Switched fonts to support other languages

Author: m-messer

Dockerfile

@@ -42,7 +42,22 @@ RUN dnf install -y \
   texlive-collection-latexrecommended.noarch \
   texlive-iftex.noarch \
   texlive-braket.noarch \
-  texlive-cancel.noarch
+  texlive-cancel.noarch \
+  texlive-xecjk.noarch
+
+# Install Noto Sans fonts: base (Latin/Greek/Cyrillic) + CJK (Korean/Chinese/Japanese)
+# These are the default fonts used by the template for broad Unicode coverage.
+RUN curl -fsSL https://github.com/notofonts/latin-greek-cyrillic/releases/download/NotoSans-v2.013/NotoSans-v2.013.zip \
+      -o /tmp/NotoSans.zip \
+  && unzip -j /tmp/NotoSans.zip "*/unhinted/ttf/NotoSans-Regular.ttf" \
+                                "*/unhinted/ttf/NotoSans-Bold.ttf" \
+                                "*/unhinted/ttf/NotoSans-Italic.ttf" \
+                                "*/unhinted/ttf/NotoSans-BoldItalic.ttf" \
+      -d /usr/share/fonts/noto \
+  && rm /tmp/NotoSans.zip \
+  && curl -fsSL https://github.com/googlefonts/noto-cjk/releases/download/Sans2.004R/NotoSansCJKkr-Regular.otf \
+      -o /usr/share/fonts/noto/NotoSansCJKkr-Regular.otf \
+  && fc-cache -fv
 
 # Copy the LaTeX template
 COPY ./src/template.latex template.latex

README.md

@@ -65,6 +65,7 @@ In addition to the runtime dependencies above, running the full test suite local
 On macOS these can be installed via Homebrew:
 ```bash
 brew install pandoc mactex poppler
+brew install --cask font-noto-sans font-noto-sans-cjk
 ```
 
 ### Running tests

index.test.ts

@@ -52,6 +52,16 @@ describe("schema", () => {
     const data = { userId: "u1", fileName: "doc", typeOfFile: "PDF", markdown: "text" };
     expect(schema.safeParse(data).success).toBe(false);
   });
+
+  it("validates a request with a variables map", () => {
+    const data = [{ userId: "u1", fileName: "doc", typeOfFile: "PDF", markdown: "text", variables: { lang: "ko", CJKmainfont: "Noto Sans CJK KR" } }];
+    expect(schema.safeParse(data).success).toBe(true);
+  });
+
+  it("rejects variables with non-string values", () => {
+    const data = [{ userId: "u1", fileName: "doc", typeOfFile: "PDF", markdown: "text", variables: { lang: 42 } }];
+    expect(schema.safeParse(data).success).toBe(false);
+  });
 });
 
 describe("handler", () => {
@@ -95,6 +105,18 @@ describe("handler", () => {
     expect(result.statusCode).toBe(200);
   });
 
+  it("passes variables as --variable flags to Pandoc", async () => {
+    const executeMock = vi.fn().mockResolvedValue("");
+    vi.mocked(PdcTs).mockImplementationOnce(() => ({ Execute: executeMock }) as any);
+
+    const event = [{ userId: "user1", fileName: "test-doc", typeOfFile: "TEX", markdown: "# Hello", variables: { lang: "ko", CJKmainfont: "Noto Sans CJK KR" } }];
+    await handler(event as any);
+
+    const calledArgs: string[] = executeMock.mock.calls[0]?.[0]?.pandocArgs ?? [];
+    expect(calledArgs).toContain("--variable=lang:ko");
+    expect(calledArgs).toContain("--variable=CJKmainfont:Noto Sans CJK KR");
+  });
+
   it("returns 500 when Pandoc execution fails", async () => {
     vi.mocked(PdcTs).mockImplementationOnce(() => ({
       Execute: vi.fn()

index.ts

@@ -14,6 +14,7 @@ export const schema = z.array(
     typeOfFile: TypeOfFileSchema,
     markdown: z.string(),
     implicitFigures: z.boolean().optional(),
+    variables: z.record(z.string(), z.string()).optional(),
   })
 );
 
@@ -147,13 +148,15 @@ export const handler = async function (
   for (let eachRequestData of requestData) {
     const markdown = eachRequestData.markdown;
     const implicitFigures = eachRequestData.implicitFigures;
+    const variableArgs = Object.entries(eachRequestData.variables ?? {})
+      .map(([k, v]) => `--variable=${k}:${v}`);
 
     switch (eachRequestData.typeOfFile) {
       case "PDF":
         const filenamePDF = `${eachRequestData.fileName}.pdf`;
         const localPathPDF = `/tmp/${filenamePDF}`;
         const generatePDFResult = await generateFile(
-          ["--pdf-engine=xelatex", `--template=./template.latex`],
+          ["--pdf-engine=xelatex", `--template=./template.latex`, ...variableArgs],
           localPathPDF,
           markdown,
           implicitFigures
@@ -170,7 +173,7 @@ export const handler = async function (
         const filenameTEX = `${eachRequestData.fileName}.tex`;
         const localPathTEX = `/tmp/${filenameTEX}`;
         await generateFile(
-          [`--template=./template.latex`],
+          [`--template=./template.latex`, ...variableArgs],
           localPathTEX,
           markdown,
           implicitFigures

src/pandoc.test.ts

@@ -78,6 +78,21 @@ describe("Pandoc markdown → LaTeX output", () => {
     });
   });
 
+  describe("pandoc variables", () => {
+    it("accepts a --variable flag without error", async () => {
+      // Verifies the variable-passing mechanism works; mainfont is a safe no-op variable
+      // since the template only applies it under XeLaTeX, which isn't invoked here
+      const latex = await new PdcTs().Execute({
+        from: "markdown",
+        to: "latex",
+        outputToFile: false,
+        sourceText: "# Hello",
+        pandocArgs: ["--variable=mainfont:Latin Modern Roman"],
+      });
+      expect(latex).toContain("\\section{Hello}");
+    });
+  });
+
   describe("unicode characters", () => {
     it("passes Greek letters in prose through to LaTeX", async () => {
       const latex = await toLatex("Lowercase: α, β, γ, Δ, Σ");

src/template.latex

@@ -33,6 +33,8 @@ $if(euro)$
 $endif$
 $if(mainfont)$
     \setmainfont[$for(mainfontoptions)$$mainfontoptions$$sep$,$endfor$]{$mainfont$}
+$else$
+    \setmainfont{Noto Sans}
 $endif$
 $if(sansfont)$
     \setsansfont[$for(sansfontoptions)$$sansfontoptions$$sep$,$endfor$]{$sansfont$}
@@ -43,10 +45,8 @@ $endif$
 $if(mathfont)$
     \setmathfont(Digits,Latin,Greek)[$for(mathfontoptions)$$mathfontoptions$$sep$,$endfor$]{$mathfont$}
 $endif$
-$if(CJKmainfont)$
-    \usepackage{xeCJK}
-    \setCJKmainfont[$for(CJKoptions)$$CJKoptions$$sep$,$endfor$]{$CJKmainfont$}
-$endif$
+\usepackage{xeCJK}
+\setCJKmainfont[$for(CJKoptions)$$CJKoptions$$sep$,$endfor$]{$if(CJKmainfont)$$CJKmainfont$$else$Noto Sans CJK KR$endif$}
 \fi
 % use upquote if available, for straight quotes in verbatim environments
 \IfFileExists{upquote.sty}{\usepackage{upquote}}{}
@@ -260,8 +260,8 @@ $endif$
 
 
 \usepackage[a4paper,top=2cm,bottom=2cm,left=2cm,right=2cm]{geometry}%
-\usepackage{lmodern}%                     font style must be sans serif to comply with
-\renewcommand*\familydefault{\sfdefault}% the college's regulations with respect to disabilities!
+% Noto Sans (sans-serif) is set as the default mainfont above, satisfying the
+% college's accessibility requirement for sans-serif fonts throughout the document.
 
 
 \begin{document}