add unfinished experimental tar support (may not lead anywhere)

7 files changed, 499 insertions, 5 deletions

diff --git a/ext/phar/config.m4 b/ext/phar/config.m4
index 75c26eae52..4782889d2f 100644
--- a/ext/phar/config.m4
+++ b/ext/phar/config.m4
@@ -5,7 +5,7 @@ PHP_ARG_ENABLE(phar, for phar support/phar zlib support,
 [  --enable-phar           Enable phar support, use --with-zlib-dir if zlib detection fails])
 
 if test "$PHP_PHAR" != "no"; then
-  PHP_NEW_EXTENSION(phar, phar.c phar_object.c phar_path_check.c, $ext_shared)
+  PHP_NEW_EXTENSION(phar, phar.c phar_object.c phar_path_check.c phar2.c phar2_openhash.c, $ext_shared)
   PHP_ADD_EXTENSION_DEP(phar, zlib, false)
   PHP_ADD_EXTENSION_DEP(phar, bz2, false)
   PHP_ADD_EXTENSION_DEP(phar, spl, false)
diff --git a/ext/phar/phar.c b/ext/phar/phar.c
index 436943b36c..1183a799ab 100644
--- a/ext/phar/phar.c
+++ b/ext/phar/phar.c
@@ -21,6 +21,7 @@
 
 #define PHAR_MAIN
 #include "phar_internal.h"
+#include "phar2.h"
 #include "SAPI.h"
 
 ZEND_DECLARE_MODULE_GLOBALS(phar)
@@ -947,6 +948,10 @@ int phar_open_file(php_stream *fp, char *fname, int fname_len, char *alias, int
 		}
 		return FAILURE;
 	}
+	if ((manifest_ver & PHAR_API_MAJORVERSION) == PHAR_API_MAJORVERSION) {
+		/* this is a phar in tar format */
+		return phar2_open_file(fp, fname, fname_len, alias, alias_len, halt_offset, pphar, savebuf, error TSRMLS_CC);
+	}
 
 	PHAR_GET_32(buffer, manifest_flags);
 
diff --git a/ext/phar/phar2.c b/ext/phar/phar2.c
new file mode 100644
index 0000000000..7dae30c30d
--- /dev/null
+++ b/ext/phar/phar2.c
@@ -0,0 +1,244 @@
+/*
+  +----------------------------------------------------------------------+
+  | phar php single-file executable PHP extension                        |
+  +----------------------------------------------------------------------+
+  | Copyright (c) 2006-2007 The PHP Group                                |
+  +----------------------------------------------------------------------+
+  | This source file is subject to version 3.01 of the PHP license,      |
+  | that is bundled with this package in the file LICENSE, and is        |
+  | available through the world-wide-web at the following url:           |
+  | http://www.php.net/license/3_01.txt.                                 |
+  | If you did not receive a copy of the PHP license and are unable to   |
+  | obtain it through the world-wide-web, please send a note to          |
+  | license@php.net so we can mail you a copy immediately.               |
+  +----------------------------------------------------------------------+
+  | Authors: Gregory Beaver <cellog@php.net>                             |
+  |          Marcus Boerger <helly@php.net>                              |
+  |          based on work by Dmitry Stogov <dmitry@zend.com>            |
+  +----------------------------------------------------------------------+
+*/
+
+/* $Id$ */
+
+#include "phar_internal.h"
+
+
+#define TAR_FILE    '0'
+#define TAR_LINK    '1'
+#define TAR_SYMLINK '2'
+#define TAR_DIR     '5'
+#define TAR_NEW     '8'
+
+/**
+ * the format of the header block for a file, in the older UNIX-compatible
+ * TAR format
+ */
+typedef struct _old_tar_header {  /* {{{ */
+	char name[100];     /* name of file;
+	                         directory is indicated by a trailing slash (/) */
+	char mode[8];       /* file mode */
+	char uid[8];        /* owner user ID */
+	char gid[8];        /* owner group ID */
+	char size[12];      /* length of file in bytes */
+	char mtime[12];     /* modify time of file */
+	char chksum[8];     /* checksum for header */
+	char link;          /* indicator for links;
+	                         1 for a linked file,
+	                         2 for a symbolic link,
+	                         0 otherwise */
+	char linkname[100]; /* name of linked file */
+} old_tar_header;
+/* }}} */
+
+/**
+ * the new USTAR header format.
+ * Note that tar can determine that the USTAR format is being used by the
+ * presence of the null-terminated string "ustar" in the magic field.
+ */
+typedef struct _tar_header {  /* {{{ */
+	char name[100];     /* name of file */
+	char mode[8];       /* file mode   */
+	char uid[8];        /* owner user ID */
+	char gid[8];        /* owner group ID */
+	char size[12];      /* length of file in bytes */
+	char mtime[12];     /* modify time of file */
+	char chksum[8];     /* checksum for header */
+	char typeflag;      /* type of file
+	                         0 Regular file
+	                         1 Link to another file already archived
+	                         2 Symbolic link
+	                         3 Character special device
+	                         4 Block special device
+	                         5 Directory
+	                         6 FIFO special file
+	                         7 Reserved */
+	char linkname[100]; /* name of linked file */
+	char magic[6];      /* USTAR indicator */
+	char version[2];    /* USTAR version */
+	char uname[32];     /* owner user name */
+	char gname[32];     /* owner group name */
+	char devmajor[8];   /* device major number */
+	char devminor[8];   /* device minor number */
+	char prefix[155];   /* prefix for file name;
+	                       the value of the prefix field, if non-null,
+	                       is prefixed to the name field to allow names
+	                       longer then 100 characters */
+} tar_header;
+/* }}} */
+
+typedef struct _tar_file tar_file;
+
+typedef struct _tar_entry { /* {{{ */
+	tar_file *tar;
+	char      type;
+	off_t	  start;
+	off_t	  pos;
+	off_t	  size;
+	mode_t	  mode;
+	time_t    mtime;
+	uid_t	  uid;
+	gid_t	  gid;
+	union {
+		HashTable *content;
+		char      *link;
+	} u;
+} tar_entry;
+/* }}} */
+
+struct _tar_file { /* {{{ */
+	php_stream *stream;
+	char       *real_name;
+	int         refcount;
+	int         opened;
+	tar_entry   root;
+};
+/* }}} */
+
+static int tar_number(char* buf, int len) /* {{{ */
+{
+	int num = 0;
+	int i = 0;
+
+	while (i < len && buf[i] == ' ') {
+		i++;
+	}
+	while (i < len &&
+	       buf[i] >= '0' &&
+	       buf[i] <= '7') {
+		num = num * 8 + (buf[i] - '0');
+		i++;
+	}
+	return num;
+}
+/* }}} */
+
+static int tar_checksum(char* buf, int len) /* {{{ */
+{
+	int sum = 0;
+	char *end = buf + len;
+
+	while (buf != end) {
+		sum += (unsigned char)*buf;
+		buf++;
+	}
+	return sum;
+}
+/* }}} */
+#define MAPPHAR_ALLOC_FAIL(msg) \
+	php_stream_close(fp);\
+	if (error) {\
+		spprintf(error, 0, msg, fname);\
+	}\
+	return FAILURE;
+
+#ifdef WORDS_BIGENDIAN
+# define PHAR_GET_32(buffer, var) \
+	var = ((((unsigned char*)(buffer))[3]) << 24) \
+		| ((((unsigned char*)(buffer))[2]) << 16) \
+		| ((((unsigned char*)(buffer))[1]) <<  8) \
+		| (((unsigned char*)(buffer))[0]); \
+	(buffer) += 4
+# define PHAR_GET_16(buffer, var) \
+	var = ((((unsigned char*)(buffer))[1]) <<  8) \
+		| (((unsigned char*)(buffer))[0]); \
+	(buffer) += 2
+#else
+# define PHAR_GET_32(buffer, var) \
+	var = *(php_uint32*)(buffer); \
+	buffer += 4
+# define PHAR_GET_16(buffer, var) \
+	var = *(php_uint16*)(buffer); \
+	buffer += 2
+#endif
+
+int phar2_open_file(php_stream *fp, char *fname, int fname_len, char *alias, int alias_len, long halt_offset, phar_archive_data** pphar, char *buffer, char **error TSRMLS_DC) /* {{{ */
+{
+	/* The structure of a tar-based phar archive is as follows:
+	   - the first file contains the loader stub and file hash.  The file hash is a mapping
+	     of filename hash to location of manifest data within the official phar manifest using
+             an implementation of open hash that allows the entire thing to be read in without any
+             processing beyond signature check.
+	   - the second file contains the official phar manifest as used in phar 1.x
+	   - third file to second-to-last file contain the phar's files
+	   - final file contains the phar signature
+	*/
+	php_uint32 openhash_len;
+	phar_archive_data *mydata;
+
+	buffer += 6;
+	PHAR_GET_32(buffer), openhash_len);
+	if (openhash_len != sizeof(phar2_openhash) && openhash_len != sizeof(phar2_openhash_large)) {
+		MAPPHAR_ALLOC_FAIL("Corrupted phar2 archive, filename hash is wrong size in phar \"%s\"");
+	}
+	buffer -= 6;
+	buffer = (char *) erealloc((void *) buffer, openhash_len);
+	if (!buffer) {
+		MAPPHAR_ALLOC_FAIL("Could not allocate space for fast filename hash in phar \"%s\"");
+	}
+	if (openhash_len != php_stream_read(fp, buffer, openhash_len)) {
+		MAPPHAR_ALLOC_FAIL("Could not read fast filename hash in phar \"%s\"");
+	}
+
+	PHAR_G(fast_fname_map) = (phar2_openhash *) buffer;
+	mydata = ecalloc(sizeof(phar_archive_data), 1);
+
+	/* check whether we have meta data, zero check works regardless of byte order */
+	if (phar_parse_metadata(fp, &buffer, endbuffer, &mydata->metadata TSRMLS_CC) == FAILURE) {
+		MAPPHAR_FAIL("unable to read phar metadata in .phar file \"%s\"");
+	}
+	
+	/* set up our manifest - entries will be created JIT as accessed */
+	zend_hash_init(&mydata->manifest, sizeof(phar_entry_info),
+		zend_get_hash_value, destroy_phar_manifest, 0);
+	snprintf(mydata->version, sizeof(mydata->version), "%u.%u.%u", manifest_ver >> 12, (manifest_ver >> 8) & 0xF, (manifest_ver >> 4) & 0xF);
+	mydata->internal_file_start = halt_offset + manifest_len + 4;
+	mydata->halt_offset = halt_offset;
+	mydata->flags = manifest_flags;
+	mydata->fp = fp;
+	mydata->fname = estrndup(fname, fname_len);
+#ifdef PHP_WIN32
+	phar_unixify_path_separators(mydata->fname, fname_len);
+#endif
+	mydata->fname_len = fname_len;
+	mydata->alias = alias ? estrndup(alias, alias_len) : mydata->fname;
+	mydata->alias_len = alias ? alias_len : fname_len;
+	mydata->sig_flags = sig_flags;
+	mydata->sig_len = sig_len;
+	mydata->signature = signature;
+	phar_request_initialize(TSRMLS_C);
+	zend_hash_add(&(PHAR_GLOBALS->phar_fname_map), mydata->fname, fname_len, (void*)&mydata, sizeof(phar_archive_data*),  NULL);
+	if (register_alias) {
+		mydata->is_explicit_alias = 1;
+		zend_hash_add(&(PHAR_GLOBALS->phar_alias_map), alias, alias_len, (void*)&mydata, sizeof(phar_archive_data*), NULL);
+	} else {
+		mydata->is_explicit_alias = 0;
+	}
+	efree(savebuf);
+	
+	if (pphar) {
+		*pphar = mydata;
+	}
+
+	return SUCCESS;
+}
+/* }}} */
\ No newline at end of file
diff --git a/ext/phar/phar2.h b/ext/phar/phar2.h
new file mode 100644
index 0000000000..b35e636638
--- /dev/null
+++ b/ext/phar/phar2.h
@@ -0,0 +1,34 @@
+/*
+  +----------------------------------------------------------------------+
+  | phar php single-file executable PHP extension                        |
+  +----------------------------------------------------------------------+
+  | Copyright (c) 2006-2007 The PHP Group                                |
+  +----------------------------------------------------------------------+
+  | This source file is subject to version 3.01 of the PHP license,      |
+  | that is bundled with this package in the file LICENSE, and is        |
+  | available through the world-wide-web at the following url:           |
+  | http://www.php.net/license/3_01.txt.                                 |
+  | If you did not receive a copy of the PHP license and are unable to   |
+  | obtain it through the world-wide-web, please send a note to          |
+  | license@php.net so we can mail you a copy immediately.               |
+  +----------------------------------------------------------------------+
+  | Authors: Gregory Beaver <cellog@php.net>                             |
+  |          Marcus Boerger <helly@php.net>                              |
+  +----------------------------------------------------------------------+
+*/
+
+/* $Id$ */
+BEGIN_EXTERN_C()
+
+int phar2_open_file(php_stream *fp, char *fname, int fname_len, char *alias, int alias_len, long halt_offset, phar_archive_data** pphar, char *buffer, char **error TSRMLS_DC);
+
+END_EXTERN_C()
+
+/*
+ * Local variables:
+ * tab-width: 4
+ * c-basic-offset: 4
+ * End:
+ * vim600: noet sw=4 ts=4 fdm=marker
+ * vim<600: noet sw=4 ts=4
+ */
diff --git a/ext/phar/phar2_openhash.c b/ext/phar/phar2_openhash.c
new file mode 100644
index 0000000000..47c7be656d
--- /dev/null
+++ b/ext/phar/phar2_openhash.c
@@ -0,0 +1,150 @@
+/*
+  +----------------------------------------------------------------------+
+  | phar open hash implementation for phar2 tar-based archives           |
+  +----------------------------------------------------------------------+
+  | Copyright (c) 2006-2007 The PHP Group                                |
+  +----------------------------------------------------------------------+
+  | This source file is subject to version 3.01 of the PHP license,      |
+  | that is bundled with this package in the file LICENSE, and is        |
+  | available through the world-wide-web at the following url:           |
+  | http://www.php.net/license/3_01.txt.                                 |
+  | If you did not receive a copy of the PHP license and are unable to   |
+  | obtain it through the world-wide-web, please send a note to          |
+  | license@php.net so we can mail you a copy immediately.               |
+  +----------------------------------------------------------------------+
+  | Authors: Gregory Beaver <cellog@php.net>                             |
+  +----------------------------------------------------------------------+
+*/
+
+/* $Id$ */
+
+#include "zend_hash.h"
+
+static inline int phar2_openhash_doublehash(ulong k, int m, int i) /* {{{ */
+{
+	/* implement double hash probing */
+	return ((k % m) + (i + (k % (m - 1)))) % m;
+}
+/* }}} */
+
+phar2_openhash *phar2_openhash_init(int large) /* {{{ */
+{
+	phar2_openhash *hash;
+	if (large) {
+		hash = (phar2_openhash *) ecalloc(sizeof(phar2_openhash_large));
+	} else {
+		hash = (phar2_openhash *) ecalloc(sizeof(phar2_openhash));
+		hash->is_large = '\001';
+	}
+	return hash;
+}
+/* }}} */
+
+int phar2_openhash_destroy(phar2_openhash *hash) /* {{{ */
+{
+	return efree(hash);
+}
+/* }}} */
+
+/* retrieve the offset of this file's manifest entry.  Returns offset or -1 */
+long phar2_openhash_getoffset(phar2_openhash *hash, char *fname, int len) /* {{{ */
+{
+	int myhash, hashsize;
+	char substr[20];
+
+	memset((void *) substr, 0, 20);
+	if (len < 20) {
+		memcpy((void *) &substr, (void *) fname, len);
+	} else {
+		memcpy((void *) &substr, (void *) fname + (len - 20), 20);
+	}
+	hashsize = (hash->is_large ? PHAR2_HASHSIZE : PHAR2_HASHSIZE_SMALL);
+	for (i = 0; i < hashsize; i ++) {
+		myhash = phar2_openhash_doublehash(zhash, hashsize, i);
+		if (hash->info[myhash].filepart[0]) {
+			if (!memcmp(hash->info[myhash].filepart, substr, 20)) {
+				/* found it */
+				return hash->info[myhash].offset;
+			}
+		} else {
+			/* empty - does not exist */
+			return -1;
+		}
+	}
+	return -1;
+}
+/* }}} */
+
+static int phar2_openhash_insert_ex(phar2_openhash *hash, char* fname, ulong zhash, long offset) /* {{{ */
+{
+	int myhash, hashsize;
+	hashsize = (hash->is_large ? PHAR2_HASHSIZE : PHAR2_HASHSIZE_SMALL);
+	for (i = 0; i < hashsize; i ++) {
+		myhash = phar2_openhash_doublehash(zhash, hashsize, i);
+		if (hash->info[myhash].filepart[0]) {
+			continue;
+		} else {
+			memcpy((void *) &hash->info[myhash].filepart, (void *) fname, 20);
+			hash->info[myhash].offset = offset;
+			hash->info[myhash].zhash = zhash;
+			hash->count++;
+			return SUCCESS;
+		}
+	}
+	return FAILURE;
+}
+/* }}} */
+
+/* a super-large phar is being created, enlarge the hash and re-populate it */
+phar2_openhash_large *phar2_openhash_enlarge(phar2_openhash *hash) /* {{{ */
+{
+	phar_openhash_large *newhash = phar2_openhash_init(1);
+	for (phar2_openhash_entry s = newhash->info, int i = 0; i < hash->count; i++) {
+		if (!s[i].filepart[0]) {
+			continue;
+		}
+		phar2_openhash_insert_ex((phar2_openhash *)newhash, &(s[i].filepart), s[i].zhash, s[i].offset);
+	}
+	phar2_openhash_destroy(hash);
+	return newhash;
+}
+/* }}} */
+
+int phar2_openhash_insert(phar2_openhash *hash, char* fname, int len, long offset) /* {{{ */
+{
+	ulong zhash = zend_inline_hash_func(string, len);
+	char substr[20];
+
+	if (-1 != phar2_openhash_getoffset(hash, fname, len)) {
+		/* already exists, so fail */
+		return FAILURE;
+	}
+	if (hash->is_large) {
+		if (hash->count + 1 == PHAR_HASHSIZE) {
+			return FAILURE;
+		}
+	} else {
+		if (hash->count + 1 == PHAR_HASHSIZE_SMALL) {
+			/* resize for mega-huge phars */
+			hash = (phar2_openhash *) phar2_openhash_enlarge(hash);
+		}
+	}
+
+	memset((void *) substr, 0, 20);
+	if (len < 20) {
+		memcpy((void *) &substr, (void *) fname, len);
+	} else {
+		memcpy((void *) &substr, (void *) fname + (len - 20), 20);
+	}
+	return phar2_openhash_insert_ex(hash, substr, zhash, offset);
+}
+/* }}} */
+
+/*
+ * Local variables:
+ * tab-width: 4
+ * c-basic-offset: 4
+ * End:
+ * vim600: noet sw=4 ts=4 fdm=marker
+ * vim<600: noet sw=4 ts=4
+ */
diff --git a/ext/phar/phar2_openhash.h b/ext/phar/phar2_openhash.h
new file mode 100644
index 0000000000..f8ef10006e
--- /dev/null
+++ b/ext/phar/phar2_openhash.h
@@ -0,0 +1,59 @@
+/*
+  +----------------------------------------------------------------------+
+  | phar open hash implementation for phar2 tar-based archives           |
+  +----------------------------------------------------------------------+
+  | Copyright (c) 2006-2007 The PHP Group                                |
+  +----------------------------------------------------------------------+
+  | This source file is subject to version 3.01 of the PHP license,      |
+  | that is bundled with this package in the file LICENSE, and is        |
+  | available through the world-wide-web at the following url:           |
+  | http://www.php.net/license/3_01.txt.                                 |
+  | If you did not receive a copy of the PHP license and are unable to   |
+  | obtain it through the world-wide-web, please send a note to          |
+  | license@php.net so we can mail you a copy immediately.               |
+  +----------------------------------------------------------------------+
+  | Authors: Gregory Beaver <cellog@php.net>                             |
+  +----------------------------------------------------------------------+
+*/
+
+/* $Id$ */
+
+/* maximum number of files that will fit into a phar */
+#define PHAR2_HASHSIZE_SMALL 997
+#define PHAR2_HASHSIZE 4999
+
+typedef struct _phar2_openhash_entry {
+	char[20] filepart; /* last 10 characters of filename */
+	long    offset;   /* offset of this file's info in the manifest */
+	ulong    zhash;    /* value of zend_inline_hash_func() on the filename */
+} phar2_openhash_entry;
+
+typedef struct _phar2_openhash {
+	char                 is_large; /* always 0 */
+	int                  count;
+	phar2_openhash_entry info[PHAR2_HASHSIZE_SMALL];
+} phar2_openhash;
+
+typedef struct _phar2_openhash {
+	char                 is_large; /* always 1 */
+	int                  count;
+	phar2_openhash_entry info[PHAR2_HASHSIZE];
+} phar2_openhash_large;
+
+BEGIN_EXTERN_C()
+
+phar2_openhash *phar2_openhash_init(int large);
+int phar2_openhash_destroy(phar2_openhash *hash);
+long phar2_openhash_getoffset(char *fname, int len);
+int phar2_openhash_insert(phar2_openhash *hash, char* fname, int len, long offset);
+
+END_EXTERN_C()
+
+/*
+ * Local variables:
+ * tab-width: 4
+ * c-basic-offset: 4
+ * End:
+ * vim600: noet sw=4 ts=4 fdm=marker
+ * vim<600: noet sw=4 ts=4
+ */
diff --git a/ext/phar/phar_internal.h b/ext/phar/phar_internal.h
index 0c4e85c436..e075ebdf01 100755
--- a/ext/phar/phar_internal.h
+++ b/ext/phar/phar_internal.h
@@ -30,6 +30,7 @@
 #include "zend_execute.h"
 #include "zend_exceptions.h"
 #include "zend_hash.h"
+#include "ext/phar/phar2_openhash.h"
 #include "zend_interfaces.h"
 #include "zend_operators.h"
 #include "zend_qsort.h"
@@ -65,12 +66,12 @@
 #define E_RECOVERABLE_ERROR E_ERROR
 #endif
 
-#define PHAR_EXT_VERSION_STR      "1.3.0"
-#define PHAR_API_VERSION_STR      "1.1.0"
+#define PHAR_EXT_VERSION_STR      "2.0.0"
+#define PHAR_API_VERSION_STR      "2.0.0"
 /* x.y.z maps to 0xyz0 */
-#define PHAR_API_VERSION          0x1100
+#define PHAR_API_VERSION          0x2000
 #define PHAR_API_MIN_READ         0x1000
-#define PHAR_API_MAJORVERSION     0x1000
+#define PHAR_API_MAJORVERSION     0x2000
 #define PHAR_API_MAJORVER_MASK    0xF000
 #define PHAR_API_VER_MASK         0xFFF0
 
@@ -106,6 +107,7 @@ ZEND_BEGIN_MODULE_GLOBALS(phar)
 	HashTable   phar_fname_map;
 	HashTable   phar_alias_map;
 	HashTable   phar_plain_map;
+	phar2_openhash *fast_fname_map;
 	char*       extract_list;
 	int         readonly;
 	zend_bool   readonly_orig;