convert - handle cells with both yaml and implicit titles

Author: cscheid

src/core/jupyter/jupyter-fixups.ts

@@ -12,7 +12,7 @@ import { Metadata } from "../../config/types.ts";
 import { lines } from "../lib/text.ts";
 import { markdownWithExtractedHeading } from "../pandoc/pandoc-partition.ts";
 import { partitionYamlFrontMatter, readYamlFromMarkdown } from "../yaml.ts";
-import { JupyterNotebook, JupyterOutput } from "./types.ts";
+import { JupyterCell, JupyterNotebook, JupyterOutput } from "./types.ts";
 import {
   jupyterCellSrcAsLines,
   jupyterCellSrcAsStr,
@@ -149,6 +149,34 @@ export function fixupFrontMatter(nb: JupyterNotebook): JupyterNotebook {
     return lns.map((line) => line.endsWith("\n") ? line : `${line}\n`);
   };
 
+  // https://github.com/quarto-dev/quarto-cli/issues/12440
+  // first, we need to find cells that have front matter _and_ markdown content,
+  // and split them into two cells.
+  const newCells: JupyterCell[] = [];
+  nb.cells.forEach((cell) => {
+    if (cell.cell_type === "raw" || cell.cell_type === "markdown") {
+      const partitioned = partitionYamlFrontMatter(jupyterCellSrcAsStr(cell)) ||
+        undefined;
+      if (partitioned?.yaml.trim()) {
+        newCells.push({
+          cell_type: "raw",
+          source: nbLines(partitioned.yaml.split("\n")),
+          metadata: {},
+        });
+      }
+      if (partitioned?.markdown.trim()) {
+        newCells.push({
+          cell_type: "markdown",
+          source: nbLines(partitioned.markdown.split("\n")),
+          metadata: {},
+        });
+      }
+    } else {
+      newCells.push(cell);
+    }
+  });
+  nb.cells = newCells;
+
   // look for the first raw block that has a yaml object
   let partitioned: { yaml: string; markdown: string } | undefined;
   const frontMatterCellIndex = nb.cells.findIndex((cell) => {

src/core/pandoc/pandoc-partition.ts

@@ -84,7 +84,7 @@ export function markdownWithExtractedHeading(markdown: string) {
         const parsedHeading = parsePandocTitle(line);
         headingText = parsedHeading.heading;
         headingAttr = parsedHeading.attr;
-        contentBeforeHeading = mdLines.length !== 0;
+        contentBeforeHeading = (mdLines.join("").trim()).length !== 0;
       } else if (line.match(/^=+\s*$/) || line.match(/^-+\s*$/)) {
         const prevLine = mdLines[mdLines.length - 1];
         if (prevLine) {