@embedpdf/engines 1.0.5 → 1.0.6

package/dist/pdfium-direct-engine.d.ts

@@ -1,4 +1,4 @@
-import { PdfEngine, Logger, Task, PdfErrorReason, PdfFileUrl, PdfUrlOptions, PdfDocumentObject, PdfFile, PdfFileLoader, PdfSignatureObject, PdfBookmarkObject, PdfPageObject, Rotation, PdfRenderOptions, ImageConversionTypes, PdfTask, Rect, PdfAnnotationObject, PdfAnnotationTransformation, PdfTextRectObject, PdfAttachmentObject, PdfWidgetAnnoObject, FormFieldValue, PdfPageFlattenFlag, PdfPageFlattenResult, PdfInkListObject, PdfStampAnnoObjectContents, Position, PdfPageGeometry, PdfGlyphObject, MatchFlag, SearchAllPagesResult, PdfImage } from '@embedpdf/models';
+import { PdfEngine, Logger, Task, PdfErrorReason, PdfFileUrl, PdfUrlOptions, PdfDocumentObject, PdfFile, PdfFileLoader, PdfSignatureObject, PdfBookmarkObject, PdfPageObject, Rotation, PdfRenderOptions, ImageConversionTypes, PdfTask, Rect, PdfAnnotationObject, PdfAnnotationTransformation, PdfTextRectObject, PdfAttachmentObject, PdfWidgetAnnoObject, FormFieldValue, PdfPageFlattenFlag, PdfPageFlattenResult, PageTextSlice, PdfInkListObject, PdfStampAnnoObjectContents, Position, PdfPageGeometry, PdfGlyphObject, MatchFlag, SearchAllPagesResult, PdfImage } from '@embedpdf/models';
 import { WrappedPdfiumModule } from '@embedpdf/pdfium';
 
 /**
@@ -213,6 +213,12 @@ declare class PdfiumEngine<T = Blob> implements PdfEngine<T> {
      * @public
      */
     extractText(doc: PdfDocumentObject, pageIndexes: number[]): Task<any, PdfErrorReason> | Task<string, PdfErrorReason>;
+    /**
+     * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
+     *
+     * @public
+     */
+    getTextSlices(doc: PdfDocumentObject, slices: PageTextSlice[]): PdfTask<string[]>;
     /**
      * {@inheritDoc @embedpdf/models!PdfEngine.merge}
      *
@@ -986,6 +992,6 @@ declare class PdfiumEngine<T = Blob> implements PdfEngine<T> {
     private searchAllInPage;
 }
 
-declare function createPdfiumEngine(wasmUrl: string): Promise<PdfiumEngine>;
+declare function createPdfiumEngine(wasmUrl: string, logger?: Logger): Promise<PdfiumEngine>;
 
 export { createPdfiumEngine };

package/dist/pdfium-direct-engine.js

@@ -1,5 +1,5 @@
-import { NoopLogger, PdfTaskHelper, PdfErrorCode, Task, Rotation, PdfAnnotationSubtype, PdfPageObjectType, PdfAnnotationObjectStatus, quadToRect, PDF_FORM_FIELD_TYPE, toIntRect, transformRect, toIntSize, transformSize, PdfActionType, PdfZoomMode, AppearanceMode, MatchFlag } from '@embedpdf/models';
 import { init } from '@embedpdf/pdfium';
+import { NoopLogger, PdfTaskHelper, PdfErrorCode, Task, Rotation, PdfAnnotationSubtype, stripPdfUnwantedMarkers, PdfPageObjectType, PdfAnnotationObjectStatus, quadToRect, PDF_FORM_FIELD_TYPE, toIntRect, transformRect, toIntSize, transformSize, PdfActionType, PdfZoomMode, AppearanceMode, MatchFlag } from '@embedpdf/models';
 
 /**
  * Read string from WASM heap
@@ -1362,6 +1362,62 @@ class PdfiumEngine {
         this.logger.perf(LOG_SOURCE, LOG_CATEGORY, `ExtractText`, 'End', doc.id);
         return PdfTaskHelper.resolve(text);
     }
+    /**
+     * {@inheritDoc @embedpdf/models!PdfEngine.getTextSlices}
+     *
+     * @public
+     */
+    getTextSlices(doc, slices) {
+        this.logger.debug(LOG_SOURCE, LOG_CATEGORY, 'getTextSlices', doc, slices);
+        this.logger.perf(LOG_SOURCE, LOG_CATEGORY, 'GetTextSlices', 'Begin', doc.id);
+        /* ⚠︎ 1 — trivial case */
+        if (slices.length === 0) {
+            this.logger.perf(LOG_SOURCE, LOG_CATEGORY, 'GetTextSlices', 'End', doc.id);
+            return PdfTaskHelper.resolve([]);
+        }
+        /* ⚠︎ 2 — document must be open */
+        const ctx = this.cache.getContext(doc.id);
+        if (!ctx) {
+            this.logger.perf(LOG_SOURCE, LOG_CATEGORY, 'GetTextSlices', 'End', doc.id);
+            return PdfTaskHelper.reject({
+                code: PdfErrorCode.DocNotOpen,
+                message: 'document does not open',
+            });
+        }
+        try {
+            /* keep caller order */
+            const out = new Array(slices.length);
+            /* group → open each page once */
+            const byPage = new Map();
+            slices.forEach((s, i) => {
+                (byPage.get(s.pageIndex) ?? byPage.set(s.pageIndex, []).get(s.pageIndex)).push({
+                    slice: s,
+                    pos: i,
+                });
+            });
+            for (const [pageIdx, list] of byPage) {
+                const pageCtx = ctx.acquirePage(pageIdx);
+                const textPagePtr = pageCtx.getTextPage();
+                for (const { slice, pos } of list) {
+                    const bufPtr = this.malloc(2 * (slice.charCount + 1)); // UTF-16 + NIL
+                    this.pdfiumModule.FPDFText_GetText(textPagePtr, slice.charIndex, slice.charCount, bufPtr);
+                    out[pos] = stripPdfUnwantedMarkers(this.pdfiumModule.pdfium.UTF16ToString(bufPtr));
+                    this.free(bufPtr);
+                }
+                pageCtx.release();
+            }
+            this.logger.perf(LOG_SOURCE, LOG_CATEGORY, 'GetTextSlices', 'End', doc.id);
+            return PdfTaskHelper.resolve(out);
+        }
+        catch (e) {
+            this.logger.error(LOG_SOURCE, LOG_CATEGORY, 'getTextSlices error', e);
+            this.logger.perf(LOG_SOURCE, LOG_CATEGORY, 'GetTextSlices', 'End', doc.id);
+            return PdfTaskHelper.reject({
+                code: PdfErrorCode.Unknown,
+                message: String(e),
+            });
+        }
+    }
     /**
      * {@inheritDoc @embedpdf/models!PdfEngine.merge}
      *
@@ -1891,14 +1947,12 @@ class PdfiumEngine {
         const runs = [];
         let current = null;
         let curObjPtr = null;
+        let bounds = null;
         /** ── main loop ──────────────────────────────────────────── */
         for (let i = 0; i < glyphs.length; i++) {
             const g = glyphs[i];
             /* 1 — find the CPDF_TextObject this glyph belongs to */
             const objPtr = this.pdfiumModule.FPDFText_GetTextObject(textPagePtr, i);
-            if (g.isEmpty) {
-                continue;
-            }
             /* 2 — start a new run when the text object changes */
             if (objPtr !== curObjPtr) {
                 curObjPtr = objPtr;
@@ -1912,6 +1966,12 @@ class PdfiumEngine {
                     charStart: i,
                     glyphs: [],
                 };
+                bounds = {
+                    minX: g.origin.x,
+                    minY: g.origin.y,
+                    maxX: g.origin.x + g.size.width,
+                    maxY: g.origin.y + g.size.height,
+                };
                 runs.push(current);
             }
             /* 3 — append the slim glyph record */
@@ -1920,16 +1980,24 @@ class PdfiumEngine {
                 y: g.origin.y,
                 width: g.size.width,
                 height: g.size.height,
-                flags: g.isSpace ? 1 : 0,
+                flags: g.isEmpty ? 2 : g.isSpace ? 1 : 0,
             });
             /* 4 — expand the run's bounding rect */
+            if (g.isEmpty) {
+                continue;
+            }
             const right = g.origin.x + g.size.width;
             const bottom = g.origin.y + g.size.height;
-            current.rect.width =
-                Math.max(current.rect.x + current.rect.width, right) - current.rect.x;
-            current.rect.y = Math.min(current.rect.y, g.origin.y);
-            current.rect.height =
-                Math.max(current.rect.y + current.rect.height, bottom) - current.rect.y;
+            // Update bounds
+            bounds.minX = Math.min(bounds.minX, g.origin.x);
+            bounds.minY = Math.min(bounds.minY, g.origin.y);
+            bounds.maxX = Math.max(bounds.maxX, right);
+            bounds.maxY = Math.max(bounds.maxY, bottom);
+            // Calculate final rect from bounds
+            current.rect.x = bounds.minX;
+            current.rect.y = bounds.minY;
+            current.rect.width = bounds.maxX - bounds.minX;
+            current.rect.height = bounds.maxY - bounds.minY;
         }
         return runs;
     }
@@ -4024,11 +4092,11 @@ class PdfiumEngine {
     }
 }
 
-async function createPdfiumEngine(wasmUrl) {
+async function createPdfiumEngine(wasmUrl, logger) {
     const response = await fetch(wasmUrl);
     const wasmBinary = await response.arrayBuffer();
     const wasmModule = await init({ wasmBinary });
-    return new PdfiumEngine(wasmModule);
+    return new PdfiumEngine(wasmModule, logger);
 }
 
 export { createPdfiumEngine };