@stll/folio-core 0.45.0 → 0.47.0

package/dist/docx/paragraphTraversal.js

@@ -16,7 +16,7 @@ const visitParagraphRuns = (paragraph, visit) => {
 			case "moveFrom":
 			case "moveTo":
 			case "inlineSdt":
-			case "bidiWrapper":
+			case "inlineWrapper":
 				for (const child of content.content) visitParagraphContent(child);
 				return;
 			case "complexField":
@@ -32,7 +32,8 @@ const visitParagraphRuns = (paragraph, visit) => {
 			case "moveFromRangeEnd":
 			case "moveToRangeStart":
 			case "moveToRangeEnd":
-			case "mathEquation": return;
+			case "mathEquation":
+			case "preservedInline": return;
 			default: return content;
 		}
 	};
@@ -69,7 +70,7 @@ const visitInlineContentSlots = (paragraph, visit) => {
 				case "deletion":
 				case "moveFrom":
 				case "moveTo":
-				case "bidiWrapper":
+				case "inlineWrapper":
 					visitContent(item.content);
 					break;
 				case "run":
@@ -83,41 +84,57 @@ const visitInlineContentSlots = (paragraph, visit) => {
 				case "moveFromRangeEnd":
 				case "moveToRangeStart":
 				case "moveToRangeEnd":
-				case "mathEquation": break;
+				case "mathEquation":
+				case "preservedInline": break;
 				default: panic(`Unsupported paragraph content: ${JSON.stringify(item)}`);
 			}
 		}
 	};
 	visitContent(paragraph.content);
 };
+/** What the walk does below a record it has just visited. */
+const BLOCK_TREE_DESCENT = {
+	descend: "descend",
+	prune: "prune"
+};
 /**
-* Positions marked for removal, per inline-content array.
+* Every record a block tree owns in document order: a paragraph, table, and
+* table row. These are the records that carry authored state by identity.
 *
-* 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.
+* A block container is not always reached through a block. `w:txbxContent`
+* hangs off a shape inside a run, so a walk that recurses on block children
+* alone enters the body, a cell and an `w:sdt` and never enters a text box:
+* every record inside one is then invisible to whatever the walk is deciding.
 */
-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. */
-	apply() {
-		let removed = 0;
-		for (const [content, indexes] of this.#byContent) {
-			if (indexes.size === 0) continue;
-			const next = content.filter((_, index) => !indexes.has(index));
-			removed += content.length - next.length;
-			content.length = 0;
-			content.push(...next);
+const visitBlockTreeRecords = (blocks, visit) => {
+	const visitParagraph = (paragraph) => {
+		if (visit(paragraph) === BLOCK_TREE_DESCENT.prune) return;
+		visitParagraphRuns(paragraph, (run) => {
+			for (const content of run.content) if (content.type === "shape" && content.shape.textBody) visitBlocks(content.shape.textBody.content);
+		});
+	};
+	const visitBlocks = (nested) => {
+		for (const block of nested) switch (block.type) {
+			case "paragraph":
+				visitParagraph(block);
+				break;
+			case "table":
+				if (visit(block) === BLOCK_TREE_DESCENT.prune) break;
+				for (const row of block.rows) {
+					if (visit(row) === BLOCK_TREE_DESCENT.prune) continue;
+					for (const cell of row.cells) visitBlocks(cell.content);
+				}
+				break;
+			case "blockSdt":
+				visitBlocks(block.content);
+				break;
+			case "preservedBlock":
+			case "bookmarkStart":
+			case "bookmarkEnd": break;
+			default: panic(`Unsupported block content: ${JSON.stringify(block)}`);
 		}
-		return removed;
-	}
+	};
+	visitBlocks(blocks);
 };
 const visitDocxParagraphs = ({ documentBody, headers, footers, footnotes, endnotes }, visit) => {
 	const seenParagraphs = /* @__PURE__ */ new WeakSet();
@@ -125,30 +142,14 @@ const visitDocxParagraphs = ({ documentBody, headers, footers, footnotes, endnot
 		if (seenParagraphs.has(paragraph)) return;
 		seenParagraphs.add(paragraph);
 		visit(paragraph);
-		visitParagraphRuns(paragraph, visitRun);
-	};
-	const visitRun = (run) => {
-		for (const content of run.content) {
-			if (content.type !== "shape" || !content.shape.textBody) continue;
-			for (const block of content.shape.textBody.content) visitBlock(block);
-		}
-	};
-	const visitTable = (table) => {
-		for (const row of table.rows) for (const cell of row.cells) visitBlocks(cell.content);
-	};
-	const visitBlock = (block) => {
-		if (block.type === "paragraph") {
-			visitParagraph(block);
-			return;
-		}
-		if (block.type === "table") {
-			visitTable(block);
-			return;
-		}
-		visitBlocks(block.content);
 	};
 	const visitBlocks = (blocks) => {
-		for (const block of blocks) visitBlock(block);
+		visitBlockTreeRecords(blocks, (record) => {
+			if (record.type !== "paragraph") return BLOCK_TREE_DESCENT.descend;
+			if (seenParagraphs.has(record)) return BLOCK_TREE_DESCENT.prune;
+			visitParagraph(record);
+			return BLOCK_TREE_DESCENT.descend;
+		});
 	};
 	visitBlocks(documentBody.content);
 	for (const section of documentBody.sections ?? []) visitBlocks(section.content);
@@ -156,7 +157,7 @@ const visitDocxParagraphs = ({ documentBody, headers, footers, footnotes, endnot
 	for (const footer of footers?.values() ?? []) visitBlocks(footer.content);
 	for (const footnote of footnotes ?? []) visitBlocks(footnote.content);
 	for (const endnote of endnotes ?? []) visitBlocks(endnote.content);
-	for (const comment of documentBody.comments ?? []) for (const paragraph of comment.content) visitParagraph(paragraph);
+	for (const comment of documentBody.comments ?? []) visitBlocks(comment.content);
 };
 //#endregion
-export { InlineContentRemovals, visitDocxParagraphs, visitInlineContentSlots, visitParagraphRuns };
+export { BLOCK_TREE_DESCENT, visitBlockTreeRecords, visitDocxParagraphs, visitInlineContentSlots, visitParagraphRuns };

package/dist/docx/parseWarningMessage.js

@@ -34,9 +34,13 @@ const PARSE_WARNING_MESSAGES = {
 	[PARSE_WARNING_CODES.unbalancedMoveRange]: (warning) => `Removed ${plural(warning.count, "unbalanced tracked move range marker")}.`,
 	[PARSE_WARNING_CODES.headerFooterTypeOutsideEnum]: (warning) => `Read header/footer type${quoted(warning.value)} as "default"${where(warning)}; ST_HdrFtr is even, default or first.`,
 	[PARSE_WARNING_CODES.unrecognisedOnOffValue]: (warning) => `Ignored on/off value${quoted(warning.value)}${where(warning)}; ST_OnOff is 1, 0, true, false, on or off.`,
+	[PARSE_WARNING_CODES.unrecognisedThemeColor]: (warning) => `Kept theme colour${quoted(warning.value)}${where(warning)} without painting it; ST_ThemeColor does not enumerate it.`,
 	[PARSE_WARNING_CODES.borderWithoutValue]: (warning) => `Read a border with no w:val${where(warning)} as having no border style.`,
+	[PARSE_WARNING_CODES.borderStyleOutsideEnum]: (warning) => `Kept border style${quoted(warning.value)} verbatim${where(warning)}; ST_Border does not declare it.`,
+	[PARSE_WARNING_CODES.outlineDashOutsideEnum]: (warning) => `Kept outline dash${quoted(warning.value)} verbatim${where(warning)}; ST_PresetLineDashVal does not declare it.`,
 	[PARSE_WARNING_CODES.styleSetDuplicateStyleId]: (warning) => `Dropped a style repeating the id${quoted(warning.value)} another style in the set already defines.`,
-	[PARSE_WARNING_CODES.styleSetInitialStyleMissing]: (warning) => `The style set names initial paragraph style${quoted(warning.value)}, which it does not contain; used the set's default instead${where(warning)}.`
+	[PARSE_WARNING_CODES.styleSetInitialStyleMissing]: (warning) => `The style set names initial paragraph style${quoted(warning.value)}, which it does not contain; used the set's default instead${where(warning)}.`,
+	[PARSE_WARNING_CODES.pageBreakProjectionApproximated]: (warning) => `${warning.detail ?? "An explicit page break is laid out approximately"}${where(warning)}.`
 };
 const formatParseWarning = (warning) => PARSE_WARNING_MESSAGES[warning.code](warning);
 const formatParseWarnings = (warnings) => warnings.map(formatParseWarning);

package/dist/docx/parserEnums.d.ts

@@ -7,17 +7,15 @@ import * as v from "valibot";
  * no `as` at the call site.
  */
 declare const narrowEnum: <T extends string>(value: string | null | undefined, schema: v.PicklistSchema<readonly T[], undefined>) => T | undefined;
-declare const ThemeColorSlotSchema: v.PicklistSchema<readonly ["dk1", "lt1", "dk2", "lt2", "accent1", "accent2", "accent3", "accent4", "accent5", "accent6", "hlink", "folHlink", "background1", "text1", "background2", "text2"], undefined>;
-declare const BorderStyleSchema: v.PicklistSchema<readonly ["none", "single", "double", "dotted", "dashed", "thick", "triple", "thinThickSmallGap", "thickThinSmallGap", "thinThickMediumGap", "thickThinMediumGap", "thinThickLargeGap", "thickThinLargeGap", "wave", "doubleWave", "dashSmallGap", "dashDotStroked", "threeDEmboss", "threeDEngrave", "outset", "inset", "nil"], undefined>;
 declare const UnderlineStyleSchema: v.PicklistSchema<readonly ["none", "single", "words", "double", "thick", "dotted", "dottedHeavy", "dash", "dashedHeavy", "dashLong", "dashLongHeavy", "dotDash", "dashDotHeavy", "dotDotDash", "dashDotDotHeavy", "wave", "wavyHeavy", "wavyDouble"], undefined>;
 declare const HighlightColorSchema: v.PicklistSchema<readonly ["black", "blue", "cyan", "darkBlue", "darkCyan", "darkGray", "darkGreen", "darkMagenta", "darkRed", "darkYellow", "green", "lightGray", "magenta", "none", "red", "white", "yellow"], undefined>;
 declare const TextEffectSchema: v.PicklistSchema<readonly ["none", "blinkBackground", "lights", "antsBlack", "antsRed", "shimmer", "sparkle"], undefined>;
 declare const EmphasisMarkSchema: v.PicklistSchema<readonly ["none", "dot", "comma", "circle", "underDot"], undefined>;
 declare const FontThemeSchema: v.PicklistSchema<readonly ["majorAscii", "majorHAnsi", "majorEastAsia", "majorBidi", "minorAscii", "minorHAnsi", "minorEastAsia", "minorBidi"], undefined>;
 declare const FontHintSchema: v.PicklistSchema<readonly ["default", "eastAsia", "cs"], undefined>;
-declare const ParagraphAlignmentSchema: v.PicklistSchema<readonly ["left", "center", "right", "both", "distribute", "mediumKashida", "highKashida", "lowKashida", "thaiDistribute"], undefined>;
+declare const ParagraphAlignmentSchema: v.PicklistSchema<readonly ["start", "center", "end", "both", "mediumKashida", "distribute", "numTab", "highKashida", "lowKashida", "thaiDistribute", "left", "right"], undefined>;
 declare const LineSpacingRuleSchema: v.PicklistSchema<readonly ["auto", "exact", "atLeast"], undefined>;
-declare const TabStopAlignmentSchema: v.PicklistSchema<readonly ["left", "center", "right", "decimal", "bar", "clear", "num"], undefined>;
+declare const TabStopAlignmentSchema: v.PicklistSchema<readonly ["clear", "start", "center", "end", "decimal", "bar", "num", "left", "right"], undefined>;
 declare const TabLeaderSchema: v.PicklistSchema<readonly ["none", "dot", "hyphen", "underscore", "heavy", "middleDot"], undefined>;
 declare const PositionalTabRelativeToSchema: v.PicklistSchema<readonly ["margin", "indent"], undefined>;
 declare const PositionalTabAlignmentSchema: v.PicklistSchema<readonly ["left", "center", "right"], undefined>;
@@ -29,9 +27,10 @@ declare const SdtLockSchema: v.PicklistSchema<readonly ["sdtLocked", "contentLoc
 declare const FieldTypeSchema: v.PicklistSchema<readonly ["PAGE", "NUMPAGES", "NUMWORDS", "NUMCHARS", "DATE", "TIME", "CREATEDATE", "SAVEDATE", "PRINTDATE", "AUTHOR", "TITLE", "SUBJECT", "KEYWORDS", "COMMENTS", "FILENAME", "FILESIZE", "TEMPLATE", "DOCPROPERTY", "DOCVARIABLE", "REF", "PAGEREF", "NOTEREF", "HYPERLINK", "TOC", "TOA", "INDEX", "SEQ", "STYLEREF", "AUTONUM", "AUTONUMLGL", "AUTONUMOUT", "LISTNUM", "IF", "MERGEFIELD", "NEXT", "NEXTIF", "ASK", "SET", "QUOTE", "INCLUDETEXT", "INCLUDEPICTURE", "SYMBOL", "ADVANCE", "EDITTIME", "REVNUM", "SECTION", "SECTIONPAGES", "USERADDRESS", "USERNAME", "USERINITIALS", "UNKNOWN"], undefined>;
 declare const StyleTypeSchema: v.PicklistSchema<readonly ["paragraph", "character", "numbering", "table"], undefined>;
 declare const ConditionalStyleTypeSchema: v.PicklistSchema<readonly ["band1Horz", "band1Vert", "band2Horz", "band2Vert", "firstCol", "firstRow", "lastCol", "lastRow", "neCell", "nwCell", "seCell", "swCell", "wholeTable"], undefined>;
+declare const TableAlignmentSchema: v.PicklistSchema<readonly ["center", "end", "left", "right", "start"], undefined>;
 declare const TableWidthTypeSchema: v.PicklistSchema<readonly ["auto", "dxa", "nil", "pct"], undefined>;
 declare const TableRowHeightRuleSchema: v.PicklistSchema<readonly ["auto", "atLeast", "exact"], undefined>;
-declare const TableCellTextDirectionSchema: v.PicklistSchema<readonly ["lr", "lrV", "rl", "rlV", "tb", "tbV", "tbRl", "tbRlV", "btLr"], undefined>;
+declare const TextDirectionSchema: v.PicklistSchema<readonly ["tb", "rl", "lr", "tbV", "rlV", "lrV", "btLr", "lrTb", "lrTbV", "tbLrV", "tbRl", "tbRlV"], undefined>;
 declare const ShadingPatternSchema: v.PicklistSchema<readonly ["clear", "solid", "horzStripe", "vertStripe", "reverseDiagStripe", "diagStripe", "horzCross", "diagCross", "thinHorzStripe", "thinVertStripe", "thinReverseDiagStripe", "thinDiagStripe", "thinHorzCross", "thinDiagCross", "pct5", "pct10", "pct12", "pct15", "pct20", "pct25", "pct30", "pct35", "pct37", "pct40", "pct45", "pct50", "pct55", "pct60", "pct62", "pct65", "pct70", "pct75", "pct80", "pct85", "pct87", "pct90", "pct95", "nil"], undefined>;
 declare const FloatingTableXSpecSchema: v.PicklistSchema<readonly ["left", "center", "right", "inside", "outside"], undefined>;
 declare const FloatingTableYSpecSchema: v.PicklistSchema<readonly ["top", "center", "bottom", "inside", "outside", "inline"], undefined>;
@@ -40,9 +39,8 @@ declare const ImageHorizontalAlignmentSchema: v.PicklistSchema<readonly ["left",
 declare const ImageVerticalRelativeToSchema: v.PicklistSchema<readonly ["insideMargin", "line", "margin", "outsideMargin", "page", "paragraph", "topMargin", "bottomMargin"], undefined>;
 declare const ImageVerticalAlignmentSchema: v.PicklistSchema<readonly ["top", "bottom", "center", "inside", "outside"], undefined>;
 declare const ImageWrapTextSchema: v.PicklistSchema<readonly ["bothSides", "left", "right", "largest"], undefined>;
-declare const ShapeOutlineStyleSchema: v.PicklistSchema<readonly ["solid", "dot", "dash", "lgDash", "dashDot", "lgDashDot", "lgDashDotDot", "sysDot", "sysDash", "sysDashDot", "sysDashDotDot"], undefined>;
 declare const ShapeTypeSchema: v.PicklistSchema<readonly ["rect", "roundRect", "ellipse", "triangle", "rtTriangle", "parallelogram", "trapezoid", "pentagon", "hexagon", "heptagon", "octagon", "decagon", "dodecagon", "star4", "star5", "star6", "star7", "star8", "star10", "star12", "star16", "star24", "star32", "line", "straightConnector1", "bentConnector2", "bentConnector3", "bentConnector4", "bentConnector5", "curvedConnector2", "curvedConnector3", "curvedConnector4", "curvedConnector5", "rightArrow", "leftArrow", "upArrow", "downArrow", "leftRightArrow", "upDownArrow", "quadArrow", "leftRightUpArrow", "bentArrow", "uturnArrow", "leftUpArrow", "bentUpArrow", "curvedRightArrow", "curvedLeftArrow", "curvedUpArrow", "curvedDownArrow", "stripedRightArrow", "notchedRightArrow", "homePlate", "chevron", "rightArrowCallout", "downArrowCallout", "leftArrowCallout", "upArrowCallout", "leftRightArrowCallout", "quadArrowCallout", "circularArrow", "flowChartProcess", "flowChartAlternateProcess", "flowChartDecision", "flowChartInputOutput", "flowChartPredefinedProcess", "flowChartInternalStorage", "flowChartDocument", "flowChartMultidocument", "flowChartTerminator", "flowChartPreparation", "flowChartManualInput", "flowChartManualOperation", "flowChartConnector", "flowChartOffpageConnector", "flowChartPunchedCard", "flowChartPunchedTape", "flowChartSummingJunction", "flowChartOr", "flowChartCollate", "flowChartSort", "flowChartExtract", "flowChartMerge", "flowChartOnlineStorage", "flowChartDelay", "flowChartMagneticTape", "flowChartMagneticDisk", "flowChartMagneticDrum", "flowChartDisplay", "wedgeRectCallout", "wedgeRoundRectCallout", "wedgeEllipseCallout", "cloudCallout", "borderCallout1", "borderCallout2", "borderCallout3", "accentCallout1", "accentCallout2", "accentCallout3", "callout1", "callout2", "callout3", "accentBorderCallout1", "accentBorderCallout2", "accentBorderCallout3", "actionButtonBlank", "actionButtonHome", "actionButtonHelp", "actionButtonInformation", "actionButtonBackPrevious", "actionButtonForwardNext", "actionButtonBeginning", "actionButtonEnd", "actionButtonReturn", "actionButtonDocument", "actionButtonSound", "actionButtonMovie", "irregularSeal1", "irregularSeal2", "frame", "halfFrame", "corner", "diagStripe", "chord", "arc", "bracketPair", "bracePair", "leftBracket", "rightBracket", "leftBrace", "rightBrace", "can", "cube", "bevel", "donut", "noSmoking", "blockArc", "foldedCorner", "smileyFace", "heart", "lightningBolt", "sun", "moon", "cloud", "snip1Rect", "snip2SameRect", "snip2DiagRect", "snipRoundRect", "round1Rect", "round2SameRect", "round2DiagRect", "plaque", "teardrop", "mathPlus", "mathMinus", "mathMultiply", "mathDivide", "mathEqual", "mathNotEqual", "gear6", "gear9", "funnel", "pieWedge", "pie", "leftCircularArrow", "leftRightCircularArrow", "swooshArrow", "textBox"], undefined>;
-declare const NumberFormatSchema: v.PicklistSchema<readonly ["decimal", "upperRoman", "lowerRoman", "upperLetter", "lowerLetter", "ordinal", "cardinalText", "ordinalText", "hex", "chicago", "ideographDigital", "japaneseCounting", "aiueo", "iroha", "decimalFullWidth", "decimalHalfWidth", "japaneseLegal", "japaneseDigitalTenThousand", "decimalEnclosedCircle", "decimalFullWidth2", "aiueoFullWidth", "irohaFullWidth", "decimalZero", "decimalZero3", "decimalZero4", "decimalZero5", "bullet", "ganada", "chosung", "decimalEnclosedFullstop", "decimalEnclosedParen", "decimalEnclosedCircleChinese", "ideographEnclosedCircle", "ideographTraditional", "ideographZodiac", "ideographZodiacTraditional", "taiwaneseCounting", "ideographLegalTraditional", "taiwaneseCountingThousand", "taiwaneseDigital", "chineseCounting", "chineseLegalSimplified", "chineseCountingThousand", "koreanDigital", "koreanCounting", "koreanLegal", "koreanDigital2", "vietnameseCounting", "russianLower", "russianUpper", "none", "numberInDash", "hebrew1", "hebrew2", "arabicAlpha", "arabicAbjad", "hindiVowels", "hindiConsonants", "hindiNumbers", "hindiCounting", "thaiLetters", "thaiNumbers", "thaiCounting"], undefined>;
+declare const NumberFormatSchema: v.PicklistSchema<readonly ["decimal", "upperRoman", "lowerRoman", "upperLetter", "lowerLetter", "ordinal", "cardinalText", "ordinalText", "hex", "chicago", "ideographDigital", "japaneseCounting", "aiueo", "iroha", "decimalFullWidth", "decimalHalfWidth", "japaneseLegal", "japaneseDigitalTenThousand", "decimalEnclosedCircle", "decimalFullWidth2", "aiueoFullWidth", "irohaFullWidth", "decimalZero", "bullet", "ganada", "chosung", "decimalEnclosedFullstop", "decimalEnclosedParen", "decimalEnclosedCircleChinese", "ideographEnclosedCircle", "ideographTraditional", "ideographZodiac", "ideographZodiacTraditional", "taiwaneseCounting", "ideographLegalTraditional", "taiwaneseCountingThousand", "taiwaneseDigital", "chineseCounting", "chineseLegalSimplified", "chineseCountingThousand", "koreanDigital", "koreanCounting", "koreanLegal", "koreanDigital2", "vietnameseCounting", "russianLower", "russianUpper", "none", "numberInDash", "hebrew1", "hebrew2", "arabicAlpha", "arabicAbjad", "hindiVowels", "hindiConsonants", "hindiNumbers", "hindiCounting", "thaiLetters", "thaiNumbers", "thaiCounting", "bahtText", "dollarText", "custom"], undefined>;
 declare const LevelSuffixSchema: v.PicklistSchema<readonly ["tab", "space", "nothing"], undefined>;
 //#endregion
-export { BorderStyleSchema, ConditionalStyleTypeSchema, EmphasisMarkSchema, FieldTypeSchema, FloatingTableXSpecSchema, FloatingTableYSpecSchema, FontHintSchema, FontThemeSchema, FrameWrapSchema, FrameXAlignSchema, FrameYAlignSchema, HighlightColorSchema, ImageHorizontalAlignmentSchema, ImageHorizontalRelativeToSchema, ImageVerticalAlignmentSchema, ImageVerticalRelativeToSchema, ImageWrapTextSchema, LevelSuffixSchema, LineSpacingRuleSchema, NumberFormatSchema, ParagraphAlignmentSchema, PositionalTabAlignmentSchema, PositionalTabLeaderSchema, PositionalTabRelativeToSchema, SdtLockSchema, ShadingPatternSchema, ShapeOutlineStyleSchema, ShapeTypeSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableCellTextDirectionSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextEffectSchema, ThemeColorSlotSchema, UnderlineStyleSchema, narrowEnum };
+export { ConditionalStyleTypeSchema, EmphasisMarkSchema, FieldTypeSchema, FloatingTableXSpecSchema, FloatingTableYSpecSchema, FontHintSchema, FontThemeSchema, FrameWrapSchema, FrameXAlignSchema, FrameYAlignSchema, HighlightColorSchema, ImageHorizontalAlignmentSchema, ImageHorizontalRelativeToSchema, ImageVerticalAlignmentSchema, ImageVerticalRelativeToSchema, ImageWrapTextSchema, LevelSuffixSchema, LineSpacingRuleSchema, NumberFormatSchema, ParagraphAlignmentSchema, PositionalTabAlignmentSchema, PositionalTabLeaderSchema, PositionalTabRelativeToSchema, SdtLockSchema, ShadingPatternSchema, ShapeTypeSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableAlignmentSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextDirectionSchema, TextEffectSchema, UnderlineStyleSchema, narrowEnum };

package/dist/docx/parserEnums.js

@@ -1,4 +1,4 @@
-import { BORDER_STYLE_VALUES, CONDITIONAL_STYLE_TYPE_VALUES, EMPHASIS_MARK_VALUES, FIELD_TYPE_VALUES, FLOATING_TABLE_X_SPEC_VALUES, FLOATING_TABLE_Y_SPEC_VALUES, FONT_HINT_VALUES, FONT_THEME_VALUES, FRAME_WRAP_VALUES, FRAME_X_ALIGN_VALUES, FRAME_Y_ALIGN_VALUES, HIGHLIGHT_COLOR_VALUES, IMAGE_HORIZONTAL_ALIGNMENT_VALUES, IMAGE_HORIZONTAL_RELATIVE_TO_VALUES, IMAGE_VERTICAL_ALIGNMENT_VALUES, IMAGE_VERTICAL_RELATIVE_TO_VALUES, IMAGE_WRAP_TEXT_VALUES, LEVEL_SUFFIX_VALUES, LINE_SPACING_RULE_VALUES, NUMBER_FORMAT_VALUES, PARAGRAPH_ALIGNMENT_VALUES, POSITIONAL_TAB_ALIGNMENT_VALUES, POSITIONAL_TAB_LEADER_VALUES, POSITIONAL_TAB_RELATIVE_TO_VALUES, SDT_LOCK_VALUES, SHADING_PATTERN_VALUES, SHAPE_OUTLINE_STYLE_VALUES, SHAPE_TYPE_VALUES, STYLE_TYPE_VALUES, TABLE_CELL_TEXT_DIRECTION_VALUES, TABLE_ROW_HEIGHT_RULE_VALUES, TABLE_WIDTH_TYPE_VALUES, TAB_LEADER_VALUES, TAB_STOP_ALIGNMENT_VALUES, TEXT_EFFECT_VALUES, THEME_COLOR_SLOT_VALUES, UNDERLINE_STYLE_VALUES } from "../types/documentEnumValues.js";
+import { CONDITIONAL_STYLE_TYPE_VALUES, EMPHASIS_MARK_VALUES, FIELD_TYPE_VALUES, FLOATING_TABLE_X_SPEC_VALUES, FLOATING_TABLE_Y_SPEC_VALUES, FONT_HINT_VALUES, FONT_THEME_VALUES, FRAME_WRAP_VALUES, FRAME_X_ALIGN_VALUES, FRAME_Y_ALIGN_VALUES, HIGHLIGHT_COLOR_VALUES, IMAGE_HORIZONTAL_ALIGNMENT_VALUES, IMAGE_HORIZONTAL_RELATIVE_TO_VALUES, IMAGE_VERTICAL_ALIGNMENT_VALUES, IMAGE_VERTICAL_RELATIVE_TO_VALUES, IMAGE_WRAP_TEXT_VALUES, LEVEL_SUFFIX_VALUES, LINE_SPACING_RULE_VALUES, NUMBER_FORMAT_VALUES, PARAGRAPH_ALIGNMENT_VALUES, POSITIONAL_TAB_ALIGNMENT_VALUES, POSITIONAL_TAB_LEADER_VALUES, POSITIONAL_TAB_RELATIVE_TO_VALUES, SDT_LOCK_VALUES, SHADING_PATTERN_VALUES, SHAPE_TYPE_VALUES, STYLE_TYPE_VALUES, TABLE_JUSTIFICATION_VALUES, TABLE_ROW_HEIGHT_RULE_VALUES, TABLE_WIDTH_TYPE_VALUES, TAB_LEADER_VALUES, TAB_STOP_ALIGNMENT_VALUES, TEXT_DIRECTION_VALUES, TEXT_EFFECT_VALUES, UNDERLINE_STYLE_VALUES } from "../types/documentEnumValues.js";
 import * as v from "valibot";
 //#region src/docx/parserEnums.ts
 /**
@@ -33,8 +33,6 @@ const narrowEnum = (value, schema) => {
 	const result = v.safeParse(schema, value);
 	return result.success ? result.output : void 0;
 };
-const ThemeColorSlotSchema = v.picklist(THEME_COLOR_SLOT_VALUES);
-const BorderStyleSchema = v.picklist(BORDER_STYLE_VALUES);
 const UnderlineStyleSchema = v.picklist(UNDERLINE_STYLE_VALUES);
 const HighlightColorSchema = v.picklist(HIGHLIGHT_COLOR_VALUES);
 const TextEffectSchema = v.picklist(TEXT_EFFECT_VALUES);
@@ -55,9 +53,10 @@ const SdtLockSchema = v.picklist(SDT_LOCK_VALUES);
 const FieldTypeSchema = v.picklist(FIELD_TYPE_VALUES);
 const StyleTypeSchema = v.picklist(STYLE_TYPE_VALUES);
 const ConditionalStyleTypeSchema = v.picklist(CONDITIONAL_STYLE_TYPE_VALUES);
+const TableAlignmentSchema = v.picklist(TABLE_JUSTIFICATION_VALUES);
 const TableWidthTypeSchema = v.picklist(TABLE_WIDTH_TYPE_VALUES);
 const TableRowHeightRuleSchema = v.picklist(TABLE_ROW_HEIGHT_RULE_VALUES);
-const TableCellTextDirectionSchema = v.picklist(TABLE_CELL_TEXT_DIRECTION_VALUES);
+const TextDirectionSchema = v.picklist(TEXT_DIRECTION_VALUES);
 const ShadingPatternSchema = v.picklist(SHADING_PATTERN_VALUES);
 const FloatingTableXSpecSchema = v.picklist(FLOATING_TABLE_X_SPEC_VALUES);
 const FloatingTableYSpecSchema = v.picklist(FLOATING_TABLE_Y_SPEC_VALUES);
@@ -66,9 +65,8 @@ const ImageHorizontalAlignmentSchema = v.picklist(IMAGE_HORIZONTAL_ALIGNMENT_VAL
 const ImageVerticalRelativeToSchema = v.picklist(IMAGE_VERTICAL_RELATIVE_TO_VALUES);
 const ImageVerticalAlignmentSchema = v.picklist(IMAGE_VERTICAL_ALIGNMENT_VALUES);
 const ImageWrapTextSchema = v.picklist(IMAGE_WRAP_TEXT_VALUES);
-const ShapeOutlineStyleSchema = v.picklist(SHAPE_OUTLINE_STYLE_VALUES);
 const ShapeTypeSchema = v.picklist(SHAPE_TYPE_VALUES);
 const NumberFormatSchema = v.picklist(NUMBER_FORMAT_VALUES);
 const LevelSuffixSchema = v.picklist(LEVEL_SUFFIX_VALUES);
 //#endregion
-export { BorderStyleSchema, ConditionalStyleTypeSchema, EmphasisMarkSchema, FieldTypeSchema, FloatingTableXSpecSchema, FloatingTableYSpecSchema, FontHintSchema, FontThemeSchema, FrameWrapSchema, FrameXAlignSchema, FrameYAlignSchema, HighlightColorSchema, ImageHorizontalAlignmentSchema, ImageHorizontalRelativeToSchema, ImageVerticalAlignmentSchema, ImageVerticalRelativeToSchema, ImageWrapTextSchema, LevelSuffixSchema, LineSpacingRuleSchema, NumberFormatSchema, ParagraphAlignmentSchema, PositionalTabAlignmentSchema, PositionalTabLeaderSchema, PositionalTabRelativeToSchema, SdtLockSchema, ShadingPatternSchema, ShapeOutlineStyleSchema, ShapeTypeSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableCellTextDirectionSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextEffectSchema, ThemeColorSlotSchema, UnderlineStyleSchema, narrowEnum };
+export { ConditionalStyleTypeSchema, EmphasisMarkSchema, FieldTypeSchema, FloatingTableXSpecSchema, FloatingTableYSpecSchema, FontHintSchema, FontThemeSchema, FrameWrapSchema, FrameXAlignSchema, FrameYAlignSchema, HighlightColorSchema, ImageHorizontalAlignmentSchema, ImageHorizontalRelativeToSchema, ImageVerticalAlignmentSchema, ImageVerticalRelativeToSchema, ImageWrapTextSchema, LevelSuffixSchema, LineSpacingRuleSchema, NumberFormatSchema, ParagraphAlignmentSchema, PositionalTabAlignmentSchema, PositionalTabLeaderSchema, PositionalTabRelativeToSchema, SdtLockSchema, ShadingPatternSchema, ShapeTypeSchema, StyleTypeSchema, TabLeaderSchema, TabStopAlignmentSchema, TableAlignmentSchema, TableRowHeightRuleSchema, TableWidthTypeSchema, TextDirectionSchema, TextEffectSchema, UnderlineStyleSchema, narrowEnum };

package/dist/docx/preservedRunContent.d.ts

@@ -0,0 +1,32 @@
+import { document_d_exports } from "../types/document.js";
+import { XmlElement } from "./xmlParser.js";
+//#region src/docx/preservedRunContent.d.ts
+/**
+ * The visible text a preserved run child contributes, or `""` when it shows
+ * nothing. Unknown markup contributes nothing on purpose: guessing at the text
+ * of an element folio has never seen would put invented words in a document.
+ */
+declare const preservedRunChildText: (element: XmlElement) => string;
+/**
+ * Capture one unmodelled run child.
+ *
+ * `captureVerbatimXml` materialises exactly the namespace bindings the
+ * fragment's own prefixes need and leaves the canonical ones to the rebuilt
+ * root, so ordinary `w:` markup comes back byte for byte while a foreign
+ * prefix still arrives bound. Copying the whole root scope onto the fragment
+ * instead would rewrite every capture with declarations it does not use.
+ */
+declare const preserveRunChild: (element: XmlElement) => document_d_exports.PreservedXmlContent;
+/**
+ * One capture from the shared dispatcher's sink, as a member of the
+ * container's own inline union.
+ *
+ * The sink holds markup the schema does not declare for the container at all,
+ * so nothing is known about it beyond its bytes: `text` is empty because a
+ * name folio has never seen shows nothing folio can read.
+ */
+declare const preservedInlineCapture: (xml: string) => document_d_exports.PreservedInline;
+/** Capture one unmodelled inline child of a paragraph or an inline wrapper. */
+declare const preserveInlineChild: (element: XmlElement) => document_d_exports.PreservedInline;
+//#endregion
+export { preserveInlineChild, preserveRunChild, preservedInlineCapture, preservedRunChildText };

package/dist/docx/preservedRunContent.js

@@ -0,0 +1,96 @@
+import { captureVerbatimXml } from "./verbatimCapture.js";
+import { findChildrenByLocalName, getChildElements, getLocalName, getTextContent } from "./xmlParser.js";
+//#region src/docx/preservedRunContent.ts
+/**
+* Where a preserved run child hides text a reader sees, by local name: the
+* child elements whose `w:t` descendants are on the line rather than above,
+* beside, or nowhere.
+*
+* `w:ruby` is the whole list today. Its `w:rt` is the annotation printed above
+* the base and is not the sentence's text; its `w:rubyBase` is.
+*/
+const VISIBLE_TEXT_CHILDREN = /* @__PURE__ */ new Map([["ruby", ["rubyBase"]]]);
+/** Cap on the text one preserved child contributes, mirroring `w:t` handling. */
+const MAX_PRESERVED_TEXT_LENGTH = 1e5;
+const collectTextContent = (element, into) => {
+	for (const child of getChildElements(element)) {
+		if (getLocalName(child.name) === "t") {
+			into.push(getTextContent(child));
+			continue;
+		}
+		collectTextContent(child, into);
+	}
+};
+/**
+* The visible text a preserved run child contributes, or `""` when it shows
+* nothing. Unknown markup contributes nothing on purpose: guessing at the text
+* of an element folio has never seen would put invented words in a document.
+*/
+const preservedRunChildText = (element) => {
+	const sources = VISIBLE_TEXT_CHILDREN.get(getLocalName(element.name));
+	if (sources === void 0) return "";
+	const parts = [];
+	for (const localName of sources) for (const source of findChildrenByLocalName(element, localName)) collectTextContent(source, parts);
+	return parts.join("").slice(0, MAX_PRESERVED_TEXT_LENGTH);
+};
+/**
+* Capture one unmodelled run child.
+*
+* `captureVerbatimXml` materialises exactly the namespace bindings the
+* fragment's own prefixes need and leaves the canonical ones to the rebuilt
+* root, so ordinary `w:` markup comes back byte for byte while a foreign
+* prefix still arrives bound. Copying the whole root scope onto the fragment
+* instead would rewrite every capture with declarations it does not use.
+*/
+const preserveRunChild = (element) => ({
+	type: "preservedXml",
+	xml: captureVerbatimXml(element),
+	text: preservedRunChildText(element)
+});
+/**
+* Where an unmodelled *inline* child hides text a reader sees, by local name.
+*
+* `w:customXml` and `w:smartTag` are transparent wrappers (ECMA-376 §17.5.1,
+* §17.5.1.9): their content is ordinary inline content, so their `w:t`
+* descendants are on the line. A `w:hyperlink` and a `w:fldSimple` are here
+* for the same reason and only in one place — inside another link or another
+* field, where folio captures the inner one rather than modelling the nesting.
+* Their runs print like any other; a capture that kept the markup and not the
+* words would take a linked clause off the page.
+*
+* Every other inline child folio captures — `w:permStart`, `w:proofErr` and
+* the custom-XML revision ranges — is an empty marker, and an element folio
+* has never seen contributes nothing on purpose, because guessing at its text
+* would put invented words in a document.
+*/
+const VISIBLE_TEXT_INLINE_CHILDREN = /* @__PURE__ */ new Set([
+	"customXml",
+	"fldSimple",
+	"hyperlink",
+	"smartTag"
+]);
+/**
+* One capture from the shared dispatcher's sink, as a member of the
+* container's own inline union.
+*
+* The sink holds markup the schema does not declare for the container at all,
+* so nothing is known about it beyond its bytes: `text` is empty because a
+* name folio has never seen shows nothing folio can read.
+*/
+const preservedInlineCapture = (xml) => ({
+	type: "preservedInline",
+	xml,
+	text: ""
+});
+/** Capture one unmodelled inline child of a paragraph or an inline wrapper. */
+const preserveInlineChild = (element) => {
+	const parts = [];
+	if (VISIBLE_TEXT_INLINE_CHILDREN.has(getLocalName(element.name))) collectTextContent(element, parts);
+	return {
+		type: "preservedInline",
+		xml: captureVerbatimXml(element),
+		text: parts.join("").slice(0, MAX_PRESERVED_TEXT_LENGTH)
+	};
+};
+//#endregion
+export { preserveInlineChild, preserveRunChild, preservedInlineCapture, preservedRunChildText };

package/dist/docx/previewBudget.d.ts

@@ -15,44 +15,65 @@
  * would have stopped the other charging for it without anything failing. The
  * table below is that agreement written once, and a producer builds its image
  * out of the entry the budget matches against.
+ *
+ * An entry says what backs the preview, because that decides what there is to
+ * agree on. A source-backed preview is image data in `src`, recognized by the
+ * strings its producer stamps on it and charged by the character. A
+ * descriptor-backed preview is no picture at all, so it carries none of those
+ * strings: a table that offered them would be offering a cap over nothing and
+ * a prefix nothing can start with.
  */
-declare const VML_PREVIEW_DATA_URL_PREFIX = "data:image/svg+xml;charset=utf-8,";
+type PreviewKindName = "vmlShape" | "wpGroup" | "smartArt";
 declare const PREVIEW_KINDS: {
   /** A VML shape folio renders rather than projects (`v:shape`, `v:rect`, ...). */
   readonly vmlShape: {
+    readonly backing: "source";
     readonly mimeType: "image/svg+xml";
     readonly filename: "vml-shape-preview.svg";
     readonly srcPrefix: "data:image/svg+xml;charset=utf-8,";
     readonly maxPackageCharacters: number;
   };
   /**
-   * A SmartArt diagram: its extent filled with one flat rectangle per shape.
+   * A WordprocessingGroup (`wpg:wgp`) folio renders rather than projects.
    *
-   * A raster rather than a vector because the display list decodes only base64
-   * PNG and JPEG, so a vector preview would be missing from every display-list
-   * backend (PDF among them) while still showing in the DOM. That is what
-   * makes this preview expensive: one of them is 7.3 MB of data URL whatever
-   * the package weighs, because its cost follows the extent the author chose
-   * rather than anything the drawing contains.
+   * The group's shapes, their text and any picture they carry are drawn into
+   * one SVG, so a package with many groups retains as much generated text as a
+   * package of VML shapes does, from a producer bounded per drawing the same
+   * way. It was missing from this table, which is not a preview with a generous
+   * cap but a preview with none: the matcher below never named it, so no
+   * package was ever charged for one.
+   */
+  readonly wpGroup: {
+    readonly backing: "source";
+    readonly mimeType: "image/svg+xml";
+    readonly filename: "wordprocessing-group.svg";
+    readonly srcPrefix: "data:image/svg+xml;charset=utf-8,";
+    readonly maxPackageCharacters: number;
+  };
+  /**
+   * A SmartArt diagram: its extent filled with one flat rectangle per shape.
    *
-   * Across the public corpus, the fifty packages that produce one retain a
-   * median of 7.3 MB and a maximum of 51.3 MB (ten previews, from a package
-   * under a megabyte). The cap is set above that maximum: it refuses no
-   * legitimate file in the corpus while bounding what had no bound at all, and
-   * it is a ceiling rather than a fix. The fix is for the preview to be a
-   * descriptor the renderer rasterizes, which needs the display-list contract
-   * to carry one.
+   * It used to be a raster in `src`, because the display list decoded only
+   * base64 PNG and JPEG and a vector preview would have been missing from
+   * every backend but the DOM. That cost 7.3 MB of data URL per diagram
+   * whatever the package weighed, up to 51.3 MB in one corpus package, because
+   * it followed the extent the author chose rather than anything the drawing
+   * contained. It is a `PreviewDescriptor` now, which both backends draw, and
+   * the bound on it is the shape cap the parse applies to the description.
    */
   readonly smartArt: {
-    readonly mimeType: "image/png";
-    readonly filename: "smartart-preview.png";
-    readonly srcPrefix: "data:image/png;base64,";
-    readonly maxPackageCharacters: number;
+    readonly backing: "descriptor";
   };
 };
-type PreviewKindName = keyof typeof PREVIEW_KINDS;
-/** Per-kind allowances for one package, defaulting to the table's caps. */
-type PreviewBudgetOverrides = Partial<Record<PreviewKindName, number>>;
+/** The kinds a package charges: those whose preview is image data it retains. */
+type SourceBackedKindName = { [Name in PreviewKindName]: (typeof PREVIEW_KINDS)[Name]["backing"] extends "source" ? Name : never; }[PreviewKindName];
+/**
+ * Per-kind allowances for one package, defaulting to the table's caps.
+ *
+ * Only a source-backed kind has an allowance: there is no number that would
+ * change what the budget does to a descriptor.
+ */
+type PreviewBudgetOverrides = Partial<Record<SourceBackedKindName, number>>;
 /**
  * Charge every generated preview in the model against its kind's allowance and
  * drop the `src` of those past it. Dropping leaves the image in place with its
@@ -61,4 +82,4 @@ type PreviewBudgetOverrides = Partial<Record<PreviewKindName, number>>;
  */
 declare const enforcePackagePreviewBudget: (root: unknown, overrides?: PreviewBudgetOverrides) => void;
 //#endregion
-export { PREVIEW_KINDS, PreviewBudgetOverrides, VML_PREVIEW_DATA_URL_PREFIX, enforcePackagePreviewBudget };
+export { PREVIEW_KINDS, PreviewBudgetOverrides, enforcePackagePreviewBudget };