@embedpdf/engines 1.0.5 → 1.0.6

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.d.ts CHANGED
@@ -1,4 +1,4 @@
1
- import { SearchTarget, PdfImage, ImageConversionTypes, PdfEngine, Logger, Task, PdfErrorReason, PdfFileUrl, PdfUrlOptions, PdfDocumentObject, PdfFile, PdfFileLoader, PdfSignatureObject, PdfBookmarkObject, PdfPageObject, Rotation, PdfRenderOptions, PdfTask, Rect, PdfAnnotationObject, PdfAnnotationTransformation, PdfTextRectObject, PdfAttachmentObject, PdfWidgetAnnoObject, FormFieldValue, PdfPageFlattenFlag, PdfPageFlattenResult, PdfInkListObject, PdfStampAnnoObjectContents, Position, PdfPageGeometry, PdfGlyphObject, MatchFlag, SearchAllPagesResult, PdfEngineMethodName, PdfEngineMethodArgs, TaskReturn, PdfEngineMethodReturnType, PdfMetadataObject, PdfBookmarksObject } from '@embedpdf/models';
1
+ import { SearchTarget, PdfImage, ImageConversionTypes, PdfEngine, Logger, Task, PdfErrorReason, PdfFileUrl, PdfUrlOptions, PdfDocumentObject, PdfFile, PdfFileLoader, PdfSignatureObject, PdfBookmarkObject, PdfPageObject, Rotation, PdfRenderOptions, PdfTask, Rect, PdfAnnotationObject, PdfAnnotationTransformation, PdfTextRectObject, PdfAttachmentObject, PdfWidgetAnnoObject, FormFieldValue, PdfPageFlattenFlag, PdfPageFlattenResult, PageTextSlice, PdfInkListObject, PdfStampAnnoObjectContents, Position, PdfPageGeometry, PdfGlyphObject, MatchFlag, SearchAllPagesResult, PdfEngineMethodName, PdfEngineMethodArgs, TaskReturn, PdfEngineMethodReturnType, PdfMetadataObject, PdfBookmarksObject } from '@embedpdf/models';
2
2
  import { WrappedPdfiumModule, PdfiumRuntimeMethods, PdfiumModule } from '@embedpdf/pdfium';
3
3
 
4
4
  /**
@@ -269,6 +269,12 @@ declare class PdfiumEngine<T = Blob> implements PdfEngine<T> {
269
269
  * @public
270
270
  */
271
271
  extractText(doc: PdfDocumentObject, pageIndexes: number[]): Task<any, PdfErrorReason> | Task<string, PdfErrorReason>;
272
+ /**
273
+ * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
274
+ *
275
+ * @public
276
+ */
277
+ getTextSlices(doc: PdfDocumentObject, slices: PageTextSlice[]): PdfTask<string[]>;
272
278
  /**
273
279
  * {@inheritDoc @embedpdf/models!PdfEngine.merge}
274
280
  *
@@ -1431,6 +1437,12 @@ declare class WebWorkerEngine implements PdfEngine {
1431
1437
  * @public
1432
1438
  */
1433
1439
  extractText(doc: PdfDocumentObject, pageIndexes: number[]): WorkerTask<string>;
1440
+ /**
1441
+ * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
1442
+ *
1443
+ * @public
1444
+ */
1445
+ getTextSlices(doc: PdfDocumentObject, slices: PageTextSlice[]): WorkerTask<string[]>;
1434
1446
  /**
1435
1447
  * {@inheritDoc @embedpdf/models!PdfEngine.getPageGlyphs}
1436
1448
  *
package/dist/index.js CHANGED
@@ -1,4 +1,4 @@
1
- import { NoopLogger, PdfTaskHelper, PdfErrorCode, Task, Rotation, PdfAnnotationSubtype, PdfPageObjectType, PdfAnnotationObjectStatus, quadToRect, PDF_FORM_FIELD_TYPE, toIntRect, transformRect, toIntSize, transformSize, PdfActionType, PdfZoomMode, AppearanceMode, MatchFlag, swap, PdfPageFlattenResult } from '@embedpdf/models';
1
+ import { NoopLogger, PdfTaskHelper, PdfErrorCode, Task, Rotation, PdfAnnotationSubtype, stripPdfUnwantedMarkers, PdfPageObjectType, PdfAnnotationObjectStatus, quadToRect, PDF_FORM_FIELD_TYPE, toIntRect, transformRect, toIntSize, transformSize, PdfActionType, PdfZoomMode, AppearanceMode, MatchFlag, swap, PdfPageFlattenResult } from '@embedpdf/models';
2
2
  import { init } from '@embedpdf/pdfium';
3
3
 
4
4
  /**
@@ -1362,6 +1362,62 @@ class PdfiumEngine {
1362
1362
  this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, `ExtractText`, 'End', doc.id);
1363
1363
  return PdfTaskHelper.resolve(text);
1364
1364
  }
1365
+ /**
1366
+ * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
1367
+ *
1368
+ * @public
1369
+ */
1370
+ getTextSlices(doc, slices) {
1371
+ this.logger.debug(LOG_SOURCE$2, LOG_CATEGORY$2, 'getTextSlices', doc, slices);
1372
+ this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, 'GetTextSlices', 'Begin', doc.id);
1373
+ /* ⚠︎ 1 — trivial case */
1374
+ if (slices.length === 0) {
1375
+ this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, 'GetTextSlices', 'End', doc.id);
1376
+ return PdfTaskHelper.resolve([]);
1377
+ }
1378
+ /* ⚠︎ 2 — document must be open */
1379
+ const ctx = this.cache.getContext(doc.id);
1380
+ if (!ctx) {
1381
+ this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, 'GetTextSlices', 'End', doc.id);
1382
+ return PdfTaskHelper.reject({
1383
+ code: PdfErrorCode.DocNotOpen,
1384
+ message: 'document does not open',
1385
+ });
1386
+ }
1387
+ try {
1388
+ /* keep caller order */
1389
+ const out = new Array(slices.length);
1390
+ /* group → open each page once */
1391
+ const byPage = new Map();
1392
+ slices.forEach((s, i) => {
1393
+ (byPage.get(s.pageIndex) ?? byPage.set(s.pageIndex, []).get(s.pageIndex)).push({
1394
+ slice: s,
1395
+ pos: i,
1396
+ });
1397
+ });
1398
+ for (const [pageIdx, list] of byPage) {
1399
+ const pageCtx = ctx.acquirePage(pageIdx);
1400
+ const textPagePtr = pageCtx.getTextPage();
1401
+ for (const { slice, pos } of list) {
1402
+ const bufPtr = this.malloc(2 * (slice.charCount + 1)); // UTF-16 + NIL
1403
+ this.pdfiumModule.FPDFText_GetText(textPagePtr, slice.charIndex, slice.charCount, bufPtr);
1404
+ out[pos] = stripPdfUnwantedMarkers(this.pdfiumModule.pdfium.UTF16ToString(bufPtr));
1405
+ this.free(bufPtr);
1406
+ }
1407
+ pageCtx.release();
1408
+ }
1409
+ this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, 'GetTextSlices', 'End', doc.id);
1410
+ return PdfTaskHelper.resolve(out);
1411
+ }
1412
+ catch (e) {
1413
+ this.logger.error(LOG_SOURCE$2, LOG_CATEGORY$2, 'getTextSlices error', e);
1414
+ this.logger.perf(LOG_SOURCE$2, LOG_CATEGORY$2, 'GetTextSlices', 'End', doc.id);
1415
+ return PdfTaskHelper.reject({
1416
+ code: PdfErrorCode.Unknown,
1417
+ message: String(e),
1418
+ });
1419
+ }
1420
+ }
1365
1421
  /**
1366
1422
  * {@inheritDoc @embedpdf/models!PdfEngine.merge}
1367
1423
  *
@@ -1891,14 +1947,12 @@ class PdfiumEngine {
1891
1947
  const runs = [];
1892
1948
  let current = null;
1893
1949
  let curObjPtr = null;
1950
+ let bounds = null;
1894
1951
  /** ── main loop ──────────────────────────────────────────── */
1895
1952
  for (let i = 0; i < glyphs.length; i++) {
1896
1953
  const g = glyphs[i];
1897
1954
  /* 1 — find the CPDF_TextObject this glyph belongs to */
1898
1955
  const objPtr = this.pdfiumModule.FPDFText_GetTextObject(textPagePtr, i);
1899
- if (g.isEmpty) {
1900
- continue;
1901
- }
1902
1956
  /* 2 — start a new run when the text object changes */
1903
1957
  if (objPtr !== curObjPtr) {
1904
1958
  curObjPtr = objPtr;
@@ -1912,6 +1966,12 @@ class PdfiumEngine {
1912
1966
  charStart: i,
1913
1967
  glyphs: [],
1914
1968
  };
1969
+ bounds = {
1970
+ minX: g.origin.x,
1971
+ minY: g.origin.y,
1972
+ maxX: g.origin.x + g.size.width,
1973
+ maxY: g.origin.y + g.size.height,
1974
+ };
1915
1975
  runs.push(current);
1916
1976
  }
1917
1977
  /* 3 — append the slim glyph record */
@@ -1920,16 +1980,24 @@ class PdfiumEngine {
1920
1980
  y: g.origin.y,
1921
1981
  width: g.size.width,
1922
1982
  height: g.size.height,
1923
- flags: g.isSpace ? 1 : 0,
1983
+ flags: g.isEmpty ? 2 : g.isSpace ? 1 : 0,
1924
1984
  });
1925
1985
  /* 4 — expand the run's bounding rect */
1986
+ if (g.isEmpty) {
1987
+ continue;
1988
+ }
1926
1989
  const right = g.origin.x + g.size.width;
1927
1990
  const bottom = g.origin.y + g.size.height;
1928
- current.rect.width =
1929
- Math.max(current.rect.x + current.rect.width, right) - current.rect.x;
1930
- current.rect.y = Math.min(current.rect.y, g.origin.y);
1931
- current.rect.height =
1932
- Math.max(current.rect.y + current.rect.height, bottom) - current.rect.y;
1991
+ // Update bounds
1992
+ bounds.minX = Math.min(bounds.minX, g.origin.x);
1993
+ bounds.minY = Math.min(bounds.minY, g.origin.y);
1994
+ bounds.maxX = Math.max(bounds.maxX, right);
1995
+ bounds.maxY = Math.max(bounds.maxY, bottom);
1996
+ // Calculate final rect from bounds
1997
+ current.rect.x = bounds.minX;
1998
+ current.rect.y = bounds.minY;
1999
+ current.rect.width = bounds.maxX - bounds.minX;
2000
+ current.rect.height = bounds.maxY - bounds.minY;
1933
2001
  }
1934
2002
  return runs;
1935
2003
  }
@@ -4175,6 +4243,9 @@ class EngineRunner {
4175
4243
  case 'extractText':
4176
4244
  task = this.engine[name](...args);
4177
4245
  break;
4246
+ case 'getTextSlices':
4247
+ task = this.engine[name](...args);
4248
+ break;
4178
4249
  case 'getPageGlyphs':
4179
4250
  task = this.engine[name](...args);
4180
4251
  break;
@@ -4935,6 +5006,26 @@ class WebWorkerEngine {
4935
5006
  this.proxy(task, request);
4936
5007
  return task;
4937
5008
  }
5009
+ /**
5010
+ * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
5011
+ *
5012
+ * @public
5013
+ */
5014
+ getTextSlices(doc, slices) {
5015
+ this.logger.debug(LOG_SOURCE, LOG_CATEGORY, 'getTextSlices', doc, slices);
5016
+ const requestId = this.generateRequestId(doc.id);
5017
+ const task = new WorkerTask(this.worker, requestId);
5018
+ const request = {
5019
+ id: requestId,
5020
+ type: 'ExecuteRequest',
5021
+ data: {
5022
+ name: 'getTextSlices',
5023
+ args: [doc, slices],
5024
+ },
5025
+ };
5026
+ this.proxy(task, request);
5027
+ return task;
5028
+ }
4938
5029
  /**
4939
5030
  * {@inheritDoc @embedpdf/models!PdfEngine.getPageGlyphs}
4940
5031
  *
@@ -5299,6 +5390,9 @@ function createMockPdfEngine(partialEngine) {
5299
5390
  extractText: (pdf, pageIndexes) => {
5300
5391
  return PdfTaskHelper.resolve('');
5301
5392
  },
5393
+ getTextSlices: (doc, slices) => {
5394
+ return PdfTaskHelper.resolve([]);
5395
+ },
5302
5396
  getPageGlyphs: (doc, page) => {
5303
5397
  return PdfTaskHelper.resolve([]);
5304
5398
  },