feat(document-schema.js): record what a node's content was in the source

packages/ooxml.js/src/typed/pptx/chart.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it } from "vitest";
 import type { XmlElement } from "../../model/node";
-import { readChartResidue } from "./chart";
+import { el, txt } from "../../xml/fragment";
+import { readChartResidue, readChartTable } from "./chart";
 
 function chartRoot(): XmlElement {
   return {
@@ -29,3 +30,58 @@ describe("readChartResidue", () => {
     expect(second.xml).toBe(first.xml);
   });
 });
+
+// One bar chart with a single series, its category labels and values in the caches PowerPoint writes
+// beside the data reference.
+function barChartRoot(): XmlElement {
+  const cachedPoint = (idx: string, value: string) =>
+    el("c:pt", { idx }, [el("c:v", {}, [txt(value)])]);
+  return el("c:chartSpace", {}, [
+    el("c:chart", {}, [
+      el("c:plotArea", {}, [
+        el("c:barChart", {}, [
+          el("c:ser", {}, [
+            el("c:tx", {}, [
+              el("c:strRef", {}, [
+                el("c:strCache", {}, [cachedPoint("0", "FY26")]),
+              ]),
+            ]),
+            el("c:cat", {}, [
+              el("c:strRef", {}, [
+                el("c:strCache", {}, [
+                  cachedPoint("0", "EMEA"),
+                  cachedPoint("1", "APAC"),
+                ]),
+              ]),
+            ]),
+            el("c:val", {}, [
+              el("c:numRef", {}, [
+                el("c:numCache", {}, [
+                  cachedPoint("0", "42"),
+                  cachedPoint("1", "51"),
+                ]),
+              ]),
+            ]),
+          ]),
+        ]),
+      ]),
+    ]),
+  ]);
+}
+
+describe("readChartTable", () => {
+  it('marks the table it produces as origin "chart"', () => {
+    // A ContentTable is a native table, a chart's cached data, or a spreadsheet range, and a consumer
+    // holding one cannot otherwise tell which. It matters: a chart's cached numbers are exact and
+    // quotable, where a vision reading of the same chart would be approximate -- so the two have to be
+    // distinguishable by something other than a consumer's guess.
+    const table = readChartTable(barChartRoot(), {
+      xPt: 0,
+      yPt: 0,
+      widthPt: 400,
+      heightPt: 300,
+    });
+
+    expect(table?.origin).toBe("chart");
+  });
+});

packages/ooxml.js/src/typed/pptx/chart.ts

@@ -130,6 +130,7 @@ export function readChartTable(
   const columnWidthPt = frame.widthPt / (series.length + 1);
   return {
     kind: "table",
+    origin: "chart",
     rows,
     columnWidthsPt: Array.from(
       { length: series.length + 1 },

packages/ooxml.js/src/typed/pptx/diagram.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it } from "vitest";
 import type { XmlElement } from "../../model/node";
-import { readDiagramResidue } from "./diagram";
+import { el, txt } from "../../xml/fragment";
+import { readDiagramResidue, readDiagramText } from "./diagram";
 
 function part(tag: string): XmlElement {
   return { type: "element", tag, attributes: [], children: [] };
@@ -31,3 +32,36 @@ describe("readDiagramResidue", () => {
     expect(readDiagramResidue(undefined, undefined, undefined)).toBeUndefined();
   });
 });
+
+// A two-node data model: a doc root, two content nodes, and the parOf connections making it a tree.
+function dataModelRoot(): XmlElement {
+  const point = (id: string, text: string, type?: string) =>
+    el("dgm:pt", type === undefined ? { modelId: id } : { modelId: id, type }, [
+      el("dgm:t", {}, [
+        el("a:p", {}, [el("a:r", {}, [el("a:t", {}, [txt(text)])])]),
+      ]),
+    ]);
+  const cxn = (srcId: string, destId: string, srcOrd: string) =>
+    el("dgm:cxn", { srcId, destId, type: "parOf", srcOrd });
+  return el("dgm:dataModel", {}, [
+    el("dgm:ptLst", {}, [
+      point("root", "", "doc"),
+      point("a", "Ad hoc"),
+      point("b", "Repeatable"),
+    ]),
+    el("dgm:cxnLst", {}, [cxn("root", "a", "0"), cxn("root", "b", "1")]),
+  ]);
+}
+
+describe("readDiagramText", () => {
+  it('marks every node paragraph as origin "diagram"', () => {
+    // SmartArt node text reaches the model as ordinary paragraphs, so nothing otherwise distinguishes a
+    // process flow's step labels from body prose -- and they are not the same thing: the relationships
+    // between the nodes (the arrows, the hierarchy) are not recovered, which a consumer reading them as
+    // prose needs to know.
+    const paragraphs = readDiagramText(dataModelRoot());
+
+    expect(paragraphs.length).toBeGreaterThan(0);
+    expect(paragraphs.every((p) => p.origin === "diagram")).toBe(true);
+  });
+});

packages/ooxml.js/src/typed/pptx/diagram.ts

@@ -29,7 +29,7 @@ function diagramTextParagraphs(
         runs.push({ text: "\n" });
       }
     }
-    return { kind: "paragraph", runs };
+    return { kind: "paragraph", origin: "diagram", runs };
   });
 }