Improve kernel language identification

Metadata fields in the Jupyter Notebook JSON are optional. Therefore, multiple JSON fields are now checked for hints about the kernel language. If no information is found, Chroma will try to infer the language from the file content

convert.go

@@ -84,9 +84,9 @@ func convertMarkdownCell(cell cell) template.HTML {
 }
 
 // convertCodeCell converts a code cell to HTML with classes for syntax highlighting.
-func convertCodeCell(cell cell, fileExtension string) template.HTML {
+func convertCodeCell(cell cell, languageID string) template.HTML {
  sourceString := strings.Join(cell.Source, "")
- cellHTML, err := renderSourceCode(sourceString, fileExtension)
+ cellHTML, err := renderSourceCode(sourceString, languageID)
 
  // Render code without syntax highlighting if an error occurred
  if err != nil {
@@ -116,12 +116,12 @@ func convertPrompt(executionCount *int) template.HTML {
 }
 
 // convertOutput converts the provided cell input to HTML.
-func convertInput(fileExtension string, cell cell) template.HTML {
+func convertInput(languageID string, cell cell) template.HTML {
  switch cell.CellType {
  case "markdown":
  return convertMarkdownCell(cell)
  case "code":
- return convertCodeCell(cell, fileExtension)
+ return convertCodeCell(cell, languageID)
  case "raw":
  return convertRawCell(cell)
  default:
@@ -187,8 +187,16 @@ func ConvertString(writer io.Writer, notebookString string) error {
  )
  }
 
- // Get format extension of Jupyter Kernel language (e.g. "py")
- fileExtension := notebook.Metadata.LanguageInfo.FileExtension[1:]
+ // Try to find information about programming language used by the notebook kernel. Metadata fields
+ // in the Jupyter Notebook JSON are optional, so multiple fields are checked
+ languageID := ""
+ if fileExtensionPtr := notebook.Metadata.LanguageInfo.FileExtension; fileExtensionPtr != nil {
+ languageID = (*fileExtensionPtr)[1:]
+ } else if kernelLanguagePtr := notebook.Metadata.KernelSpec.Language; kernelLanguagePtr != nil {
+ languageID = *kernelLanguagePtr
+ } else if kernelNamePtr := notebook.Metadata.KernelSpec.Name; kernelNamePtr != nil {
+ languageID = *kernelNamePtr
+ }
 
  t := template.New("notebook")
  t = t.Funcs(template.FuncMap{
@@ -205,15 +213,15 @@ func ConvertString(writer io.Writer, notebookString string) error {
  })
  t, err = t.Parse(`
  <div class="notebook">
- {{ $fileExtension := .fileExtension }}
+ {{ $languageID := .languageID }}
  {{ range .notebook.Cells }}
  <div class="{{ . | getCellClasses }}">
  <div class="input-wrapper">
  <div class="input-prompt">
  {{ .ExecutionCount | convertPrompt }}
  </div>
  <div class="input">
- {{ . | convertInput $fileExtension }}
+ {{ . | convertInput $languageID }}
  </div>
  </div>
  {{ range .Outputs }}
@@ -235,8 +243,8 @@ func ConvertString(writer io.Writer, notebookString string) error {
  }
 
  templateVars := map[string]interface{}{
- "fileExtension": fileExtension,
- "notebook":  notebook,
+ "languageID": languageID,
+ "notebook": notebook,
  }
  return t.Execute(writer, templateVars)
 }

parse.go

@@ -41,15 +41,22 @@ type cell struct {
 
 // languageInfo provides details about the programming language of the Jupyter Notebook kernel.
 type languageInfo struct {
- FileExtension string `json:"file_extension"`
- // Omitted fields: codemirror_mode", "mimetype", "name", "nbconvert_exporter", "pygments_lexer",
+ FileExtension *string `json:"file_extension,omitempty"`
+ // Omitted fields: "codemirror_mode", "mimetype", "name", "nbconvert_exporter", "pygments_lexer",
  // "version"
 }
 
+// kernelSpec provides details about the Jupyter Notebook kernel.
+type kernelSpec struct {
+ DisplayName *string `json:"display_name,omitempty"`
+ Language *string `json:"language,omitempty"`
+ Name *string `json:"name,omitempty"`
+}
+
 // metadata contains additional information about the Jupyter Notebook.
 type metadata struct {
  LanguageInfo languageInfo `json:"language_info"`
- // Omitted fields: "kernelspec"
+ KernelSpec kernelSpec `json:"kernelspec"`
 }
 
 // notebook represents the JSON data structure in which a Jupyter Notebook is stored.

render.go

@@ -24,11 +24,14 @@ func renderMarkdown(markdownLines []string) template.HTML {
 // renderSourceCode uses the Chroma library to convert the provided source code string to HTML.
 // Instead of inline styles, HTML classes are used for syntax highlighting, which allows the users
 // to style source code according to their needs.
-func renderSourceCode(source string, fileExtension string) (template.HTML, error) {
+func renderSourceCode(source string, languageID string) (template.HTML, error) {
  sourceBuffer := new(bytes.Buffer)
 
- // Set up lexer for file extension
- l := lexers.Get(fileExtension)
+ // Set up lexer for programming language
+ var l chroma.Lexer
+ if languageID != "" {
+ l = lexers.Get(languageID)
+ }
  if l == nil {
  l = lexers.Analyse(source)
  }

test_data.go

@@ -383,9 +383,18 @@ var testRawCellCodeInjection = cell{
  },
 }
 
+var testFileExtension = ".py"
+var testKernelLanguage = "python"
+var testKernelDisplayName = "Python 3"
+var testKernelName = "python3"
 var testMetadata = metadata{
  LanguageInfo: languageInfo{
- FileExtension: ".py",
+ FileExtension: &testFileExtension,
+ },
+ KernelSpec: kernelSpec{
+ DisplayName: &testKernelDisplayName,
+ Language: &testKernelLanguage,
+ Name: &testKernelName,
  },
 }