document-cli 1.3.0 → 1.5.0

package/dist/index.cjs

@@ -533,8 +533,158 @@ function registerConversionCommands(program) {
 	});
 }
 //#endregion
+//#region src/docx-extras-format.ts
+const INDENT$1 = "  ";
+function indent$1(depth) {
+	return INDENT$1.repeat(depth);
+}
+function commentLine(comment, position) {
+	const author = comment.author ?? "(no author)";
+	return `${indent$1(1)}[${position}] ${author}: ${comment.text}`;
+}
+function footnoteLine(footnote, position) {
+	const typeSuffix = footnote.type === void 0 ? "" : ` (${footnote.type})`;
+	return `${indent$1(1)}[${position}]${typeSuffix} ${footnote.text}`;
+}
+function commentsSection(comments) {
+	if (comments.length === 0) return [];
+	return ["comments", ...comments.map((comment, index) => commentLine(comment, index + 1))];
+}
+function footnotesSection(footnotes) {
+	if (footnotes.length === 0) return [];
+	return ["footnotes", ...footnotes.map((footnote, index) => footnoteLine(footnote, index + 1))];
+}
+function headerOrFooterSection(label, values) {
+	if (values.length === 0) return [];
+	return [label, ...values.map((text, index) => `${indent$1(1)}[${index + 1}] ${text}`)];
+}
+function numberingLevelLine(ilvl, level) {
+	const restartSuffix = level.restart === void 0 ? "" : `, restarts at level ${level.restart}`;
+	return `${indent$1(2)}level ${ilvl}: ${level.format} ${JSON.stringify(level.text)} starting at ${level.startAt}${restartSuffix}`;
+}
+function numberingSection(numbering) {
+	const numIds = Object.keys(numbering);
+	if (numIds.length === 0) return [];
+	const lines = ["numbering"];
+	for (const numId of numIds) {
+		const definition = numbering[numId];
+		if (definition === void 0) continue;
+		lines.push(`${indent$1(1)}numId ${numId}`);
+		const ilvls = Object.keys(definition.levels).sort((a, b) => Number(a) - Number(b));
+		for (const ilvl of ilvls) {
+			const level = definition.levels[ilvl];
+			if (level === void 0) continue;
+			lines.push(numberingLevelLine(ilvl, level));
+		}
+	}
+	return lines;
+}
+function formatDocxExtrasLines(extras) {
+	const nonEmptySections = [
+		commentsSection(extras.comments),
+		footnotesSection(extras.footnotes),
+		headerOrFooterSection("headers", extras.headers),
+		headerOrFooterSection("footers", extras.footers),
+		numberingSection(extras.numbering)
+	].filter((section) => section.length > 0);
+	if (nonEmptySections.length === 0) return ["This document carries no comments, footnotes, headers, footers, or numbering definitions."];
+	return nonEmptySections.flatMap((section, index) => index === 0 ? section : ["", ...section]);
+}
+//#endregion
+//#region src/commands/docx-extras.ts
+async function runDocxExtras(input, options) {
+	const command = "docx-extras";
+	const { signal, getAbortReason } = createRuntimeSignal({});
+	try {
+		const inputBytes = await readInput(input, { signal });
+		const pkg = (0, documents_js.decodePackage)(new Uint8Array(inputBytes));
+		const extras = (0, documents_js.readDocxExtras)(pkg);
+		if (options.json) {
+			process.stdout.write(`${JSON.stringify(extras)}\n`);
+			return 0;
+		}
+		for (const line of formatDocxExtrasLines(extras)) process.stdout.write(`${line}\n`);
+		return 0;
+	} catch (error) {
+		process.stderr.write(`[${command}] ${formatError(error, false)}\n`);
+		return mapErrorToExit(error, getAbortReason());
+	}
+}
+function registerDocxExtrasCommand(program) {
+	program.command("docx-extras <input>").description("print a docx's own comments, footnotes, headers, footers, and numbering definitions -- data readDocxContent's ContentDocument cannot carry").option("--json", "emit the raw DocxExtras object as JSON instead of a human-readable report", false).action(async (input, options) => {
+		process.exitCode = await runDocxExtras(input, options);
+	});
+}
+//#endregion
+//#region src/commands/fonts.ts
+const FONT_SOURCE_FORMATS = {
+	docx: true,
+	pptx: true,
+	odt: true,
+	odp: true,
+	ods: true,
+	odg: true
+};
+function isFontSourceFormat(format) {
+	return format in FONT_SOURCE_FORMATS;
+}
+function resolveFontSourcePackage(format, bytes) {
+	if (format === "docx" || format === "pptx") return {
+		kind: format,
+		package: (0, documents_js.decodePackage)(bytes)
+	};
+	return {
+		kind: "odf",
+		package: (0, odf_js.decodePackage)(bytes)
+	};
+}
+async function runFonts(input, options) {
+	const command = "fonts";
+	const { signal, getAbortReason } = createRuntimeSignal({});
+	try {
+		const format = inferFormatFromExtension(input);
+		if (format === void 0) {
+			process.stderr.write(`[${command}] cannot infer a document format from '${input}'; expected one of docx, pptx, odt, odp, ods, odg\n`);
+			return 2;
+		}
+		if (!isFontSourceFormat(format)) {
+			process.stderr.write(`[${command}] '${format}' documents carry no source-embedded font faces this command can extract; expected one of docx, pptx, odt, odp, ods, odg\n`);
+			return 2;
+		}
+		const inputBytes = await readInput(input, { signal });
+		const source = resolveFontSourcePackage(format, new Uint8Array(inputBytes));
+		const summaries = (0, documents_js.extractSourceFonts)(source).map((face) => ({
+			family: face.family,
+			bold: face.bold,
+			italic: face.italic,
+			byteLength: face.bytes.length
+		}));
+		if (options.json) {
+			process.stdout.write(`${JSON.stringify(summaries)}\n`);
+			return 0;
+		}
+		if (summaries.length === 0) {
+			process.stdout.write("This document embeds no source fonts.\n");
+			return 0;
+		}
+		for (const face of summaries) {
+			const style = [face.bold ? "bold" : void 0, face.italic ? "italic" : void 0].filter((value) => value !== void 0).join(" ");
+			process.stdout.write(`${face.family}${style === "" ? "" : ` (${style})`} -- ${face.byteLength} bytes\n`);
+		}
+		return 0;
+	} catch (error) {
+		process.stderr.write(`${formatError(error, false)}\n`);
+		return mapErrorToExit(error, getAbortReason());
+	}
+}
+function registerFontsCommand(program) {
+	program.command("fonts <input>").description("list every source-embedded font face a docx/pptx/odt/odp/ods/odg document carries (family, weight/style, byte length)").option("--json", "emit the face list as a JSON array instead of a human-readable report", false).action(async (input, options) => {
+		process.exitCode = await runFonts(input, options);
+	});
+}
+//#endregion
 //#region src/commands/formats.ts
-const COMMANDS_NOT_LISTED = "odm-to-pdf, odb-to-csv, odb-to-xlsx, odb-tables, odb-forms, odb-reports, pdf-inspect, from-package";
+const COMMANDS_NOT_LISTED = "odm-to-pdf, odb-to-csv, odb-to-xlsx, odb-tables, odb-forms, odb-reports, pdf-inspect, from-package, fonts, docx-extras, metadata, set-metadata";
 function registerFormatsCommand(program) {
 	program.command("formats").description("list every source -> target conversion this CLI supports via a <source>-to-<target> command").option("--json", "emit the conversion list as a JSON array instead of a human-readable table", false).action((options) => {
 		const { conversions } = (0, documents_js.createLocalDocumentConverter)();
@@ -627,6 +777,78 @@ function registerFromPackageCommand(program) {
 	});
 }
 //#endregion
+//#region src/runtime/metadata-format.ts
+const METADATA_KEYS = [
+	"title",
+	"author",
+	"subject",
+	"keywords",
+	"creator",
+	"producer",
+	"createdIso",
+	"modifiedIso"
+];
+function isPresent(entry) {
+	return entry[1] !== void 0;
+}
+function formatMetadataValue(value) {
+	return typeof value === "string" ? value : value.join(", ");
+}
+function presentMetadataEntries(metadata) {
+	return METADATA_KEYS.map((key) => [key, metadata[key]]).filter(isPresent);
+}
+function formatMetadataLines(metadata) {
+	return presentMetadataEntries(metadata).map(([key, value]) => `${key}: ${formatMetadataValue(value)}`);
+}
+//#endregion
+//#region src/commands/metadata.ts
+function readMetadataForFormat(format, bytes, signal) {
+	switch (format) {
+		case "docx": return (0, documents_js.readDocxContent)((0, documents_js.decodePackage)(bytes)).metadata;
+		case "pptx": return (0, documents_js.readPptxContent)((0, documents_js.decodePackage)(bytes)).metadata;
+		case "odt": return (0, documents_js.readOdtContent)((0, odf_js.decodePackage)(bytes)).metadata;
+		case "odp": return (0, documents_js.readOdpContent)((0, odf_js.decodePackage)(bytes)).metadata;
+		case "ods": return (0, documents_js.readOdsContent)((0, odf_js.decodePackage)(bytes)).metadata;
+		case "odg": return (0, documents_js.readOdgContent)((0, odf_js.decodePackage)(bytes)).metadata;
+		case "odf": return (0, documents_js.readOdfFormulaContent)((0, odf_js.decodePackage)(bytes)).metadata;
+		case "markdown": return (0, documents_js.readMarkdownContent)((0, documents_js.decodeMarkdownText)(bytes)).metadata;
+		case "pdf": return (0, documents_js.readPdf)(bytes, { signal }).metadata;
+		case "xlsx": return (0, documents_js.readPdf)((0, documents_js.xlsxToPdf)(bytes, { signal }), { signal }).metadata;
+	}
+}
+async function runMetadata(input, options) {
+	const command = "metadata";
+	const source = inferFormatFromExtension(input);
+	if (source === void 0) {
+		process.stderr.write(`[${command}] cannot infer a source format from '${input}'; rename the file with a recognised extension (${KNOWN_DOCUMENT_FORMATS})\n`);
+		return 2;
+	}
+	const { signal, getAbortReason } = createRuntimeSignal({});
+	try {
+		const inputBytes = await readInput(input, { signal });
+		const metadata = readMetadataForFormat(source, new Uint8Array(inputBytes), signal);
+		if (options.json) {
+			process.stdout.write(`${JSON.stringify(metadata)}\n`);
+			return 0;
+		}
+		const lines = formatMetadataLines(metadata);
+		if (lines.length === 0) {
+			process.stdout.write("This document carries no metadata.\n");
+			return 0;
+		}
+		for (const line of lines) process.stdout.write(`${line}\n`);
+		return 0;
+	} catch (error) {
+		process.stderr.write(`[${command}] ${formatError(error, false)}\n`);
+		return mapErrorToExit(error, getAbortReason());
+	}
+}
+function registerMetadataCommand(program) {
+	program.command("metadata <input>").description(`print a document's own title/author/subject/keywords/creator/producer/created/modified metadata (${KNOWN_DOCUMENT_FORMATS})`).option("--json", "emit the metadata as a JSON object instead of a human-readable report", false).action(async (input, options) => {
+		process.exitCode = await runMetadata(input, options);
+	});
+}
+//#endregion
 //#region src/odb-structure.ts
 const INDENT = "  ";
 function indent(depth) {
@@ -1230,12 +1452,6 @@ function countImagesByFormat(images) {
 	for (const asset of Object.values(images)) counts.set(asset.format, (counts.get(asset.format) ?? 0) + 1);
 	return counts;
 }
-function isPresent(entry) {
-	return entry[1] !== void 0;
-}
-function formatMetadataValue(value) {
-	return typeof value === "string" ? value : value.join(", ");
-}
 async function runPdfInspect(input, options) {
 	const command = "pdf-inspect";
 	const { signal, getAbortReason } = createRuntimeSignal({});
@@ -1253,7 +1469,7 @@ async function runPdfInspect(input, options) {
 			}
 		});
 		if (options.full) {
-			process.stdout.write(`${JSON.stringify(layout, void 0, 2)}\n`);
+			process.stdout.write(`${JSON.stringify((0, documents_js.layoutDocumentWithSchema)(layout), void 0, 2)}\n`);
 			return 0;
 		}
 		const imagesByFormat = countImagesByFormat(layout.images);
@@ -1277,19 +1493,9 @@ async function runPdfInspect(input, options) {
 			const histogramText = Array.from(histogram.entries()).map(([kind, count]) => `${kind}=${count}`).join(", ");
 			process.stdout.write(`  page ${index + 1}: ${page.widthPt}pt x ${page.heightPt}pt${histogramText === "" ? "" : ` (${histogramText})`}\n`);
 		});
-		const presentMetadata = [
-			["title", layout.metadata.title],
-			["author", layout.metadata.author],
-			["subject", layout.metadata.subject],
-			["keywords", layout.metadata.keywords],
-			["creator", layout.metadata.creator],
-			["producer", layout.metadata.producer],
-			["createdIso", layout.metadata.createdIso],
-			["modifiedIso", layout.metadata.modifiedIso]
-		].filter(isPresent);
-		if (presentMetadata.length > 0) {
+		if (presentMetadataEntries(layout.metadata).length > 0) {
 			process.stdout.write("metadata:\n");
-			for (const [key, value] of presentMetadata) process.stdout.write(`  ${key}: ${formatMetadataValue(value)}\n`);
+			for (const line of formatMetadataLines(layout.metadata)) process.stdout.write(`  ${line}\n`);
 		}
 		if (imagesByFormat.size > 0) {
 			process.stdout.write("images:\n");
@@ -1307,8 +1513,155 @@ function registerPdfInspectCommand(program) {
 	});
 }
 //#endregion
+//#region src/commands/set-metadata.ts
+const REBUILD_FORMATS = {
+	docx: true,
+	pptx: true,
+	odt: true,
+	odp: true,
+	ods: true,
+	odg: true,
+	markdown: true
+};
+function isRebuildFormat(format) {
+	return format in REBUILD_FORMATS;
+}
+function readContentForFormat(format, bytes) {
+	switch (format) {
+		case "docx": return (0, documents_js.readDocxContent)((0, documents_js.decodePackage)(bytes));
+		case "pptx": return (0, documents_js.readPptxContent)((0, documents_js.decodePackage)(bytes));
+		case "odt": return (0, documents_js.readOdtContent)((0, odf_js.decodePackage)(bytes));
+		case "odp": return (0, documents_js.readOdpContent)((0, odf_js.decodePackage)(bytes));
+		case "ods": return (0, documents_js.readOdsContent)((0, odf_js.decodePackage)(bytes));
+		case "odg": return (0, documents_js.readOdgContent)((0, odf_js.decodePackage)(bytes));
+		case "markdown": return (0, documents_js.readMarkdownContent)((0, documents_js.decodeMarkdownText)(bytes));
+	}
+}
+function buildBytesForRebuildFormat(format, content) {
+	switch (format) {
+		case "docx": return (0, documents_js.encodePackage)((0, documents_js.buildDocxPackage)(content));
+		case "pptx": return (0, documents_js.encodePackage)((0, documents_js.buildPptxPackage)(content));
+		case "odt": return (0, odf_js.encodePackage)((0, documents_js.buildOdtPackage)(content));
+		case "odp": return (0, odf_js.encodePackage)((0, documents_js.buildOdpPackage)(content));
+		case "ods": return (0, odf_js.encodePackage)((0, documents_js.buildOdsPackage)(content));
+		case "odg": return (0, odf_js.encodePackage)((0, documents_js.buildOdgPackage)(content));
+		case "markdown": return (0, documents_js.encodeMarkdownText)((0, documents_js.buildMarkdownText)(content));
+	}
+}
+function mergeMetadata(current, overrides) {
+	return {
+		...current,
+		...overrides.title !== void 0 ? { title: overrides.title } : {},
+		...overrides.author !== void 0 ? { author: overrides.author } : {},
+		...overrides.subject !== void 0 ? { subject: overrides.subject } : {},
+		...overrides.keywords !== void 0 ? { keywords: overrides.keywords } : {}
+	};
+}
+function parseKeywords(csv) {
+	return csv.split(",").map((entry) => entry.trim()).filter((entry) => entry.length > 0);
+}
+function classifyWritePath(source, target) {
+	if (source === "pdf" && target === "pdf") return { kind: "pdf" };
+	if (target === "xlsx" || source === "xlsx") return { errorMessage: "'xlsx' is not a supported set-metadata source or target -- documents.js does not re-export a ContentDocument-to-xlsx builder or a readXlsxContent from its own public surface (see that package's own README, Architecture section); convert with 'xlsx-to-ods'/'ods-to-xlsx' first, then set metadata on the ods" };
+	if (target === "odf" || source === "odf") return { errorMessage: "'odf' (a standalone formula document) is not a supported set-metadata source or target -- it has no write path back out at all" };
+	if (!isRebuildFormat(source) || !isRebuildFormat(target)) return { errorMessage: `set-metadata only patches metadata in place; it does not convert format -- source ('${source}') and target ('${target}') must be the same format (or both 'pdf'). Run 'convert'/'from-package' first if you need a different target format.` };
+	if (source !== target) return { errorMessage: `set-metadata only patches metadata in place; it does not convert format -- source ('${source}') and target ('${target}') must be the same format. Run 'convert'/'from-package' first if you need a different target format.` };
+	return {
+		kind: "rebuild",
+		format: source
+	};
+}
+async function runSetMetadata(input, output, options) {
+	const command = "set-metadata";
+	if (output !== void 0 && options.out !== void 0 && output !== options.out) {
+		process.stderr.write(`[${command}] conflicting output destinations: positional '${output}' and --out '${options.out}'\n`);
+		return 2;
+	}
+	const target = resolveTargetFormat(output, options.out, options.to);
+	if ("errorMessage" in target) {
+		process.stderr.write(`[${command}] ${target.errorMessage}\n`);
+		return 2;
+	}
+	const source = inferFormatFromExtension(input);
+	if (source === void 0) {
+		process.stderr.write(`[${command}] cannot infer a source format from '${input}'; rename the file with a recognised extension (${KNOWN_DOCUMENT_FORMATS})\n`);
+		return 2;
+	}
+	const writePath = classifyWritePath(source, target.format);
+	if ("errorMessage" in writePath) {
+		process.stderr.write(`[${command}] ${writePath.errorMessage}\n`);
+		return 2;
+	}
+	const overrides = {
+		title: options.setTitle,
+		author: options.setAuthor,
+		subject: options.setSubject,
+		keywords: options.setKeywords === void 0 ? void 0 : parseKeywords(options.setKeywords)
+	};
+	const resolvedOutput = output ?? options.out ?? (input === "-" ? "-" : resolveDefaultOutputPath(input, target.format));
+	const { signal, getAbortReason } = createRuntimeSignal({ timeoutMs: options.timeout });
+	try {
+		const inputBytes = await readInput(input, { signal });
+		const bytes = writePath.kind === "pdf" ? (() => {
+			const layout = (0, documents_js.readPdf)(new Uint8Array(inputBytes), { signal });
+			const patched = {
+				...layout,
+				metadata: mergeMetadata(layout.metadata, overrides)
+			};
+			return (0, documents_js.writePdf)(patched, { signal });
+		})() : (() => {
+			const content = readContentForFormat(writePath.format, new Uint8Array(inputBytes));
+			const nextContent = {
+				...content,
+				metadata: mergeMetadata(content.metadata, overrides)
+			};
+			return buildBytesForRebuildFormat(writePath.format, nextContent);
+		})();
+		await writeOutput(resolvedOutput, bytes);
+		createDiagnosticReporter({
+			json: options.json,
+			quiet: options.quiet,
+			command
+		}).summarize({
+			output: resolvedOutput,
+			bytes: bytes.byteLength,
+			diagnosticCount: 0
+		});
+		return 0;
+	} catch (error) {
+		process.stderr.write(`${formatError(error, options.verbose)}\n`);
+		return mapErrorToExit(error, getAbortReason());
+	}
+}
+function registerSetMetadataCommand(program) {
+	const command = program.command("set-metadata <input> [output]").description("patch a document's own title/author/subject/keywords, leaving every other field and every other flag as-is").addHelpText("after", [
+		"",
+		"Two write paths: a pdf source/target patches the metadata directly on the parsed PDF (writePdf), with no layout engine",
+		"involved at all -- genuinely lossless for everything else on the page. Every other supported format (docx, pptx, odt,",
+		"odp, ods, odg, markdown) rebuilds a fresh package from that format's own ContentDocument -- for docx specifically,",
+		"this is LOSSY: it drops anything docx-extras covers (comments, footnotes, headers/footers, numbering definitions),",
+		"since buildDocxPackage builds a fresh package from the ContentDocument alone, with no way to carry that data through.",
+		"",
+		"set-metadata does not convert format -- source and target must match. Run convert/from-package first, then",
+		"set-metadata on the result, if you need a different target format."
+	].join("\n"));
+	addOutOption(command);
+	addTimeoutOption(command);
+	addJsonOption(command);
+	addQuietOption(command);
+	addVerboseOption(command);
+	command.option("--to <format>", `target format when it cannot be inferred from the output path (${KNOWN_DOCUMENT_FORMATS})`);
+	command.option("--set-title <text>", "set the title field");
+	command.option("--set-author <text>", "set the author field");
+	command.option("--set-subject <text>", "set the subject field");
+	command.option("--set-keywords <csv>", "set the keywords field, comma-separated (trimmed, empty entries dropped)");
+	command.action(async (input, output, options) => {
+		process.exitCode = await runSetMetadata(input, output, options);
+	});
+}
+//#endregion
 //#region package.json
-var version = "1.3.0";
+var version = "1.5.0";
 //#endregion
 //#region src/program.ts
 function createProgram() {
@@ -1325,6 +1678,10 @@ function createProgram() {
 	registerOdmCommand(program);
 	registerOdbCommands(program);
 	registerPdfInspectCommand(program);
+	registerFontsCommand(program);
+	registerDocxExtrasCommand(program);
+	registerMetadataCommand(program);
+	registerSetMetadataCommand(program);
 	return program;
 }
 //#endregion