@stll/folio-core 0.32.0 → 0.32.1

package/dist/markdown/fromMarkdown.js

@@ -1,227 +1,24 @@
 import { createEmptyDocument } from "../utils/createDocument.js";
-import { sanitizeExternalUrl } from "../utils/urlSecurity.js";
-import { marked } from "marked";
+import { compileMarkdownToContent } from "@stll/docx-core";
 //#region src/markdown/fromMarkdown.ts
 /**
 * Markdown → DOCX-document import — the inverse of {@link toMarkdown} and the
-* second half of the skills bridge. Parses the GFM subset skill bodies use
-* (headings, paragraphs, bold/italic/strike, inline code, bullet + ordered
-* lists incl. nesting, pipe tables, blockquotes, links) into the docx
-* `Document` model so a skill's markdown can be edited in the Folio editor and
-* re-exported with {@link toMarkdown} without drift.
+* second half of the skills bridge. The parsing lives in `@stll/docx-core`
+* (`compileMarkdownToContent`), the same GFM reader the legal-source compiler
+* uses; this wrapper only places the blocks into an editor-ready `Document`.
 *
 * Round-trip notes:
-* - Lists are emitted as real list paragraphs (`listRendering`), so the editor
-*   shows a marker and {@link toMarkdown} re-derives `- ` / `1. ` rather than
-*   leaking a literal bullet glyph into the text.
-* - Inline code uses `Courier New` — a whitelisted monospace family that
-*   {@link toMarkdown} infers back to a backtick span (Folio renders it via its
-*   bundled Cousine substitute).
+* - Lists arrive as real list paragraphs (`listRendering` plus `numPr`) with a
+*   matching `document.package.numbering`, so the editor shows a marker and
+*   {@link toMarkdown} re-derives `- ` / `1. ` rather than leaking a literal
+*   bullet glyph. Merging this content onto another document that has its own
+*   numbering (e.g. a styled preset) needs `mergeDocumentContent` to renumber
+*   the two numbering namespaces apart.
 * - Markdown carries no page geometry, so the section is flattened to a
 *   continuous, header/footer-free band (a skill body is a document, not a
 *   Word page). Headers/footers live outside `document.content` and are never
 *   produced here.
-* - Every markdown list also gets a matching `w:abstractNum`/`w:num` pair in
-*   `document.package.numbering` (see {@link buildNumbering}), so the result
-*   is self-consistent and `createDocx` never has to fail with a missing
-*   numbering definition. Merging this content onto another document that
-*   has its own numbering (e.g. a styled preset) needs `mergeDocumentContent`
-*   to renumber the two numbering namespaces apart — appending
-*   `document.package.document.content` directly can collide.
 */
-const MONO_FONT = {
-	ascii: "Courier New",
-	hAnsi: "Courier New"
-};
-const isTokenType = (token, type) => token.type === type;
-const textRun = (text, fmt = {}) => {
-	const formatting = {
-		...fmt.bold ? { bold: true } : {},
-		...fmt.italic ? { italic: true } : {},
-		...fmt.strike ? { strike: true } : {},
-		...fmt.mono ? { fontFamily: MONO_FONT } : {}
-	};
-	const segments = text.split("\n");
-	const content = [];
-	for (const [index, segment] of segments.entries()) {
-		if (index > 0) content.push({ type: "break" });
-		if (segment.length > 0) content.push({
-			type: "text",
-			text: segment
-		});
-	}
-	if (content.length === 0) content.push({
-		type: "text",
-		text: ""
-	});
-	return {
-		type: "run",
-		formatting,
-		content
-	};
-};
-const sanitizeMarkdownHref = (rawHref) => {
-	const trimmed = rawHref.trim();
-	if (!trimmed) return;
-	if (trimmed.startsWith("#")) {
-		const anchor = trimmed.slice(1);
-		if (!anchor || hasUnsafeAnchorCharacter(anchor)) return;
-		return `#${anchor}`;
-	}
-	return sanitizeExternalUrl(trimmed);
-};
-const hasUnsafeAnchorCharacter = (anchor) => {
-	for (const char of anchor) {
-		const codePoint = char.codePointAt(0) ?? 0;
-		if (codePoint <= 32 || codePoint === 127 || char.trim() === "") return true;
-	}
-	return false;
-};
-const inlineToRuns = (tokens, fallback, base) => {
-	if (!tokens || tokens.length === 0) return [textRun(fallback, base)];
-	const runs = [];
-	for (const token of tokens) if (isTokenType(token, "strong")) runs.push(...inlineToRuns(token.tokens, token.text, {
-		...base,
-		bold: true
-	}));
-	else if (isTokenType(token, "em")) runs.push(...inlineToRuns(token.tokens, token.text, {
-		...base,
-		italic: true
-	}));
-	else if (isTokenType(token, "del")) runs.push(...inlineToRuns(token.tokens, token.text, {
-		...base,
-		strike: true
-	}));
-	else if (isTokenType(token, "codespan")) runs.push(textRun(token.text, {
-		...base,
-		mono: true
-	}));
-	else if (isTokenType(token, "link")) {
-		const children = inlineToRuns(token.tokens, token.text, base).filter((child) => child.type === "run");
-		const href = sanitizeMarkdownHref(token.href);
-		const linkChildren = children.length > 0 ? children : [textRun(token.text, base)];
-		if (!href) {
-			runs.push(...linkChildren);
-			continue;
-		}
-		const anchor = href.startsWith("#") ? href.slice(1) : void 0;
-		runs.push({
-			type: "hyperlink",
-			href,
-			...anchor ? { anchor } : {},
-			children: linkChildren
-		});
-	} else if (isTokenType(token, "paragraph")) runs.push(...inlineToRuns(token.tokens, token.text, base));
-	else if (token.type === "br") runs.push({
-		type: "run",
-		content: [{ type: "break" }]
-	});
-	else if (token.type === "space") {
-		if (runs.length > 0 && token.raw.includes("\n")) runs.push(textRun("\n", base));
-	} else if (isTokenType(token, "text")) {
-		const nested = token.tokens;
-		if (nested && nested.length > 0) runs.push(...inlineToRuns(nested, token.text, base));
-		else runs.push(textRun(token.text, base));
-	} else if ("text" in token && typeof token.text === "string") runs.push(textRun(token.text, base));
-	return runs.length > 0 ? runs : [textRun(fallback, base)];
-};
-const para = (runs, styleId) => ({
-	type: "paragraph",
-	formatting: styleId ? { styleId } : {},
-	content: runs.length > 0 ? runs : [textRun("")]
-});
-const listPara = (runs, rendering) => ({
-	type: "paragraph",
-	formatting: { numPr: {
-		numId: rendering.numId,
-		ilvl: rendering.level
-	} },
-	listRendering: rendering,
-	content: runs.length > 0 ? runs : [textRun("")]
-});
-const cellOf = (cell) => ({
-	type: "tableCell",
-	content: [para(inlineToRuns(cell.tokens, cell.text, {}))]
-});
-const tableFromToken = (token) => ({
-	type: "table",
-	rows: [{
-		type: "tableRow",
-		cells: token.header.map((c) => cellOf(c))
-	}, ...token.rows.map((row) => ({
-		type: "tableRow",
-		cells: row.map((c) => cellOf(c))
-	}))]
-});
-const LIST_INDENT_STEP_TWIPS = 720;
-const buildListLevel = (ilvl, isBullet, start) => ({
-	ilvl,
-	...!isBullet && { start },
-	numFmt: isBullet ? "bullet" : "decimal",
-	lvlText: isBullet ? "•" : `%${ilvl + 1}.`,
-	suffix: "tab",
-	pPr: {
-		indentLeft: LIST_INDENT_STEP_TWIPS * (ilvl + 1),
-		indentFirstLine: -360,
-		hangingIndent: true
-	}
-});
-const listBlocks = (list, level, numId, levels) => {
-	const out = [];
-	const start = Number(list.start) || 1;
-	const decimalLevels = Array.from({ length: level + 1 }, () => "decimal");
-	if (!levels.has(level)) levels.set(level, buildListLevel(level, !list.ordered, start));
-	for (const item of list.items) {
-		const rendering = list.ordered ? {
-			marker: `%${level + 1}.`,
-			level,
-			numId,
-			isBullet: false,
-			numFmt: "decimal",
-			levelNumFmts: decimalLevels,
-			...start !== 1 && { startOverride: start }
-		} : {
-			marker: "•",
-			level,
-			numId,
-			isBullet: true
-		};
-		const inlineTokens = [];
-		const nestedLists = [];
-		for (const child of item.tokens) if (isTokenType(child, "list")) nestedLists.push(child);
-		else inlineTokens.push(child);
-		out.push(listPara(inlineToRuns(inlineTokens, item.text, {}), rendering));
-		for (const nested of nestedLists) out.push(...listBlocks(nested, level + 1, numId, levels));
-	}
-	return out;
-};
-const blocksFromTokens = (tokens, numIds) => {
-	const blocks = [];
-	for (const token of tokens ?? []) if (isTokenType(token, "heading")) {
-		const level = Math.min(Math.max(token.depth, 1), 4);
-		blocks.push(para(inlineToRuns(token.tokens, token.text, {}), `Heading${level}`));
-	} else if (isTokenType(token, "paragraph")) blocks.push(para(inlineToRuns(token.tokens, token.text, {})));
-	else if (isTokenType(token, "list")) {
-		const numId = numIds.next++;
-		const levels = /* @__PURE__ */ new Map();
-		numIds.levels.set(numId, levels);
-		blocks.push(...listBlocks(token, 0, numId, levels));
-	} else if (isTokenType(token, "table")) blocks.push(tableFromToken(token));
-	else if (isTokenType(token, "code")) for (const line of token.text.split("\n")) blocks.push(para([textRun(line.length > 0 ? line : " ", { mono: true })]));
-	else if (isTokenType(token, "blockquote")) for (const inner of blocksFromTokens(token.tokens, numIds)) {
-		const styled = inner.type === "paragraph" ? {
-			...inner,
-			formatting: {
-				...inner.formatting,
-				styleId: "Quote"
-			}
-		} : inner;
-		blocks.push(styled);
-	}
-	else if (token.type === "hr") blocks.push(para([textRun("———")]));
-	else if (token.type !== "space" && "text" in token && typeof token.text === "string" && token.text.trim().length > 0) blocks.push(para([textRun(token.text)]));
-	return blocks;
-};
 const applyMarkdownPageGeometry = (document) => {
 	const section = document.package.document.finalSectionProperties;
 	if (!section) return;
@@ -232,26 +29,6 @@ const applyMarkdownPageGeometry = (document) => {
 	section.headerDistance = 0;
 	section.footerDistance = 0;
 };
-const buildNumbering = (numIdLevels) => {
-	const abstractNums = [];
-	const nums = [];
-	for (const [numId, levels] of numIdLevels) {
-		const sortedLevels = [...levels.entries()].sort(([a], [b]) => a - b).map(([, lvl]) => lvl);
-		abstractNums.push({
-			abstractNumId: numId,
-			multiLevelType: sortedLevels.length > 1 ? "multilevel" : "singleLevel",
-			levels: sortedLevels
-		});
-		nums.push({
-			numId,
-			abstractNumId: numId
-		});
-	}
-	return {
-		abstractNums,
-		nums
-	};
-};
 /**
 * Convert a markdown string to a parsed `Document`. Synchronous. The result is
 * ready to hand to the editor (`<DocxEditor document={…} />`) and to re-export
@@ -259,13 +36,9 @@ const buildNumbering = (numIdLevels) => {
 */
 function fromMarkdown(markdown) {
 	const document = createEmptyDocument();
-	const numIds = {
-		next: 1,
-		levels: /* @__PURE__ */ new Map()
-	};
-	const blocks = blocksFromTokens(marked.lexer(markdown), numIds);
-	if (blocks.length > 0) document.package.document.content = blocks;
-	if (numIds.levels.size > 0) document.package.numbering = buildNumbering(numIds.levels);
+	const { content, numbering } = compileMarkdownToContent(markdown);
+	if (content.length > 0) document.package.document.content = content;
+	if (numbering) document.package.numbering = numbering;
 	applyMarkdownPageGeometry(document);
 	return document;
 }

package/dist/utils/urlSecurity.d.ts

@@ -1,5 +1,5 @@
+import { sanitizeExternalUrl } from "@stll/docx-core";
 //#region src/utils/urlSecurity.d.ts
-declare function sanitizeExternalUrl(rawUrl: string | undefined): string | undefined;
 declare function normalizeUserUrl(rawUrl: string): string;
 declare function isAllowedUserUrl(rawUrl: string): boolean;
 declare function sanitizeLinkTarget(target: string | undefined): string;

package/dist/utils/urlSecurity.js

@@ -1,29 +1,11 @@
+import { sanitizeExternalUrl } from "@stll/docx-core";
 //#region src/utils/urlSecurity.ts
-const ALLOWED_URL_PROTOCOLS = /* @__PURE__ */ new Set([
-	"http:",
-	"https:",
-	"mailto:",
-	"tel:"
-]);
 const ALLOWED_TARGETS = /* @__PURE__ */ new Set([
 	"_blank",
 	"_self",
 	"_parent",
 	"_top"
 ]);
-function sanitizeExternalUrl(rawUrl) {
-	if (!rawUrl) return;
-	const trimmed = rawUrl.trim();
-	if (!trimmed) return;
-	try {
-		const parsed = new URL(trimmed);
-		if (!ALLOWED_URL_PROTOCOLS.has(parsed.protocol)) return;
-		if ((parsed.protocol === "mailto:" || parsed.protocol === "tel:") && parsed.pathname.trim() === "") return;
-		return parsed.href;
-	} catch {
-		return;
-	}
-}
 function normalizeUserUrl(rawUrl) {
 	const trimmed = rawUrl.trim();
 	if (!trimmed) return "";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@stll/folio-core",
-  "version": "0.32.0",
+  "version": "0.32.1",
   "description": "Headless, framework-neutral core of folio: the OOXML (.docx) parser, document model, ProseMirror integration, and page-layout engine. No React.",
   "keywords": [
     "document-model",
@@ -113,7 +113,7 @@
     "perf": "bun scripts/profile-editor.ts"
   },
   "dependencies": {
-    "@stll/docx-core": "^0.17.3",
+    "@stll/docx-core": "^0.18.0",
     "@stll/docx-utils": "^0.1.0",
     "@stll/template-conditions": "^0.1.0",
     "better-result": "3.0.1",
@@ -122,7 +122,6 @@
     "fast-xml-parser": "^5.10.1",
     "hyphen": "1.14.1",
     "jszip": "3.10.1",
-    "marked": "^18.0.5",
     "prosemirror-commands": "^1.7.1",
     "prosemirror-dropcursor": "^1.8.2",
     "prosemirror-gapcursor": "^1.4.1",