Bugfix/issue 3159 (#4009)

Closes #3159

Author: cscheid

news/changelog-1.3.md

@@ -111,6 +111,7 @@
 - remove scaffolding div from conditional content in final output ([#3847](https://github.com/quarto-dev/quarto-cli/issues/3847)).
 - ensure proof titles are appended to paragraph nodes ([#3772](https://github.com/quarto-dev/quarto-cli/issues/3772)).
 - Support parsing markdown in table captions in LaTeX and HTML tables ([#2573](https://github.com/quarto-dev/quarto-cli/issues/2573)).
+- Improve parsing of include shortcodes ([#3159](https://github.com/quarto-dev/quarto-cli/issues/3159)).
 
 ## Pandoc filter changes
 

src/core/lib/parse-shortcode.ts

@@ -30,31 +30,75 @@ export interface Shortcode {
   params: string[];
 }
 
-export function parseShortcode(shortCodeCapture: string): Shortcode {
-  const [name, ...args] = shortCodeCapture.trim().split(" ");
-  const namedParams: Record<string, string> = {};
-  const params: string[] = [];
-  const rawParams = args.map((v) => {
-    const p = v.indexOf("=");
-    let name: string | undefined = undefined;
-    let value: string;
-    if (p === -1) {
-      value = v;
-      params.push(value);
+// shortcode capture BNF:
+// shortcode = shortcode-name shortcode-params?
+// shortcode-name = [a-zA-Z0-9_]+
+// shortcode-params = shortcode-param*
+// shortcode-param = shortcode-param-value | shortcode-param-name "=" shortcode-param-value
+// shortcode-param-name = [a-zA-Z0-9_-]+
+// shortcode-param-value = [^"'\s]+ | '"' [^"]* '"' | "'" [^']* "'"
+function parseShortcodeCapture(capture: string): Shortcode | undefined {
+  // match shortcode name
+  const nameMatch = capture.match(/^[a-zA-Z0-9_]+/);
+  if (!nameMatch) {
+    return;
+  }
+  const params: Shortcode["params"] = [];
+  const namedParams: Shortcode["namedParams"] = {};
+  const rawParams: Shortcode["rawParams"] = [];
+
+  const name = nameMatch[0];
+  let paramStr = capture.slice(name.length).trim();
+  // match params
+  const paramName = "([a-zA-Z0-9_-]+)";
+  const paramValue1 = "([^\"'\\s]+)";
+  const paramValue2 = `"([^"]*)"`;
+  const paramValue3 = `'([^\']*)'`;
+
+  const paramValue = `(?:${paramValue1})|(?:${paramValue2})|(?:${paramValue3})`;
+  const paramNameAndValue =
+    `(?:${paramName}\\s*=\\s*${paramValue1})|(?:${paramName}\\s*=\\s*${paramValue2})|(?:${paramName}\\s*=\\s*${paramValue3})`;
+
+  const paramRe = new RegExp(`(?:${paramValue}|${paramNameAndValue})`);
+
+  while (paramStr.length) {
+    const paramMatch = paramStr.match(paramRe);
+    if (!paramMatch) {
+      throw new Error("invalid shortcode: " + capture);
+    }
+
+    const captures = paramMatch.slice(1).filter((x) => x !== undefined);
+    if (captures.length === 1) {
+      params.push(captures[0]);
+      rawParams.push({
+        value: captures[0],
+      });
+      // value only
+    } else if (captures.length === 2) {
+      namedParams[captures[0]] = captures[1];
+      rawParams.push({
+        name: captures[0],
+        value: captures[1],
+      });
     } else {
-      name = v.slice(0, p);
-      value = v.slice(p + 1);
-      namedParams[name] = value;
+      throw new Error(
+        "Internal Error, could not determine correct shortcode capture for " +
+          capture,
+      );
     }
-    return { name, value };
-  });
 
-  return {
-    name,
-    rawParams,
-    namedParams,
-    params,
-  };
+    paramStr = paramStr.slice(paramMatch[0].length).trim();
+  }
+  return { name, params, namedParams, rawParams };
+}
+
+// TODO this should be handled by a pandoc parser.
+export function parseShortcode(shortCodeCapture: string): Shortcode {
+  const result = parseShortcodeCapture(shortCodeCapture);
+  if (!result) {
+    throw new Error("invalid shortcode: " + shortCodeCapture);
+  }
+  return result;
 }
 
 export function getShortcodeUnnamedParams(shortcode: Shortcode): string[] {

tests/docs/smoke-all/2023/01/18/3159.qmd

@@ -0,0 +1,5 @@
+  ---
+  title: 3159
+  ---
+
+  {{< include "file with spaces.md" >}}

tests/docs/smoke-all/2023/01/18/file with spaces.md

@@ -0,0 +1 @@
+This is some content