@simple-mark/model 0.1.0 → 0.3.0

package/index.d.ts

@@ -1,11 +1,80 @@
 import { MarkdownExtension } from "@lezer/markdown";
-import { Tree } from "@lezer/common";
+import { SyntaxNode, Tree } from "@lezer/common";
 //#region packages/model/src/parser.d.ts
 /** The Markdown flavor every SimpleMark renderer parses: CommonMark + GFM + $math$. */
 export declare const markdownExtensions: MarkdownExtension[];
 /** A Lezer parser for that flavor, for hosts without CodeMirror (e.g. a canvas renderer). */
 export declare const markdownParser: import("@lezer/markdown").MarkdownParser;
 //#endregion
+//#region packages/model/src/links.d.ts
+interface LinkAt {
+  /** An image (`![alt](src)`) or anything else that links (links, autolinks, bare URLs). */
+  kind: 'link' | 'image';
+  /** The target as written; null when a reference (`[text][label]`) has no definition. */
+  target: string | null;
+  /** Where the link is in the text. */
+  from: number;
+  to: number;
+}
+/**
+ * The link at `pos`: a link, image, autolink, bare URL, or the URL of a definition
+ * (`[label]: url`). `text` must use one character per line break, so offsets match the tree.
+ */
+export declare function linkAt(tree: Tree, text: string, pos: number): LinkAt | null;
+/**
+ * What a link target is:
+ * - `web`: http(s), mailto, or a bare `www.` address (given back as a full URL)
+ * - `anchor`: `#heading` in the same document
+ * - `file`: a relative or absolute path (an image, another Markdown file, …)
+ * - `unsafe`: any other scheme (javascript:, data:, file:, …), which should not be opened
+ */
+export declare function linkKind(target: string): {
+  kind: 'web' | 'anchor' | 'file' | 'unsafe';
+  url: string;
+};
+/**
+ * The anchor of a heading, as GitHub makes it: lower case, punctuation removed, spaces as
+ * hyphens ("Next Steps!" → "next-steps", "다음 단계" → "다음-단계").
+ */
+export declare function headingSlug(heading: string): string;
+/**
+ * The headings of a document with their anchors; repeated anchors get -1, -2, … like GitHub.
+ * `from` is where each heading starts.
+ */
+export declare function headingAnchors(tree: Tree, text: string): {
+  anchor: string;
+  from: number;
+}[];
+//#endregion
+//#region packages/model/src/previews.d.ts
+/** An image as written in the document. */
+interface ImageSource {
+  alt: string;
+  /** The path as written, or null when a reference has no definition. */
+  src: string | null;
+  /** For messages: the path, or the reference label (`[logo]`) that has no definition. */
+  raw: string;
+}
+/** A preview drawn below the line that ends at `at` (an offset in the text). */
+type PreviewBlock = {
+  at: number;
+  images: ImageSource[];
+} | {
+  at: number;
+  tex: string;
+};
+/**
+ * The TeX of a math node (`InlineMath` or `BlockMath`), without delimiters or quote markers;
+ * null while it is not closed. `slice` reads the document.
+ */
+export declare function mathSource(node: SyntaxNode, slice: (from: number, to: number) => string): string | null;
+/**
+ * The previews of a document: the images of each line (one entry per line, in order) and
+ * closed, non-empty `$$` blocks. `text` must use one character per line break ("\n"), so
+ * offsets match the tree.
+ */
+export declare function previewBlocks(tree: Tree, text: string): PreviewBlock[];
+//#endregion
 //#region packages/model/src/styles.d.ts
 /** Styles of whole lines, in precedence order (later wins when two set the same property). */
 export declare const BLOCK_STYLES: readonly ['quote', 'codeblock', 'mathblock', 'table', 'tableHeader', 'tableRowAlt', 'hr', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'];
@@ -113,4 +182,4 @@ export declare function resolveLineLook(blocks: Iterable<BlockStyle>, theme?: Th
 /** The look of text with these block and inline styles (any order). */
 export declare function resolveLook(blocks: Iterable<BlockStyle>, inline: Iterable<InlineStyle>, theme?: Theme): Look;
 //#endregion
-export type { BlockRange, BlockStyle, ColumnAlign, InlineStyle, LineLook, Look, MarkdownStyles, RenderedTable, RowCell, Span, Table, Theme };
+export type { BlockRange, BlockStyle, ColumnAlign, ImageSource, InlineStyle, LineLook, LinkAt, Look, MarkdownStyles, PreviewBlock, RenderedTable, RowCell, Span, Table, Theme };

package/index.js

@@ -103,6 +103,215 @@ const markdownExtensions = [GFM, {
 /** A Lezer parser for that flavor, for hosts without CodeMirror (e.g. a canvas renderer). */
 const markdownParser = parser.configure(markdownExtensions);
 //#endregion
+//#region packages/model/src/previews.ts
+/** A reference label (`[Logo]`) as definitions are looked up: case and spacing ignored. */
+const normalizeLabel = (label) => label.slice(1, -1).trim().replace(/\s+/g, " ").toLowerCase();
+/** A link destination as written, without the optional angle brackets (`<a b.png>`). */
+const destination = (text, url) => text.slice(url.from, url.to).replace(/^<(.*)>$/, "$1");
+/** Link reference definitions (`[label]: url`) of the document, by normalized label. */
+function linkDefinitions(tree, text) {
+	const definitions = /* @__PURE__ */ new Map();
+	for (let node = tree.topNode.firstChild; node; node = node.nextSibling) {
+		if (node.name !== "LinkReference") continue;
+		const label = node.getChild("LinkLabel");
+		const url = node.getChild("URL");
+		if (label && url) {
+			const key = normalizeLabel(text.slice(label.from, label.to));
+			if (!definitions.has(key)) definitions.set(key, destination(text, url));
+		}
+	}
+	return definitions;
+}
+function imageSource(node, text, definitions) {
+	const marks = node.getChildren("LinkMark");
+	const alt = marks.length >= 2 ? text.slice(marks[0].to, marks[1].from) : "";
+	const url = node.getChild("URL");
+	if (url) {
+		const src = destination(text, url);
+		return {
+			alt,
+			src,
+			raw: src
+		};
+	}
+	const label = node.getChild("LinkLabel");
+	const labelText = label && label.to - label.from > 2 ? text.slice(label.from, label.to) : `[${alt}]`;
+	const src = definitions.get(normalizeLabel(labelText));
+	return src === void 0 ? {
+		alt,
+		src: null,
+		raw: labelText
+	} : {
+		alt,
+		src,
+		raw: src
+	};
+}
+/**
+* The TeX of a math node (`InlineMath` or `BlockMath`), without delimiters or quote markers;
+* null while it is not closed. `slice` reads the document.
+*/
+function mathSource(node, slice) {
+	const marks = node.getChildren("MathMark");
+	if (marks.length < 2) return null;
+	if (node.name === "InlineMath") return slice(marks[0].to, marks[1].from);
+	return node.getChildren("MathText").map((part) => slice(part.from, part.to)).join("\n");
+}
+/**
+* The previews of a document: the images of each line (one entry per line, in order) and
+* closed, non-empty `$$` blocks. `text` must use one character per line break ("\n"), so
+* offsets match the tree.
+*/
+function previewBlocks(tree, text) {
+	const definitions = linkDefinitions(tree, text);
+	const lineEnd = (pos) => {
+		const end = text.indexOf("\n", pos);
+		return end === -1 ? text.length : end;
+	};
+	const imagesByLine = /* @__PURE__ */ new Map();
+	const blocks = [];
+	tree.iterate({ enter(node) {
+		if (node.name === "Image") {
+			const at = lineEnd(node.to);
+			let images = imagesByLine.get(at);
+			if (!images) {
+				imagesByLine.set(at, images = []);
+				blocks.push({
+					at,
+					images
+				});
+			}
+			images.push(imageSource(node.node, text, definitions));
+			return false;
+		}
+		if (node.name === "BlockMath") {
+			const tex = mathSource(node.node, (from, to) => text.slice(from, to));
+			if (tex !== null && tex.trim()) blocks.push({
+				at: lineEnd(node.to),
+				tex
+			});
+			return false;
+		}
+	} });
+	return blocks;
+}
+//#endregion
+//#region packages/model/src/links.ts
+/** Where a URL node is a link's target rather than a bare URL of its own. */
+const LINK_PARENTS = /* @__PURE__ */ new Set([
+	"Link",
+	"Image",
+	"Autolink",
+	"LinkReference"
+]);
+function targetOf(node, text, definitions) {
+	const url = node.name === "URL" ? node : node.getChild("URL");
+	if (url) return destination(text, url);
+	const marks = node.getChildren("LinkMark");
+	const content = marks.length >= 2 ? text.slice(marks[0].to, marks[1].from) : "";
+	const label = node.getChild("LinkLabel");
+	const labelText = label && label.to - label.from > 2 ? text.slice(label.from, label.to) : `[${content}]`;
+	return definitions().get(normalizeLabel(labelText)) ?? null;
+}
+/**
+* The link at `pos`: a link, image, autolink, bare URL, or the URL of a definition
+* (`[label]: url`). `text` must use one character per line break, so offsets match the tree.
+*/
+function linkAt(tree, text, pos) {
+	let definitions;
+	const lazyDefinitions = () => definitions ??= linkDefinitions(tree, text);
+	for (const side of [1, -1]) for (let node = tree.resolveInner(pos, side); node; node = node.parent) {
+		if (node.from > pos || node.to < pos) continue;
+		if (node.name === "Link" || node.name === "Image" || node.name === "Autolink") return {
+			kind: node.name === "Image" ? "image" : "link",
+			target: targetOf(node, text, lazyDefinitions),
+			from: node.from,
+			to: node.to
+		};
+		if (node.name === "URL" && !LINK_PARENTS.has(node.parent?.name ?? "")) return {
+			kind: "link",
+			target: destination(text, node),
+			from: node.from,
+			to: node.to
+		};
+		if (node.name === "URL" && node.parent?.name === "LinkReference") return {
+			kind: "link",
+			target: destination(text, node),
+			from: node.from,
+			to: node.to
+		};
+	}
+	return null;
+}
+/**
+* What a link target is:
+* - `web`: http(s), mailto, or a bare `www.` address (given back as a full URL)
+* - `anchor`: `#heading` in the same document
+* - `file`: a relative or absolute path (an image, another Markdown file, …)
+* - `unsafe`: any other scheme (javascript:, data:, file:, …), which should not be opened
+*/
+function linkKind(target) {
+	const url = target.trim();
+	if (/^(https?:\/\/|mailto:)/i.test(url)) return {
+		kind: "web",
+		url
+	};
+	if (/^www\./i.test(url)) return {
+		kind: "web",
+		url: `https://${url}`
+	};
+	if (/^[^\s/:@]+@[^\s/:@]+\.[^\s/:@]+$/.test(url)) return {
+		kind: "web",
+		url: `mailto:${url}`
+	};
+	if (url.startsWith("#")) return {
+		kind: "anchor",
+		url
+	};
+	if (/^[a-z][a-z0-9+.-]+:/i.test(url)) return {
+		kind: "unsafe",
+		url
+	};
+	return {
+		kind: "file",
+		url
+	};
+}
+/**
+* The anchor of a heading, as GitHub makes it: lower case, punctuation removed, spaces as
+* hyphens ("Next Steps!" → "next-steps", "다음 단계" → "다음-단계").
+*/
+function headingSlug(heading) {
+	return heading.trim().toLowerCase().replace(/[^\p{L}\p{N}\p{M}\s_-]/gu, "").replace(/\s/g, "-");
+}
+const HEADINGS = /^(ATXHeading[1-6]|SetextHeading[12])$/;
+/**
+* The headings of a document with their anchors; repeated anchors get -1, -2, … like GitHub.
+* `from` is where each heading starts.
+*/
+function headingAnchors(tree, text) {
+	const out = [];
+	const seen = /* @__PURE__ */ new Map();
+	for (let node = tree.topNode.firstChild; node; node = node.nextSibling) {
+		if (!HEADINGS.test(node.name)) continue;
+		let title = "";
+		let pos = node.from;
+		for (const mark of node.getChildren("HeaderMark")) {
+			title += text.slice(pos, mark.from);
+			pos = mark.to;
+		}
+		title += text.slice(pos, node.to);
+		const base = headingSlug(title.split("\n")[0]);
+		const count = seen.get(base) ?? 0;
+		seen.set(base, count + 1);
+		out.push({
+			anchor: count ? `${base}-${count}` : base,
+			from: node.from
+		});
+	}
+	return out;
+}
+//#endregion
 //#region packages/model/src/styles.ts
 /** Styles of whole lines, in precedence order (later wins when two set the same property). */
 const BLOCK_STYLES = [
@@ -505,4 +714,4 @@ function resolveLook(blocks, inline, theme = defaultTheme) {
 	return look;
 }
 //#endregion
-export { BLOCK_STYLES, INLINE_STYLES, cellAt, charWidth, defaultTheme, displayWidth, flattenSpans, markdownExtensions, markdownParser, markdownStyles, parseTable, renderTable, resolveLineLook, resolveLook, rowLine, splitRow };
+export { BLOCK_STYLES, INLINE_STYLES, cellAt, charWidth, defaultTheme, displayWidth, flattenSpans, headingAnchors, headingSlug, linkAt, linkKind, markdownExtensions, markdownParser, markdownStyles, mathSource, parseTable, previewBlocks, renderTable, resolveLineLook, resolveLook, rowLine, splitRow };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@simple-mark/model",
-  "version": "0.1.0",
+  "version": "0.3.0",
   "description": "Markdown parsing and styling model shared by every renderer. Pure TypeScript, no DOM.",
   "license": "MIT",
   "type": "module",