Infer via shebang (#5149)

If no file type can be inferred, this attempts to read the first line of a file and extract a shebang, which can be checked against a known list.

package.json

@@ -49,6 +49,7 @@
     "mem": "1.1.0",
     "minimatch": "3.0.4",
     "minimist": "1.2.0",
+    "n-readlines": "1.0.0",
     "normalize-path": "3.0.0",
     "parse5-htmlparser2-tree-adapter": "5.0.0",
     "postcss-less": "1.1.5",

src/language-js/index.js

@@ -11,6 +11,9 @@ const languages = [
       since: "0.0.0",
       parsers: ["babylon", "flow"],
       vscodeLanguageIds: ["javascript"]
+    },
+    extend: {
+      interpreters: ["nodejs"]
     }
   }),
   createLanguage(require("linguist-languages/data/javascript"), {

src/main/options.js

@@ -1,6 +1,8 @@
 "use strict";
 
+const fs = require("fs");
 const normalizePath = require("normalize-path");
+const readlines = require("n-readlines");
 const UndefinedParserError = require("../common/errors").UndefinedParserError;
 const getSupportInfo = require("../main/support").getSupportInfo;
 const normalizer = require("./options-normalizer");
@@ -114,10 +116,56 @@ function getPlugin(options) {
   return printerPlugin;
 }
 
+function getInterpreter(filepath) {
+  if (typeof filepath !== "string") {
+    return "";
+  }
+
+  let fd;
+  try {
+    fd = fs.openSync(filepath, "r");
+  } catch (err) {
+    return "";
+  }
+
+  try {
+    const liner = new readlines(fd);
+    const firstLine = liner.next().toString("utf8");
+
+    // #!/bin/env node, #!/usr/bin/env node
+    const m1 = firstLine.match(/^#!\/(?:usr\/)?bin\/env\s+(\S+)/);
+    if (m1) {
+      return m1[1];
+    }
+
+    // #!/bin/node, #!/usr/bin/node, #!/usr/local/bin/node
+    const m2 = firstLine.match(/^#!\/(?:usr\/(?:local\/)?)?bin\/(\S+)/);
+    if (m2) {
+      return m2[1];
+    }
+    return "";
+  } catch (err) {
+    // There are some weird cases where paths are missing, causing Jest
+    // failures. It's unclear what these correspond to in the real world.
+    return "";
+  } finally {
+    try {
+      // There are some weird cases where paths are missing, causing Jest
+      // failures. It's unclear what these correspond to in the real world.
+      fs.closeSync(fd);
+    } catch (err) {
+      // nop
+    }
+  }
+}
+
 function inferParser(filepath, plugins) {
   const filepathParts = normalizePath(filepath).split("/");
   const filename = filepathParts[filepathParts.length - 1].toLowerCase();
 
+  // If the file has no extension, we can try to infer the language from the
+  // interpreter in the shebang line, if any; but since this requires FS access,
+  // do it last.
   const language = getSupportInfo(null, {
     plugins
   }).languages.find(
@@ -126,7 +174,10 @@ function inferParser(filepath, plugins) {
       ((language.extensions &&
         language.extensions.some(extension => filename.endsWith(extension))) ||
         (language.filenames &&
-          language.filenames.find(name => name.toLowerCase() === filename)))
+          language.filenames.find(name => name.toLowerCase() === filename)) ||
+        (filename.indexOf(".") === -1 &&
+          language.interpreters &&
+          language.interpreters.indexOf(getInterpreter(filepath)) !== -1))
   );
 
   return language && language.parsers[0];

tests_integration/__tests__/__snapshots__/support-info.js.snap

@@ -566,7 +566,7 @@ exports[`CLI --support-info (stdout) 1`] = `
         \\".xsjslib\\"
       ],
       \\"filenames\\": [\\"Jakefile\\"],
-      \\"interpreters\\": [\\"node\\"],
+      \\"interpreters\\": [\\"node\\", \\"nodejs\\"],
       \\"linguistLanguageId\\": 183,
       \\"name\\": \\"JavaScript\\",
       \\"parsers\\": [\\"babylon\\", \\"flow\\"],

tests_integration/__tests__/file-info.js

@@ -194,6 +194,33 @@ test("API getFileInfo with withNodeModules", () => {
   });
 });
 
+describe("extracts file-info for a JS file with no extension but a standard shebang", () => {
+  expect(
+    prettier.getFileInfo.sync("tests_integration/cli/shebang/node-shebang")
+  ).toMatchObject({
+    ignored: false,
+    inferredParser: "babylon"
+  });
+});
+
+describe("extracts file-info for a JS file with no extension but an env-based shebang", () => {
+  expect(
+    prettier.getFileInfo.sync("tests_integration/cli/shebang/env-node-shebang")
+  ).toMatchObject({
+    ignored: false,
+    inferredParser: "babylon"
+  });
+});
+
+describe("returns null parser for unknown shebang", () => {
+  expect(
+    prettier.getFileInfo.sync("tests_integration/cli/shebang/nonsense-shebang")
+  ).toMatchObject({
+    ignored: false,
+    inferredParser: null
+  });
+});
+
 test("API getFileInfo with plugins loaded using pluginSearchDir", () => {
   const file = "file.foo";
   const pluginsPath = path.resolve(

tests_integration/cli/shebang/env-node-shebang

@@ -0,0 +1,2 @@
+#!/usr/bin/env node
+'use strict';

tests_integration/cli/shebang/node-shebang

@@ -0,0 +1,2 @@
+#!/usr/bin/node
+'use strict';

tests_integration/cli/shebang/nonsense-shebang

@@ -0,0 +1 @@
+#!/bin/frobble

yarn.lock

@@ -4728,6 +4728,10 @@ mute-stream@0.0.7:
   resolved "https://registry.yarnpkg.com/mute-stream/-/mute-stream-0.0.7.tgz#3075ce93bc21b8fab43e1bc4da7e8115ed1e7bab"
   integrity sha1-MHXOk7whuPq0PhvE2n6BFe0ee6s=
 
+n-readlines@1.0.0:
+  version "1.0.0"
+  resolved "https://registry.yarnpkg.com/n-readlines/-/n-readlines-1.0.0.tgz#c353797f216c253fdfef7e91da4e8b17c29a91a6"
+
 nan@^2.9.2:
   version "2.10.0"
   resolved "https://registry.yarnpkg.com/nan/-/nan-2.10.0.tgz#96d0cd610ebd58d4b4de9cc0c6828cda99c7548f"