fix(pdf-codec): hold the text state parameters in the graphics state

packages/pdf-codec/src/interpret.test.ts

@@ -201,6 +201,161 @@ describe("interpretContentStream: text", () => {
   });
 });
 
+// ISO 32000-1 Table 52 lists the text state parameters (font and size, character spacing, word spacing, horizontal scaling, leading, rise) among the device-independent graphics state parameters, so `q` saves them and `Q` restores them exactly as it does the CTM or the fill colour. Only the text matrix and text line matrix are excluded -- those are text object state (9.4.1), reset by BT and untouched by q/Q.
+describe("interpretContentStream: text state is graphics state", () => {
+  it("restores the font a Q's matching q selected, so a Tf inside the pair does not leak past it", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes(
+        "BT /F1 12 Tf (First) Tj ET q /F2 24 Tf Q BT 0 -20 Td (Second) Tj ET",
+      ),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    expect(items).toHaveLength(2);
+    const [first, second] = items;
+    if (first?.kind !== "text" || second?.kind !== "text") {
+      throw new Error("expected two text items");
+    }
+    expect(first.fontResourceName).toBe("F1");
+    expect(first.sizePt).toBe(12);
+    expect(second.fontResourceName).toBe("F1");
+    expect(second.sizePt).toBe(12);
+  });
+
+  it("restores character spacing across a q/Q pair", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("/F1 10 Tf 0 Tc q 100 Tc Q BT 0 0 Td (AB) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    // Two glyphs at width 5 each (500/1000 * 10) with the restored Tc=0: 10, not the 210 a leaked Tc=100 would give.
+    expect(item.endMatrix).toEqual([10, 0, 0, 10, 10, 0]);
+  });
+
+  it("restores word spacing across a q/Q pair", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("/F1 10 Tf q 40 Tw Q BT 0 0 Td (A B) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    expect(item.endMatrix).toEqual([10, 0, 0, 10, 15, 0]);
+  });
+
+  it("restores horizontal scaling across a q/Q pair", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("/F1 10 Tf q 50 Tz Q BT 0 0 Td (A) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    expect(item.startMatrix).toEqual([10, 0, 0, 10, 0, 0]);
+  });
+
+  it("restores leading across a q/Q pair, so T* advances by the outer TL", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("/F1 10 Tf 20 TL q 5 TL Q BT 0 0 Td (A) Tj T* (B) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [, second] = items;
+    if (second?.kind !== "text") {
+      throw new Error("expected a second text item");
+    }
+    expect(second.startMatrix).toEqual([10, 0, 0, 10, 0, -20]);
+  });
+
+  it("restores text rise across a q/Q pair", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("/F1 10 Tf q 5 Ts Q BT 0 0 Td (A) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    expect(item.startMatrix).toEqual([10, 0, 0, 10, 0, 0]);
+  });
+
+  it("keeps a text-state change made outside any q/Q pair, which no Q ever unwinds", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("q 1 0 0 1 0 0 cm Q /F2 24 Tf BT 0 0 Td (X) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    expect(item.fontResourceName).toBe("F2");
+    expect(item.sizePt).toBe(24);
+  });
+
+  it("leaves the text matrix untouched by Q, since Tm/Tlm are text object state rather than graphics state", () => {
+    const { sink } = collectDiagnostics();
+    const items = interpretContentStream(
+      textBytes("BT /F1 10 Tf q 0 0 Td Q 30 40 Td (X) Tj ET"),
+      EMPTY_RESOURCES,
+      {
+        fontMetrics: fixedWidthFontMetrics(),
+        resolver: makeResolver(new Map()),
+        sink,
+      },
+    );
+    const [item] = items;
+    if (item?.kind !== "text") {
+      throw new Error("expected a text item");
+    }
+    expect(item.startMatrix).toEqual([10, 0, 0, 10, 30, 40]);
+  });
+});
+
 describe("interpretContentStream: axis-aligned rectangles", () => {
   it("recovers a rectangle painted under a non-rotated CTM", () => {
     const { sink } = collectDiagnostics();
@@ -865,6 +1020,68 @@ describe("interpretContentStream: XObjects", () => {
     ]);
   });
 
+  // ISO 32000-1 8.10.2: a form XObject's content stream executes in the graphics state in effect at the moment of Do, as if it were nested inline inside an implicit q/Q pair -- so the text state travels inward, and the form's own changes to it do not travel back out.
+  it("runs a Form XObject in the caller's text state, so a form with no Tf of its own draws in the inherited font", () => {
+    const { sink } = collectDiagnostics();
+    const formDict = pdfDict({
+      Type: pdfName("XObject"),
+      Subtype: pdfName("Form"),
+    });
+    const objects = new Map<number, PdfObject>([
+      [30, pdfStream(formDict, textBytes("BT (FormText) Tj ET"))],
+    ]);
+    const resources = pdfDict({ XObject: pdfDict({ Fm2: pdfRef(30, 0) }) });
+    const items = interpretContentStream(
+      textBytes("BT /F1 12 Tf (Outer) Tj ET /Fm2 Do"),
+      resources,
+      {
+        fontMetrics: fixedWidthFontMetrics(),
+        resolver: makeResolver(objects),
+        sink,
+      },
+    );
+    expect(items).toHaveLength(2);
+    const [outer, inner] = items;
+    if (outer?.kind !== "text" || inner?.kind !== "text") {
+      throw new Error("expected two text items");
+    }
+    expect(Array.from(outer.codes)).toEqual(Array.from(textBytes("Outer")));
+    expect(Array.from(inner.codes)).toEqual(Array.from(textBytes("FormText")));
+    expect(inner.fontResourceName).toBe("F1");
+    expect(inner.sizePt).toBe(12);
+  });
+
+  it("does not let a Form XObject's own text-state changes survive back into the caller", () => {
+    const { sink } = collectDiagnostics();
+    const formDict = pdfDict({
+      Type: pdfName("XObject"),
+      Subtype: pdfName("Form"),
+    });
+    const objects = new Map<number, PdfObject>([
+      [31, pdfStream(formDict, textBytes("BT /F2 24 Tf 8 Tc (Inner) Tj ET"))],
+    ]);
+    const resources = pdfDict({ XObject: pdfDict({ Fm3: pdfRef(31, 0) }) });
+    const items = interpretContentStream(
+      textBytes("BT /F1 10 Tf (Outer) Tj ET /Fm3 Do BT 0 0 Td (After) Tj ET"),
+      resources,
+      {
+        fontMetrics: fixedWidthFontMetrics(500, 1),
+        resolver: makeResolver(objects),
+        sink,
+      },
+    );
+    expect(items).toHaveLength(3);
+    const [, inner, after] = items;
+    if (inner?.kind !== "text" || after?.kind !== "text") {
+      throw new Error("expected text items from the form and after it");
+    }
+    expect(inner.fontResourceName).toBe("F2");
+    expect(after.fontResourceName).toBe("F1");
+    expect(after.sizePt).toBe(10);
+    // Five glyphs at width 5 each with the caller's own Tc=0, not the form's Tc=8.
+    expect(after.endMatrix).toEqual([10, 0, 0, 10, 25, 0]);
+  });
+
   it("stops a self-referential chain of forms at the recursion depth limit, with a diagnostic", () => {
     const { sink, diagnostics } = collectDiagnostics();
     const selfResources = pdfDict({