web-doc 0.6.2 → 0.7.0

package/dist/edit/docx/structure-ops.js

@@ -0,0 +1,494 @@
+import { resolveBinary } from "../assets.js";
+import { extensionForMime, IMAGE_RELATIONSHIP_TYPE } from "../ooxml/opc.js";
+import { patches } from "../ooxml/patch.js";
+import { relativeTarget } from "../ooxml/transaction.js";
+import { OFFICE_RELATIONSHIPS, W_NS } from "./ids.js";
+import { firstTextItem } from "./text.js";
+import { attributeProblem, changedRunProperties, colorProblem, namespacePatches, normalizeText, paragraphPropertiesWithoutSection, paragraphXml, runContentXml, runXml, sliceOf, textProblem, } from "./write.js";
+/*
+ * Structure: paragraphs inserted next to other elements, paragraphs,
+ * tables and inline pictures deleted, paragraphs and tables moved within
+ * their container, inline pictures inserted as a paragraph of their own.
+ * Every operation is a patch of the body part; a moved or new paragraph
+ * carries its id, so ids survive the move.
+ */
+const MEDIA_FOLDER = "/word/media";
+const WP_NS = "http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing";
+const A_NS = "http://schemas.openxmlformats.org/drawingml/2006/main";
+const PIC_NS = "http://schemas.openxmlformats.org/drawingml/2006/picture";
+const R_NS = OFFICE_RELATIONSHIPS.slice(0, -1);
+/** EMU per point. */
+const EMU_PER_PT = 12_700;
+/** The node a block occupies among its siblings: the paragraph or table itself. */
+function blockNode(record) {
+    return record.node;
+}
+/** Exactly one of `before` and `after`, resolved to its record. */
+function placement(operation, context, issue) {
+    const named = [operation.before, operation.after].filter((value) => value !== undefined);
+    if (named.length !== 1) {
+        issue("/before", "invalid-value", "Exactly one of before and after names the reference element");
+        return undefined;
+    }
+    const side = operation.before !== undefined ? "before" : "after";
+    const id = operation[side];
+    const record = context.model.byId.get(id);
+    if (!record || (record.kind !== "paragraph" && record.kind !== "table")) {
+        issue(`/${side}`, record ? "invalid-target" : "unknown-target", record
+            ? `Element ${id} is not a paragraph or a table`
+            : `No element ${id}`);
+        return undefined;
+    }
+    return { record, side };
+}
+/** The `w:rPr` bytes new text next to a reference takes: its first text run's, else its paragraph mark's. */
+function referenceRunProperties(part, record) {
+    const paragraph = record.kind === "paragraph" ? record : record.rows[0]?.[0]?.paragraphs[0];
+    if (!paragraph)
+        return undefined;
+    const first = firstTextItem(paragraph.text);
+    if (first?.rPr)
+        return first.rPr;
+    return paragraph.pPr?.children.find((child) => child.local === "rPr" && child.namespace === W_NS);
+}
+/** Ids of every `w:p` under a node, the node included when it is one. */
+function paragraphIdsUnder(model, node) {
+    const out = [];
+    if (node.local === "p" && node.namespace === W_NS) {
+        const own = model.paragraphIds.get(node);
+        if (own)
+            out.push(own);
+    }
+    for (const paragraph of model.document.findAll("p", node)) {
+        if (paragraph === node || paragraph.namespace !== W_NS)
+            continue;
+        const id = model.paragraphIds.get(paragraph);
+        if (id)
+            out.push(id);
+    }
+    return out;
+}
+/** Element ids of every listed element under a node, for `removedIds`. */
+function elementIdsUnder(model, node) {
+    const out = [];
+    for (const record of model.records) {
+        let current = record.node;
+        while (current && current !== node)
+            current = current.parent;
+        if (current === node)
+            out.push(record.elementId);
+    }
+    return out;
+}
+/**
+ * The bytes of a node with a `w14:paraId` written on every paragraph in
+ * it that lacks one, the ids being those the engine knows them by, so a
+ * copy placed elsewhere keeps the same ids.
+ */
+function stampedSlice(model, node) {
+    const part = model.document;
+    const targets = [
+        ...(node.local === "p" && node.namespace === W_NS ? [node] : []),
+        ...part
+            .findAll("p", node)
+            .filter((p) => p !== node && p.namespace === W_NS),
+    ]
+        .filter((paragraph) => {
+        const id = model.paragraphIds.get(paragraph);
+        return id !== undefined && model.unauthoredSet.has(id);
+    })
+        .sort((a, b) => b.start - a.start);
+    let xml = sliceOf(part, node);
+    const stamped = [];
+    for (const paragraph of targets) {
+        const id = model.paragraphIds.get(paragraph);
+        // A paragraph stamped by a shown base already carries the attribute.
+        if (part.attribute(paragraph, "w14:paraId")) {
+            stamped.push(id);
+            continue;
+        }
+        // After the attributes of the start tag, before ">" or "/>".
+        const last = paragraph.attributes.at(-1);
+        const from = last ? last.end : paragraph.start + 1 + paragraph.name.length;
+        const close = part.text.indexOf(">", from);
+        let at = close;
+        if (part.text[close - 1] === "/")
+            at = close - 1;
+        const offset = at - node.start;
+        xml = `${xml.slice(0, offset)} w14:paraId="${id}"${xml.slice(offset)}`;
+        stamped.push(id);
+    }
+    return { xml, stamped: stamped.reverse() };
+}
+/** The table a cell paragraph names and whether it is the table's first paragraph. */
+function firstCellParagraph(record) {
+    const table = record.table;
+    return table && table.rows[0]?.[0]?.paragraphs[0] === record
+        ? table
+        : undefined;
+}
+/**
+ * A table is named after its first paragraph; when that paragraph
+ * changes, the table's id changes with it and the receipt says so.
+ */
+function tableRemap(table, newFirstParagraphId) {
+    return table.id === newFirstParagraphId
+        ? undefined
+        : { [table.elementId]: `tbl:${newFirstParagraphId}` };
+}
+function commit(context, items, result, transaction = context.pkg.transaction()) {
+    const part = context.model.document;
+    transaction.patch(part, [...namespacePatches(part), ...items]);
+    return transaction
+        .commit()
+        .then((change) => ({ ...result, warnings: change.warnings }));
+}
+export const insertParagraphHandler = {
+    async validate(operation, context, issue) {
+        const problem = textProblem(normalizeText(operation.text));
+        if (problem)
+            issue("/text", "invalid-text", `The text holds ${problem}`);
+        if (operation.style?.color !== undefined) {
+            const colour = colorProblem(operation.style.color);
+            if (colour)
+                issue("/style/color", "invalid-value", `Colour: ${colour}`);
+        }
+        if (operation.style?.fontFamily !== undefined) {
+            const font = attributeProblem(operation.style.fontFamily);
+            if (font)
+                issue("/style/fontFamily", "invalid-value", `Font: ${font}`);
+        }
+        placement(operation, context, issue);
+    },
+    async apply(operation, context) {
+        const { record, side } = placement(operation, context, () => { });
+        const part = context.model.document;
+        const node = blockNode(record);
+        const sourceRPr = referenceRunProperties(part, record);
+        const rPr = operation.style
+            ? changedRunProperties(part, sourceRPr, operation.style, context.model.styles)
+            : sliceOf(part, sourceRPr);
+        const pPr = record.kind === "paragraph"
+            ? paragraphPropertiesWithoutSection(part, record.pPr)
+            : "";
+        const segments = normalizeText(operation.text).split("\n");
+        const createdIds = [];
+        const xml = segments.map((segment) => {
+            const id = context.freshParagraphId();
+            createdIds.push(`p:${id}`);
+            return paragraphXml(undefined, id, pPr, runXml(rPr, runContentXml(segment)));
+        });
+        // Inserts at one position land in call order, so the paragraphs
+        // keep their order on either side of the reference.
+        const items = side === "before"
+            ? xml.map((paragraph) => patches.insertBefore(part, node, paragraph))
+            : xml.map((paragraph) => patches.insertAfter(part, node, paragraph));
+        const table = side === "before" && record.kind === "paragraph"
+            ? firstCellParagraph(record)
+            : undefined;
+        const remappedIds = table
+            ? tableRemap(table, createdIds[0].slice(2))
+            : undefined;
+        return commit(context, items, {
+            createdIds,
+            reflowFrom: record.id,
+            ...(remappedIds ? { remappedIds } : {}),
+        });
+    },
+};
+/**
+ * The patch that keeps a body from ending with a table, or from holding
+ * no block at all, once `remaining` are its blocks: an empty paragraph
+ * after the last of them, or where the body's section properties start.
+ */
+function bodyTailPatch(context, remaining) {
+    const { model } = context;
+    const last = remaining.at(-1);
+    if (last && last.kind === "paragraph")
+        return undefined;
+    const id = context.freshParagraphId();
+    const xml = paragraphXml(undefined, id, "", "");
+    const part = model.document;
+    const patch = last
+        ? patches.insertAfter(part, last.node, xml)
+        : model.bodySectPr
+            ? patches.insertBefore(part, model.bodySectPr, xml)
+            : patches.appendChild(part, model.body, xml);
+    return { patch, id };
+}
+/** Whether the container of a block would keep a paragraph without it. */
+function otherParagraphsIn(model, record) {
+    if (record.table) {
+        for (const row of record.table.rows)
+            for (const cell of row)
+                if (cell.paragraphs.includes(record))
+                    return cell.paragraphs.some((candidate) => candidate !== record);
+        return false;
+    }
+    return model.blocks.some((block) => block.kind === "paragraph" && block !== record);
+}
+export const deleteElementHandler = {
+    async validate(operation, context, issue) {
+        const record = context.model.byId.get(operation.target);
+        if (!record || record.kind === "other") {
+            issue("/target", record ? "invalid-target" : "unknown-target", record
+                ? `Element ${operation.target} cannot be deleted`
+                : `No element ${operation.target}`);
+            return;
+        }
+        if (record.kind === "paragraph") {
+            if (record.sectPr)
+                issue("/target", "section-break", "A paragraph that ends a section cannot be deleted");
+            else if (record.readOnlyReason)
+                issue("/target", "invalid-target", `Paragraph ${operation.target} is read-only (${record.readOnlyReason})`);
+            else if (!otherParagraphsIn(context.model, record))
+                issue("/target", "last-paragraph", "The last paragraph of a body or a cell cannot be deleted");
+        }
+    },
+    async apply(operation, context) {
+        const { model } = context;
+        const found = model.byId.get(operation.target);
+        const part = model.document;
+        if (found.kind !== "paragraph" && found.kind !== "table")
+            return deleteInline(context, found);
+        const record = found;
+        const node = blockNode(record);
+        const items = [patches.removeElement(part, node)];
+        const createdIds = [];
+        if (model.blocks.includes(record)) {
+            // The body must not end with a table or stand empty.
+            const tail = bodyTailPatch(context, model.blocks.filter((block) => block !== record));
+            if (tail) {
+                createdIds.push(`p:${tail.id}`);
+                items.push(tail.patch);
+            }
+        }
+        const removedParagraphIds = paragraphIdsUnder(model, node);
+        const table = record.kind === "paragraph" ? firstCellParagraph(record) : undefined;
+        const successor = table?.rows[0]?.[0]?.paragraphs[1];
+        const remappedIds = table && successor ? tableRemap(table, successor.id) : undefined;
+        return commit(context, items, {
+            createdIds,
+            removedIds: elementIdsUnder(model, node),
+            removedParagraphIds,
+            reflowFrom: record.id,
+            ...(remappedIds ? { remappedIds } : {}),
+        });
+    },
+};
+/** Removes an inline object's drawing and the relationship only it used. */
+async function deleteInline(context, record) {
+    const { model } = context;
+    const part = model.document;
+    const transaction = context.pkg.transaction();
+    const blip = part.find("blip", record.node);
+    const rId = blip
+        ? (part.attribute(blip, "r:embed") ?? part.attribute(blip, "r:link"))
+        : undefined;
+    if (rId) {
+        const uses = part
+            .findAll("blip")
+            .filter((candidate) => candidate !== blip &&
+            (part.attribute(candidate, "r:embed") === rId ||
+                part.attribute(candidate, "r:link") === rId));
+        const elsewhere = new RegExp(`r:(?:id|embed|link|pict)="${rId.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}"`);
+        const others = uses.length > 0 || elsewhere.test(stripNode(part, record.node));
+        if (!others)
+            transaction.removeRelationship(model.mainPart, rId);
+    }
+    const removedParagraphIds = paragraphIdsUnder(model, record.node);
+    return commit(context, [patches.removeElement(part, record.node)], {
+        createdIds: [],
+        removedIds: [record.elementId],
+        ...(removedParagraphIds.length > 0 ? { removedParagraphIds } : {}),
+        reflowFrom: record.paragraph.id,
+    }, transaction);
+}
+/** The part's text without a node, to look for other uses of a relationship. */
+function stripNode(part, node) {
+    return part.text.slice(0, node.start) + part.text.slice(node.end);
+}
+export const moveElementHandler = {
+    async validate(operation, context, issue) {
+        const record = context.model.byId.get(operation.target);
+        if (!record || (record.kind !== "paragraph" && record.kind !== "table")) {
+            issue("/target", record ? "invalid-target" : "unknown-target", record
+                ? `Element ${operation.target} is not a paragraph or a table`
+                : `No element ${operation.target}`);
+            return;
+        }
+        if (record.kind === "paragraph" && record.sectPr) {
+            issue("/target", "section-break", "A paragraph that ends a section cannot move");
+            return;
+        }
+        if (record.kind === "paragraph" && record.readOnlyReason) {
+            issue("/target", "invalid-target", `Paragraph ${operation.target} is read-only (${record.readOnlyReason})`);
+            return;
+        }
+        const reference = placement(operation, context, issue);
+        if (!reference)
+            return;
+        if (reference.record === record) {
+            issue(`/${reference.side}`, "invalid-target", "An element cannot move next to itself");
+            return;
+        }
+        if (blockNode(reference.record).parent !== blockNode(record).parent)
+            issue(`/${reference.side}`, "invalid-target", "The reference must be in the same body or cell as the element");
+    },
+    async apply(operation, context) {
+        const { model } = context;
+        const record = model.byId.get(operation.target);
+        const { record: reference, side } = placement(operation, context, () => { });
+        const part = model.document;
+        const node = blockNode(record);
+        const target = blockNode(reference);
+        const { xml, stamped } = stampedSlice(model, node);
+        const items = [
+            patches.removeElement(part, node),
+            side === "before"
+                ? patches.insertBefore(part, target, xml)
+                : patches.insertAfter(part, target, xml),
+        ];
+        const createdIds = [];
+        if (model.blocks.includes(record) && model.blocks.includes(reference)) {
+            // The body must not end with a table once the blocks are reordered.
+            const without = model.blocks.filter((block) => block !== record);
+            const at = without.indexOf(reference);
+            without.splice(side === "before" ? at : at + 1, 0, record);
+            const tail = bodyTailPatch(context, without);
+            if (tail) {
+                createdIds.push(`p:${tail.id}`);
+                items.push(tail.patch);
+            }
+        }
+        // The document reflows from whichever of the two comes first.
+        const order = (candidate) => model.records.indexOf(candidate);
+        const first = order(record) < order(reference) ? record : reference;
+        // Within a table's first cell, the move may change which paragraph
+        // comes first, and so the table's id.
+        let remappedIds;
+        if (record.kind === "paragraph" && record.table) {
+            const cell = record.table.rows[0]?.[0];
+            if (cell && cell.paragraphs.includes(record)) {
+                const without = cell.paragraphs.filter((p) => p !== record);
+                const at = without.indexOf(reference);
+                const reordered = [...without];
+                if (at >= 0)
+                    reordered.splice(side === "before" ? at : at + 1, 0, record);
+                const head = reordered[0];
+                if (head)
+                    remappedIds = tableRemap(record.table, head.id);
+            }
+        }
+        return commit(context, items, {
+            createdIds,
+            stamped,
+            reflowFrom: first.id,
+            ...(remappedIds ? { remappedIds } : {}),
+        });
+    },
+};
+/** A media part of the package with these bytes and this type's extension, if any. */
+async function existingMedia(context, bytes, mimeType) {
+    const extension = `.${extensionForMime(mimeType)}`;
+    for (const name of context.pkg.currentPartNames) {
+        if (!name.startsWith(`${MEDIA_FOLDER}/`) || !name.endsWith(extension))
+            continue;
+        const entry = context.pkg.entry(name);
+        if (entry && entry.uncompressedSize !== bytes.length)
+            continue;
+        const candidate = await context.pkg.part(name);
+        if (candidate.length !== bytes.length)
+            continue;
+        let same = true;
+        for (let index = 0; index < bytes.length; index += 1)
+            if (candidate[index] !== bytes[index]) {
+                same = false;
+                break;
+            }
+        if (same)
+            return name;
+    }
+    return undefined;
+}
+/** Whether the bytes carry the signature of the declared image type. */
+function matchesMime(bytes, mimeType) {
+    if (mimeType === "image/png")
+        return (bytes.length >= 8 &&
+            bytes[0] === 0x89 &&
+            bytes[1] === 0x50 &&
+            bytes[2] === 0x4e &&
+            bytes[3] === 0x47);
+    return (bytes.length >= 3 &&
+        bytes[0] === 0xff &&
+        bytes[1] === 0xd8 &&
+        bytes[2] === 0xff);
+}
+/** One above every `wp:docPr/@id` of the part. */
+function nextDrawingId(part) {
+    let max = 0;
+    for (const node of part.findAll("docPr")) {
+        if (node.namespace !== WP_NS)
+            continue;
+        const id = Number(part.attribute(node, "id") ?? NaN);
+        if (Number.isInteger(id))
+            max = Math.max(max, id);
+    }
+    return max + 1;
+}
+/** Patches that declare the namespaces an inline picture's markup uses on the root. */
+function drawingNamespacePatches(part) {
+    const root = part.root;
+    const has = (name) => root.attributes.some((attribute) => attribute.name === name);
+    const items = [];
+    if (!has("xmlns:wp"))
+        items.push(patches.setAttribute(part, root, "xmlns:wp", WP_NS));
+    if (!has("xmlns:r"))
+        items.push(patches.setAttribute(part, root, "xmlns:r", R_NS));
+    return items;
+}
+export const insertImageHandler = {
+    async validate(operation, context, issue) {
+        placement(operation, context, issue);
+        try {
+            const bytes = resolveBinary(operation.data, context.assets);
+            if (bytes.length === 0)
+                issue("/data", "invalid-value", "The image has no bytes");
+            else if (!matchesMime(bytes, operation.mimeType))
+                issue("/data", "invalid-value", `The bytes are not a ${operation.mimeType} image`);
+        }
+        catch {
+            issue("/data", "unknown-asset", "The asset is not known to the session");
+        }
+    },
+    async apply(operation, context) {
+        const { model } = context;
+        const { record, side } = placement(operation, context, () => { });
+        const part = model.document;
+        const bytes = resolveBinary(operation.data, context.assets);
+        const transaction = context.pkg.transaction();
+        // The same bytes already in the package are related, not stored again.
+        const existing = await existingMedia(context, bytes, operation.mimeType);
+        const rId = existing
+            ? await transaction.addRelationship(model.mainPart, IMAGE_RELATIONSHIP_TYPE, relativeTarget(model.mainPart, existing))
+            : (await transaction.addMedia(model.mainPart, MEDIA_FOLDER, bytes, operation.mimeType)).rId;
+        const cx = Math.round(operation.size.width * EMU_PER_PT);
+        const cy = Math.round(operation.size.height * EMU_PER_PT);
+        const drawingId = nextDrawingId(part);
+        const id = context.freshParagraphId();
+        const drawing = `<w:r><w:drawing><wp:inline distT="0" distB="0" distL="0" distR="0"><wp:extent cx="${cx}" cy="${cy}"/><wp:effectExtent l="0" t="0" r="0" b="0"/><wp:docPr id="${drawingId}" name="Picture ${drawingId}"/>` +
+            `<wp:cNvGraphicFramePr><a:graphicFrameLocks xmlns:a="${A_NS}" noChangeAspect="1"/></wp:cNvGraphicFramePr>` +
+            `<a:graphic xmlns:a="${A_NS}"><a:graphicData uri="${PIC_NS}"><pic:pic xmlns:pic="${PIC_NS}"><pic:nvPicPr><pic:cNvPr id="0" name="Picture ${drawingId}"/><pic:cNvPicPr/></pic:nvPicPr>` +
+            `<pic:blipFill><a:blip r:embed="${rId}"/><a:stretch><a:fillRect/></a:stretch></pic:blipFill>` +
+            `<pic:spPr><a:xfrm><a:off x="0" y="0"/><a:ext cx="${cx}" cy="${cy}"/></a:xfrm><a:prstGeom prst="rect"><a:avLst/></a:prstGeom></pic:spPr></pic:pic></a:graphicData></a:graphic></wp:inline></w:drawing></w:r>`;
+        const xml = paragraphXml(undefined, id, "", drawing);
+        const node = blockNode(record);
+        const items = [
+            ...drawingNamespacePatches(part),
+            side === "before"
+                ? patches.insertBefore(part, node, xml)
+                : patches.insertAfter(part, node, xml),
+        ];
+        return commit(context, items, { createdIds: [`p:${id}`, `img:${id}.0`], reflowFrom: record.id }, transaction);
+    },
+};

package/dist/edit/docx/style.d.ts

@@ -0,0 +1,45 @@
+import type { OoxmlPackage } from "../ooxml/package.js";
+import type { XmlElement, XmlPart } from "../ooxml/xml.js";
+import type { DocxParagraphStyle, DocxTextAlign, DocxTextStyle } from "./types.js";
+export interface StyleRecord {
+    readonly id: string;
+    readonly type: string;
+    readonly basedOn?: string;
+    readonly isDefault: boolean;
+    readonly rPr?: XmlElement;
+    readonly pPr?: XmlElement;
+}
+export interface ThemeFonts {
+    readonly major: string;
+    readonly minor: string;
+}
+/** `#RRGGBB` per colour scheme slot (`dk1`, `lt1`, `accent1`, …). */
+export type ThemeColors = ReadonlyMap<string, string>;
+/** Whether a name is one Word accepts in `w:themeColor`. */
+export declare function isThemeColorName(name: string): boolean;
+/** The styles and theme of a document, read once per revision. */
+export declare class DocxStyles {
+    readonly part: XmlPart | undefined;
+    readonly styles: ReadonlyMap<string, StyleRecord>;
+    readonly defaultParagraph: StyleRecord | undefined;
+    readonly defaultCharacter: StyleRecord | undefined;
+    readonly docDefaultsRPr: XmlElement | undefined;
+    readonly docDefaultsPPr: XmlElement | undefined;
+    readonly fonts: ThemeFonts;
+    readonly colors: ThemeColors;
+    private constructor();
+    /** The `#RRGGBB` a `w:themeColor` name resolves to in this theme, when it has one. */
+    themeColor(name: string): string | undefined;
+    static load(pkg: OoxmlPackage, documentPart: string, signal?: AbortSignal): Promise<DocxStyles>;
+    /** A style and the styles it is based on, nearest first; cycles stop. */
+    chain(id: string | undefined): StyleRecord[];
+}
+/** `w:b`, `w:i`, `w:default`-style toggles: present without a value means on. */
+export declare function isOn(value: string | undefined): boolean;
+/** The run property holders in resolution order for a run of a paragraph. */
+export declare function runPropertySources(styles: DocxStyles, pPr: XmlElement | undefined, rPr: XmlElement | undefined): readonly XmlElement[];
+/** The paragraph property holders in resolution order. */
+export declare function paragraphPropertySources(styles: DocxStyles, pPr: XmlElement | undefined): readonly XmlElement[];
+export declare function resolveTextStyle(styles: DocxStyles, pPr: XmlElement | undefined, rPr: XmlElement | undefined): DocxTextStyle;
+export declare function alignOf(value: string | undefined): DocxTextAlign | undefined;
+export declare function resolveParagraphStyle(styles: DocxStyles, pPr: XmlElement | undefined): DocxParagraphStyle;