- use php's stream in libmagic

- remove duplicate code (like mime_content_type() compatibility function
  now use the same base as finfo_file())
- make it portable (works now on windows too, belongs other OSes)

- ([20:28]  <lsmith> Pierre: ok please commit)

ext/fileinfo/fileinfo.c

@@ -38,6 +38,10 @@
 #include "php_fileinfo.h"
 #include "fopen_wrappers.h" /* needed for is_url */
 
+#ifndef _S_IFDIR
+# define _S_IFDIR		S_IFDIR
+#endif
+
 /* {{{ macros and type definitions */
 struct php_fileinfo {
 	long options;
@@ -405,83 +409,148 @@ PHP_FUNCTION(finfo_set_flags)
 }
 /* }}} */
 
-static void _php_finfo_get_type(INTERNAL_FUNCTION_PARAMETERS, int mode) /* {{{ */
+#define FILEINFO_MODE_BUFFER 0
+#define FILEINFO_MODE_STREAM 1
+#define FILEINFO_MODE_FILE 2
+
+static void _php_finfo_get_type(INTERNAL_FUNCTION_PARAMETERS, int mode, int mimetype_emu) /* {{{ */
 {
 	long options = 0;
-	char *tmp, *ret_val, *buffer = NULL;
+	char *ret_val = NULL, *buffer = NULL;
 	int buffer_len;
 	struct php_fileinfo *finfo;
 	zval *zfinfo, *zcontext = NULL;
+	zval *what;
+	char mime_directory[] = "directory";
+
+	struct magic_set *magic = NULL;
 	FILEINFO_DECLARE_INIT_OBJECT(object)
 
-	if (object) {
+	if (mimetype_emu) {
+
+		/* mime_content_type(..) emulation */
+		if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "z", &what) == FAILURE) {
+			return;
+		}
+
+		switch (Z_TYPE_P(what)) {
+			case IS_STRING:
+				buffer = Z_STRVAL_P(what);
+				buffer_len = Z_STRLEN_P(what);
+				break;
+
+			case IS_RESOURCE:
+				mode = FILEINFO_MODE_STREAM;
+				break;
+
+			default:
+				php_error_docref(NULL TSRMLS_CC, E_WARNING, "Can only process string or stream arguments");
+				RETURN_FALSE;
+		}
+
+		magic = magic_open(MAGIC_MIME);
+		if (magic_load(magic, NULL) == -1) {
+			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to load magic database.");
+			magic_close(magic);
+			RETURN_FALSE;
+		}
+
+	} else if (object) {
 		if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|lr", &buffer, &buffer_len, &options, &zcontext) == FAILURE) {
 			RETURN_FALSE;
 		}
 		FILEINFO_FROM_OBJECT(finfo, object);
+		magic = finfo->magic;
 	} else {
 		if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rs|lr", &zfinfo, &buffer, &buffer_len, &options, &zcontext) == FAILURE) {
 			RETURN_FALSE;
 		}
 		ZEND_FETCH_RESOURCE(finfo, struct php_fileinfo *, &zfinfo, -1, "file_info", le_fileinfo);
+		magic = finfo->magic;
 	}	
 
 	/* Set options for the current file/buffer. */
 	if (options) {
-		FINFO_SET_OPTION(finfo->magic, options)
+		FINFO_SET_OPTION(magic, options)
 	}
 
-	if (mode) { /* file */
+	switch (mode) {
+		case FILEINFO_MODE_BUFFER:
+		{
+			ret_val = (char *) magic_buffer(magic, buffer, buffer_len);
+			break;
+		}
+
+		case FILEINFO_MODE_STREAM:
+		{
+				php_stream *stream;
+				off_t streampos;
+
+				php_stream_from_zval_no_verify(stream, &what);
+				if (!stream) {
+					goto common;
+				}
+
+				streampos = php_stream_tell(stream); /* remember stream position for restoration */
+				php_stream_seek(stream, 0, SEEK_SET);
+
+				ret_val = (char *) magic_stream(magic, stream);
+				break;
+		}
+
+		case FILEINFO_MODE_FILE:
+		{
 			/* determine if the file is a local file or remote URL */
 			char *tmp2;
-		php_stream_wrapper *wrap = php_stream_locate_url_wrapper(buffer, &tmp2, 0 TSRMLS_CC);
+			php_stream_wrapper *wrap;
-		if (wrap && wrap->is_url) {
+			struct stat sb;
-#ifdef ZEND_ENGINE_2
+
+			if (buffer_len < 1) {
+				php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty filename or path");
+				RETURN_FALSE;
+			}
+
+			if (php_sys_stat(buffer, &sb) == 0 && (sb.st_mode & _S_IFDIR)) {
+				ret_val = mime_directory;
+				goto common;
+			}
+
+			wrap = php_stream_locate_url_wrapper(buffer, &tmp2, 0 TSRMLS_CC);
+
+			if (wrap) {
 				php_stream_context *context = php_stream_context_from_zval(zcontext, 0);
-#else 
+
-			php_stream_context *context = NULL;
+				php_stream *stream = php_stream_open_wrapper_ex(buffer, "rb", ENFORCE_SAFE_MODE | REPORT_ERRORS, NULL, context);
-#endif			
+
-			php_stream *stream = php_stream_open_wrapper_ex(buffer, "rb", 
-					ENFORCE_SAFE_MODE | REPORT_ERRORS, NULL, context);
 				if (!stream) {
 					RETURN_FALSE;
 				}
-			buffer_len = php_stream_copy_to_mem(stream, &tmp, HOWMANY, 0);
+
+				ret_val = magic_stream(magic, stream);
 				php_stream_close(stream);
-
-			if (buffer_len == 0) {
-				RETURN_FALSE;
 			}
-		} else { /* local file */
+			break;
-			char resolved_path[MAXPATHLEN];
-
-			if (*buffer && VCWD_REALPATH(buffer, resolved_path)) {
-				if ((PG(safe_mode) && (!php_checkuid(resolved_path, NULL, CHECKUID_CHECK_FILE_AND_DIR))) || php_check_open_basedir(resolved_path TSRMLS_CC)) {
-					RETURN_FALSE;
-				}
-				ret_val = (char *) magic_file(finfo->magic, resolved_path);
-			} else {
-				RETURN_FALSE;
-			}
-			goto common;
-		}
-	} else { /* buffer */
-		tmp = buffer;
 		}
 
-	ret_val = (char *) magic_buffer(finfo->magic, tmp, buffer_len);
+		default:
-	if (mode) {
+			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Can only process string or stream arguments");
-		efree(tmp);
+			RETURN_FALSE;
 	}
+
 common:
+	if (mimetype_emu) {
+		if (magic) {
+			magic_close(magic);
+		}
+	}
+
 	/* Restore options */
 	if (options) {
-		FINFO_SET_OPTION(finfo->magic, finfo->options)
+		FINFO_SET_OPTION(magic, finfo->options)
 	}
 
 	if (!ret_val) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed identify data %d:%s",
+		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed identify data %d:%s", magic_errno(magic), magic_error(magic));
-				magic_errno(finfo->magic), magic_error(finfo->magic));
 		RETURN_FALSE;
 	} else {
 		RETURN_STRING(ret_val, 1);
@@ -493,7 +562,7 @@ common:
    Return information about a file. */
 PHP_FUNCTION(finfo_file)
 {
-	_php_finfo_get_type(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
+	_php_finfo_get_type(INTERNAL_FUNCTION_PARAM_PASSTHRU, FILEINFO_MODE_FILE, 0);
 }
 /* }}} */
 
@@ -501,7 +570,7 @@ PHP_FUNCTION(finfo_file)
    Return infromation about a string buffer. */
 PHP_FUNCTION(finfo_buffer)
 {
-	_php_finfo_get_type(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
+	_php_finfo_get_type(INTERNAL_FUNCTION_PARAM_PASSTHRU, FILEINFO_MODE_BUFFER, 0);
 }
 /* }}} */
 
@@ -509,90 +578,7 @@ PHP_FUNCTION(finfo_buffer)
    Return content-type for file */
 PHP_FUNCTION(mime_content_type)
 {
-	zval *what;
+	_php_finfo_get_type(INTERNAL_FUNCTION_PARAM_PASSTHRU, -1, 1);
-	magic_t magic;
-	char *tmp, *ret_val;
-	int buffer_len;
-	char *tmp2;
-	php_stream_wrapper *wrap; 
-	zval *zcontext = NULL;
-
-	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "z", &what) == FAILURE) {
-		return;
-	}
-
-	RETVAL_FALSE;
-
-	magic = magic_open(MAGIC_MIME);
-	if (magic_load(magic, NULL) == -1) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to load magic database.");
-		goto cleanup;
-	}
-
-	switch (Z_TYPE_P(what)) {
-		case IS_STRING:
-			wrap = php_stream_locate_url_wrapper(Z_STRVAL_P(what), &tmp2, 0 TSRMLS_CC);
-			/* determine if the file is a local file or remote URL */
-			if (wrap && wrap->is_url) {
-				php_stream_context *context = php_stream_context_from_zval(zcontext, 0);
-				php_stream *stream = php_stream_open_wrapper_ex(Z_STRVAL_P(what), "rb", 
-						ENFORCE_SAFE_MODE | REPORT_ERRORS, NULL, context);
-				if (!stream) {
-					goto cleanup;
-				}
-				buffer_len = php_stream_copy_to_mem(stream, &tmp, 4096, 0);
-				php_stream_close(stream);
-
-				if (buffer_len == 0) {
-					goto cleanup;
-				}
-				ret_val = (char *) magic_buffer(magic, tmp, buffer_len);
-			} else { /* local file */
-				char resolved_path[MAXPATHLEN];
-
-				if (*Z_STRVAL_P(what) && VCWD_REALPATH(Z_STRVAL_P(what), resolved_path)) {
-					if ((PG(safe_mode) && (!php_checkuid(resolved_path, NULL, CHECKUID_CHECK_FILE_AND_DIR))) || php_check_open_basedir(resolved_path TSRMLS_CC)) {
-						goto cleanup;
-					}
-					ret_val = (char *) magic_file(magic, resolved_path);
-				} else {
-					goto cleanup;
-				}
-			}
-			break;
-		case IS_RESOURCE:
-			{
-				php_stream *stream;
-				off_t streampos;
-
-				php_stream_from_zval_no_verify(stream, &what);
-				if (!stream) {
-					goto cleanup;
-				}
-				streampos = php_stream_tell(stream); /* remember stream position for restoration */
-				php_stream_seek(stream, 0, SEEK_SET);
-
-				buffer_len = php_stream_copy_to_mem(stream, &tmp, 4096, 0);
-		        php_stream_seek(stream, streampos, SEEK_SET); 
-
-				if (buffer_len == 0) {
-					goto cleanup;
-				}
-				ret_val = (char *) magic_buffer(magic, tmp, buffer_len);
-			}
-			break;
-		default:
-			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Can only process string or stream arguments");
-			goto cleanup;
-	}
-	if (!ret_val) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed identify data %d:%s",
-				magic_errno(magic), magic_error(magic));
-	} else {
-		RETVAL_STRING(ret_val, 1);
-	}
-cleanup:
-	magic_close(magic);
 }
 /* }}} */
 

ext/fileinfo/libmagic/apprentice.c

@@ -29,21 +29,32 @@
  * apprentice - make one pass through /etc/magic, learning its secrets.
  */
 
+#include "php.h"
+
 #include "file.h"
 #include "magic.h"
 #include "patchlevel.h"
 #include <stdlib.h>
-//#ifdef HAVE_UNISTD_H
+
+#ifdef PHP_WIN32
+#include "win32/unistd.h"
+#define strtoull _strtoui64
+#else
 #include <unistd.h>
-//#endif
+#endif
+
 #include <string.h>
 #include <assert.h>
 #include <ctype.h>
 #include <fcntl.h>
 #include <sys/stat.h>
+#ifndef PHP_WIN32
 #include <sys/param.h>
+#endif
 #include <sys/types.h>
+#ifndef PHP_WIN32
 #include <dirent.h>
+#endif
 
 #ifndef	lint
 FILE_RCSID("@(#)$File: apprentice.c,v 1.132 2008/03/28 18:19:30 christos Exp $")
@@ -174,6 +185,10 @@ static const struct type_tbl_s {
 # undef XX_NULL
 };
 
+#ifndef S_ISDIR
+#define S_ISDIR(mode) ((mode) & _S_IFDIR)
+#endif
+
 private int
 get_type(const char *l, const char **t)
 {
@@ -528,30 +543,43 @@ private void
 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
    struct magic_entry **marray, uint32_t *marraycount)
 {
-	char line[BUFSIZ];
+	char buffer[BUFSIZ + 1];
+	char *line;
+	size_t line_len;
 	size_t lineno = 0;
-	FILE *f = fopen(ms->file = fn, "r");
+
-	if (f == NULL) {
+	php_stream *stream;
+
+	TSRMLS_FETCH();
+
+#if (PHP_MAJOR_VERSION < 6)
+	stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
+#else
+	stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
+#endif
+
+	if (stream == NULL) {
 		if (errno != ENOENT)
 			file_error(ms, errno, "cannot read magic file `%s'",
 				   fn);
 		(*errs)++;
 	} else {
+
 		/* read and parse this file */
-		for (ms->line = 1; fgets(line, sizeof(line), f) != NULL; ms->line++) {
+		for (ms->line = 1; (line = php_stream_get_line(stream, buffer , BUFSIZ, &line_len)) != NULL; ms->line++) {
-			size_t len;
+			if (line_len == 0 ||		/* null line, garbage, etc */
-			len = strlen(line);
+					line[0] == '\0' ||	/* empty*/
-			if (len == 0) /* null line, garbage, etc */
+					line[0] == '#' || 	/* comment */
+					line[0] == '\n' || line[0] == '\r') {	/* New Line */
 				continue;
-			if (line[len - 1] == '\n') {
-				lineno++;
-				line[len - 1] = '\0'; /* delete newline */
 			}
-			if (line[0] == '\0')	/* empty, do not parse */
+
-				continue;
+			if (line[line_len - 1] == '\n') {
-			if (line[0] == '#')	/* comment, do not parse */
+				lineno++;
-				continue;
+				line[line_len - 1] = '\0'; /* delete newline */
-			if (len > mime_marker_len &&
+			}
+
+			if (line_len > mime_marker_len &&
 			    memcmp(line, mime_marker, mime_marker_len) == 0) {
 				/* MIME type */
 				if (parse_mime(ms, marray, marraycount,
@@ -563,7 +591,7 @@ load_1(struct magic_set *ms, int action, const char *fn, int *errs,
 				(*errs)++;
 		}
 
-		(void)fclose(f);
+		php_stream_close(stream);
 	}
 }
 
@@ -594,7 +622,7 @@ apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
 		(void)fprintf(stderr, "%s\n", usg_hdr);
 
 	/* load directory or file */
-	if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
+	if (php_sys_stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
 		dir = opendir(fn);
 		if (dir) {
 			while ((d = readdir(dir))) {
@@ -1816,17 +1844,20 @@ private int
 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
     const char *fn)
 {
-	int fd;
-	struct stat st;
 	uint32_t *ptr;
 	uint32_t version;
 	int needsbyteswap;
 	char *dbname = NULL;
 	void *mm = NULL;
 	int   ret = 0;
+	php_stream *stream;
+	php_stream_statbuf st;
+
+
+	TSRMLS_FETCH();
 
 	if (fn == NULL) {
-		mm = &php_magic_database;
+		mm = (void *)&php_magic_database;
 		ret = 3;
 		goto internal_loaded;
 	}
@@ -1835,27 +1866,34 @@ apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
 	if (dbname == NULL)
 		goto error2;
 
-	if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
+#if (PHP_MAJOR_VERSION < 6)
-		goto error2;
+		stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
+#else
+		stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
+#endif
 
-	if (fstat(fd, &st) == -1) {
+	if (!stream) {
+		goto error2;
+	}
+
+	if (php_stream_stat(stream, &st) < 0) {
 		file_error(ms, errno, "cannot stat `%s'", dbname);
 		goto error1;
 	}
-	if (st.st_size < 8) {
+
+	if (st.sb.st_size < 8) {
 		file_error(ms, 0, "file `%s' is too small", dbname);
 		goto error1;
 	}
 
-	mm = emalloc((size_t)st.st_size);
+	mm = emalloc((size_t)st.sb.st_size);
-	if (read(fd, mm, (size_t)st.st_size) != (size_t)st.st_size) {
+	if (php_stream_read(stream, mm, (size_t)st.sb.st_size) != (size_t)st.sb.st_size) {
 		file_badread(ms);
 		goto error1;
 	}
 	ret = 1;
 
-	(void)close(fd);
+	php_stream_close(stream);
-	fd = -1;
 
 internal_loaded:
 	*magicp = mm;
@@ -1882,7 +1920,7 @@ internal_loaded:
 	if (fn == NULL) {
 		*nmagicp = (sizeof(php_magic_database) / sizeof(struct magic));
 	} else {
-		*nmagicp = (uint32_t)(st.st_size / sizeof(struct magic));
+		*nmagicp = (uint32_t)(st.sb.st_size / sizeof(struct magic));
 	}
 	if (*nmagicp > 0) {
 		(*nmagicp)--;
@@ -1898,8 +1936,9 @@ internal_loaded:
 	return ret;
 
 error1:
-	if (fd != -1)
+	if (stream) {
-		(void)close(fd);
+		php_stream_close(stream);
+	}
 	if (mm) {
 		efree(mm);
 	} else {
@@ -1921,38 +1960,47 @@ private int
 apprentice_compile(struct magic_set *ms, struct magic **magicp,
     uint32_t *nmagicp, const char *fn)
 {
-	int fd;
 	char *dbname;
 	int rv = -1;
+	php_stream *stream;
+
+	TSRMLS_FETCH();
 
 	mkdbname(fn, &dbname, 1);
 
-	if (dbname == NULL) 
+	if (dbname == NULL) {
 		goto out;
+	}
 
-	if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
+/* wb+ == O_WRONLY|O_CREAT|O_TRUNC|O_BINARY */
+#if (PHP_MAJOR_VERSION < 6)
+	stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
+#else
+	stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS, NULL);
+#endif
+
+	if (!stream) {
 		file_error(ms, errno, "cannot open `%s'", dbname);
 		goto out;
 	}
 
-	if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
+	if (php_stream_write(stream, (char *)ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
 		file_error(ms, errno, "error writing `%s'", dbname);
 		goto out;
 	}
 
-	if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
+	if (php_stream_seek(stream,(off_t)sizeof(struct magic), SEEK_SET) != sizeof(struct magic)) {
-	    != sizeof(struct magic)) {
 		file_error(ms, errno, "error seeking `%s'", dbname);
 		goto out;
 	}
 
-	if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp)) 
+	if (php_stream_write(stream, (char *)*magicp, (sizeof(struct magic) * *nmagicp) != (ssize_t)(sizeof(struct magic) * *nmagicp))) {
-	    != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
 		file_error(ms, errno, "error writing `%s'", dbname);
 		goto out;
 	}
 
-	(void)close(fd);
+	php_stream_close(stream);
+
 	rv = 0;
 out:
 	efree(dbname);

ext/fileinfo/libmagic/compress.c

@@ -32,6 +32,7 @@
  *	uncompress(method, old, n, newch) - uncompress old into new, 
  *					    using method, return sizeof new
  */
+
 #include "config.h"
 #include "file.h"
 #include "magic.h"
@@ -43,7 +44,9 @@
 #include <string.h>
 #include <errno.h>
 #include <sys/types.h>
+#ifndef PHP_WIN32
 #include <sys/ioctl.h>
+#endif
 #ifdef HAVE_SYS_WAIT_H
 #include <sys/wait.h>
 #endif
@@ -55,6 +58,7 @@
 #include <zlib.h>
 #endif
 
+#undef FIONREAD
 
 #ifndef lint
 FILE_RCSID("@(#)$File: compress.c,v 1.56 2008/02/07 00:58:52 christos Exp $")
@@ -86,6 +90,7 @@ private size_t ncompr = sizeof(compr) / sizeof(compr[0]);
 
 
 private ssize_t swrite(int, const void *, size_t);
+#ifdef PHP_FILEINFO_UNCOMPRESS
 private size_t uncompressbuf(struct magic_set *, int, size_t,
     const unsigned char *, unsigned char **, size_t);
 #ifdef BUILTIN_DECOMPRESS
@@ -138,6 +143,7 @@ error:
 	ms->flags |= MAGIC_COMPRESS;
 	return rv;
 }
+#endif
 
 /*
  * `safe' write for sockets and pipes.
@@ -295,6 +301,7 @@ file_pipe2file(struct magic_set *ms, int fd, const void *startbuf,
 	return fd;
 }
 
+#ifdef PHP_FILEINFO_UNCOMPRESS
 #ifdef BUILTIN_DECOMPRESS
 
 #define FHCRC		(1 << 1)
@@ -365,6 +372,7 @@ uncompressgzipped(struct magic_set *ms, const unsigned char *old,
 }
 #endif
 
+
 private size_t
 uncompressbuf(struct magic_set *ms, int fd, size_t method,
     const unsigned char *old, unsigned char **newch, size_t n)
@@ -483,3 +491,4 @@ err:
 		return n;
 	}
 }
+#endif /* if PHP_FILEINFO_UNCOMPRESS */

ext/fileinfo/libmagic/file.h

@@ -33,9 +33,7 @@
 #ifndef __file_h__
 #define __file_h__
 
-//#ifdef HAVE_CONFIG_H
 #include "config.h"
-//#endif */
 
 #include <stdio.h>	/* Include that here, to make sure __P gets defined */
 #include <errno.h>
@@ -46,6 +44,9 @@
 #ifdef HAVE_INTTYPES_H
 #include <inttypes.h>
 #endif
+#ifdef PHP_WIN32
+#include "win32/php_stdint.h"
+#endif
 
 #include "php.h"
 #include "ext/standard/php_string.h"
@@ -62,7 +63,7 @@
 #define MAGIC "/etc/magic"
 #endif
 
-#ifdef __EMX__
+#if defined(__EMX__) || defined(PHP_WIN32)
 #define PATHSEP	';'
 #else
 #define PATHSEP	':'
@@ -323,16 +324,18 @@ typedef unsigned long unichar;
 
 struct stat;
 protected const char *file_fmttime(uint32_t, int);
-protected int file_buffer(struct magic_set *, int, const char *, const void *,
+protected int file_buffer(struct magic_set *, php_stream *, const char *, const void *,
     size_t);
-protected int file_fsmagic(struct magic_set *, const char *, struct stat *);
+protected int file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb, php_stream *stream);
 protected int file_pipe2file(struct magic_set *, int, const void *, size_t);
 protected int file_printf(struct magic_set *, const char *, ...);
 protected int file_reset(struct magic_set *);
 protected int file_tryelf(struct magic_set *, int, const unsigned char *,
     size_t);
+#ifdef PHP_FILEINFO_UNCOMPRESS 
 protected int file_zmagic(struct magic_set *, int, const char *,
     const unsigned char *, size_t);
+#endif
 protected int file_ascmagic(struct magic_set *, const unsigned char *, size_t);
 protected int file_is_tar(struct magic_set *, const unsigned char *, size_t);
 protected int file_softmagic(struct magic_set *, const unsigned char *, size_t, int);

ext/fileinfo/libmagic/fsmagic.c

@@ -60,69 +60,64 @@
 FILE_RCSID("@(#)$File: fsmagic.c,v 1.50 2008/02/12 17:22:54 rrt Exp $")
 #endif	/* lint */
 
-private int
+#ifdef PHP_WIN32
-bad_link(struct magic_set *ms, int err, char *buf)
+
-{
+# undef S_IFIFO
-	char *errfmt;
+#endif
-	if (err == ELOOP)
+
-		errfmt = "symbolic link in a loop";
+
-	else
+#ifndef S_ISDIR
-		errfmt = "broken symbolic link to `%s'";
+#define S_ISDIR(mode) ((mode) & _S_IFDIR)
-	if (ms->flags & MAGIC_ERROR) {
+#endif
-		file_error(ms, err, errfmt, buf);
+
-		return -1;
+#ifndef S_ISREG
-	} 
+#define S_ISREG(mode) ((mode) & _S_IFREG)
-	if (file_printf(ms, errfmt, buf) == -1)
+#endif
-		return -1;
-	return 1;
-}
 
 protected int
-file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb)
+file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb, php_stream *stream)
 {
-	int ret = 0;
 	int mime = ms->flags & MAGIC_MIME;
-#ifdef	S_IFLNK
+	TSRMLS_FETCH();
-	char buf[BUFSIZ+4];
-	int nch;
-	struct stat tstatbuf;
-#endif
 
-	if (fn == NULL)
+	if (!fn && !stream) {
-		return 0;
-
-	/*
-	 * Fstat is cheaper but fails for files you don't have read perms on.
-	 * On 4.2BSD and similar systems, use lstat() to identify symlinks.
-	 */
-#ifdef	S_IFLNK
-	if ((ms->flags & MAGIC_SYMLINK) == 0)
-		ret = lstat(fn, sb);
-	else
-#endif
-	ret = stat(fn, sb);	/* don't merge into if; see "ret =" above */
-
-	if (ret) {
-		if (ms->flags & MAGIC_ERROR) {
-			file_error(ms, errno, "cannot stat `%s'", fn);
 		return -1;
 	}
-		if (file_printf(ms, "cannot open `%s' (%s)",
+
-		    fn, strerror(errno)) == -1)
+	if (stream) {
-			return -1;
+		php_stream_statbuf ssb;
+		if (php_stream_stat(stream, &ssb) < 0) {
+			if (ms->flags & MAGIC_ERROR) {
+				file_error(ms, errno, "cannot stat `%s'", fn);
+			}
 			return 1;
 		}
+		memcpy(sb, &ssb.sb, sizeof(struct stat));
+	} else {
+		if (php_sys_stat(fn, sb) != 0) {
+			if (ms->flags & MAGIC_ERROR) {
+				file_error(ms, errno, "cannot stat `%s'", fn);
+			}
+			return 1;
+		}
+	}
 
 	if (mime) {
-		if ((sb->st_mode & S_IFMT) != S_IFREG) {
+			if (!S_ISREG(sb->st_mode)) {
 				if ((mime & MAGIC_MIME_TYPE) &&
-			    file_printf(ms, "application/x-not-regular-file")
+						file_printf(ms, "application/x-not-regular-file") == -1) {
-			    == -1)
 					return -1;
+				}
 				return 1;
 			}
+
+			if (S_ISDIR(sb->st_mode)) {
+				if (file_printf(ms, "directory") == -1) {
+					return -1;
 				}
-	else {
+				return 1;
+			}
+	} else {
 #ifdef S_ISUID
 		if (sb->st_mode & S_ISUID) 
 			if (file_printf(ms, "setuid ") == -1)
@@ -141,10 +136,7 @@ file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb)
 	}
 
 	switch (sb->st_mode & S_IFMT) {
-	case S_IFDIR:
+#ifndef PHP_WIN32
-		if (file_printf(ms, "directory") == -1)
-			return -1;
-		return 1;
 # ifdef S_IFCHR
 		case S_IFCHR:
 			/* 
@@ -152,52 +144,31 @@ file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb)
 			 * like ordinary files.  Otherwise, just report that they
 			 * are block special files and go on to the next file.
 			 */
-		if ((ms->flags & MAGIC_DEVICES) != 0)
+			if ((ms->flags & MAGIC_DEVICES) != 0) {
 				break;
+			}
 #  ifdef HAVE_STAT_ST_RDEV
 #   ifdef dv_unit
 			if (file_printf(ms, "character special (%d/%d/%d)",
 					major(sb->st_rdev), dv_unit(sb->st_rdev),
-		    dv_subunit(sb->st_rdev)) == -1)
+					dv_subunit(sb->st_rdev)) == -1) {
 				return -1;
+			}
 #   else
 			if (file_printf(ms, "character special (%ld/%ld)",
-		    (long) major(sb->st_rdev), (long) minor(sb->st_rdev)) == -1)
+					(long) major(sb->st_rdev), (long) minor(sb->st_rdev)) == -1) {
 				return -1;
+			}
 #   endif
 #  else
-		if (file_printf(ms, "character special") == -1)
+			if (file_printf(ms, "character special") == -1) {
 				return -1;
+			}
 #  endif
 			return 1;
 # endif
-#ifdef S_IFBLK
-	case S_IFBLK:
-		/* 
-		 * If -s has been specified, treat block special files
-		 * like ordinary files.  Otherwise, just report that they
-		 * are block special files and go on to the next file.
-		 */
-		if ((ms->flags & MAGIC_DEVICES) != 0)
-			break;
-#ifdef HAVE_STAT_ST_RDEV
-# ifdef dv_unit
-		if (file_printf(ms, "block special (%d/%d/%d)",
-		    major(sb->st_rdev), dv_unit(sb->st_rdev),
-		    dv_subunit(sb->st_rdev)) == -1)
-			return -1;
-# else
-		if (file_printf(ms, "block special (%ld/%ld)",
-		    (long)major(sb->st_rdev), (long)minor(sb->st_rdev)) == -1)
-			return -1;
 #endif
-#else
+			
-		if (file_printf(ms, "block special") == -1)
-			return -1;
-#endif
-		return 1;
-#endif
-	/* TODO add code to handle V7 MUX and Blit MUX files */
 #ifdef	S_IFIFO
 	case S_IFIFO:
 		if((ms->flags & MAGIC_DEVICES) != 0)
@@ -212,67 +183,17 @@ file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb)
 			return -1;
 		return 1;
 #endif
+
 #ifdef	S_IFLNK
 	case S_IFLNK:
-		if ((nch = readlink(fn, buf, BUFSIZ-1)) <= 0) {
+		/* stat is used, if it made here then the link is broken */
 			if (ms->flags & MAGIC_ERROR) {
-			    file_error(ms, errno, "unreadable symlink `%s'",
+			    file_error(ms, errno, "unreadable symlink `%s'", fn);
-				fn);
-			    return -1;
-			}
-			if (file_printf(ms,
-			    "unreadable symlink `%s' (%s)", fn,
-			    strerror(errno)) == -1)
-				return -1;
-			return 1;
-		}
-		buf[nch] = '\0';	/* readlink(2) does not do this */
-
-		/* If broken symlink, say so and quit early. */
-		if (*buf == '/') {
-			if (stat(buf, &tstatbuf) < 0)
-				return bad_link(ms, errno, buf);
-		} else {
-			char *tmp;
-			char buf2[BUFSIZ+BUFSIZ+4];
-
-			if ((tmp = strrchr(fn,  '/')) == NULL) {
-				tmp = buf; /* in current directory anyway */
-			} else {
-				if (tmp - fn + 1 > BUFSIZ) {
-					if (ms->flags & MAGIC_ERROR) {
-						file_error(ms, 0, 
-						    "path too long: `%s'", buf);
-						return -1;
-					}
-					if (file_printf(ms,
-					    "path too long: `%s'", fn) == -1)
-						return -1;
-					return 1;
-				}
-				(void)strcpy(buf2, fn);  /* take dir part */
-				buf2[tmp - fn + 1] = '\0';
-				(void)strcat(buf2, buf); /* plus (rel) link */
-				tmp = buf2;
-			}
-			if (stat(tmp, &tstatbuf) < 0)
-				return bad_link(ms, errno, buf);
-		}
-
-		/* Otherwise, handle it. */
-		if ((ms->flags & MAGIC_SYMLINK) != 0) {
-			const char *p;
-			ms->flags &= MAGIC_SYMLINK;
-			p = magic_file(ms, buf);
-			ms->flags |= MAGIC_SYMLINK;
-			return p != NULL ? 1 : -1;
-		} else { /* just print what it points to */
-			if (file_printf(ms, "symbolic link to `%s'",
-			    buf) == -1)
 			    return -1;
 			}
 	return 1;
 #endif
+
 #ifdef	S_IFSOCK
 #ifndef __COHERENT__
 	case S_IFSOCK:
@@ -281,8 +202,10 @@ file_fsmagic(struct magic_set *ms, const char *fn, struct stat *sb)
 		return 1;
 #endif
 #endif
+
 		case S_IFREG:
 			break;
+
 		default:
 			file_error(ms, 0, "invalid mode 0%o", sb->st_mode);
 			return -1;

ext/fileinfo/libmagic/funcs.c

@@ -146,7 +146,7 @@ file_badread(struct magic_set *ms)
 }
 
 protected int
-file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
+file_buffer(struct magic_set *ms, php_stream *stream, const char *inname, const void *buf,
     size_t nb)
 {
 	int m;
@@ -166,7 +166,7 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 		return 1;
 	}
 
-#ifdef __EMX__
+#if defined(__EMX__)
 	if ((ms->flags & MAGIC_NO_CHECK_APPTYPE) == 0 && inname) {
 		switch (file_os2_apptype(ms, inname, buf, nb)) {
 		case -1:
@@ -179,31 +179,30 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 	}
 #endif
 
+#if PHP_FILEINFO_UNCOMPRESS
 	/* try compression stuff */
 	if ((ms->flags & MAGIC_NO_CHECK_COMPRESS) != 0 ||
-	    (m = file_zmagic(ms, fd, inname, buf, nb)) == 0) {
+	    (m = file_zmagic(ms, stream, inname, buf, nb)) == 0) 
+#endif
+	{
 		/* Check if we have a tar file */
-	    if ((ms->flags & MAGIC_NO_CHECK_TAR) != 0 ||
+		if ((ms->flags & MAGIC_NO_CHECK_TAR) != 0 || (m = file_is_tar(ms, buf, nb)) == 0) {
-		(m = file_is_tar(ms, buf, nb)) == 0) {
 			/* try tests in /etc/magic (or surrogate magic file) */
-		if ((ms->flags & MAGIC_NO_CHECK_SOFT) != 0 ||
+			if ((ms->flags & MAGIC_NO_CHECK_SOFT) != 0 || (m = file_softmagic(ms, buf, nb, BINTEST)) == 0) {
-		    (m = file_softmagic(ms, buf, nb, BINTEST)) == 0) {
 				/* try known keywords, check whether it is ASCII */
-		    if ((ms->flags & MAGIC_NO_CHECK_ASCII) != 0 ||
+				if ((ms->flags & MAGIC_NO_CHECK_ASCII) != 0 || (m = file_ascmagic(ms, buf, nb)) == 0) {
-			(m = file_ascmagic(ms, buf, nb)) == 0) {
 					/* abandon hope, all ye who remain here */
-			if ((!mime || (mime & MAGIC_MIME_TYPE)) &&
+					if ((!mime || (mime & MAGIC_MIME_TYPE)) && file_printf(ms, mime ? "application/octet-stream" : "data") == -1) {
-			    file_printf(ms, mime ? "application/octet-stream" :
-				"data") == -1)
 						return -1;
+					}
 					m = 1;
 				}
 			}
 		}
 	}
+
 #ifdef BUILTIN_ELF
-	if ((ms->flags & MAGIC_NO_CHECK_ELF) == 0 && m == 1 &&
+	if ((ms->flags & MAGIC_NO_CHECK_ELF) == 0 && m == 1 && nb > 5 && fd != -1) {
-	    nb > 5 && fd != -1) {
 		/*
 		 * We matched something in the file, so this *might*
 		 * be an ELF file, and the file is at least 5 bytes
@@ -212,7 +211,7 @@ file_buffer(struct magic_set *ms, int fd, const char *inname, const void *buf,
 		 * information from the ELF headers that cannot easily
 		 * be extracted with rules in the magic file.
 		 */
-		(void)file_tryelf(ms, fd, buf, nb);
+		(void)file_tryelf(ms, stream, buf, nb);
 	}
 #endif
 	return m;

ext/fileinfo/libmagic/magic.c

@@ -30,10 +30,19 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#ifdef PHP_WIN32
+#include "win32/unistd.h"
+#else
 #include <unistd.h>
+#endif
 #include <string.h>
 #include <sys/types.h>
-#include <sys/param.h>	/* for MAXPATHLEN */
+#ifdef PHP_WIN32
+# include "config.w32.h"
+#else
+# include "php_config.h"
+#endif
+
 #include <sys/stat.h>
 #include <limits.h>	/* for PIPE_BUF */
 
@@ -54,9 +63,9 @@
 #ifdef HAVE_LOCALE_H
 #include <locale.h>
 #endif
-
+#ifndef PHP_WIN32
 # include <netinet/in.h>		/* for byte swapping */
-
+#endif
 #include "patchlevel.h"
 
 #ifndef	lint
@@ -72,6 +81,11 @@ FILE_RCSID("@(#)$File: magic.c,v 1.50 2008/02/19 00:58:59 rrt Exp $")
 #endif
 #endif
 
+#ifdef PHP_WIN32
+# undef S_IFLNK
+# undef S_IFIFO
+#endif
+
 #ifdef __EMX__
 private char *apptypeName = NULL;
 protected int file_os2_apptype(struct magic_set *ms, const char *fn,
@@ -82,7 +96,7 @@ private void free_mlist(struct mlist *);
 private void close_and_restore(const struct magic_set *, const char *, int,
     const struct stat *);
 private int info_from_stat(struct magic_set *, mode_t);
-private const char *file_or_fd(struct magic_set *, const char *, int);
+private const char *file_or_stream(struct magic_set *, const char *, php_stream *);
 
 #ifndef	STDIN_FILENO
 #define	STDIN_FILENO	0
@@ -204,10 +218,6 @@ private void
 close_and_restore(const struct magic_set *ms, const char *name, int fd,
     const struct stat *sb)
 {
-	if (fd == STDIN_FILENO)
-		return;
-	(void) close(fd);
-
 	if ((ms->flags & MAGIC_PRESERVE_ATIME) != 0) {
 		/*
 		 * Try to restore access, modification times if read it.
@@ -240,7 +250,7 @@ close_and_restore(const struct magic_set *ms, const char *name, int fd,
 public const char *
 magic_descriptor(struct magic_set *ms, int fd)
 {
-	return file_or_fd(ms, NULL, fd);
+	return file_or_stream(ms, NULL, NULL);
 }
 
 /*
@@ -249,17 +259,28 @@ magic_descriptor(struct magic_set *ms, int fd)
 public const char *
 magic_file(struct magic_set *ms, const char *inname)
 {
-	return file_or_fd(ms, inname, STDIN_FILENO);
+	return file_or_stream(ms, inname, NULL);
+}
+
+public const char *
+magic_stream(struct magic_set *ms, php_stream *stream)
+{
+	return file_or_stream(ms, NULL, stream);
 }
 
 private const char *
-file_or_fd(struct magic_set *ms, const char *inname, int fd)
+file_or_stream(struct magic_set *ms, const char *inname, php_stream *stream)
 {
 	int	rv = -1;
 	unsigned char *buf;
 	struct stat	sb;
 	ssize_t nbytes = 0;	/* number of bytes read from a datafile */
 	int	ispipe = 0;
+	TSRMLS_FETCH();
+
+	if (!inname && !stream) {
+		return NULL;
+	}
 
 	/*
 	 * one extra for terminating '\0', and
@@ -268,10 +289,11 @@ file_or_fd(struct magic_set *ms, const char *inname, int fd)
 #define SLOP (1 + sizeof(union VALUETYPE))
 	buf = emalloc(HOWMANY + SLOP);
 
-	if (file_reset(ms) == -1)
+	if (file_reset(ms) == -1) {
 		goto done;
+	}
 
-	switch (file_fsmagic(ms, inname, &sb)) {
+	switch (file_fsmagic(ms, inname, &sb, stream)) {
 		case -1:		/* error */
 			goto done;
 		case 0:			/* nothing found */
@@ -281,76 +303,51 @@ file_or_fd(struct magic_set *ms, const char *inname, int fd)
 			goto done;
 	}
 
-	if (inname == NULL) {
+		errno = 0;
-		if (fstat(fd, &sb) == 0 && S_ISFIFO(sb.st_mode))
-			ispipe = 1;
-	} else {
-		int flags = O_RDONLY|O_BINARY;
 
-		if (stat(inname, &sb) == 0 && S_ISFIFO(sb.st_mode)) {
+		if (!stream && inname) {
-			flags |= O_NONBLOCK;
+#if (PHP_MAJOR_VERSION < 6)
-			ispipe = 1;
+		stream = php_stream_open_wrapper(inname, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
+#else
+		stream = php_stream_open_wrapper(inname, "rb", REPORT_ERRORS, NULL);
+#endif
 		}
 
-		errno = 0;
+		if (!stream) {
-		if ((fd = open(inname, flags)) < 0) {
-#ifdef __CYGWIN__
-			/* FIXME: Do this with EXEEXT from autotools */
-			char *tmp = alloca(strlen(inname) + 5);
-			(void)strcat(strcpy(tmp, inname), ".exe");
-			if ((fd = open(tmp, flags)) < 0) {
-#endif
 				fprintf(stderr, "couldn't open file\n");
 				if (info_from_stat(ms, sb.st_mode) == -1)
 					goto done;
 				rv = 0;
 				goto done;
-#ifdef __CYGWIN__
-			}
-#endif
 		}
+
 #ifdef O_NONBLOCK
-		if ((flags = fcntl(fd, F_GETFL)) != -1) {
+/* we should be already be in non blocking mode for network socket 
-			flags &= ~O_NONBLOCK;
+ * leaving the comment/#ifdef as documentation
-			(void)fcntl(fd, F_SETFL, flags);
+ */
-		}
 #endif
-	}
+
 
 	/*
 	 * try looking at the first HOWMANY bytes
 	 */
-	if (ispipe) {
+	if ((nbytes = php_stream_read(stream, (char *)buf, HOWMANY)) < 0) {
-		ssize_t r = 0;
-
-		while ((r = sread(fd, (void *)&buf[nbytes],
-		    (size_t)(HOWMANY - nbytes), 1)) > 0) {
-			nbytes += r;
-			if (r < PIPE_BUF) break;
-		}
-
-		if (nbytes == 0) {
-			/* We can not read it, but we were able to stat it. */
-			if (info_from_stat(ms, sb.st_mode) == -1)
-				goto done;
-			rv = 0;
-			goto done;
-		}
-
-	} else {
-		if ((nbytes = read(fd, (char *)buf, HOWMANY)) == -1) {
 		file_error(ms, errno, "cannot read `%s'", inname);
 		goto done;
 	}
-	}
 
 	(void)memset(buf + nbytes, 0, SLOP); /* NUL terminate */
-	if (file_buffer(ms, fd, inname, buf, (size_t)nbytes) == -1)
+	if (file_buffer(ms, stream, inname, buf, (size_t)nbytes) == -1)
 		goto done;
 	rv = 0;
 done:
 	efree(buf);
-	close_and_restore(ms, inname, fd, &sb);
+
+	if (stream) {
+		php_stream_close(stream);
+	}
+
+	close_and_restore(ms, inname, 0, &sb);
 	return rv == 0 ? file_getbuffer(ms) : NULL;
 }
 
@@ -364,7 +361,7 @@ magic_buffer(struct magic_set *ms, const void *buf, size_t nb)
 	 * The main work is done here!
 	 * We have the file name and/or the data buffer to be identified. 
 	 */
-	if (file_buffer(ms, -1, NULL, buf, nb) == -1) {
+	if (file_buffer(ms, NULL, NULL, buf, nb) == -1) {
 		return NULL;
 	}
 	return file_getbuffer(ms);

ext/fileinfo/libmagic/magic.h

@@ -64,6 +64,7 @@ magic_t magic_open(int);
 void magic_close(magic_t);
 
 const char *magic_file(magic_t, const char *);
+const char *magic_stream(magic_t, php_stream *);
 const char *magic_descriptor(magic_t, int);
 const char *magic_buffer(magic_t, const void *, size_t);
 

ext/fileinfo/libmagic/readelf.c

@@ -943,7 +943,6 @@ dophn_exec(struct magic_set *ms, int class, int swap, int fd, off_t off,
 	return 0;
 }
 
-
 protected int
 file_tryelf(struct magic_set *ms, int fd, const unsigned char *buf,
     size_t nbytes)
@@ -963,6 +962,7 @@ file_tryelf(struct magic_set *ms, int fd, const unsigned char *buf,
 
 	if (ms->flags & MAGIC_MIME)
 		return 0;
+
 	/*
 	 * ELF executables have multiple section headers in arbitrary
 	 * file locations and thus file(1) cannot determine it from easily.

ext/fileinfo/tests/finfo_file_001.phpt

@@ -11,9 +11,15 @@ var_dump(finfo_file($fp, '.'));
 var_dump(finfo_file($fp, '&'));
 
 ?>
---EXPECT--
+--EXPECTF--
 bool(false)
+
+Warning: finfo_file(): Empty filename or path in %s on line %d
 bool(false)
+
+Warning: finfo_file(): Empty filename or path in %s on line %d
 bool(false)
 string(9) "directory"
+
+Warning: finfo_file(&): failed to open stream: No such file or directory in %s on line %d
 bool(false)