wpd-codec 3.1.1 → 3.2.0

package/dist/read.cjs

@@ -10,6 +10,8 @@ const require_stream_eol = require("./stream/eol.cjs");
 const require_stream_page = require("./stream/page.cjs");
 const require_stream_style = require("./stream/style.cjs");
 const require_stream_table = require("./stream/table.cjs");
+const require_stream_box = require("./stream/box.cjs");
+const require_stream_formula = require("./stream/formula.cjs");
 const require_stream_tab = require("./stream/tab.cjs");
 const require_stream_tokenise = require("./stream/tokenise.cjs");
 let document_schema_js = require("document-schema.js");
@@ -30,6 +32,8 @@ const CROSS_REFERENCE_GROUP = 213;
 const HEADER_FOOTER_GROUP = 214;
 const FOOTNOTE_ENDNOTE_GROUP = 215;
 const MERGE_GROUP = 222;
+const MERGE_FIELD_ON = 76;
+const MERGE_FIELD_OFF = 77;
 const BOX_GROUP = 223;
 const PARAGRAPH_GROUP = 211;
 const PARAGRAPH_SET_JUSTIFICATION = 5;
@@ -49,6 +53,7 @@ const JUSTIFICATION = [
 const THREE_THOUSAND_SIX_HUNDREDTHS_PER_POINT = 50;
 const COLOR_COMPONENT_MAX = 255;
 const NO_SPAN = 1;
+const MAX_STYLE_RESOLUTION_DEPTH = 16;
 function sameAttributes(a, b) {
 	return a.bold === b.bold && a.italic === b.italic && a.underline === b.underline && a.strike === b.strike;
 }
@@ -80,29 +85,35 @@ function flushRun(state) {
 function targetBlocks(state) {
 	return state.table === void 0 ? state.blocks : state.table.cellBlocks;
 }
-function flushParagraph(state) {
+function flushParagraph(state, sink) {
 	flushRun(state);
+	if (state.openMergeFieldStartRun !== void 0) {
+		state.openMergeFieldStartRun = void 0;
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.MergeFieldSpansParagraphs, "A merge field's own On/Off pair straddled a paragraph boundary, which the run-scoped field construct cannot express; its text became ordinary paragraph text with no field tag.");
+	}
 	const alignment = state.pendingAlignment ?? state.alignment;
+	const constructs = state.pendingConstructs.splice(0, state.pendingConstructs.length);
 	const paragraph = {
 		kind: "paragraph",
 		runs: state.runs.splice(0, state.runs.length),
 		...alignment === void 0 ? {} : { alignment },
 		...state.pendingHeadingLevel === void 0 ? {} : { headingLevel: state.pendingHeadingLevel },
-		...state.pendingListLevel === void 0 ? {} : { list: { level: state.pendingListLevel } }
+		...state.pendingListLevel === void 0 ? {} : { list: { level: state.pendingListLevel } },
+		...constructs.length === 0 ? {} : { constructs }
 	};
 	state.pendingHeadingLevel = void 0;
 	state.pendingListLevel = void 0;
 	state.pendingAlignment = void 0;
 	targetBlocks(state).push(paragraph);
 }
-function flushParagraphIfContent(state) {
+function flushParagraphIfContent(state, sink) {
 	if (state.text.length === 0 && state.runs.length === 0) return;
-	flushParagraph(state);
+	flushParagraph(state, sink);
 }
 function effectiveStyle(state) {
 	for (let index = state.styleScopes.length - 1; index >= 0; index -= 1) {
-		const scope = state.styleScopes[index];
-		if (scope !== void 0) return scope;
+		const semantics = state.styleScopes[index]?.semantics;
+		if (semantics !== void 0) return semantics;
 	}
 }
 function appendText(state, text) {
@@ -123,6 +134,9 @@ function readCellAttributes(state, nonDeletable, sink) {
 	const spanning = require_stream_table.findEmbeddedSubfunction(subfunctions, 133);
 	const fill = require_stream_table.findEmbeddedSubfunction(subfunctions, 134);
 	const row = require_stream_table.findEmbeddedSubfunction(subfunctions, 128);
+	const formulaData = require_stream_table.findEmbeddedSubfunction(subfunctions, 129);
+	const formula = formulaData === void 0 ? void 0 : require_stream_formula.readTableFormula(formulaData);
+	if (formulaData !== void 0 && formula === void 0) reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.TableFormulaUnresolved, "A table cell carries a formula this reader could not decode with confidence, so the cell keeps its displayed text but not the formula that produced it.");
 	const cellSpanning = spanning === void 0 ? void 0 : require_stream_table.readCellSpanning(spanning);
 	const cellFill = fill === void 0 ? void 0 : require_stream_table.readCellFill(fill);
 	if (cellFill?.blended === true) reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.CellFillBlended, "A cell is filled with a shaded blend of two colours, resolved to a 'pattern' fill whose density is this reader's own best-effort derivation, not a value confirmed against a specification.");
@@ -132,13 +146,14 @@ function readCellAttributes(state, nonDeletable, sink) {
 			background: cellFill?.fill,
 			columnSpan: cellSpanning?.columnSpan ?? NO_SPAN,
 			rowSpan: cellSpanning?.rowSpan ?? NO_SPAN,
-			covered: cellSpanning?.coveredFromLeft === true || cellSpanning?.coveredFromAbove === true
+			covered: cellSpanning?.coveredFromLeft === true || cellSpanning?.coveredFromAbove === true,
+			formula
 		},
 		rowHeightPt: row === void 0 ? void 0 : require_stream_table.readRowInformation(row)?.heightPt
 	};
 }
-function closeCell(state, table, attributes) {
-	flushParagraphIfContent(state);
+function closeCell(state, table, attributes, sink) {
+	flushParagraphIfContent(state, sink);
 	const blocks = table.cellBlocks;
 	table.cellBlocks = [];
 	if (attributes.covered) return;
@@ -149,7 +164,8 @@ function closeCell(state, table, attributes) {
 		blocks,
 		...attributes.columnSpan > NO_SPAN ? { colSpan: attributes.columnSpan } : {},
 		...attributes.rowSpan > NO_SPAN ? { rowSpan: attributes.rowSpan } : {},
-		...attributes.background === void 0 ? {} : { background: attributes.background }
+		...attributes.background === void 0 ? {} : { background: attributes.background },
+		...attributes.formula === void 0 ? {} : { formula: attributes.formula }
 	};
 	table.cells.push(cell);
 }
@@ -181,14 +197,14 @@ function applyEolMapping(state, mapping, nonDeletable, sink) {
 			appendText(state, " ");
 			return;
 		case "hardReturn":
-			flushParagraph(state);
+			flushParagraph(state, sink);
 			return;
 		case "hardEndOfColumn":
 			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.ColumnBreakFlattened, "A column break became a paragraph break.");
-			flushParagraph(state);
+			flushParagraph(state, sink);
 			return;
 		case "hardEndOfPage":
-			flushParagraph(state);
+			flushParagraph(state, sink);
 			targetBlocks(state).push({ kind: "pageBreak" });
 			return;
 		case "tableCell":
@@ -198,13 +214,13 @@ function applyEolMapping(state, mapping, nonDeletable, sink) {
 			const table = state.table;
 			if (table === void 0) {
 				reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.TableFlattened, "A table cell or row boundary appeared with no table definition open; its text became a paragraph.");
-				flushParagraph(state);
+				flushParagraph(state, sink);
 				return;
 			}
 			const { attributes, rowHeightPt } = readCellAttributes(state, nonDeletable, sink);
 			if (rowHeightPt !== void 0) table.rowHeightPt = rowHeightPt;
 			const cellIsOpen = state.text.length > 0 || state.runs.length > 0 || table.cellBlocks.length > 0;
-			if (mapping !== "tableOff" || cellIsOpen) closeCell(state, table, attributes);
+			if (mapping !== "tableOff" || cellIsOpen) closeCell(state, table, attributes, sink);
 			if (mapping === "tableCell") return;
 			closeRow(table);
 			if (mapping === "tableOff") closeTable(state);
@@ -233,13 +249,13 @@ function applySingleByteFunction(state, code, sink) {
 			appendText(state, "-");
 			return;
 		case DORMANT_HARD_RETURN:
-			flushParagraph(state);
+			flushParagraph(state, sink);
 			return;
 		case SOFT_END_OF_CENTER_ALIGN:
 			appendText(state, " ");
 			return;
 		case HARD_END_OF_CENTER_ALIGN:
-			flushParagraph(state);
+			flushParagraph(state, sink);
 			return;
 		case START_OF_TEXT_TO_SKIP:
 			state.skipDepth += 1;
@@ -301,13 +317,56 @@ function applyColumnGroup(state, token, sink) {
 	if (token.subgroup === 0) setPageDimension(state, "leftPt", margin, sink);
 	else if (token.subgroup === 1) setPageDimension(state, "rightPt", margin, sink);
 }
-function applyStyleGroup(state, token) {
+function snapshotFormatting(state) {
+	return {
+		activeAttributes: new Set(state.activeAttributes),
+		fontFamily: state.fontFamily,
+		sizePt: state.sizePt,
+		color: state.color
+	};
+}
+function restoreFormattingSnapshot(state, snapshot) {
+	flushRun(state);
+	state.activeAttributes.clear();
+	for (const attribute of snapshot.activeAttributes) state.activeAttributes.add(attribute);
+	state.attributes = require_stream_attributes.runAttributesFrom(state.activeAttributes);
+	state.fontFamily = snapshot.fontFamily;
+	state.sizePt = snapshot.sizePt;
+	state.color = snapshot.color;
+}
+function applyStylePacketBegin(state, token, container, sink) {
+	const prefixId = token.prefixIds[0];
+	if (prefixId === void 0) return;
+	const packet = require_container_prefix.packetByPrefixId(container.packets, prefixId);
+	if (packet?.packetType !== 48) return;
+	const beginBlock = require_stream_style.readStyleBeginBlock(packet.bytes);
+	if (beginBlock === void 0) return;
+	if (state.styleResolutionDepth >= MAX_STYLE_RESOLUTION_DEPTH) {
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.StyleResolutionDepthExceeded, "A chain of styles resolving one another's own packets ran deeper than this reader will follow, so the deepest style's own direct formatting was not applied.");
+		return;
+	}
+	const snapshot = snapshotFormatting(state);
+	const tokens = require_stream_tokenise.tokeniseDocumentArea(beginBlock, 0, beginBlock.length);
+	state.styleResolutionDepth += 1;
+	for (const beginToken of tokens) applyToken(state, beginToken, container, sink);
+	state.styleResolutionDepth -= 1;
+	return snapshot;
+}
+function applyStyleGroup(state, token, container, sink) {
 	if (require_stream_style.isStyleScopeOpener(token.subgroup)) {
 		const systemStyleNumber = require_stream_style.readSystemStyleNumber(token.nonDeletable);
-		state.styleScopes.push(systemStyleNumber === void 0 ? void 0 : require_stream_style.styleSemanticsFor(systemStyleNumber));
+		const semantics = systemStyleNumber === void 0 ? void 0 : require_stream_style.styleSemanticsFor(systemStyleNumber);
+		const snapshot = applyStylePacketBegin(state, token, container, sink);
+		state.styleScopes.push({
+			semantics,
+			snapshot
+		});
 		return;
 	}
-	if (require_stream_style.isStyleScopeCloser(token.subgroup)) state.styleScopes.pop();
+	if (require_stream_style.isStyleScopeCloser(token.subgroup)) {
+		const entry = state.styleScopes.pop();
+		if (entry?.snapshot !== void 0) restoreFormattingSnapshot(state, entry.snapshot);
+	}
 }
 function applyDisplayNumberGroup(state, token, sink) {
 	if (require_stream_style.isParagraphNumberDisplayOn(token.subgroup)) {
@@ -323,7 +382,7 @@ function applyCharacterGroup(state, token, container, sink) {
 	switch (token.subgroup) {
 		case 42:
 			closeTable(state);
-			flushParagraphIfContent(state);
+			flushParagraphIfContent(state, sink);
 			state.table = {
 				columnWidthsPt: [],
 				rows: [],
@@ -368,6 +427,104 @@ function applyCharacterGroup(state, token, container, sink) {
 		default: return;
 	}
 }
+function applyMergeGroup(state, token, sink) {
+	if (token.subgroup === MERGE_FIELD_ON) {
+		flushRun(state);
+		state.openMergeFieldStartRun = state.runs.length;
+		return;
+	}
+	if (token.subgroup === MERGE_FIELD_OFF) {
+		const startRun = state.openMergeFieldStartRun;
+		state.openMergeFieldStartRun = void 0;
+		if (startRun === void 0) return;
+		flushRun(state);
+		const endRun = state.runs.length;
+		const instruction = state.runs.slice(startRun, endRun).map((run) => run.text).join("");
+		state.pendingConstructs.push({
+			descriptor: {
+				kind: "field",
+				instruction
+			},
+			startRun,
+			endRun
+		});
+		return;
+	}
+	reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.MergeCodeDropped, "This document contains merge codes, which are a form-letter template's placeholders rather than text.");
+}
+function plainTextOf(blocks) {
+	return blocks.filter((block) => block.kind === "paragraph").map((paragraph) => paragraph.runs.map((run) => run.text).join("")).join("\n");
+}
+function applyBoxGroup(state, token, container, sink) {
+	const boxContent = require_stream_box.readBoxContent(token.nonDeletable, token.prefixIds);
+	if (boxContent === void 0) {
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.BoxDropped, "This document contains a box -- a figure, text box, equation, or graphic -- whose function-level override names no content this reader can resolve.");
+		return;
+	}
+	const packet = boxContent.contentType === 1 || boxContent.contentType === 2 || boxContent.contentType === 4 ? require_container_prefix.packetByPrefixId(container.packets, boxContent.contentPrefixId) : void 0;
+	const textBlocks = packet?.packetType !== 8 ? void 0 : require_container_prefix.readGeneralWpTextBlocks(packet.bytes);
+	if (textBlocks === void 0) {
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.BoxContentUnresolved, "This document contains a box whose content this reader could not read -- an image, OLE object, or other content type this reader does not yet decode into the shared schema.");
+		return;
+	}
+	if (boxContent.frame === void 0) {
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.BoxFrameUnresolved, "This document contains a box whose content this reader could read, but whose function-level override states no width and height this reader can trust, so its content was not lifted.");
+		return;
+	}
+	flushParagraphIfContent(state, sink);
+	const nestedTokens = require_stream_tokenise.tokeniseDocumentArea(textBlocks, 0, textBlocks.length);
+	const frame = {
+		xPt: boxContent.frame.xPt,
+		yPt: boxContent.frame.yPt,
+		widthPt: boxContent.frame.widthPt,
+		heightPt: boxContent.frame.heightPt
+	};
+	if (boxContent.contentType === 4) {
+		const { blocks } = foldTokens(nestedTokens, container, sink);
+		const embed = {
+			kind: "embeddedObject",
+			objectKind: "formula",
+			frame,
+			document: {
+				kind: "formula",
+				metadata: {},
+				formula: {
+					mathml: [],
+					source: {
+						format: "wpd",
+						xml: plainTextOf(blocks)
+					}
+				}
+			}
+		};
+		targetBlocks(state).push(embed);
+		return;
+	}
+	const { blocks, page } = foldTokens(nestedTokens, container, sink);
+	const embed = {
+		kind: "embeddedObject",
+		objectKind: "wordprocessing",
+		frame,
+		document: {
+			kind: "wordprocessing",
+			metadata: {},
+			sections: [{
+				pageSize: {
+					widthPt: page.widthPt ?? 612,
+					heightPt: page.heightPt ?? 792
+				},
+				margins: {
+					topPt: page.topPt ?? 72,
+					rightPt: page.rightPt ?? 72,
+					bottomPt: page.bottomPt ?? 72,
+					leftPt: page.leftPt ?? 72
+				},
+				blocks
+			}]
+		}
+	};
+	targetBlocks(state).push(embed);
+}
 function applyVariableFunction(state, token, container, sink) {
 	switch (token.group) {
 		case 208: {
@@ -398,7 +555,7 @@ function applyVariableFunction(state, token, container, sink) {
 			return;
 		}
 		case 221:
-			applyStyleGroup(state, token);
+			applyStyleGroup(state, token, container, sink);
 			return;
 		case 218:
 			applyDisplayNumberGroup(state, token, sink);
@@ -413,10 +570,10 @@ function applyVariableFunction(state, token, container, sink) {
 			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.NoteDropped, "This document contains a footnote or endnote; its text lives in a prefix packet the flat content model has nowhere to put.");
 			return;
 		case MERGE_GROUP:
-			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.MergeCodeDropped, "This document contains merge codes, which are a form-letter template's placeholders rather than text.");
+			applyMergeGroup(state, token, sink);
 			return;
 		case BOX_GROUP:
-			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.BoxDropped, "This document contains a box -- a figure, text box, equation, or graphic -- whose contents were not read.");
+			applyBoxGroup(state, token, container, sink);
 			return;
 		default: return;
 	}
@@ -450,6 +607,32 @@ function applyFixedFunction(state, token, sink) {
 	flushRun(state);
 	state.attributes = next;
 }
+function applyToken(state, token, container, sink) {
+	switch (token.kind) {
+		case "character": {
+			const character = require_stream_characters.decodeSingleByteCharacter(token.byte);
+			if (character === void 0) {
+				sink({
+					code: require_diagnostics.WpdDiagnosticCodes.UnmappedCharacter,
+					message: `Byte ${token.byte} in the document area has no character mapping and was rendered as U+FFFD.`
+				});
+				appendText(state, "�");
+				return;
+			}
+			appendText(state, character);
+			return;
+		}
+		case "singleByteFunction":
+			applySingleByteFunction(state, token.code, sink);
+			return;
+		case "variableFunction":
+			applyVariableFunction(state, token, container, sink);
+			return;
+		case "fixedFunction":
+			applyFixedFunction(state, token, sink);
+			return;
+	}
+}
 function foldTokens(tokens, container, sink) {
 	const state = {
 		blocks: [],
@@ -463,6 +646,7 @@ function foldTokens(tokens, container, sink) {
 		alignment: void 0,
 		skipDepth: 0,
 		styleScopes: [],
+		styleResolutionDepth: 0,
 		pendingHeadingLevel: void 0,
 		pendingListLevel: void 0,
 		pendingAlignment: void 0,
@@ -477,32 +661,13 @@ function foldTokens(tokens, container, sink) {
 			changeReported: false
 		},
 		table: void 0,
-		reported: /* @__PURE__ */ new Set()
+		reported: /* @__PURE__ */ new Set(),
+		pendingConstructs: [],
+		openMergeFieldStartRun: void 0
 	};
-	for (const token of tokens) switch (token.kind) {
-		case "character": {
-			const character = require_stream_characters.decodeSingleByteCharacter(token.byte);
-			if (character === void 0) {
-				sink({
-					code: require_diagnostics.WpdDiagnosticCodes.UnmappedCharacter,
-					message: `Byte ${token.byte} in the document area has no character mapping and was rendered as U+FFFD.`
-				});
-				appendText(state, "�");
-				break;
-			}
-			appendText(state, character);
-			break;
-		}
-		case "singleByteFunction":
-			applySingleByteFunction(state, token.code, sink);
-			break;
-		case "variableFunction":
-			applyVariableFunction(state, token, container, sink);
-			break;
-		case "fixedFunction": applyFixedFunction(state, token, sink);
-	}
+	for (const token of tokens) applyToken(state, token, container, sink);
 	flushRun(state);
-	if (state.runs.length > 0) flushParagraph(state);
+	if (state.runs.length > 0) flushParagraph(state, sink);
 	if (state.table !== void 0) {
 		closeRow(state.table);
 		closeTable(state);

package/dist/read.d.cts

@@ -1,4 +1,4 @@
-import { i as WpdDiagnosticSink } from "./diagnostics-CR0pXOyb.cjs";
+import { i as WpdDiagnosticSink } from "./diagnostics-Dtb2uPw4.cjs";
 import { ContentDocument, DocumentTree } from "document-schema.js";
 //#region src/read.d.ts
 interface ReadWpdOptions {

package/dist/read.d.ts

@@ -1,4 +1,4 @@
-import { i as WpdDiagnosticSink } from "./diagnostics-CR0pXOyb.js";
+import { i as WpdDiagnosticSink } from "./diagnostics-Dtb2uPw4.js";
 import { ContentDocument, DocumentTree } from "document-schema.js";
 //#region src/read.d.ts
 interface ReadWpdOptions {