@embedpdf/engine-services 3.0.0-next.2 → 3.0.0-next.4

package/dist/index.d.cts

@@ -1,5 +1,5 @@
 import { PdfFunctions, PdfRuntimeMemory, PdfRuntimeModule, Ptr } from "@embedpdf/engine-runtime";
-import { ActionReadBudget, AnnotationActor, AnnotationAppearanceRenderOptions, AnnotationAppearancesResult, AnnotationCreateResult, AnnotationDeleteResult, AnnotationListMutationMeta, AnnotationListPageSnapshot, AnnotationListSnapshotAllPages, AnnotationMoveResult, AnnotationMutationKind, AnnotationRef, AnnotationStableId, AnnotationUpdateResult, AttachmentCreateResult, AttachmentDeleteResult, AttachmentFileInfo, AttachmentFileWorkerPayload, DocumentActionsSnapshot, DocumentEvent, DocumentEventInit, DocumentEventStream, DocumentMetadata, DocumentSecurityProbeInfo, EmbeddedFileItem, EmbeddedFileRef, FormDataFormat, FormEffect, FormEffectsResult, FormFieldDTO, FormFieldDraft, FormFieldPatch, FormFieldRef, FormFieldValue, FormImportResult, FormRepairResult, FormSetValueResult, FormSnapshot, FormWidgetRef, MetadataPatch, MetadataUpdateResult, PageDeleteResult, PageFlattenResult, PageFlattenUsage, PageGeometrySnapshot, PageInsertResult, PageListSnapshot, PageMoveResult, PageObjectNumber, PageRenderOptions, PageRotateResult, PageRotation, PageState, PageTextSnapshot, PdfActionTree, PdfSaveMode, RedactionApplyResult, RedactionApplyScope, RevisionToken, SearchRequest, SearchSlice, WeakAnnotationState, WireAnnotationDraft, WireAnnotationPatch, WireAttachmentFile, WirePack, WireResource, WireResourceMap, WorkerRequest, WorkerResponse } from "@embedpdf/engine-core/runtime";
+import { ActionReadBudget, AnnotationActor, AnnotationAppearanceRenderOptions, AnnotationAppearancesResult, AnnotationCreateResult, AnnotationDeleteResult, AnnotationListMutationMeta, AnnotationListPageSnapshot, AnnotationListSnapshotAllPages, AnnotationMoveResult, AnnotationMutationKind, AnnotationRef, AnnotationStableId, AnnotationUpdateResult, AttachmentCreateResult, AttachmentDeleteResult, AttachmentFileInfo, AttachmentFileWorkerPayload, DocumentActionsSnapshot, DocumentEvent, DocumentEventInit, DocumentEventStream, DocumentMetadata, DocumentSecurityProbeInfo, EmbeddedFileItem, EmbeddedFileRef, FormDataFormat, FormEffect, FormEffectsResult, FormFieldDTO, FormFieldDraft, FormFieldPatch, FormFieldRef, FormFieldValue, FormImportResult, FormRepairResult, FormSetValueResult, FormSnapshot, FormWidgetRef, MetadataPatch, MetadataUpdateResult, PageDeleteResult, PageFlattenResult, PageFlattenUsage, PageGeometryRun, PageGeometrySnapshot, PageInsertResult, PageListSnapshot, PageMoveResult, PageObjectNumber, PageRenderOptions, PageRotateResult, PageRotation, PageState, PageTextSnapshot, PdfActionTree, PdfQuad, PdfRect, PdfSaveMode, RedactionApplyResult, RedactionApplyScope, RevisionToken, SearchRequest, SearchSlice, WeakAnnotationState, WireAnnotationDraft, WireAnnotationPatch, WireAttachmentFile, WirePack, WireResource, WireResourceMap, WorkerRequest, WorkerResponse } from "@embedpdf/engine-core/runtime";
 //#region src/runtime/lifecycle/bootstrap.d.ts
 /**
  * Initialize the thread-confined PDFium runtime for the calling thread. Each
@@ -626,6 +626,7 @@ declare class RedactionApplier {
    */
   private preflight;
   private applyAllOnPage;
+  private pageHasRedactAnnotations;
   private applyRefsOnPage;
 }
 //#endregion
@@ -658,21 +659,93 @@ declare class PageTextReader {
 }
 //#endregion
 //#region src/features/geometry/PageGeometryReader.d.ts
+/**
+ * One glyph as read from `EPDFText_GetCharGeometry`, plus the run-grouping
+ * context sampled alongside it. Pure data — no native handles — so the
+ * grouping stage below is a pure function the tests can drive directly.
+ */
+interface RawGeometryGlyphRecord {
+  /** Grouping key: the glyph's text object handle (opaque identity). */
+  objectKey: number | bigint;
+  /** Sampled at each text object's first glyph (run splits inherit it). */
+  fontSize?: number;
+  /**
+   * Wire flags (bit 1 = space, bit 2 = empty). Synthesized-but-visible
+   * glyphs (/ActualText pieces) stay 0, exactly like the legacy reader —
+   * they represent real, selectable text.
+   */
+  flags: number;
+  /** Page-space AABB (zeroed when empty). */
+  looseBox: PdfRect;
+  tightBox?: PdfRect;
+  /** Oriented cells — real glyphs with a usable (non-singular) matrix only. */
+  looseQuad?: PdfQuad;
+  tightQuad?: PdfQuad;
+  /** Native uprightness of the char matrix (empty glyphs: true, inert). */
+  upright: boolean;
+  /** Baseline angle (radians CCW, PDF y-up); 0 when upright. */
+  rotation: number;
+  /** True when the ascent vector maps opposite the rotated +y (det < 0). */
+  ascentFlip: boolean;
+}
 /**
  * Geometry-only text layout reader.
  *
  * Emits geometry in PDF user space (y-up edges) — the canonical engine
  * geometry. The viewer converts to content/view space via the page geometry
  * matrix; this reader applies NO Y-flip or device transform.
+ *
+ * One `EPDFText_GetCharGeometry` call per glyph supplies boxes, oriented
+ * cells, and flags together; `buildRunsFromRawGlyphs` then groups glyphs
+ * into the upright/rotated run union. Upright runs are byte-identical to the
+ * pre-orientation reader (quads the native layer offers are deliberately
+ * dropped — the dominant case never pays for orientation).
  */
 declare class PageGeometryReader {
   private readonly runtime;
   private readonly session;
   constructor(runtime: PdfRuntimeModule, session: DocumentSession);
   read(pageObjectNumber: PageObjectNumber, signal: AbortSignal): PageGeometrySnapshot;
-  private buildRunsFromGlyphs;
-  private readGlyphInfo;
+  private readGlyphRaw;
 }
+/**
+ * Group per-glyph records into the run union. Pure — exported for tests.
+ *
+ * Runs split on text-object change (the legacy rule) and on orientation
+ * change between classifiable glyphs (θ drift / mixed orientation). A run's
+ * variant is decided by its FIRST classifiable glyph:
+ *   - empty glyphs never classify (they adopt the run's variant);
+ *   - real glyphs WITHOUT an oriented cell (singular matrix, synthesized
+ *     /ActualText pieces) classify as upright — box-only data degrades to
+ *     exactly the legacy behavior;
+ *   - real glyphs with a non-upright matrix classify as rotated.
+ * Runs that never see a classifiable glyph emit as upright (degenerate,
+ * legacy behavior).
+ */
+declare function buildRunsFromRawGlyphs(records: RawGeometryGlyphRecord[]): PageGeometryRun[];
+//#endregion
+//#region src/runtime/memory/structs.d.ts
+/** Generated from `EPDF_CHAR_GEOMETRY`; pinned by native static_asserts. */
+declare const EPDF_CHAR_GEOMETRY_LAYOUT: {
+  readonly bytes: 124;
+  readonly offsets: {
+    readonly looseBox: 0;
+    readonly tightBox: 16;
+    readonly looseQuad: 32;
+    readonly tightQuad: 64;
+    readonly matrix: 96;
+    readonly flags: 120;
+  };
+  readonly flags: {
+    readonly hasTightBox: number;
+    readonly hasLooseQuad: number;
+    readonly hasTightQuad: number;
+    readonly upright: number;
+    readonly space: number;
+    readonly empty: number;
+    readonly synthesized: number;
+  };
+};
 //#endregion
 //#region src/features/render/PageRenderReader.d.ts
 declare class PageRenderReader {
@@ -1477,5 +1550,5 @@ declare class WorkerHost {
   private saveLayerArtifact;
 }
 //#endregion
-export { ALL_STANDARD_SECURITY_PERMISSIONS, type AcquiredBaseDocument, ActionReadBudgetTracker, AnnotationAppearanceReader, AnnotationMutator, AnnotationReader, AttachmentMutator, AttachmentReader, BaseDocumentRegistry, CloseStack, DocumentActionsReader, PagesMutator as DocumentPagesMutator, PagesMutator, DocumentSaver, DocumentSession, EventHub, FontRegistrar, FormMutator, FormReader, FormsEffectsApplier, type ImpactInputs, type LayerSource, LocalRevisionAuthority, MetadataMutator, MetadataReader, type MutationKind, type OpenedPdfDocument, type OpenedPdfDocumentKind, PageGeometryReader, PagesReader as PageLayoutReader, PagesReader, PagePtrPool, PageRenderReader as PageRasterReader, PageRenderReader, type PageRecord, PageTextReader, PagesExtractor, PagesFlattener, PagesInserter, RawAnnotationReader, RedactionApplier, type RevisionAuthority, SearchReader, SecurityReader, SessionEventPublisher, type StartupFontSpec, WorkerHost, computeMutationImpact, destroyLibrary, disposeFormModel, ensureInitialized, formatPdfDate, generateUuid, hasAllStandardSecurityPermissions, normalizeU32, openFatMemoryDocument, openLayerDocument, pdfDateToIso, readActionModel, readAttachmentFileInfo, throwIfAborted, writeAttachmentFilePayload };
+export { ALL_STANDARD_SECURITY_PERMISSIONS, type AcquiredBaseDocument, ActionReadBudgetTracker, AnnotationAppearanceReader, AnnotationMutator, AnnotationReader, AttachmentMutator, AttachmentReader, BaseDocumentRegistry, CloseStack, DocumentActionsReader, PagesMutator as DocumentPagesMutator, PagesMutator, DocumentSaver, DocumentSession, EPDF_CHAR_GEOMETRY_LAYOUT, EventHub, FontRegistrar, FormMutator, FormReader, FormsEffectsApplier, type ImpactInputs, type LayerSource, LocalRevisionAuthority, MetadataMutator, MetadataReader, type MutationKind, type OpenedPdfDocument, type OpenedPdfDocumentKind, PageGeometryReader, PagesReader as PageLayoutReader, PagesReader, PagePtrPool, PageRenderReader as PageRasterReader, PageRenderReader, type PageRecord, PageTextReader, PagesExtractor, PagesFlattener, PagesInserter, RawAnnotationReader, type RawGeometryGlyphRecord, RedactionApplier, type RevisionAuthority, SearchReader, SecurityReader, SessionEventPublisher, type StartupFontSpec, WorkerHost, buildRunsFromRawGlyphs, computeMutationImpact, destroyLibrary, disposeFormModel, ensureInitialized, formatPdfDate, generateUuid, hasAllStandardSecurityPermissions, normalizeU32, openFatMemoryDocument, openLayerDocument, pdfDateToIso, readActionModel, readAttachmentFileInfo, throwIfAborted, writeAttachmentFilePayload };
 //# sourceMappingURL=index.d.cts.map

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-import { ActionReadBudget, AnnotationActor, AnnotationAppearanceRenderOptions, AnnotationAppearancesResult, AnnotationCreateResult, AnnotationDeleteResult, AnnotationListMutationMeta, AnnotationListPageSnapshot, AnnotationListSnapshotAllPages, AnnotationMoveResult, AnnotationMutationKind, AnnotationRef, AnnotationStableId, AnnotationUpdateResult, AttachmentCreateResult, AttachmentDeleteResult, AttachmentFileInfo, AttachmentFileWorkerPayload, DocumentActionsSnapshot, DocumentEvent, DocumentEventInit, DocumentEventStream, DocumentMetadata, DocumentSecurityProbeInfo, EmbeddedFileItem, EmbeddedFileRef, FormDataFormat, FormEffect, FormEffectsResult, FormFieldDTO, FormFieldDraft, FormFieldPatch, FormFieldRef, FormFieldValue, FormImportResult, FormRepairResult, FormSetValueResult, FormSnapshot, FormWidgetRef, MetadataPatch, MetadataUpdateResult, PageDeleteResult, PageFlattenResult, PageFlattenUsage, PageGeometrySnapshot, PageInsertResult, PageListSnapshot, PageMoveResult, PageObjectNumber, PageRenderOptions, PageRotateResult, PageRotation, PageState, PageTextSnapshot, PdfActionTree, PdfSaveMode, RedactionApplyResult, RedactionApplyScope, RevisionToken, SearchRequest, SearchSlice, WeakAnnotationState, WireAnnotationDraft, WireAnnotationPatch, WireAttachmentFile, WirePack, WireResource, WireResourceMap, WorkerRequest, WorkerResponse } from "@embedpdf/engine-core/runtime";
+import { ActionReadBudget, AnnotationActor, AnnotationAppearanceRenderOptions, AnnotationAppearancesResult, AnnotationCreateResult, AnnotationDeleteResult, AnnotationListMutationMeta, AnnotationListPageSnapshot, AnnotationListSnapshotAllPages, AnnotationMoveResult, AnnotationMutationKind, AnnotationRef, AnnotationStableId, AnnotationUpdateResult, AttachmentCreateResult, AttachmentDeleteResult, AttachmentFileInfo, AttachmentFileWorkerPayload, DocumentActionsSnapshot, DocumentEvent, DocumentEventInit, DocumentEventStream, DocumentMetadata, DocumentSecurityProbeInfo, EmbeddedFileItem, EmbeddedFileRef, FormDataFormat, FormEffect, FormEffectsResult, FormFieldDTO, FormFieldDraft, FormFieldPatch, FormFieldRef, FormFieldValue, FormImportResult, FormRepairResult, FormSetValueResult, FormSnapshot, FormWidgetRef, MetadataPatch, MetadataUpdateResult, PageDeleteResult, PageFlattenResult, PageFlattenUsage, PageGeometryRun, PageGeometrySnapshot, PageInsertResult, PageListSnapshot, PageMoveResult, PageObjectNumber, PageRenderOptions, PageRotateResult, PageRotation, PageState, PageTextSnapshot, PdfActionTree, PdfQuad, PdfRect, PdfSaveMode, RedactionApplyResult, RedactionApplyScope, RevisionToken, SearchRequest, SearchSlice, WeakAnnotationState, WireAnnotationDraft, WireAnnotationPatch, WireAttachmentFile, WirePack, WireResource, WireResourceMap, WorkerRequest, WorkerResponse } from "@embedpdf/engine-core/runtime";
 import { PdfFunctions, PdfRuntimeMemory, PdfRuntimeModule, Ptr } from "@embedpdf/engine-runtime";
 //#region src/runtime/lifecycle/bootstrap.d.ts
 /**
@@ -626,6 +626,7 @@ declare class RedactionApplier {
    */
   private preflight;
   private applyAllOnPage;
+  private pageHasRedactAnnotations;
   private applyRefsOnPage;
 }
 //#endregion
@@ -658,21 +659,93 @@ declare class PageTextReader {
 }
 //#endregion
 //#region src/features/geometry/PageGeometryReader.d.ts
+/**
+ * One glyph as read from `EPDFText_GetCharGeometry`, plus the run-grouping
+ * context sampled alongside it. Pure data — no native handles — so the
+ * grouping stage below is a pure function the tests can drive directly.
+ */
+interface RawGeometryGlyphRecord {
+  /** Grouping key: the glyph's text object handle (opaque identity). */
+  objectKey: number | bigint;
+  /** Sampled at each text object's first glyph (run splits inherit it). */
+  fontSize?: number;
+  /**
+   * Wire flags (bit 1 = space, bit 2 = empty). Synthesized-but-visible
+   * glyphs (/ActualText pieces) stay 0, exactly like the legacy reader —
+   * they represent real, selectable text.
+   */
+  flags: number;
+  /** Page-space AABB (zeroed when empty). */
+  looseBox: PdfRect;
+  tightBox?: PdfRect;
+  /** Oriented cells — real glyphs with a usable (non-singular) matrix only. */
+  looseQuad?: PdfQuad;
+  tightQuad?: PdfQuad;
+  /** Native uprightness of the char matrix (empty glyphs: true, inert). */
+  upright: boolean;
+  /** Baseline angle (radians CCW, PDF y-up); 0 when upright. */
+  rotation: number;
+  /** True when the ascent vector maps opposite the rotated +y (det < 0). */
+  ascentFlip: boolean;
+}
 /**
  * Geometry-only text layout reader.
  *
  * Emits geometry in PDF user space (y-up edges) — the canonical engine
  * geometry. The viewer converts to content/view space via the page geometry
  * matrix; this reader applies NO Y-flip or device transform.
+ *
+ * One `EPDFText_GetCharGeometry` call per glyph supplies boxes, oriented
+ * cells, and flags together; `buildRunsFromRawGlyphs` then groups glyphs
+ * into the upright/rotated run union. Upright runs are byte-identical to the
+ * pre-orientation reader (quads the native layer offers are deliberately
+ * dropped — the dominant case never pays for orientation).
  */
 declare class PageGeometryReader {
   private readonly runtime;
   private readonly session;
   constructor(runtime: PdfRuntimeModule, session: DocumentSession);
   read(pageObjectNumber: PageObjectNumber, signal: AbortSignal): PageGeometrySnapshot;
-  private buildRunsFromGlyphs;
-  private readGlyphInfo;
+  private readGlyphRaw;
 }
+/**
+ * Group per-glyph records into the run union. Pure — exported for tests.
+ *
+ * Runs split on text-object change (the legacy rule) and on orientation
+ * change between classifiable glyphs (θ drift / mixed orientation). A run's
+ * variant is decided by its FIRST classifiable glyph:
+ *   - empty glyphs never classify (they adopt the run's variant);
+ *   - real glyphs WITHOUT an oriented cell (singular matrix, synthesized
+ *     /ActualText pieces) classify as upright — box-only data degrades to
+ *     exactly the legacy behavior;
+ *   - real glyphs with a non-upright matrix classify as rotated.
+ * Runs that never see a classifiable glyph emit as upright (degenerate,
+ * legacy behavior).
+ */
+declare function buildRunsFromRawGlyphs(records: RawGeometryGlyphRecord[]): PageGeometryRun[];
+//#endregion
+//#region src/runtime/memory/structs.d.ts
+/** Generated from `EPDF_CHAR_GEOMETRY`; pinned by native static_asserts. */
+declare const EPDF_CHAR_GEOMETRY_LAYOUT: {
+  readonly bytes: 124;
+  readonly offsets: {
+    readonly looseBox: 0;
+    readonly tightBox: 16;
+    readonly looseQuad: 32;
+    readonly tightQuad: 64;
+    readonly matrix: 96;
+    readonly flags: 120;
+  };
+  readonly flags: {
+    readonly hasTightBox: number;
+    readonly hasLooseQuad: number;
+    readonly hasTightQuad: number;
+    readonly upright: number;
+    readonly space: number;
+    readonly empty: number;
+    readonly synthesized: number;
+  };
+};
 //#endregion
 //#region src/features/render/PageRenderReader.d.ts
 declare class PageRenderReader {
@@ -1477,5 +1550,5 @@ declare class WorkerHost {
   private saveLayerArtifact;
 }
 //#endregion
-export { ALL_STANDARD_SECURITY_PERMISSIONS, type AcquiredBaseDocument, ActionReadBudgetTracker, AnnotationAppearanceReader, AnnotationMutator, AnnotationReader, AttachmentMutator, AttachmentReader, BaseDocumentRegistry, CloseStack, DocumentActionsReader, PagesMutator as DocumentPagesMutator, PagesMutator, DocumentSaver, DocumentSession, EventHub, FontRegistrar, FormMutator, FormReader, FormsEffectsApplier, type ImpactInputs, type LayerSource, LocalRevisionAuthority, MetadataMutator, MetadataReader, type MutationKind, type OpenedPdfDocument, type OpenedPdfDocumentKind, PageGeometryReader, PagesReader as PageLayoutReader, PagesReader, PagePtrPool, PageRenderReader as PageRasterReader, PageRenderReader, type PageRecord, PageTextReader, PagesExtractor, PagesFlattener, PagesInserter, RawAnnotationReader, RedactionApplier, type RevisionAuthority, SearchReader, SecurityReader, SessionEventPublisher, type StartupFontSpec, WorkerHost, computeMutationImpact, destroyLibrary, disposeFormModel, ensureInitialized, formatPdfDate, generateUuid, hasAllStandardSecurityPermissions, normalizeU32, openFatMemoryDocument, openLayerDocument, pdfDateToIso, readActionModel, readAttachmentFileInfo, throwIfAborted, writeAttachmentFilePayload };
+export { ALL_STANDARD_SECURITY_PERMISSIONS, type AcquiredBaseDocument, ActionReadBudgetTracker, AnnotationAppearanceReader, AnnotationMutator, AnnotationReader, AttachmentMutator, AttachmentReader, BaseDocumentRegistry, CloseStack, DocumentActionsReader, PagesMutator as DocumentPagesMutator, PagesMutator, DocumentSaver, DocumentSession, EPDF_CHAR_GEOMETRY_LAYOUT, EventHub, FontRegistrar, FormMutator, FormReader, FormsEffectsApplier, type ImpactInputs, type LayerSource, LocalRevisionAuthority, MetadataMutator, MetadataReader, type MutationKind, type OpenedPdfDocument, type OpenedPdfDocumentKind, PageGeometryReader, PagesReader as PageLayoutReader, PagesReader, PagePtrPool, PageRenderReader as PageRasterReader, PageRenderReader, type PageRecord, PageTextReader, PagesExtractor, PagesFlattener, PagesInserter, RawAnnotationReader, type RawGeometryGlyphRecord, RedactionApplier, type RevisionAuthority, SearchReader, SecurityReader, SessionEventPublisher, type StartupFontSpec, WorkerHost, buildRunsFromRawGlyphs, computeMutationImpact, destroyLibrary, disposeFormModel, ensureInitialized, formatPdfDate, generateUuid, hasAllStandardSecurityPermissions, normalizeU32, openFatMemoryDocument, openLayerDocument, pdfDateToIso, readActionModel, readAttachmentFileInfo, throwIfAborted, writeAttachmentFilePayload };
 //# sourceMappingURL=index.d.ts.map