCanvasFactory(canvas?: () => Promise<typeof canvas>): Promise<{
33
+ _createCanvas(width: number, height: number): Canvas | HTMLCanvasElement;
34
+ create(width: number, height: number): {
35
+ canvas: Canvas | HTMLCanvasElement;
36
+ context: CanvasRenderingContext2D;
37
+ };
38
+ reset(ctx: {
39
+ canvas?: Canvas;
40
+ context?: CanvasRenderingContext2D;
41
+ }, width: number, height: number): void;
42
+ destroy(ctx: {
43
+ canvas?: Canvas;
44
+ context?: CanvasRenderingContext2D;
45
+ }): void;
46
+ }>;
25
47
 
26
48
  type PDFJS = typeof _PDFJS;
27
49
  interface UnPDFConfiguration {
@@ -37,8 +59,6 @@ interface UnPDFConfiguration {
37
59
  pdfjs?: () => Promise<any>;
38
60
  }
39
61
 
40
- declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
41
-
42
62
  /**
43
63
  * Returns a PDFDocumentProxy instance from a given binary data.
44
64
  *
@@ -46,18 +66,17 @@ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
46
66
  * - `isEvalSupported: false`
47
67
  * - `useSystemFonts: true`
48
68
  */
49
- declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
69
+ declare function getDocumentProxy(data: DocumentInitParameters["data"], options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
50
70
  declare function getResolvedPDFJS(): Promise<PDFJS>;
51
71
  declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<any>, { force }?: {
52
72
  force?: boolean | undefined;
53
73
  }): Promise<void>;
54
74
 
55
- /** @deprecated Use `configureUnPDF` instead */
56
- declare const defineUnPDFConfig: typeof configureUnPDF;
75
+ declare function configureUnPDF(options: UnPDFConfiguration): Promise<void>;
57
76
 
58
77
  declare const getMeta: typeof getMeta$1;
59
78
  declare const extractText: typeof extractText$1;
60
79
  declare const extractImages: typeof extractImages$1;
61
80
  declare const renderPageAsImage: typeof renderPageAsImage$1;
62
81
 
63
- export { configureUnPDF, defineUnPDFConfig, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
82
+ export { configureUnPDF, createIsomorphicCanvasFactory, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
package/dist/index.mjs CHANGED
@@ -67,20 +67,15 @@ async function extractText$1(data, options = {}) {
67
67
  const texts = await Promise.all(
68
68
  Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
69
69
  );
70
- const filteredTexts = texts.filter(Boolean);
71
70
  return {
72
71
  totalPages: pdf.numPages,
73
- text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
72
+ text: mergePages ? texts.join("\n").replace(/\s+/g, " ") : texts
74
73
  };
75
74
  }
76
- async function getPageText(pdf, pageNumber) {
77
- const page = await pdf.getPage(pageNumber);
75
+ async function getPageText(document, pageNumber) {
76
+ const page = await document.getPage(pageNumber);
78
77
  const content = await page.getTextContent();
79
- const items = content.items;
80
- if (items.length === 0) {
81
- return;
82
- }
83
- return items.map(({ str }) => str).join("\n");
78
+ return content.items.filter((item) => item.str != null).map((item) => item.str + (item.hasEOL ? "\n" : "")).join("");
84
79
  }
85
80
 
86
81
  async function extractImages$1(data, pageNumber) {
@@ -107,7 +102,7 @@ async function extractImages$1(data, pageNumber) {
107
102
  }
108
103
  async function renderPageAsImage$1(data, pageNumber, options = {}) {
109
104
  const canvasFactory = await createIsomorphicCanvasFactory(options.canvas);
110
- const pdf = await getDocumentProxy(data, { canvasFactory });
105
+ const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data, { canvasFactory });
111
106
  const page = await pdf.getPage(pageNumber);
112
107
  if (pageNumber < 1 || pageNumber > pdf.numPages) {
113
108
  throw new Error(
@@ -185,7 +180,6 @@ async function configureUnPDF(options) {
185
180
  }
186
181
  }
187
182
 
188
- const defineUnPDFConfig = configureUnPDF;
189
183
  const getMeta = async (...args) => {
190
184
  await resolvePDFJSImports();
191
185
  return await getMeta$1(...args);
@@ -203,4 +197,4 @@ const renderPageAsImage = async (...args) => {
203
197
  return await renderPageAsImage$1(...args);
204
198
  };
205
199
 
206
- export { configureUnPDF, defineUnPDFConfig, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
200
+ export { configureUnPDF, createIsomorphicCanvasFactory, extractImages, extractText, getDocumentProxy, getMeta, getResolvedPDFJS, renderPageAsImage, resolvePDFJSImports };
package/dist/pdfjs.d.mts CHANGED
@@ -1,3 +1,4 @@
1
1
  import * as PDFJS from './types/src/pdf'
2
2
  declare function resolvePDFJS(): Promise<typeof PDFJS>
3
3
  export { resolvePDFJS }
4
+ export * from './types/src/pdf'
package/dist/pdfjs.d.ts CHANGED
@@ -1,3 +1,4 @@
1
1
  import * as PDFJS from './types/src/pdf'
2
2
  declare function resolvePDFJS(): Promise<typeof PDFJS>
3
3
  export { resolvePDFJS }
4
+ export * from './types/src/pdf'