web-doc 0.11.0 → 0.12.0

package/dist/edit/pdf/range-map.js

@@ -10,17 +10,23 @@ export function mapRangeThrough(range, records) {
     return { start: start.position, end: end.position };
 }
 function carry(position, records) {
-    const carried = { position, gone: false };
+    const carried = { position, gone: false, migrations: [] };
     for (const record of records) {
         const forward = record.kind === "apply" || record.kind === "redo";
-        const operations = forward
-            ? record.operations
-            : [...record.operations].reverse();
-        for (const operation of operations) {
-            if (forward)
+        const indexed = record.operations.map((operation, index) => ({
+            operation,
+            index,
+        }));
+        for (const { operation, index } of forward ? indexed : indexed.reverse()) {
+            const migrations = (record.textAnchorMigrations ?? []).filter((migration) => migration.operationIndex === index);
+            if (forward) {
+                migrateForward(carried, migrations);
                 applyForward(carried, operation);
-            else
+            }
+            else {
                 applyBackward(carried, operation);
+                migrateBackward(carried, migrations);
+            }
         }
         const { receipt } = record;
         const id = carried.position.elementId;
@@ -34,6 +40,30 @@ function carry(position, records) {
     }
     return carried;
 }
+function migrateForward(carried, migrations) {
+    const migration = migrations.find((entry) => entry.sourceElementId === carried.position.elementId);
+    if (!migration || carried.gone)
+        return;
+    carried.migrations.push(migration);
+    carried.position = {
+        elementId: migration.targetElementId,
+        offset: migration.targetOffset +
+            Math.min(carried.position.offset, migration.sourceLength),
+    };
+}
+function migrateBackward(carried, migrations) {
+    const previous = carried.migrations.at(-1);
+    if (!previous ||
+        carried.position.elementId !== previous.targetElementId ||
+        !migrations.some((entry) => entry.sourceElementId === previous.sourceElementId &&
+            entry.targetElementId === previous.targetElementId))
+        return;
+    carried.migrations.pop();
+    carried.position = {
+        elementId: previous.sourceElementId,
+        offset: Math.max(0, Math.min(previous.sourceLength, carried.position.offset - previous.targetOffset)),
+    };
+}
 function applyForward(carried, operation) {
     const id = carried.position.elementId;
     if (operation.op === "deleteElement" && targetOf(operation) === id) {
@@ -84,7 +114,7 @@ function applyBackward(carried, operation) {
     };
 }
 function asReplace(operation, elementId) {
-    if (operation.op !== "replaceText")
+    if (operation.op !== "replaceText" && operation.op !== "replaceParagraphText")
         return undefined;
     const candidate = operation;
     if (candidate.target !== elementId || typeof candidate.text !== "string")

package/dist/edit/pdf/schemas.d.ts

@@ -9,3 +9,5 @@ export declare const tableMarkSchema: JsonSchema;
 /** The mark on a table's text objects: the id alone. */
 export declare const tableMemberMarkSchema: JsonSchema;
 export declare const textBoxMarkSchema: JsonSchema;
+/** Persisted native imported paragraph; no unvalidated layout inputs are trusted. */
+export declare const paragraphMarkSchema: JsonSchema;

package/dist/edit/pdf/schemas.js

@@ -143,6 +143,11 @@ export const pdfOperationSchemas = Object.freeze({
             target: { $ref: "#/$defs/target" },
             style: { $ref: "#/$defs/textBoxStyle" },
         }, ["target", "style"]),
+        replaceParagraphText: operation("replaceParagraphText", {
+            target: { $ref: "#/$defs/target" },
+            text: { type: "string", maxLength: 20000 },
+            range: { $ref: "#/$defs/textRange" },
+        }, ["target", "text"]),
         resizeElement: operation("resizeElement", {
             target: { $ref: "#/$defs/target" },
             rect: { $ref: "#/$defs/rect" },
@@ -292,3 +297,35 @@ export const textBoxMarkSchema = {
     },
     $defs: definitions,
 };
+/** Persisted native imported paragraph; no unvalidated layout inputs are trusted. */
+export const paragraphMarkSchema = {
+    type: "object",
+    required: ["kind", "id", "rect", "text", "lines", "style", "baselineOffset"],
+    additionalProperties: false,
+    properties: {
+        kind: { const: "paragraph" },
+        id: { $ref: "#/$defs/target" },
+        rect: { $ref: "#/$defs/rect" },
+        text: { type: "string", maxLength: 20000 },
+        baselineOffset: { type: "number", minimum: 0, maximum: 2 },
+        lines: {
+            type: "array",
+            maxItems: 20000,
+            items: { type: "string", maxLength: 20000 },
+        },
+        style: {
+            type: "object",
+            required: [
+                "fontFamily",
+                "fontSize",
+                "bold",
+                "italic",
+                "color",
+                "lineHeight",
+            ],
+            additionalProperties: false,
+            properties: definitions.textBoxStyle.properties,
+        },
+    },
+    $defs: definitions,
+};

package/dist/edit/pdf/session.d.ts

@@ -2,7 +2,7 @@ import type { TextSelection } from "../../contracts.js";
 import type { DescribeOptions, DocumentDescription, EditCheckpoint, OutlineOptions, OutlineResult, TargetCandidate, TargetQuery, ToolCall, ToolCallOptions, ToolResult, ToolSet } from "../ai/types.js";
 import type { EditSessionCore } from "../engine.js";
 import type { ApplyOptions, AssetOptions, EditFindOptions, EditOperation, EditReceipt, EditState, ElementQuery, HistoryOptions, OperationSchemaSet, PagePoint, PageRect, ReadItem, ReadOptions, ReadResult, SavedDocument, TextPosition, TextRange, TextTarget } from "../types.js";
-import type { DeleteElementOperation, DeletePageOperation, Fields, InsertImageOperation, InsertPageOperation, InsertShapeOperation, InsertTableOperation, InsertTextBoxOperation, MoveElementOperation, MovePageOperation, PdfEditSession, PdfElement, PdfOperation, PdfSaveOptions, ReplaceTextOperation, ResizeElementOperation, RotatePageOperation, SetShapeStyleOperation, SetTableCellOperation, PageBitmap, PageLayout, RenderOptions, SetTextStyleOperation, TextFont, TextLayout } from "./types.js";
+import type { DeleteElementOperation, DeletePageOperation, Fields, InsertImageOperation, InsertPageOperation, InsertShapeOperation, InsertTableOperation, InsertTextBoxOperation, MoveElementOperation, MovePageOperation, PdfEditSession, PdfElement, PdfOperation, PdfSaveOptions, PdfTextParagraph, ReplaceParagraphTextOperation, ReplaceTextOperation, ResizeElementOperation, RotatePageOperation, SetShapeStyleOperation, SetTableCellOperation, PageBitmap, PageLayout, RenderOptions, SetTextStyleOperation, TextFont, TextLayout } from "./types.js";
 /**
  * The PDF session: the core session narrowed to PDF operations and elements.
  * Typed methods are added here as their operations ship; each is `apply()`
@@ -38,6 +38,8 @@ export declare class PdfSession implements PdfEditSession {
     dropCheckpoint(id: string): void;
     get tools(): ToolSet;
     callTool(call: ToolCall, options?: ToolCallOptions): Promise<ToolResult>;
+    getTextParagraph(elementId: string, options?: ReadOptions): Promise<ReadItem<PdfTextParagraph>>;
+    replaceParagraphText(fields: Fields<ReplaceParagraphTextOperation>, options?: ApplyOptions): Promise<EditReceipt>;
     getTextLayout(elementId: string, options?: ReadOptions): Promise<ReadItem<TextLayout>>;
     getTextFont(elementId: string, options?: ReadOptions): Promise<ReadItem<TextFont>>;
     positionAt(pageIndex: number, point: PagePoint, options?: ReadOptions): Promise<ReadItem<TextPosition>>;

package/dist/edit/pdf/session.js

@@ -66,22 +66,25 @@ export class PdfSession {
         if (receipt.dryRun || receipt.revision === last)
             return;
         this.#refreshGeometry(receipt);
-        let involved = operations;
+        let involved = {
+            operations,
+            textAnchorMigrations: receipt.textAnchorMigrations ?? [],
+        };
         switch (kind) {
             case "apply":
-                this.#applied.push(operations);
+                this.#applied.push(involved);
                 this.#undone = [];
                 break;
             case "undo":
-                involved = this.#applied.pop() ?? [];
+                involved = this.#applied.pop() ?? involved;
                 this.#undone.push(involved);
                 break;
             case "redo":
-                involved = this.#undone.pop() ?? [];
+                involved = this.#undone.pop() ?? involved;
                 this.#applied.push(involved);
                 break;
             case "reset":
-                involved = this.#applied.flat();
+                involved = combineBatches(this.#applied);
                 this.#applied = [];
                 this.#undone = [];
                 break;
@@ -95,7 +98,8 @@ export class PdfSession {
         this.#log.push({
             revision: receipt.revision,
             kind,
-            operations: involved,
+            operations: involved.operations,
+            textAnchorMigrations: involved.textAnchorMigrations,
             receipt,
         });
         if (this.#log.length > LOG_LIMIT)
@@ -195,6 +199,17 @@ export class PdfSession {
     callTool(call, options) {
         return runTool(this, this.tools, call, options);
     }
+    getTextParagraph(elementId, options) {
+        return this.#core.readItem(options, (engine, signal) => {
+            const reads = pdfReads(engine);
+            if (!reads.textParagraph)
+                throw new ViewerError("edit-unsupported", "This engine does not resolve PDF paragraphs");
+            return reads.textParagraph(elementId, signal);
+        });
+    }
+    replaceParagraphText(fields, options) {
+        return this.apply([{ op: "replaceParagraphText", ...fields }], options);
+    }
     getTextLayout(elementId, options) {
         return this.#core.readItem(options, (engine, signal) => pdfReads(engine).textLayout(elementId, signal));
     }
@@ -302,6 +317,20 @@ export class PdfSession {
 }
 /** How many committed calls `mapRange` can look back over. */
 const LOG_LIMIT = 512;
+/** Reset reverses all committed batches, retaining each migration's operation. */
+function combineBatches(batches) {
+    const operations = [];
+    const textAnchorMigrations = [];
+    for (const batch of batches) {
+        for (const migration of batch.textAnchorMigrations)
+            textAnchorMigrations.push({
+                ...migration,
+                operationIndex: operations.length + migration.operationIndex,
+            });
+        operations.push(...batch.operations);
+    }
+    return { operations, textAnchorMigrations };
+}
 /** The engine behind a PDF session answers the overlay reads; a stand-in may not. */
 function pdfReads(engine) {
     const reads = engine;

package/dist/edit/pdf/types.d.ts

@@ -1,6 +1,6 @@
 import type { TextSelection } from "../../contracts.js";
 import type { ApplyOptions, BinaryData, EditElement, EditReceipt, EditSessionBase, PagePoint, PageRect, ReadItem, ReadOptions, ReadResult, SavedDocument, SaveOptions, TextPosition, TextRange } from "../types.js";
-export type PdfElementKind = "text" | "image" | "shape" | "textBox" | "table" | "other";
+export type PdfElementKind = "text" | "image" | "shape" | "textBox" | "paragraph" | "table" | "other";
 export type PdfTextAlign = "left" | "center" | "right";
 export interface PdfTextStyle {
     readonly fontFamily: string;
@@ -12,7 +12,7 @@ export interface PdfTextStyle {
     readonly color: string;
     /** Text boxes only. */
     readonly align?: PdfTextAlign;
-    /** Text boxes only; a multiple of the font size. */
+    /** Text boxes and paragraphs; a multiple of the font size. */
     readonly lineHeight?: number;
 }
 export interface PdfShapeStyle {
@@ -26,7 +26,9 @@ export interface PdfShapeStyle {
 }
 export interface PdfElement extends EditElement {
     readonly kind: PdfElementKind;
-    /** Present for `text` and `textBox`. */
+    /** Canonical native paragraph target, when this imported row belongs to one. */
+    readonly textEditingTarget?: string;
+    /** Present for `text`, `textBox` and `paragraph`. */
     readonly textStyle?: PdfTextStyle;
     /** Present for `shape` and `table`. */
     readonly shapeStyle?: PdfShapeStyle;
@@ -57,7 +59,7 @@ export interface InsertTextBoxOperation {
 }
 export interface ReplaceTextOperation {
     readonly op: "replaceText";
-    /** A `textBox` or `text` element. */
+    /** A `textBox`, `text` or `paragraph` element; supported fields depend on its capabilities. */
     readonly target: string;
     readonly text: string;
     /**
@@ -68,9 +70,33 @@ export interface ReplaceTextOperation {
      */
     readonly range?: TextRange;
 }
+/** A confidently resolved homogeneous paragraph of imported horizontal text. */
+export interface PdfTextParagraph {
+    readonly id: string;
+    readonly pageIndex: number;
+    readonly text: string;
+    readonly bounds: PageRect;
+    readonly textStyle: PdfTextStyle & {
+        readonly lineHeight: number;
+    };
+    readonly memberIds: readonly string[];
+    /** Half-open UTF-16 spans in the paragraph's logical text. */
+    readonly members: readonly {
+        readonly elementId: string;
+        readonly start: number;
+        readonly end: number;
+    }[];
+}
+export interface ReplaceParagraphTextOperation {
+    readonly op: "replaceParagraphText";
+    readonly target: string;
+    readonly text: string;
+    /** Both endpoints name the paragraph id, not an individual visual row. */
+    readonly range?: TextRange;
+}
 export interface SetTextStyleOperation {
     readonly op: "setTextStyle";
-    /** A `textBox` or `text` element. */
+    /** A `textBox`, `text`, or `paragraph` element. */
     readonly target: string;
     readonly style: PdfTextBoxStyle;
 }
@@ -211,7 +237,7 @@ export type RotatePageOperation = {
     readonly rotation?: never;
 };
 /** The PDF operation union; operations are added as they ship. */
-export type PdfOperation = InsertTextBoxOperation | ReplaceTextOperation | SetTextStyleOperation | ResizeElementOperation | MoveElementOperation | DeleteElementOperation | InsertPageOperation | DeletePageOperation | MovePageOperation | RotatePageOperation | InsertShapeOperation | SetShapeStyleOperation | InsertImageOperation | InsertTableOperation | SetTableCellOperation;
+export type PdfOperation = InsertTextBoxOperation | ReplaceTextOperation | ReplaceParagraphTextOperation | SetTextStyleOperation | ResizeElementOperation | MoveElementOperation | DeleteElementOperation | InsertPageOperation | DeletePageOperation | MovePageOperation | RotatePageOperation | InsertShapeOperation | SetShapeStyleOperation | InsertImageOperation | InsertTableOperation | SetTableCellOperation;
 export interface PdfSaveOptions extends SaveOptions {
     /**
      * `full` rewrites the file, so deleted content is gone and the bytes do not
@@ -315,10 +341,14 @@ export type Fields<T extends PdfOperation> = T extends unknown ? Omit<T, "op"> :
 export interface PdfEditSession extends EditSessionBase<PdfOperation, PdfElement> {
     readonly format: "pdf";
     save(options?: PdfSaveOptions): Promise<SavedDocument>;
-    /** Lines, glyph boxes and styles of a `text`, `textBox` or `table` element. */
+    /** Resolves an imported row or a paragraph id; undefined when grouping is unsafe. */
+    getTextParagraph(elementId: string, options?: ReadOptions): Promise<ReadItem<PdfTextParagraph>>;
+    /** Reflows the complete paragraph as one atomic history change. */
+    replaceParagraphText(fields: Fields<ReplaceParagraphTextOperation>, options?: ApplyOptions): Promise<EditReceipt>;
+    /** Lines, glyph boxes and styles of a `text`, `textBox`, `paragraph` or `table` element. */
     getTextLayout(elementId: string, options?: ReadOptions): Promise<ReadItem<TextLayout>>;
     /**
-     * The font a `text` or `textBox` element is drawn in, as a browser loads it,
+     * The font a `text`, `textBox` or `paragraph` element is drawn in, as a browser loads it,
      * or why it cannot be loaded; see `TextFont`.
      */
     getTextFont(elementId: string, options?: ReadOptions): Promise<ReadItem<TextFont>>;

package/dist/edit/session.js

@@ -739,6 +739,11 @@ export class EditSessionController {
             ...(change.remappedIds
                 ? { remappedIds: Object.freeze({ ...change.remappedIds }) }
                 : {}),
+            ...(change.textAnchorMigrations?.length
+                ? {
+                    textAnchorMigrations: Object.freeze(change.textAnchorMigrations.map((migration) => Object.freeze({ ...migration }))),
+                }
+                : {}),
             changedPages: Object.freeze([...change.changedPages]),
             pageCount: change.pageCount,
             warnings: Object.freeze(change.warnings.map((warning) => Object.freeze({ ...warning }))),

package/dist/edit/types.d.ts

@@ -214,6 +214,17 @@ export interface AssetOptions {
     readonly mimeType?: string;
     readonly signal?: AbortSignal;
 }
+/** A source text object incorporated into another logical text element. */
+export interface TextAnchorMigration {
+    /** Zero-based operation in the applied batch, before its text replacement. */
+    readonly operationIndex: number;
+    readonly sourceElementId: string;
+    readonly targetElementId: string;
+    /** UTF-16 length retained from the source, excluding generated trailing whitespace. */
+    readonly sourceLength: number;
+    /** UTF-16 start of the source text in the target before the operation. */
+    readonly targetOffset: number;
+}
 export interface EditReceipt {
     readonly sessionId: string;
     /** `state.revision` after the call; unchanged for a dry run or a no-op. */
@@ -226,6 +237,8 @@ export interface EditReceipt {
     readonly removedIds: readonly string[];
     /** Old id → new id, when a format has to rename an element; absent for formats that never do. */
     readonly remappedIds?: Readonly<Record<string, string>>;
+    /** Text anchors migrated by an apply (or preview); absent when none migrated. */
+    readonly textAnchorMigrations?: readonly TextAnchorMigration[];
     /** Page indexes in the resulting document whose content may have changed; a superset. */
     readonly changedPages: readonly number[];
     readonly pageCount: number;

package/dist/viewport.js

@@ -155,6 +155,15 @@ export class ViewerViewport {
         this.schedule();
     }
     replaceDocument(info, changedPages, revision) {
+        // The viewer retires the old adapter after this synchronous handoff.
+        // Unfinished paints still belong to that adapter, even on unchanged pages.
+        for (const slot of this.#slots.values()) {
+            if (!slot.controller)
+                continue;
+            slot.controller.abort();
+            delete slot.controller;
+            delete slot.renderKey;
+        }
         const previous = this.#info;
         const painted = this.#painted.get(this.#contentRevision);
         this.#painted.clear();
@@ -533,6 +542,8 @@ export class ViewerViewport {
             await rendering?.catch(() => undefined);
             canvas.width = 0;
             canvas.height = 0;
+            if (slot.controller === controller)
+                delete slot.controller;
         }
     }
     async #buildTextLayers(textLayer, highlightLayer, pageIndex, runs, zoom, signal) {

package/dist/worker-protocol.d.ts

@@ -2,7 +2,7 @@ import type { DocumentFormat, DocumentInfo, ResourceLimits, TextRun, ViewerError
 import type { EditableFormat } from "./edit/types.js";
 export type DocumentWorkerOperation = "init" | "open" | "get-info" | "render" | "get-text-map" | "close" | "destroy";
 /** Operations of an edit engine worker; payloads mirror the engine interface. */
-export type EditWorkerOperation = "edit-init" | "edit-open" | "edit-validate" | "edit-apply" | "edit-materialize" | "edit-restore" | "edit-put-asset" | "edit-elements" | "edit-element" | "edit-elements-at" | "edit-find-text" | "edit-text-layout" | "edit-text-font" | "edit-position-at" | "edit-range-rects" | "edit-render-without" | "edit-page-layout" | "edit-pptx-slides" | "edit-pptx-layouts" | "edit-docx-revisions" | "edit-docx-text-style" | "edit-pptx-text-style" | "edit-dispose";
+export type EditWorkerOperation = "edit-init" | "edit-open" | "edit-validate" | "edit-apply" | "edit-materialize" | "edit-restore" | "edit-put-asset" | "edit-elements" | "edit-element" | "edit-elements-at" | "edit-find-text" | "edit-text-layout" | "edit-text-paragraph" | "edit-text-font" | "edit-position-at" | "edit-range-rects" | "edit-render-without" | "edit-page-layout" | "edit-pptx-slides" | "edit-pptx-layouts" | "edit-docx-revisions" | "edit-docx-text-style" | "edit-pptx-text-style" | "edit-dispose";
 export type WorkerOperation = DocumentWorkerOperation | EditWorkerOperation;
 export interface WorkerRequest {
     readonly kind: "request";