@stll/folio-core 0.46.0 → 0.47.0

package/dist/docx/imageParser.js

@@ -2,15 +2,44 @@ import { normalizeImageLuminancePercent } from "../utils/imageLuminance.js";
 import { sanitizeImageSrc } from "../utils/sanitizeImageSrc.js";
 import { emuToPixels } from "../utils/units.js";
 import { sanitizeExternalUrl } from "../utils/urlSecurity.js";
-import { WRAP_ELEMENT_NAMES, parseAnchorBehindDoc, parsePositionH, parsePositionV, parseWrapElement } from "./drawingUtils.js";
+import { WORDPROCESSING_DRAWING_NAMESPACE_URIS, parseDrawingAnchor } from "./drawingAnchor.js";
+import { findWrapElement, parseAnchorBehindDoc, parseDrawingEffectExtent, parseInlineWrap, parsePositionH, parsePositionV, parseWrapElement } from "./drawingUtils.js";
 import { parseGraphicFrameLocks } from "./graphicFrameLocks.js";
 import { parseNonVisualDrawingNames } from "./nonVisualDrawingProps.js";
 import { RELATIONSHIP_TYPES, resolveRelationshipIdOfType } from "./relsParser.js";
 import { isTextBoxDrawing } from "./textBoxParser.js";
 import { percentageSpelling } from "./transitionalSpelling.js";
 import { captureVerbatimXml } from "./verbatimCapture.js";
-import { findByFullName, findChild, findChildByNamespaceUri, findChildrenByNamespaceUri, getAttribute, getChildElements, getLocalName, getNamespaceUri, parseNumericAttribute, parseOnOffAttribute, parseOnOffValue } from "./xmlParser.js";
+import { findByFullName, findChildByNamespaceUri, findChildrenByNamespaceUri, getAttribute, getChildElements, getLocalName, getNamespaceUri, parseNumericAttribute, parseOnOffAttribute, parseOnOffValue } from "./xmlParser.js";
+import { relationshipIdOf } from "@stll/docx-core/model";
 //#region src/docx/imageParser.ts
+/**
+* Image Parser - Parse embedded images from w:drawing elements
+*
+* DOCX images are contained in <w:drawing> elements with either:
+* - wp:inline - Inline images that flow with text
+* - wp:anchor - Floating/anchored images with text wrapping
+*
+* OOXML Structure:
+* w:drawing
+*   ├── wp:inline or wp:anchor
+*   │   ├── wp:extent (size: cx, cy in EMUs)
+*   │   ├── wp:effectExtent (effect margins)
+*   │   ├── wp:docPr (document properties: id, name, descr, title)
+*   │   ├── wp:positionH / wp:positionV (for anchor only)
+*   │   ├── wp:wrap* (wrapping mode for anchor: wrapNone, wrapSquare, etc.)
+*   │   └── a:graphic
+*   │       └── a:graphicData
+*   │           └── pic:pic
+*   │               ├── pic:nvPicPr (non-visual properties)
+*   │               ├── pic:blipFill
+*   │               │   └── a:blip (r:embed = rId)
+*   │               └── pic:spPr
+*   │                   └── a:xfrm (transform: rotation, flip)
+*
+* EMU (English Metric Units): 914400 EMU = 1 inch
+* Conversion: pixels = (emu * 96) / 914400
+*/
 const DRAWINGML_MAIN_NAMESPACE_URIS = /* @__PURE__ */ new Set(["http://schemas.openxmlformats.org/drawingml/2006/main", "http://purl.oclc.org/ooxml/drawingml/main"]);
 const FIXED_PERCENTAGE = /^-?\d+$/u;
 /**
@@ -26,13 +55,13 @@ function rotToDegrees(rot) {
 	return val / 6e4;
 }
 /**
-* Find any of the specified elements
+* A child of `wp:inline` or `wp:anchor`, resolved by namespace.
+*
+* The `wp` prefix is the producer's choice, not the document's meaning: a
+* package that binds the WordprocessingDrawing namespace to another prefix
+* carries the same `wp:extent` and a prefix-matched read finds none of it.
 */
-function findAnyOf(parent, names) {
-	const children = getChildElements(parent);
-	for (const child of children) if (names.includes(child.name || "")) return child;
-	return null;
-}
+const findDrawingChild = (parent, localName) => findChildByNamespaceUri(parent, WORDPROCESSING_DRAWING_NAMESPACE_URIS, localName);
 /**
 * Parse extent element for image size
 *
@@ -50,26 +79,6 @@ function parseExtent(extent) {
 	};
 }
 /**
-* Parse effect extent for shadow/effect margins
-*
-* @param effectExtent - wp:effectExtent element
-* @returns Padding for effects
-*/
-function parseEffectExtent(effectExtent) {
-	if (!effectExtent) return;
-	const l = parseNumericAttribute(effectExtent, null, "l") ?? 0;
-	const t = parseNumericAttribute(effectExtent, null, "t") ?? 0;
-	const r = parseNumericAttribute(effectExtent, null, "r") ?? 0;
-	const b = parseNumericAttribute(effectExtent, null, "b") ?? 0;
-	if (l === 0 && t === 0 && r === 0 && b === 0) return;
-	return {
-		left: l,
-		top: t,
-		right: r,
-		bottom: b
-	};
-}
-/**
 * The `a:ext` uri under which Word records that a drawing is decorative. The
 * extension holds `<adec:decorative val="…"/>`; `CT_NonVisualDrawingProps` has
 * no `@decorative` attribute for it to be written as.
@@ -111,6 +120,22 @@ const parseDocPropsExtensions = (docPr) => {
 	return result;
 };
 /**
+* One of `wp:docPr`'s two `CT_Hyperlink` children, as the source wrote it.
+*
+* Resolved by namespace URI rather than by the `a:` prefix: a producer is free
+* to bind DrawingML's namespace to another prefix, and a prefix-matched read
+* would drop the whole link for a document that does.
+*/
+const parseDocPrLink = (docPr, localName) => {
+	const element = findChildByNamespaceUri(docPr, DRAWINGML_NAMESPACE_URIS, localName);
+	if (!element) return;
+	const rId = getAttribute(element, "r", "id");
+	return {
+		xml: captureVerbatimXml(element),
+		...rId == null ? {} : { rId }
+	};
+};
+/**
 * Parse document properties (wp:docPr)
 *
 * @param docPr - wp:docPr element
@@ -122,17 +147,35 @@ function parseDocProps(docPr) {
 	const names = parseNonVisualDrawingNames(docPr);
 	const hidden = parseOnOffAttribute(docPr, null, "hidden");
 	const { decorative, other: docPrExtensions } = parseDocPropsExtensions(docPr);
-	const hlinkClickEl = findChild(docPr, "a", "hlinkClick");
-	const hlinkRId = hlinkClickEl ? getAttribute(hlinkClickEl, "r", "id") : null;
+	const hlinkClickSource = parseDocPrLink(docPr, "hlinkClick");
+	const hlinkHoverSource = parseDocPrLink(docPr, "hlinkHover");
 	return {
 		...id != null ? { id } : {},
 		...names,
 		...decorative === void 0 ? {} : { decorative },
 		...hidden === void 0 ? {} : { hidden },
 		...docPrExtensions.length > 0 ? { docPrExtensions } : {},
-		...hlinkRId != null ? { hlinkRId } : {}
+		...hlinkClickSource === void 0 ? {} : { hlinkClickSource },
+		...hlinkHoverSource === void 0 ? {} : { hlinkHoverSource }
 	};
 }
+const safeDocPrLinks = ({ props, rels }) => {
+	const targetOf = (link) => {
+		if (link === void 0) return;
+		if (link.rId === void 0) return "";
+		const resolved = resolveRelationshipIdOfType(rels, link.rId, RELATIONSHIP_TYPES.hyperlink);
+		return resolved.status === "resolved" ? sanitizeExternalUrl(resolved.relationship.target) : void 0;
+	};
+	const click = props.hlinkClickSource;
+	const clickHref = targetOf(click);
+	const hoverKept = targetOf(props.hlinkHoverSource) !== void 0;
+	return {
+		...clickHref ? { hlinkHref: clickHref } : {},
+		...clickHref !== void 0 && click?.rId !== void 0 ? { hlinkRId: click.rId } : {},
+		...clickHref === void 0 || click === void 0 ? {} : { hlinkClickSource: click },
+		...hoverKept && props.hlinkHoverSource !== void 0 ? { hlinkHoverXml: props.hlinkHoverSource.xml } : {}
+	};
+};
 /**
 * Parse transform properties from a:xfrm
 */
@@ -157,13 +200,22 @@ function parseTransform(xfrm) {
 * `a:srcRect` crop element, so callers that need either share this walk.
 */
 function findBlipFillElement(container) {
-	const graphic = findByFullName(container, "a:graphic");
-	if (!graphic) return null;
-	const graphicData = findByFullName(graphic, "a:graphicData");
-	if (!graphicData) return null;
-	const pic = findByFullName(graphicData, "pic:pic");
-	if (!pic) return null;
-	return findByFullName(pic, "pic:blipFill");
+	return findByFullName(findPictureElement(container), "pic:blipFill");
+}
+/** `a:graphic > a:graphicData > pic:pic`, the one walk every picture read shares. */
+function findPictureElement(container) {
+	return findByFullName(findByFullName(findByFullName(container, "a:graphic"), "a:graphicData"), "pic:pic");
+}
+/**
+* `pic:cNvPr` — the picture's own name, alt text and title.
+*
+* Not `wp:docPr`'s. They are two `CT_NonVisualDrawingProps` elements on the
+* same drawing and a reader names the object from whichever one it is looking
+* at, so folding them together loses whichever the source did not repeat.
+*/
+function findPictureNonVisualProps(container) {
+	const pic = findPictureElement(container);
+	return pic ? findByFullName(findByFullName(pic, "pic:nvPicPr"), "pic:cNvPr") : null;
 }
 /**
 * Parse `<a:srcRect l="..." t="..." r="..." b="..."/>` inside `pic:blipFill`.
@@ -194,16 +246,6 @@ function parseImageCrop(blipFill) {
 	return crop;
 }
 /**
-* Parse an OOXML `ST_OnOff` attribute on an element. Accepts the full
-* set of literals the spec allows (`"1"`/`"true"`/`"on"` and
-* `"0"`/`"false"`/`"off"`); anything else (including an absent
-* attribute) folds back to `undefined` so callers can apply the
-* spec-defined default.
-*/
-function parseOnOffAttr(element, name) {
-	return parseOnOffValue(getAttribute(element, null, name));
-}
-/**
 * Parse `<a:alphaModFix amt="..."/>` inside the `a:blip` element. The
 * `amt` value is in 1/100000; convert to a fraction in [0, 1] for CSS
 * `opacity`. Returns undefined when no alpha modifier is present (fully
@@ -249,12 +291,7 @@ function parseImageLuminance(blip) {
 */
 function extractBlipRId(blip) {
 	if (!blip) return;
-	const rEmbed = getAttribute(blip, "r", "embed");
-	if (rEmbed) return rEmbed;
-	const embed = getAttribute(blip, null, "embed");
-	if (embed) return embed;
-	const rLink = getAttribute(blip, "r", "link");
-	if (rLink) return rLink;
+	return relationshipIdOf(getAttribute(blip, "r", "embed")) ?? relationshipIdOf(getAttribute(blip, null, "embed")) ?? relationshipIdOf(getAttribute(blip, "r", "link"));
 }
 /**
 * Find transform (a:xfrm) from picture shape properties
@@ -262,15 +299,7 @@ function extractBlipRId(blip) {
 * Path: a:graphic > a:graphicData > pic:pic > pic:spPr > a:xfrm
 */
 function findPictureTransform(container) {
-	const graphic = findByFullName(container, "a:graphic");
-	if (!graphic) return null;
-	const graphicData = findByFullName(graphic, "a:graphicData");
-	if (!graphicData) return null;
-	const pic = findByFullName(graphicData, "pic:pic");
-	if (!pic) return null;
-	const spPr = findByFullName(pic, "pic:spPr");
-	if (!spPr) return null;
-	return findByFullName(spPr, "a:xfrm");
+	return findByFullName(findByFullName(findPictureElement(container), "pic:spPr"), "a:xfrm");
 }
 /**
 * Normalize a target path to the standard word/media/... format
@@ -362,10 +391,11 @@ function resolveImageData(rId, rels, media) {
 * @returns Parsed Image object
 */
 function parseInline(inlineEl, rels, media) {
-	const size = parseExtent(findByFullName(inlineEl, "wp:extent"));
-	const padding = parseEffectExtent(findByFullName(inlineEl, "wp:effectExtent"));
-	const props = parseDocProps(findByFullName(inlineEl, "wp:docPr"));
+	const size = parseExtent(findDrawingChild(inlineEl, "extent"));
+	const padding = parseDrawingEffectExtent(findDrawingChild(inlineEl, "effectExtent"));
+	const props = parseDocProps(findDrawingChild(inlineEl, "docPr"));
 	const frameLocks = parseGraphicFrameLocks(inlineEl);
+	const pictureNames = parseNonVisualDrawingNames(findPictureNonVisualProps(inlineEl));
 	const blipFill = findBlipFillElement(inlineEl);
 	const blip = blipFill ? findByFullName(blipFill, "a:blip") : null;
 	const rId = extractBlipRId(blip);
@@ -374,15 +404,7 @@ function parseInline(inlineEl, rels, media) {
 	const effects = parseImageLuminance(blip);
 	const imageData = resolveImageData(rId, rels, media);
 	const transform = parseTransform(findPictureTransform(inlineEl));
-	const distT = parseNumericAttribute(inlineEl, null, "distT") ?? void 0;
-	const distB = parseNumericAttribute(inlineEl, null, "distB") ?? void 0;
-	const distL = parseNumericAttribute(inlineEl, null, "distL") ?? void 0;
-	const distR = parseNumericAttribute(inlineEl, null, "distR") ?? void 0;
-	const wrap = { type: "inline" };
-	if (distT !== void 0) wrap.distT = distT;
-	if (distB !== void 0) wrap.distB = distB;
-	if (distL !== void 0) wrap.distL = distL;
-	if (distR !== void 0) wrap.distR = distR;
+	const wrap = parseInlineWrap(inlineEl);
 	const image = {
 		type: "image",
 		...rId === void 0 ? {} : { rId },
@@ -393,6 +415,7 @@ function parseInline(inlineEl, rels, media) {
 	if (props.name !== void 0) image.docPrName = props.name;
 	if (props.alt !== void 0) image.alt = props.alt;
 	if (props.title !== void 0) image.title = props.title;
+	if (Object.keys(pictureNames).length > 0) image.pictureNames = pictureNames;
 	if (props.decorative !== void 0) image.decorative = props.decorative;
 	if (props.hidden !== void 0) image.hidden = props.hidden;
 	if (props.docPrExtensions !== void 0) image.docPrExtensions = props.docPrExtensions;
@@ -406,14 +429,10 @@ function parseInline(inlineEl, rels, media) {
 	if (opacity !== void 0) image.opacity = opacity;
 	if (effects !== void 0) image.effects = effects;
 	if (frameLocks) image.frameLocks = frameLocks;
-	const hlink = resolveRelationshipIdOfType(rels, props.hlinkRId, RELATIONSHIP_TYPES.hyperlink);
-	if (hlink.status === "resolved") {
-		const safeHref = sanitizeExternalUrl(hlink.relationship.target);
-		if (safeHref) {
-			image.hlinkHref = safeHref;
-			image.hlinkRId = hlink.relationship.id;
-		}
-	}
+	Object.assign(image, safeDocPrLinks({
+		props,
+		rels
+	}));
 	return image;
 }
 /**
@@ -425,13 +444,14 @@ function parseInline(inlineEl, rels, media) {
 * @returns Parsed Image object
 */
 function parseAnchor(anchorEl, rels, media) {
-	const size = parseExtent(findByFullName(anchorEl, "wp:extent"));
-	const padding = parseEffectExtent(findByFullName(anchorEl, "wp:effectExtent"));
-	const props = parseDocProps(findByFullName(anchorEl, "wp:docPr"));
+	const size = parseExtent(findDrawingChild(anchorEl, "extent"));
+	const effectExtent = findDrawingChild(anchorEl, "effectExtent");
+	const padding = parseDrawingEffectExtent(effectExtent);
+	const props = parseDocProps(findDrawingChild(anchorEl, "docPr"));
 	const frameLocks = parseGraphicFrameLocks(anchorEl);
+	const pictureNames = parseNonVisualDrawingNames(findPictureNonVisualProps(anchorEl));
 	const behindDoc = parseAnchorBehindDoc(anchorEl);
-	const layoutInCell = parseOnOffAttr(anchorEl, "layoutInCell");
-	const allowOverlap = parseOnOffAttr(anchorEl, "allowOverlap");
+	const anchor = parseDrawingAnchor(anchorEl);
 	const anchorDistT = parseNumericAttribute(anchorEl, null, "distT");
 	const anchorDistB = parseNumericAttribute(anchorEl, null, "distB");
 	const anchorDistL = parseNumericAttribute(anchorEl, null, "distL");
@@ -442,9 +462,15 @@ function parseAnchor(anchorEl, rels, media) {
 		...anchorDistL != null ? { distL: anchorDistL } : {},
 		...anchorDistR != null ? { distR: anchorDistR } : {}
 	};
-	const wrap = parseWrapElement(findAnyOf(anchorEl, WRAP_ELEMENT_NAMES), behindDoc, anchorDistances);
-	const posH = findByFullName(anchorEl, "wp:positionH");
-	const posV = findByFullName(anchorEl, "wp:positionV");
+	const anchorEffectExtent = parseDrawingEffectExtent(effectExtent);
+	const wrap = parseWrapElement({
+		wrapEl: findWrapElement(anchorEl),
+		behindDoc,
+		anchorDistances,
+		...anchorEffectExtent === void 0 ? {} : { anchorEffectExtent }
+	});
+	const posH = findDrawingChild(anchorEl, "positionH");
+	const posV = findDrawingChild(anchorEl, "positionV");
 	const horizontal = parsePositionH(posH);
 	const vertical = parsePositionV(posV);
 	let position;
@@ -470,6 +496,7 @@ function parseAnchor(anchorEl, rels, media) {
 	if (props.name !== void 0) image.docPrName = props.name;
 	if (props.alt !== void 0) image.alt = props.alt;
 	if (props.title !== void 0) image.title = props.title;
+	if (Object.keys(pictureNames).length > 0) image.pictureNames = pictureNames;
 	if (props.decorative !== void 0) image.decorative = props.decorative;
 	if (props.hidden !== void 0) image.hidden = props.hidden;
 	if (props.docPrExtensions !== void 0) image.docPrExtensions = props.docPrExtensions;
@@ -484,16 +511,11 @@ function parseAnchor(anchorEl, rels, media) {
 	if (opacity !== void 0) image.opacity = opacity;
 	if (effects !== void 0) image.effects = effects;
 	if (frameLocks) image.frameLocks = frameLocks;
-	if (layoutInCell !== void 0) image.layoutInCell = layoutInCell;
-	if (allowOverlap !== void 0) image.allowOverlap = allowOverlap;
-	const hlink = resolveRelationshipIdOfType(rels, props.hlinkRId, RELATIONSHIP_TYPES.hyperlink);
-	if (hlink.status === "resolved") {
-		const safeHref = sanitizeExternalUrl(hlink.relationship.target);
-		if (safeHref) {
-			image.hlinkHref = safeHref;
-			image.hlinkRId = hlink.relationship.id;
-		}
-	}
+	if (anchor !== void 0) image.anchor = anchor;
+	Object.assign(image, safeDocPrLinks({
+		props,
+		rels
+	}));
 	return image;
 }
 /**
@@ -508,12 +530,10 @@ function parseAnchor(anchorEl, rels, media) {
 */
 function parseDrawing(drawingEl, rels, media) {
 	if (isTextBoxDrawing(drawingEl)) return null;
-	const children = getChildElements(drawingEl);
-	for (const child of children) {
-		const name = child.name || "";
-		if (name === "wp:inline" || name === "wp:anchor") return name === "wp:inline" ? parseInline(child, rels, media) : parseAnchor(child, rels, media);
-	}
-	return null;
+	const inline = findDrawingChild(drawingEl, "inline");
+	if (inline) return parseInline(inline, rels, media);
+	const anchor = findDrawingChild(drawingEl, "anchor");
+	return anchor ? parseAnchor(anchor, rels, media) : null;
 }
 /**
 * Parse an image from a w:drawing element

package/dist/docx/index.d.ts

@@ -1,5 +1,6 @@
 import { getCachedNumberingMap } from "./numberingParser.js";
 import { DOCX_ENCRYPTION_ERROR_CODES, DocxEncryptionError, DocxEncryptionErrorCode, isDocxEncryptionError } from "./encryption/errors.js";
 import { DecryptDocxOptions, DecryptDocxResult, decryptDocxIfNeeded, openDocxBuffer } from "./encryption/openEncryptedDocx.js";
+import { NO_NUMBERING_NUM_ID, NO_PARAGRAPH_NUMBERING, ParagraphNumberingOverride, ParagraphNumberingSlots, ResolvedParagraphNumbering, isNumberingReference, mergeParagraphNumbering, paragraphNumberingFromSlots, paragraphNumberingLevel, paragraphNumberingReferenceId, paragraphNumberingSlots, readParagraphNumbering, resolveParagraphNumbering, sameEffectiveParagraphNumbering, sameStatedParagraphNumbering } from "./numberingReference.js";
 import { FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, FolioDocumentMetadataProperty, FolioDocumentPrivacyArchiveError, FolioDocumentPrivacyOptions, FolioDocumentPrivacyReport, FolioDocumentPrivacyTransform, InvalidFolioDocumentPrivacyOptionsError, RewriteDocxMetadataPrivacyResult, isFolioDocumentPrivacyTransform, rewriteDocxMetadataPrivacy } from "./metadataPrivacy.js";
-export { DOCX_ENCRYPTION_ERROR_CODES, type DecryptDocxOptions, type DecryptDocxResult, DocxEncryptionError, type DocxEncryptionErrorCode, FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, type FolioDocumentMetadataProperty, FolioDocumentPrivacyArchiveError, type FolioDocumentPrivacyOptions, type FolioDocumentPrivacyReport, type FolioDocumentPrivacyTransform, InvalidFolioDocumentPrivacyOptionsError, type RewriteDocxMetadataPrivacyResult, decryptDocxIfNeeded, getCachedNumberingMap, isDocxEncryptionError, isFolioDocumentPrivacyTransform, openDocxBuffer, rewriteDocxMetadataPrivacy };
+export { DOCX_ENCRYPTION_ERROR_CODES, type DecryptDocxOptions, type DecryptDocxResult, DocxEncryptionError, type DocxEncryptionErrorCode, FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, type FolioDocumentMetadataProperty, FolioDocumentPrivacyArchiveError, type FolioDocumentPrivacyOptions, type FolioDocumentPrivacyReport, type FolioDocumentPrivacyTransform, InvalidFolioDocumentPrivacyOptionsError, NO_NUMBERING_NUM_ID, NO_PARAGRAPH_NUMBERING, type ParagraphNumberingOverride, type ParagraphNumberingSlots, type ResolvedParagraphNumbering, type RewriteDocxMetadataPrivacyResult, decryptDocxIfNeeded, getCachedNumberingMap, isDocxEncryptionError, isFolioDocumentPrivacyTransform, isNumberingReference, mergeParagraphNumbering, openDocxBuffer, paragraphNumberingFromSlots, paragraphNumberingLevel, paragraphNumberingReferenceId, paragraphNumberingSlots, readParagraphNumbering, resolveParagraphNumbering, rewriteDocxMetadataPrivacy, sameEffectiveParagraphNumbering, sameStatedParagraphNumbering };

package/dist/docx/index.js

@@ -2,4 +2,5 @@ import { DOCX_ENCRYPTION_ERROR_CODES, DocxEncryptionError, isDocxEncryptionError
 import { decryptDocxIfNeeded, openDocxBuffer } from "./encryption/openEncryptedDocx.js";
 import { FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, FolioDocumentPrivacyArchiveError, InvalidFolioDocumentPrivacyOptionsError, isFolioDocumentPrivacyTransform, rewriteDocxMetadataPrivacy } from "./metadataPrivacy.js";
 import { getCachedNumberingMap } from "./numberingParser.js";
-export { DOCX_ENCRYPTION_ERROR_CODES, DocxEncryptionError, FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, FolioDocumentPrivacyArchiveError, InvalidFolioDocumentPrivacyOptionsError, decryptDocxIfNeeded, getCachedNumberingMap, isDocxEncryptionError, isFolioDocumentPrivacyTransform, openDocxBuffer, rewriteDocxMetadataPrivacy };
+import { NO_NUMBERING_NUM_ID, NO_PARAGRAPH_NUMBERING, isNumberingReference, mergeParagraphNumbering, paragraphNumberingFromSlots, paragraphNumberingLevel, paragraphNumberingReferenceId, paragraphNumberingSlots, readParagraphNumbering, resolveParagraphNumbering, sameEffectiveParagraphNumbering, sameStatedParagraphNumbering } from "./numberingReference.js";
+export { DOCX_ENCRYPTION_ERROR_CODES, DocxEncryptionError, FOLIO_DOCUMENT_METADATA_PROPERTIES, FOLIO_DOCUMENT_PRIVACY_TRANSFORMS, FolioDocumentPrivacyArchiveError, InvalidFolioDocumentPrivacyOptionsError, NO_NUMBERING_NUM_ID, NO_PARAGRAPH_NUMBERING, decryptDocxIfNeeded, getCachedNumberingMap, isDocxEncryptionError, isFolioDocumentPrivacyTransform, isNumberingReference, mergeParagraphNumbering, openDocxBuffer, paragraphNumberingFromSlots, paragraphNumberingLevel, paragraphNumberingReferenceId, paragraphNumberingSlots, readParagraphNumbering, resolveParagraphNumbering, rewriteDocxMetadataPrivacy, sameEffectiveParagraphNumbering, sameStatedParagraphNumbering };

package/dist/docx/inlineContentRemovals.d.ts

@@ -0,0 +1,28 @@
+import { InlineContentSlot } from "./paragraphTraversal.js";
+//#region src/docx/inlineContentRemovals.d.ts
+/**
+ * Positions marked for removal, per inline-content array.
+ *
+ * Removal shifts every later index in that array, so a normaliser records the
+ * positions while it reads and drops them once, after it has finished reading.
+ */
+declare class InlineContentRemovals {
+  #private;
+  mark({ content, index }: Pick<InlineContentSlot, "content" | "index">): void;
+  /**
+   * Applies every marked removal and answers how many items were dropped.
+   *
+   * Consolidation is re-run over each array a removal touched, because every
+   * inline item that is not a run is a merge boundary: `parseParagraph`
+   * consolidated the array while the removed item still stood between its
+   * neighbours, so dropping it leaves two runs adjacent that the next parse
+   * merges. That is an oscillation rather than a loss, save 1 writes the pair
+   * and save 2 writes one run, and it reaches every normaliser that removes an
+   * item, not only the one it was found through. Restoring the invariant here
+   * makes the parse a fixed point by construction; an array nothing was
+   * removed from is not touched.
+   */
+  apply(): number;
+}
+//#endregion
+export { InlineContentRemovals };

package/dist/docx/inlineContentRemovals.js

@@ -0,0 +1,46 @@
+import { consolidateParagraphContent } from "./runConsolidator.js";
+//#region src/docx/inlineContentRemovals.ts
+/**
+* Positions marked for removal, per inline-content array.
+*
+* Removal shifts every later index in that array, so a normaliser records the
+* positions while it reads and drops them once, after it has finished reading.
+*/
+var InlineContentRemovals = class {
+	#byContent = /* @__PURE__ */ new Map();
+	mark({ content, index }) {
+		const indexes = this.#byContent.get(content);
+		if (indexes) {
+			indexes.add(index);
+			return;
+		}
+		this.#byContent.set(content, /* @__PURE__ */ new Set([index]));
+	}
+	/**
+	* Applies every marked removal and answers how many items were dropped.
+	*
+	* Consolidation is re-run over each array a removal touched, because every
+	* inline item that is not a run is a merge boundary: `parseParagraph`
+	* consolidated the array while the removed item still stood between its
+	* neighbours, so dropping it leaves two runs adjacent that the next parse
+	* merges. That is an oscillation rather than a loss, save 1 writes the pair
+	* and save 2 writes one run, and it reaches every normaliser that removes an
+	* item, not only the one it was found through. Restoring the invariant here
+	* makes the parse a fixed point by construction; an array nothing was
+	* removed from is not touched.
+	*/
+	apply() {
+		let removed = 0;
+		for (const [content, indexes] of this.#byContent) {
+			if (indexes.size === 0) continue;
+			const kept = content.filter((_, index) => !indexes.has(index));
+			removed += content.length - kept.length;
+			const next = consolidateParagraphContent(kept);
+			content.length = 0;
+			content.push(...next);
+		}
+		return removed;
+	}
+};
+//#endregion
+export { InlineContentRemovals };

package/dist/docx/inlineWrapperContent.d.ts

@@ -45,11 +45,23 @@ declare const TRACKED_CHANGE_WRAPPER_CONTENT: {
  * the bidirectional wrapper: OOXML allows runs, hyperlinks, simple/complex
  * fields, nested SDTs, tracked insertions/deletions/moves, math, markup folio
  * does not model, and the bidirectional controls directly inside
- * `<w:sdtContent>`. Bookmarks and range markers are lifted out as siblings of
- * the SDT so the control itself stays valid.
+ * `<w:sdtContent>`.
+ *
+ * The bookmark boundaries are admitted for the reason the revision wrapper
+ * admits them: `CT_SdtContentRun` reaches `EG_RangeMarkupElements` through
+ * `EG_RunLevelElts`, so the marker is valid where Word wrote it, and lifting
+ * it out changes what the bookmark covers. A range whose extent was the
+ * control's content comes back covering the control plus whatever follows, so
+ * a `REF` field or a link to it resolves to the wrong text.
+ *
+ * The other range markers are not: a `w:commentRangeStart` or a
+ * `w:moveFromRangeStart` inside the control is a marker the control does not
+ * own, and the pairing passes read it as a paragraph-level sibling.
  */
 declare const INLINE_SDT_CONTENT: {
   readonly inlineWrapper: true;
+  readonly bookmarkEnd: true;
+  readonly bookmarkStart: true;
   readonly complexField: true;
   readonly deletion: true;
   readonly hyperlink: true;
@@ -61,15 +73,80 @@ declare const INLINE_SDT_CONTENT: {
   readonly preservedInline: true;
   readonly run: true;
   readonly simpleField: true;
+  readonly commentRangeEnd: false;
+  readonly commentRangeStart: false;
+  readonly commentReference: false;
+  readonly moveFromRangeEnd: false;
+  readonly moveFromRangeStart: false;
+  readonly moveToRangeEnd: false;
+  readonly moveToRangeStart: false;
+};
+/**
+ * What a `w:hyperlink` keeps inside itself (CT_Hyperlink).
+ *
+ * Runs, the bookmark boundaries a link's target sits on, markup folio does not
+ * model, and the four transparent wrappers: `CT_Hyperlink` is `EG_PContent`,
+ * which declares `w:bdo`, `w:dir`, `w:smartTag` and the run-level
+ * `w:customXml`, so `w:hyperlink > w:bdo > w:r` is a link an author may write
+ * and folio reads it as authored.
+ *
+ * A revision is not admitted, and the exclusion is the model's, not the
+ * schema's: OOXML nests `w:ins` inside `w:hyperlink` and folio nests the link
+ * inside the revision, so the paragraph parser hoists one around the link.
+ */
+declare const HYPERLINK_CONTENT: {
+  readonly bookmarkEnd: true;
+  readonly bookmarkStart: true;
+  readonly inlineWrapper: true;
+  readonly preservedInline: true;
+  readonly run: true;
+  readonly commentRangeEnd: false;
+  readonly commentRangeStart: false;
+  readonly commentReference: false;
+  readonly complexField: false;
+  readonly deletion: false;
+  readonly hyperlink: false;
+  readonly inlineSdt: false;
+  readonly insertion: false;
+  readonly mathEquation: false;
+  readonly moveFrom: false;
+  readonly moveFromRangeEnd: false;
+  readonly moveFromRangeStart: false;
+  readonly moveTo: false;
+  readonly moveToRangeEnd: false;
+  readonly moveToRangeStart: false;
+  readonly simpleField: false;
+};
+/**
+ * What a `w:fldSimple` keeps inside itself (CT_SimpleField).
+ *
+ * The cached result's runs and the link one may carry, the same four
+ * transparent wrappers `EG_PContent` declares, and markup folio does not
+ * model. `w:fldData` is the field's own child rather than its content, so it
+ * rides `dispatchChildren`'s sink like any other capture.
+ */
+declare const SIMPLE_FIELD_CONTENT: {
+  readonly hyperlink: true;
+  readonly inlineWrapper: true;
+  readonly preservedInline: true;
+  readonly run: true;
   readonly bookmarkEnd: false;
   readonly bookmarkStart: false;
   readonly commentRangeEnd: false;
   readonly commentRangeStart: false;
   readonly commentReference: false;
+  readonly complexField: false;
+  readonly deletion: false;
+  readonly inlineSdt: false;
+  readonly insertion: false;
+  readonly mathEquation: false;
+  readonly moveFrom: false;
   readonly moveFromRangeEnd: false;
   readonly moveFromRangeStart: false;
+  readonly moveTo: false;
   readonly moveToRangeEnd: false;
   readonly moveToRangeStart: false;
+  readonly simpleField: false;
 };
 /**
  * Whether a revision wrapper keeps this content inside itself.
@@ -81,5 +158,9 @@ declare const INLINE_SDT_CONTENT: {
 declare const isTrackedChangeWrapperChild: <Content extends document_d_exports.ParagraphContent>(content: Content) => content is Extract<Content, document_d_exports.TrackedRunContent>;
 /** Whether an inline content control keeps this content inside `<w:sdtContent>`. */
 declare const isInlineSdtContent: <Content extends document_d_exports.ParagraphContent>(content: Content) => content is Extract<Content, document_d_exports.InlineSdt["content"][number]>;
+/** Whether a `w:hyperlink` keeps this content inside itself. */
+declare const isHyperlinkContent: <Content extends document_d_exports.ParagraphContent>(content: Content) => content is Extract<Content, document_d_exports.Hyperlink["children"][number]>;
+/** Whether a `w:fldSimple` keeps this content inside itself. */
+declare const isSimpleFieldContent: <Content extends document_d_exports.ParagraphContent>(content: Content) => content is Extract<Content, document_d_exports.SimpleField["content"][number]>;
 //#endregion
-export { INLINE_SDT_CONTENT, TRACKED_CHANGE_WRAPPER_CONTENT, isInlineSdtContent, isTrackedChangeWrapperChild };
+export { HYPERLINK_CONTENT, INLINE_SDT_CONTENT, SIMPLE_FIELD_CONTENT, TRACKED_CHANGE_WRAPPER_CONTENT, isHyperlinkContent, isInlineSdtContent, isSimpleFieldContent, isTrackedChangeWrapperChild };