gh-94399: Restore PATH search behaviour of py.exe launcher for '/usr/bin/env' shebang lines (GH-95582)

miss-islington · zooba · web-flow · commit 2c0017b5e610 · 2022-08-03T15:00:46.000-07:00
(cherry picked from commit 67840ed) Co-authored-by: Steve Dower <steve.dower@python.org>
diff --git a/Doc/using/windows.rst b/Doc/using/windows.rst
@@ -855,6 +855,11 @@ The ``/usr/bin/env`` form of shebang line has one further special property.
 Before looking for installed Python interpreters, this form will search the
 executable :envvar:`PATH` for a Python executable. This corresponds to the
 behaviour of the Unix ``env`` program, which performs a :envvar:`PATH` search.
+If an executable matching the first argument after the ``env`` command cannot
+be found, it will be handled as described below. Additionally, the environment
+variable :envvar:`PYLAUNCHER_NO_SEARCH_PATH` may be set (to any value) to skip
+this additional search.
+
 
 Arguments in shebang lines
 --------------------------
diff --git a/Lib/test/test_launcher.py b/Lib/test/test_launcher.py
@@ -188,16 +188,21 @@ def find_py(cls):
             )
         return py_exe
 
+    def get_py_exe(self):
+        if not self.py_exe:
+            self.py_exe = self.find_py()
+        return self.py_exe
+
     def run_py(self, args, env=None, allow_fail=False, expect_returncode=0, argv=None):
         if not self.py_exe:
             self.py_exe = self.find_py()
 
         ignore = {"VIRTUAL_ENV", "PY_PYTHON", "PY_PYTHON2", "PY_PYTHON3"}
         env = {
             **{k.upper(): v for k, v in os.environ.items() if k.upper() not in ignore},
-            **{k.upper(): v for k, v in (env or {}).items()},
             "PYLAUNCHER_DEBUG": "1",
             "PYLAUNCHER_DRYRUN": "1",
+            **{k.upper(): v for k, v in (env or {}).items()},
         }
         if not argv:
             argv = [self.py_exe, *args]
@@ -497,61 +502,93 @@ def test_virtualenv_with_env(self):
 
     def test_py_shebang(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python -prearg") as script:
+            with self.script("#! /usr/bin/python -prearg") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y.exe -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py2_shebang(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python2 -prearg") as script:
+            with self.script("#! /usr/bin/python2 -prearg") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100-32", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y-32.exe -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py3_shebang(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python3 -prearg") as script:
+            with self.script("#! /usr/bin/python3 -prearg") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100-arm64", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y-arm64.exe -X fake_arg_for_test -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py_shebang_nl(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python -prearg\n") as script:
+            with self.script("#! /usr/bin/python -prearg\n") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y.exe -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py2_shebang_nl(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python2 -prearg\n") as script:
+            with self.script("#! /usr/bin/python2 -prearg\n") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100-32", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y-32.exe -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py3_shebang_nl(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python3 -prearg\n") as script:
+            with self.script("#! /usr/bin/python3 -prearg\n") as script:
                 data = self.run_py([script, "-postarg"])
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100-arm64", data["SearchInfo.tag"])
         self.assertEqual(f"X.Y-arm64.exe -X fake_arg_for_test -prearg {script} -postarg", data["stdout"].strip())
 
     def test_py_shebang_short_argv0(self):
         with self.py_ini(TEST_PY_COMMANDS):
-            with self.script("#! /usr/bin/env python -prearg") as script:
+            with self.script("#! /usr/bin/python -prearg") as script:
                 # Override argv to only pass "py.exe" as the command
                 data = self.run_py([script, "-postarg"], argv=f'"py.exe" "{script}" -postarg')
         self.assertEqual("PythonTestSuite", data["SearchInfo.company"])
         self.assertEqual("3.100", data["SearchInfo.tag"])
         self.assertEqual(f'X.Y.exe -prearg "{script}" -postarg', data["stdout"].strip())
 
+    def test_search_path(self):
+        stem = Path(sys.executable).stem
+        with self.py_ini(TEST_PY_COMMANDS):
+            with self.script(f"#! /usr/bin/env {stem} -prearg") as script:
+                data = self.run_py(
+                    [script, "-postarg"],
+                    env={"PATH": f"{Path(sys.executable).parent};{os.getenv('PATH')}"},
+                )
+        self.assertEqual(f"{sys.executable} -prearg {script} -postarg", data["stdout"].strip())
+
+    def test_search_path_exe(self):
+        # Leave the .exe on the name to ensure we don't add it a second time
+        name = Path(sys.executable).name
+        with self.py_ini(TEST_PY_COMMANDS):
+            with self.script(f"#! /usr/bin/env {name} -prearg") as script:
+                data = self.run_py(
+                    [script, "-postarg"],
+                    env={"PATH": f"{Path(sys.executable).parent};{os.getenv('PATH')}"},
+                )
+        self.assertEqual(f"{sys.executable} -prearg {script} -postarg", data["stdout"].strip())
+
+    def test_recursive_search_path(self):
+        stem = self.get_py_exe().stem
+        with self.py_ini(TEST_PY_COMMANDS):
+            with self.script(f"#! /usr/bin/env {stem}") as script:
+                data = self.run_py(
+                    [script],
+                    env={"PATH": f"{self.get_py_exe().parent};{os.getenv('PATH')}"},
+                )
+        # The recursive search is ignored and we get normal "py" behavior
+        self.assertEqual(f"X.Y.exe {script}", data["stdout"].strip())
+
     def test_install(self):
         data = self.run_py(["-V:3.10"], env={"PYLAUNCHER_ALWAYS_INSTALL": "1"}, expect_returncode=111)
         cmd = data["stdout"].strip()
diff --git a/Misc/NEWS.d/next/Windows/2022-08-03-00-49-46.gh-issue-94399.KvxHc0.rst b/Misc/NEWS.d/next/Windows/2022-08-03-00-49-46.gh-issue-94399.KvxHc0.rst
@@ -0,0 +1,3 @@
+Restores the behaviour of :ref:`launcher` for ``/usr/bin/env`` shebang
+lines, which will now search :envvar:`PATH` for an executable matching the
+given command. If none is found, the usual search process is used.
diff --git a/PC/launcher2.c b/PC/launcher2.c
@@ -36,6 +36,7 @@
 #define RC_DUPLICATE_ITEM   110
 #define RC_INSTALLING       111
 #define RC_NO_PYTHON_AT_ALL 112
+#define RC_NO_SHEBANG       113
 
 static FILE * log_fp = NULL;
 
@@ -750,6 +751,88 @@ _shebangStartsWith(const wchar_t *buffer, int bufferLength, const wchar_t *prefi
 }
 
 
+int
+searchPath(SearchInfo *search, const wchar_t *shebang, int shebangLength)
+{
+    if (isEnvVarSet(L"PYLAUNCHER_NO_SEARCH_PATH")) {
+        return RC_NO_SHEBANG;
+    }
+
+    wchar_t *command;
+    if (!_shebangStartsWith(shebang, shebangLength, L"/usr/bin/env ", &command)) {
+        return RC_NO_SHEBANG;
+    }
+
+    wchar_t filename[MAXLEN];
+    int lastDot = 0;
+    int commandLength = 0;
+    while (commandLength < MAXLEN && command[commandLength] && !isspace(command[commandLength])) {
+        if (command[commandLength] == L'.') {
+            lastDot = commandLength;
+        }
+        filename[commandLength] = command[commandLength];
+        commandLength += 1;
+    }
+
+    if (!commandLength || commandLength == MAXLEN) {
+        return RC_BAD_VIRTUAL_PATH;
+    }
+
+    filename[commandLength] = L'\0';
+
+    const wchar_t *ext = L".exe";
+    // If the command already has an extension, we do not want to add it again
+    if (!lastDot || _comparePath(&filename[lastDot], -1, ext, -1)) {
+        if (wcscat_s(filename, MAXLEN, L".exe")) {
+            return RC_BAD_VIRTUAL_PATH;
+        }
+    }
+
+    wchar_t pathVariable[MAXLEN];
+    int n = GetEnvironmentVariableW(L"PATH", pathVariable, MAXLEN);
+    if (!n) {
+        if (GetLastError() == ERROR_ENVVAR_NOT_FOUND) {
+            return RC_NO_SHEBANG;
+        }
+        winerror(0, L"Failed to read PATH\n", filename);
+        return RC_INTERNAL_ERROR;
+    }
+
+    wchar_t buffer[MAXLEN];
+    n = SearchPathW(pathVariable, filename, NULL, MAXLEN, buffer, NULL);
+    if (!n) {
+        if (GetLastError() == ERROR_FILE_NOT_FOUND) {
+            debug(L"# Did not find %s on PATH\n", filename);
+            // If we didn't find it on PATH, let normal handling take over
+            return RC_NO_SHEBANG;
+        }
+        // Other errors should cause us to break
+        winerror(0, L"Failed to find %s on PATH\n", filename);
+        return RC_BAD_VIRTUAL_PATH;
+    }
+
+    // Check that we aren't going to call ourselves again
+    // If we are, pretend there was no shebang and let normal handling take over
+    if (GetModuleFileNameW(NULL, filename, MAXLEN) &&
+        0 == _comparePath(filename, -1, buffer, -1)) {
+        debug(L"# ignoring recursive shebang command\n");
+        return RC_NO_SHEBANG;
+    }
+
+    wchar_t *buf = allocSearchInfoBuffer(search, n + 1);
+    if (!buf || wcscpy_s(buf, n + 1, buffer)) {
+        return RC_NO_MEMORY;
+    }
+
+    search->executablePath = buf;
+    search->executableArgs = &command[commandLength];
+    search->executableArgsLength = shebangLength - commandLength;
+    debug(L"# Found %s on PATH\n", buf);
+
+    return 0;
+}
+
+
 int
 _readIni(const wchar_t *section, const wchar_t *settingName, wchar_t *buffer, int bufferLength)
 {
@@ -885,6 +968,12 @@ checkShebang(SearchInfo *search)
     }
     debug(L"Shebang: %s\n", shebang);
 
+    // Handle shebangs that we should search PATH for
+    exitCode = searchPath(search, shebang, shebangLength);
+    if (exitCode != RC_NO_SHEBANG) {
+        return exitCode;
+    }
+
     // Handle some known, case-sensitive shebang templates
     const wchar_t *command;
     int commandLength;
@@ -895,6 +984,7 @@ checkShebang(SearchInfo *search)
         L"",
         NULL
     };
+
     for (const wchar_t **tmpl = shebangTemplates; *tmpl; ++tmpl) {
         if (_shebangStartsWith(shebang, shebangLength, *tmpl, &command)) {
             commandLength = 0;
@@ -910,6 +1000,22 @@ checkShebang(SearchInfo *search)
             } else if (_shebangStartsWith(command, commandLength, L"python", NULL)) {
                 search->tag = &command[6];
                 search->tagLength = commandLength - 6;
+                // If we had 'python3.12.exe' then we want to strip the suffix
+                // off of the tag
+                if (search->tagLength > 4) {
+                    const wchar_t *suffix = &search->tag[search->tagLength - 4];
+                    if (0 == _comparePath(suffix, 4, L".exe", -1)) {
+                        search->tagLength -= 4;
+                    }
+                }
+                // If we had 'python3_d' then we want to strip the '_d' (any
+                // '.exe' is already gone)
+                if (search->tagLength > 2) {
+                    const wchar_t *suffix = &search->tag[search->tagLength - 2];
+                    if (0 == _comparePath(suffix, 2, L"_d", -1)) {
+                        search->tagLength -= 2;
+                    }
+                }
                 search->oldStyleTag = true;
                 search->executableArgs = &command[commandLength];
                 search->executableArgsLength = shebangLength - commandLength;

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	+Restores the behaviour of :ref:`launcher` for ``/usr/bin/env`` shebang
	`2`	+lines, which will now search :envvar:`PATH` for an executable matching the
	`3`	`+given command. If none is found, the usual search process is used.`