Web: $_SERVER['SCRIPT_FILENAME'] for CGI routing and AOT (fixes #302)

bin/compile.php

@@ -17,10 +17,18 @@ function run(string $filename, string $code, array $options): void
     $runtime = new Runtime(Runtime::MODE_AOT);
     $queryString = $options['-q'] ?? null;
     $postBody = $options['-p'] ?? null;
+    $scriptFilename = null;
+    if ('-' !== $filename && 'Command line code' !== $filename) {
+        $resolved = realpath($filename);
+        if (false !== $resolved) {
+            $scriptFilename = $resolved;
+        }
+    }
     Superglobals::populateFromEnvironment(
         $runtime->vmContext,
         is_string($queryString) ? $queryString : null,
-        is_string($postBody) ? $postBody : null
+        is_string($postBody) ? $postBody : null,
+        $scriptFilename
     );
     $block = $runtime->parseAndCompile($code, $filename);
     if (! isset($options['-l'])) {

bin/vm.php

@@ -17,10 +17,18 @@ function run(string $filename, string $code, array $options): void
     $runtime = new Runtime();
     $queryString = $options['-q'] ?? null;
     $postBody = $options['-p'] ?? null;
+    $scriptFilename = null;
+    if ('-' !== $filename && 'Command line code' !== $filename) {
+        $resolved = realpath($filename);
+        if (false !== $resolved) {
+            $scriptFilename = $resolved;
+        }
+    }
     Superglobals::populateFromEnvironment(
         $runtime->vmContext,
         is_string($queryString) ? $queryString : null,
-        is_string($postBody) ? $postBody : null
+        is_string($postBody) ? $postBody : null,
+        $scriptFilename
     );
     $block = $runtime->parseAndCompile($code, $filename);
     if (! isset($options['-l'])) {

docs/bootstrap-inventory.md

@@ -1287,8 +1287,8 @@ These `LogicException` messages indicate CFG ops or expressions not yet lowered:
 ### `lib/JIT/SuperglobalInit.php`
 
 **Warnings** (review for bootstrap subset):
-- new Variable (line 121)
-- new VMVariable (line 145)
+- new Variable (line 123)
+- new VMVariable (line 147)
 - 7 class method(s) — PHPCfg Op\Stmt\ClassMethod not lowered in Compiler
 
 ### `lib/JIT/ValueEchoHelper.php`
@@ -1501,11 +1501,11 @@ These `LogicException` messages indicate CFG ops or expressions not yet lowered:
 ### `lib/Web/Superglobals.php`
 
 **Warnings** (review for bootstrap subset):
-- new HashTable (line 411)
-- new Variable (line 412)
-- new Variable (line 431)
-- new Variable (line 481)
-- 22 class method(s) — PHPCfg Op\Stmt\ClassMethod not lowered in Compiler
+- new HashTable (line 420)
+- new Variable (line 421)
+- new Variable (line 440)
+- new Variable (line 515)
+- 23 class method(s) — PHPCfg Op\Stmt\ClassMethod not lowered in Compiler
 
 ### `src/macro_functions.php`
 

lib/AOT/runtime/superglobals_refresh.c

@@ -439,6 +439,37 @@ static void apply_scheme_and_port(__hashtable__ *server)
     set_string_key(server, "SERVER_PORT", port_buf);
 }
 
+static void resolve_script_filename(
+    const char *script_name,
+    char *out,
+    size_t out_len
+) {
+    const char *from_env = getenv("SCRIPT_FILENAME");
+
+    out[0] = '\0';
+    if (NULL != from_env && '\0' != from_env[0]) {
+        strncpy(out, from_env, out_len - 1);
+        out[out_len - 1] = '\0';
+
+        return;
+    }
+
+    {
+        const char *document_root = getenv("DOCUMENT_ROOT");
+        size_t root_len;
+
+        if (NULL == document_root || '\0' == document_root[0]
+            || NULL == script_name || '\0' == script_name[0]) {
+            return;
+        }
+        root_len = strlen(document_root);
+        while (root_len > 0 && '/' == document_root[root_len - 1]) {
+            root_len--;
+        }
+        snprintf(out, out_len, "%.*s%s", (int) root_len, document_root, script_name);
+    }
+}
+
 static void derive_path_info(const char *script_name, const char *request_uri, char *out, size_t out_len)
 {
     char path_buf[1024];
@@ -481,6 +512,7 @@ void __superglobals__refresh(void)
     const char *script_name = env_or_empty("SCRIPT_NAME");
     const char *request_uri = getenv("REQUEST_URI");
     char path_info[512];
+    char script_filename[1024];
     char request_uri_buf[1024];
 
     if (NULL == request_uri || '\0' == request_uri[0]) {
@@ -516,6 +548,10 @@ void __superglobals__refresh(void)
     set_string_key(sg_SERVER, "QUERY_STRING", query_string);
     set_string_key(sg_SERVER, "SCRIPT_NAME", script_name);
     set_string_key(sg_SERVER, "PHP_SELF", script_name);
+    resolve_script_filename(script_name, script_filename, sizeof(script_filename));
+    if ('\0' != script_filename[0]) {
+        set_string_key(sg_SERVER, "SCRIPT_FILENAME", script_filename);
+    }
     set_string_key(sg_SERVER, "REQUEST_URI", request_uri);
     set_string_key(sg_SERVER, "GATEWAY_INTERFACE", "CGI/1.1");
     set_string_key(sg_SERVER, "SERVER_SOFTWARE", "PHP-Compiler-AOT");

lib/JIT/SuperglobalInit.php

@@ -16,12 +16,14 @@ final class SuperglobalInit
     /** @var array<string, \PHPLLVM\Value> */
     public static array $globals = [];
 
-    /** $_SERVER keys repopulated by __superglobals__refresh (issue #201, #235). */
+    /** $_SERVER keys repopulated by __superglobals__refresh (issue #201, #235, #296, #302). */
     private const RUNTIME_SERVER_KEYS = [
         'REQUEST_SCHEME',
         'HTTPS',
         'SERVER_PORT',
         'SERVER_NAME',
+        'DOCUMENT_ROOT',
+        'SCRIPT_FILENAME',
     ];
 
     public static function declareRefresh(Context $context): void

lib/Web/DevServer.php

@@ -105,6 +105,11 @@ public static function handleConnection($conn, string $docroot, callable $handle
             return;
         }
 
+        $scriptFilename = realpath($script);
+        if (false === $scriptFilename) {
+            $scriptFilename = $script;
+        }
+
         $requestUri = $scriptName.$pathInfo;
         if ('' !== $query) {
             $requestUri .= '?'.$query;
@@ -115,6 +120,7 @@ public static function handleConnection($conn, string $docroot, callable $handle
             'QUERY_STRING' => $query,
             'REQUEST_BODY' => $body,
             'SCRIPT_NAME' => $scriptName,
+            'SCRIPT_FILENAME' => $scriptFilename,
             'REQUEST_URI' => $requestUri,
             'DOCUMENT_ROOT' => $docroot,
         ];
@@ -129,6 +135,7 @@ public static function handleConnection($conn, string $docroot, callable $handle
         putenv('QUERY_STRING='.$query);
         putenv('REQUEST_BODY='.$body);
         putenv('SCRIPT_NAME='.$scriptName);
+        putenv('SCRIPT_FILENAME='.$scriptFilename);
         putenv('REQUEST_URI='.$requestUri);
         putenv('DOCUMENT_ROOT='.$docroot);
         if ('' !== $pathInfo) {

lib/Web/Superglobals.php

@@ -47,9 +47,13 @@ public static function isSuperglobalName(string $name): bool
     public static function populateFromEnvironment(
         Context $context,
         ?string $queryString = null,
-        ?string $postBody = null
+        ?string $postBody = null,
+        ?string $scriptFilename = null
     ): void {
         self::$activeContext = $context;
+        if (null !== $scriptFilename && '' !== $scriptFilename) {
+            putenv('SCRIPT_FILENAME='.$scriptFilename);
+        }
         if (null === $queryString) {
             $fromEnv = getenv('QUERY_STRING');
             $queryString = false === $fromEnv ? '' : $fromEnv;
@@ -161,6 +165,11 @@ private static function populateServer(
         self::setStringEntry($server, 'SCRIPT_NAME', $scriptName);
         self::setStringEntry($server, 'PHP_SELF', $scriptName);
 
+        $scriptFilename = self::resolveScriptFilename($scriptName);
+        if ('' !== $scriptFilename) {
+            self::setStringEntry($server, 'SCRIPT_FILENAME', $scriptFilename);
+        }
+
         $requestUri = getenv('REQUEST_URI');
         if (false === $requestUri || '' === $requestUri) {
             $requestUri = $scriptName;
@@ -456,6 +465,31 @@ private static function setScalarEntry(HashTable $ht, $key, $value): void
         }
     }
 
+    /**
+     * Resolve absolute filesystem path for the entry script (issue #302).
+     */
+    public static function resolveScriptFilename(?string $scriptName = null): string
+    {
+        $fromEnv = getenv('SCRIPT_FILENAME');
+        if (false !== $fromEnv && '' !== $fromEnv) {
+            return $fromEnv;
+        }
+
+        if (null === $scriptName) {
+            $scriptName = getenv('SCRIPT_NAME');
+            if (false === $scriptName || '' === $scriptName) {
+                $scriptName = '/index.php';
+            }
+        }
+
+        $documentRoot = getenv('DOCUMENT_ROOT');
+        if (false !== $documentRoot && '' !== $documentRoot) {
+            return rtrim($documentRoot, '/').$scriptName;
+        }
+
+        return '';
+    }
+
     /**
      * Derive PATH_INFO from REQUEST_URI path suffix after SCRIPT_NAME (front-controller pattern).
      */

test/aot/RuntimeSuperglobalRefreshTest.php

@@ -184,6 +184,60 @@ public function testCookieFromHttpCookieEnv(): void
         @unlink($outfile);
     }
 
+    public function testScriptFilenameFromCgiEnvironment(): void
+    {
+        $source = <<<'PHP'
+<?php
+declare(strict_types=1);
+header('Content-Type: text/plain; charset=UTF-8');
+echo $_SERVER['SCRIPT_FILENAME'];
+PHP;
+
+        $outfile = tempnam(sys_get_temp_dir(), 'phpc_script_fn_');
+        $this->assertNotFalse($outfile);
+        unlink($outfile);
+
+        $repoRoot = dirname(__DIR__, 2);
+        $env = $this->llvmProcessEnv($repoRoot);
+        $descriptorSpec = [
+            0 => ['pipe', 'r'],
+            1 => ['pipe', 'w'],
+            2 => ['pipe', 'w'],
+        ];
+
+        $compile = proc_open(
+            array_merge(
+                self::llvmEnvPrefix(),
+                self::phpCommand(),
+                [$this->compileBin, '-o', $outfile]
+            ),
+            $descriptorSpec,
+            $pipes,
+            $repoRoot,
+            $env
+        );
+        fwrite($pipes[0], $source);
+        fclose($pipes[0]);
+        $compileErr = stream_get_contents($pipes[2]);
+        fclose($pipes[1]);
+        fclose($pipes[2]);
+        proc_close($compile);
+        $this->assertFileExists($outfile, trim($compileErr !== false ? $compileErr : ''));
+
+        $root = realpath(sys_get_temp_dir());
+        $this->assertNotFalse($root);
+        $runEnv = $env;
+        unset($runEnv['SCRIPT_FILENAME']);
+        $runEnv['DOCUMENT_ROOT'] = $root;
+        $runEnv['SCRIPT_NAME'] = '/index.php';
+        $runEnv['REQUEST_URI'] = '/index.php';
+        $output = $this->runBinary($outfile, $runEnv);
+        $expected = $root.'/index.php';
+        $this->assertStringContainsString($expected, $this->cgiBody($output));
+
+        @unlink($outfile);
+    }
+
     public function testDocumentRootFromCgiEnvironment(): void
     {
         $source = <<<'PHP'
@@ -287,6 +341,13 @@ public function testHttpHostFromCgiEnvironment(): void
         @unlink($outfile);
     }
 
+    private function cgiBody(string $output): string
+    {
+        $parts = preg_split("/\r?\n\r?\n/", $output, 2);
+
+        return $parts[1] ?? $output;
+    }
+
     /**
      * @param array<string, string> $env
      */

test/fixtures/aot/cases/web_script_filename.phpt

@@ -0,0 +1,15 @@
+--TEST--
+AOT: SCRIPT_FILENAME from CGI env (issue #302)
+--ENV--
+REQUEST_METHOD=GET
+SCRIPT_NAME=/index.php
+REQUEST_URI=/index.php
+DOCUMENT_ROOT=/var/www/html
+SCRIPT_FILENAME=/var/www/html/index.php
+--FILE--
+<?php
+echo $_SERVER['SCRIPT_FILENAME'];
+--EXPECT--
+/var/www/html/index.php
+--EXPECT_EXIT--
+0

test/fixtures/web_echo_script_filename.php

@@ -0,0 +1,5 @@
+<?php
+
+declare(strict_types=1);
+
+echo $_SERVER['SCRIPT_FILENAME'];

test/real/ServeAotTest.php

@@ -33,6 +33,29 @@ protected function setUp(): void
         $this->phpCmd = self::phpCommand();
     }
 
+    public function testServeAotPopulatesScriptFilename(): void
+    {
+        $docroot = $this->makeDocroot([
+            'script.php' => <<<'PHP'
+<?php
+declare(strict_types=1);
+header('Content-Type: text/plain; charset=UTF-8');
+echo $_SERVER['SCRIPT_FILENAME'];
+PHP,
+        ]);
+        $script = realpath($docroot.'/script.php');
+        $this->assertNotFalse($script);
+        $binaryDir = sys_get_temp_dir().'/phpc_serve_aot_sf_'.bin2hex(random_bytes(4));
+        $this->assertTrue(mkdir($binaryDir));
+        $binary = $binaryDir.'/app';
+        $this->compileExample($docroot.'/script.php', $binary);
+        $response = $this->httpGetAot($docroot, $binary, '/script.php');
+        $this->assertStringContainsString('HTTP/1.1 200', $response);
+        $this->assertStringContainsString($script, $response);
+        @unlink($binary);
+        @rmdir($binaryDir);
+    }
+
     public function testServeAotPopulatesDocumentRoot(): void
     {
         $docroot = $this->makeDocroot([

test/real/ServeTest.php

@@ -119,6 +119,21 @@ public function testPopulatesDocumentRoot(): void
     $this->assertStringContainsString($resolved, $response);
   }
 
+  public function testPopulatesScriptFilename(): void
+  {
+    $docroot = $this->makeDocroot([
+      'script.php' => <<<'PHP'
+<?php
+echo $_SERVER['SCRIPT_FILENAME'];
+PHP,
+    ]);
+    $script = realpath($docroot.'/script.php');
+    $this->assertNotFalse($script);
+    $response = $this->httpGet($docroot, '/script.php');
+    $this->assertStringContainsString('HTTP/1.1 200', $response);
+    $this->assertStringContainsString($script, $response);
+  }
+
   public function testPopulatesCookieFromRequestHeader(): void
   {
     $docroot = $this->makeDocroot([

test/real/cases/web_script_filename.phpt

@@ -0,0 +1,15 @@
+--TEST--
+Web: SCRIPT_FILENAME from CGI env (issue #302)
+--ENV--
+REQUEST_METHOD=GET
+SCRIPT_NAME=/index.php
+REQUEST_URI=/index.php
+DOCUMENT_ROOT=/var/www/html
+SCRIPT_FILENAME=/var/www/html/index.php
+--FILE--
+<?php
+echo $_SERVER['SCRIPT_FILENAME'];
+--EXPECT--
+/var/www/html/index.php
+--EXPECT_EXIT--
+0