feat: process embedded languages in text blocks with language identifier comments

Author: jtkiesel

src/comments.ts

@@ -72,10 +72,13 @@ export function canAttachComment(node: JavaNode) {
   }
   switch (node.type) {
     case SyntaxType.EnumBodyDeclarations:
+    case SyntaxType.EscapeSequence:
     case SyntaxType.FormalParameters:
     case SyntaxType.Modifier:
+    case SyntaxType.MultilineStringFragment:
     case SyntaxType.ParenthesizedExpression:
     case SyntaxType.Program:
+    case SyntaxType.StringFragment:
     case SyntaxType.Visibility:
       return false;
     default:

src/printer.ts

@@ -8,11 +8,14 @@ import {
   willPrintOwnComments
 } from "./comments.js";
 import {
+  embedTextBlock,
+  hasType,
   printComment,
   printValue,
   type JavaComment,
   type JavaNode,
-  type JavaNodePath
+  type JavaNodePath,
+  type JavaNodeType
 } from "./printers/helpers.js";
 import { printerForNodeType } from "./printers/index.js";
 import { SyntaxType } from "./tree-sitter-java.js";
@@ -23,6 +26,11 @@ export default {
       ? printerForNodeType(path.node.type)(path, print, options, args)
       : printValue(path);
   },
+  embed(path: JavaNodePath<JavaNodeType>) {
+    return hasType(path, SyntaxType.StringLiteral)
+      ? embedTextBlock(path)
+      : null;
+  },
   hasPrettierIgnore(path) {
     return (
       path.node.comments?.some(isPrettierIgnore) === true ||
@@ -44,6 +52,9 @@ export default {
     ownLine: handleLineComment,
     endOfLine: handleLineComment,
     remaining: handleRemainingComment
+  },
+  getVisitorKeys() {
+    return ["namedChildren"];
   }
 } satisfies Printer<JavaNode>;
 

src/printers/helpers.ts

@@ -1,5 +1,5 @@
-import type { AstPath, Doc, ParserOptions } from "prettier";
-import { builders } from "prettier/doc";
+import type { AstPath, Doc, Options, ParserOptions } from "prettier";
+import { builders, utils } from "prettier/doc";
 import {
   SyntaxType,
   type NodeOfType,
@@ -9,6 +9,7 @@ import {
 } from "../tree-sitter-java.js";
 
 const { group, hardline, ifBreak, indent, join, line, softline } = builders;
+const { mapDoc } = utils;
 
 export function hasType<T extends JavaTypeString>(
   path: AstPath<JavaNode>,
@@ -315,12 +316,117 @@ export function printVariableDeclaration(
   return declaration;
 }
 
-export function findBaseIndent(lines: string[]) {
-  return lines.length
-    ? Math.min(
-        ...lines.map(line => line.search(/\S/)).filter(indent => indent >= 0)
-      )
-    : 0;
+export function printTextBlock(
+  path: JavaNodePath<SyntaxType.StringLiteral>,
+  contents: Doc
+) {
+  const parts = ['"""', hardline, contents, '"""'];
+  const parentType = (path.parent as JavaNode | null)?.type;
+  const grandparentType = (path.grandparent as JavaNode | null)?.type;
+  return parentType === SyntaxType.AssignmentExpression ||
+    parentType === SyntaxType.VariableDeclarator ||
+    (path.node.fieldName === "object" &&
+      (grandparentType === SyntaxType.AssignmentExpression ||
+        grandparentType === SyntaxType.VariableDeclarator))
+    ? indent(parts)
+    : parts;
+}
+
+export function embedTextBlock(path: JavaNodePath<SyntaxType.StringLiteral>) {
+  const hasInterpolations = path.node.namedChildren.some(
+    ({ type }) => type === SyntaxType.StringInterpolation
+  );
+  if (hasInterpolations || path.node.children[0].value === '"') {
+    return null;
+  }
+
+  const language = findEmbeddedLanguage(path);
+  if (!language) {
+    return null;
+  }
+
+  const text = unescapeTextBlockContents(textBlockContents(path.node));
+
+  return async (
+    textToDoc: (text: string, options: Options) => Promise<Doc>
+  ) => {
+    const doc = await textToDoc(text, { parser: language });
+    return printTextBlock(path, escapeDocForTextBlock(doc));
+  };
+}
+
+export function textBlockContents(node: JavaNode<SyntaxType.StringLiteral>) {
+  const lines = node.value
+    .replace(
+      /(?<=^|[^\\])((?:\\\\)*)\\u+([0-9a-fA-F]{4})/g,
+      (_, backslashPairs: string, hex: string) =>
+        backslashPairs + String.fromCharCode(parseInt(hex, 16))
+    )
+    .split("\n")
+    .slice(1);
+  const baseIndent = findBaseIndent(lines);
+  return lines
+    .map(line => line.slice(baseIndent))
+    .join("\n")
+    .slice(0, -3);
+}
+
+function findBaseIndent(lines: string[]) {
+  return Math.min(
+    ...lines.map(line => line.search(/\S/)).filter(indent => indent >= 0)
+  );
+}
+
+function findEmbeddedLanguage(path: JavaNodePath) {
+  return path.ancestors
+    .find(
+      ({ type, comments }) =>
+        type === SyntaxType.Block || comments?.some(({ leading }) => leading)
+    )
+    ?.comments?.filter(({ leading }) => leading)
+    .map(
+      ({ value }) => value.match(/^(?:\/\/|\/\*)\s*language\s*=\s*(\S+)/)?.[1]
+    )
+    .findLast(language => language)
+    ?.toLowerCase();
+}
+
+function escapeDocForTextBlock(doc: Doc) {
+  return mapDoc(doc, currentDoc =>
+    typeof currentDoc === "string"
+      ? currentDoc.replace(/\\|"""/g, match => `\\${match}`)
+      : currentDoc
+  );
+}
+
+function unescapeTextBlockContents(text: string) {
+  return text.replace(
+    /\\(?:([bstnfr"'\\])|\n|\r\n?|([0-3][0-7]{0,2}|[0-7]{1,2}))/g,
+    (_, single, octal) => {
+      if (single) {
+        switch (single) {
+          case "b":
+            return "\b";
+          case "s":
+            return " ";
+          case "t":
+            return "\t";
+          case "n":
+            return "\n";
+          case "f":
+            return "\f";
+          case "r":
+            return "\r";
+          default:
+            return single;
+        }
+      } else if (octal) {
+        return String.fromCharCode(parseInt(octal, 8));
+      } else {
+        return "";
+      }
+    }
+  );
 }
 
 export type JavaNode<T extends JavaTypeString = JavaTypeString> =

src/printers/lexical-structure.ts

@@ -1,9 +1,9 @@
 import { builders } from "prettier/doc";
 import { SyntaxType } from "../tree-sitter-java.js";
 import {
-  findBaseIndent,
+  printTextBlock,
   printValue,
-  type JavaNode,
+  textBlockContents,
   type JavaNodePrinters
 } from "./helpers.js";
 
@@ -18,25 +18,10 @@ export default {
       return path.map(print, "children");
     }
 
-    const lines = path.node.children
-      .map(({ value }) => value)
-      .join("")
-      .split("\n")
-      .slice(1);
-    const baseIndent = findBaseIndent(lines);
-    const textBlock = join(hardline, [
-      '"""',
-      ...lines.map(line => line.slice(baseIndent))
-    ]);
-    const parentType = (path.parent as JavaNode | null)?.type;
-    const grandparentType = (path.grandparent as JavaNode | null)?.type;
-    return parentType === SyntaxType.AssignmentExpression ||
-      parentType === SyntaxType.VariableDeclarator ||
-      (path.node.fieldName === "object" &&
-        (grandparentType === SyntaxType.AssignmentExpression ||
-          grandparentType === SyntaxType.VariableDeclarator))
-      ? indent(textBlock)
-      : textBlock;
+    return printTextBlock(
+      path,
+      join(hardline, textBlockContents(path.node).split("\n"))
+    );
   },
 
   string_fragment: printValue,

test/unit-test/text-blocks/_input.java

@@ -55,4 +55,26 @@ public void print(%s object) {
     );
   }
 
+  void json() {
+    // language = json
+    String someJson = """
+    {"glossary":{"title": "example glossary"}}
+    """;
+
+    // language=json
+    String config = """
+          {   "name":"example",
+      "enabled"   :true,
+            "timeout":30}
+    """;
+
+    /* language = JSON */
+    String query = """
+         {
+       "sql":"SELECT * FROM users \
+    WHERE active=1 \
+    AND deleted=0",
+       "limit":10}
+    """;
+  }
 }

test/unit-test/text-blocks/_output.java

@@ -52,4 +52,21 @@ public void print(%s object) {
             abc"""
     );
   }
+
+  void json() {
+    // language = json
+    String someJson = """
+      { "glossary": { "title": "example glossary" } }""";
+
+    // language=json
+    String config = """
+      { "name": "example", "enabled": true, "timeout": 30 }""";
+
+    /* language = JSON */
+    String query = """
+      {
+        "sql": "SELECT * FROM users WHERE active=1 AND deleted=0",
+        "limit": 10
+      }""";
+  }
 }