ochre-sdk 1.0.40 → 1.0.41

package/dist/parsers/string.mjs

@@ -16,8 +16,10 @@ const RAW_MDX_BLOCK_DELIMITER = "``md``";
 const RAW_MDX_BLOCK_PLACEHOLDER_PREFIX = "\0raw-mdx-block:";
 const RAW_MDX_BLOCK_PLACEHOLDER_SUFFIX = "\0";
 const RICH_LINE_BREAK = "<br />\n";
-const RICH_PARAGRAPH_BREAK = "\n\n";
 const MDX_QUOTED_ATTRIBUTE_ESCAPE_REGEX = /[\n\r"]/;
+const MDX_NEWLINE_RUN_REGEX = /\n{3,}/g;
+const MDX_SIMPLE_INLINE_TAG_REGEX = /<\/?(?:em|strong|u)>/g;
+const MDX_MARKDOWN_LIST_LINE_REGEX = /^\s*(?:[*+-]|\d+[).])\s+/;
 const MDX_RENDER_ELEMENTS = {
 	bold: "strong",
 	italic: "em",
@@ -26,6 +28,47 @@ const MDX_RENDER_ELEMENTS = {
 function hasNewlineWhitespace(value) {
 	return value?.split(" ").includes("newline") === true;
 }
+function hasNoRichTextPayload(item) {
+	return (item.payload == null || item.payload === "") && item.rend == null && item.links == null && item.properties == null && item.annotation == null && item.string == null;
+}
+function isRawMDXBlockDelimiter(item, index) {
+	return item.payload === RAW_MDX_BLOCK_DELIMITER && (index === 0 || hasNewlineWhitespace(item.whitespace)) && item.rend == null && item.links == null && item.properties == null && item.annotation == null && item.string == null;
+}
+function normalizeMDXNewlineRuns(value) {
+	return value.replaceAll(MDX_NEWLINE_RUN_REGEX, "\n\n");
+}
+function isMarkdownListLine(value) {
+	return MDX_MARKDOWN_LIST_LINE_REGEX.test(value);
+}
+function isStandaloneFormattedLine(value) {
+	const trimmedValue = value.trim();
+	if (trimmedValue === "" || isMarkdownListLine(trimmedValue)) return false;
+	const unwrappedValue = trimmedValue.replaceAll(MDX_SIMPLE_INLINE_TAG_REGEX, "").trim();
+	return unwrappedValue !== trimmedValue && unwrappedValue !== "";
+}
+function pushBlankLine(lines) {
+	if (lines.length > 0 && lines.at(-1) !== "") lines.push("");
+}
+function getLastNonBlankLine(lines) {
+	for (let index = lines.length - 1; index >= 0; index -= 1) {
+		const line = lines[index];
+		if (line != null && line.trim() !== "") return line;
+	}
+	return null;
+}
+function normalizeRawMDXBlock(value) {
+	const sourceLines = normalizeMDXNewlineRuns(value).split("\n");
+	const normalizedLines = [];
+	for (const [index, line] of sourceLines.entries()) {
+		const nextLine = sourceLines[index + 1];
+		const shouldStartOwnParagraph = isStandaloneFormattedLine(line);
+		const lastNonBlankLine = getLastNonBlankLine(normalizedLines);
+		if (shouldStartOwnParagraph || isMarkdownListLine(line) && (lastNonBlankLine == null || !isMarkdownListLine(lastNonBlankLine))) pushBlankLine(normalizedLines);
+		normalizedLines.push(line);
+		if (shouldStartOwnParagraph && nextLine != null && nextLine.trim() !== "" && !isMarkdownListLine(nextLine)) pushBlankLine(normalizedLines);
+	}
+	return normalizeMDXNewlineRuns(normalizedLines.join("\n"));
+}
 function isXMLRichTextLink(value) {
 	return typeof value === "object" && value != null;
 }
@@ -74,7 +117,7 @@ function applyMDXRenderElements(contentString, options) {
 function applyNewlineWhitespace(contentString, whitespace, rendering) {
 	if (whitespace == null) return contentString;
 	if (!hasNewlineWhitespace(whitespace)) return contentString;
-	if (contentString === "" && rendering !== "plain") return RICH_PARAGRAPH_BREAK;
+	if (contentString === "") return rendering === "rich" ? RICH_LINE_BREAK : "\n";
 	if (rendering === "rich") return contentString.trim() === "***" ? `${contentString}\n` : `${RICH_LINE_BREAK}${contentString}`;
 	return `\n${contentString}`;
 }
@@ -263,7 +306,7 @@ function parseNestedStringItems(items, contentItem, options) {
 	let result = "";
 	let rawMDXBlockStartIndex = null;
 	for (const [index, item] of items.entries()) {
-		if (item.payload === RAW_MDX_BLOCK_DELIMITER && (index === 0 || hasNewlineWhitespace(item.whitespace)) && item.rend == null && item.links == null && item.properties == null && item.annotation == null && item.string == null) {
+		if (isRawMDXBlockDelimiter(item, index)) {
 			if (rawMDXBlockStartIndex == null) {
 				rawMDXBlockStartIndex = index;
 				continue;
@@ -279,6 +322,7 @@ function parseNestedStringItems(items, contentItem, options) {
 				});
 			}
 			if (rawMDXBlock !== "" && !rawMDXBlock.endsWith("\n") && hasNewlineWhitespace(item.whitespace)) rawMDXBlock += "\n";
+			rawMDXBlock = normalizeRawMDXBlock(rawMDXBlock);
 			if (options.rendering === "rich" && options.rawMDXBlocks != null) {
 				const placeholder = `${RAW_MDX_BLOCK_PLACEHOLDER_PREFIX}${options.rawMDXBlocks.length}${RAW_MDX_BLOCK_PLACEHOLDER_SUFFIX}`;
 				options.rawMDXBlocks.push(rawMDXBlock);
@@ -288,6 +332,11 @@ function parseNestedStringItems(items, contentItem, options) {
 			continue;
 		}
 		if (rawMDXBlockStartIndex != null) continue;
+		const nextItem = items[index + 1];
+		if (options.rendering === "rich" && hasNoRichTextPayload(item) && hasNewlineWhitespace(item.whitespace) && nextItem != null && isRawMDXBlockDelimiter(nextItem, index + 1)) {
+			result += "\n";
+			continue;
+		}
 		const parsedItem = parseXMLStringItem(item, contentItem, options);
 		result += options.rendering === "rich" && result.endsWith("\n") && parsedItem.startsWith(RICH_LINE_BREAK) ? parsedItem.slice(6) : parsedItem;
 	}
@@ -468,6 +517,7 @@ function parseXMLContentItem(contentItem, options) {
 		rawMDXBlocks
 	}));
 	for (const [index, rawMDXBlock] of rawMDXBlocks.entries()) serializedRichText = serializedRichText.replaceAll(`${RAW_MDX_BLOCK_PLACEHOLDER_PREFIX}${index}${RAW_MDX_BLOCK_PLACEHOLDER_SUFFIX}`, rawMDXBlock);
+	serializedRichText = normalizeMDXNewlineRuns(serializedRichText);
 	return {
 		text: parseNestedStringItems(contentItem.string, contentItem, {
 			...options,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ochre-sdk",
-  "version": "1.0.40",
+  "version": "1.0.41",
   "type": "module",
   "license": "MIT",
   "description": "Node.js library for working with OCHRE (Online Cultural and Historical Research Environment) data",