POSIX GNU tar read support

- supporting tared csm.bin of Chasm: The Rift Remastered at https://store.steampowered.com/app/2061230/Chasm_The_Rift/

Author: Jon Daniel

CMakeLists.txt

@@ -92,6 +92,7 @@ set(PHYSFS_SRCS
     src/physfs_archiver_qpak.c
     src/physfs_archiver_wad.c
     src/physfs_archiver_csm.c
+    src/physfs_archiver_tar.c
     src/physfs_archiver_zip.c
     src/physfs_archiver_slb.c
     src/physfs_archiver_iso9660.c
@@ -130,6 +131,11 @@ if(NOT PHYSFS_ARCHIVE_CSM)
     add_definitions(-DPHYSFS_SUPPORTS_CSM=0)
 endif()
 
+option(PHYSFS_ARCHIVE_TAR "Enable POSIX TAR / Chasm: The Rift [Demo] Remastered csm.bin support" TRUE)
+if(NOT PHYSFS_ARCHIVE_TAR)
+    add_definitions(-DPHYSFS_SUPPORTS_TAR=0)
+endif()
+
 option(PHYSFS_ARCHIVE_HOG "Enable Descent I/II HOG support" TRUE)
 if(NOT PHYSFS_ARCHIVE_HOG)
     add_definitions(-DPHYSFS_SUPPORTS_HOG=0)
@@ -325,6 +331,7 @@ message_bool_option("7zip support" PHYSFS_ARCHIVE_7Z)
 message_bool_option("GRP support" PHYSFS_ARCHIVE_GRP)
 message_bool_option("WAD support" PHYSFS_ARCHIVE_WAD)
 message_bool_option("CSM support" PHYSFS_ARCHIVE_CSM)
+message_bool_option("TAR support" PHYSFS_ARCHIVE_TAR)
 message_bool_option("HOG support" PHYSFS_ARCHIVE_HOG)
 message_bool_option("MVL support" PHYSFS_ARCHIVE_MVL)
 message_bool_option("QPAK support" PHYSFS_ARCHIVE_QPAK)

src/Makefile.os2

@@ -26,6 +26,7 @@ SRCS = physfs.c                   &
        physfs_archiver_slb.c      &
        physfs_archiver_iso9660.c  &
        physfs_archiver_csm.c      &
+       physfs_archiver_tar.c      &
        physfs_archiver_vdf.c
 
 

src/physfs.c

@@ -1191,6 +1191,9 @@ static int initStaticArchivers(void)
     #if PHYSFS_SUPPORTS_CSM
         REGISTER_STATIC_ARCHIVER(CSM);
     #endif
+    #if PHYSFS_SUPPORTS_TAR
+        REGISTER_STATIC_ARCHIVER(TAR);
+    #endif
     #if PHYSFS_SUPPORTS_SLB
         REGISTER_STATIC_ARCHIVER(SLB);
     #endif

src/physfs_archiver_tar.c

@@ -0,0 +1,185 @@
+/*
+ * tar support routines for PhysicsFS.
+ *
+ * Please see the file LICENSE.txt in the source's root directory.
+ *
+ * uses POSIX GNU tar header: https://www.gnu.org/software/tar
+ *
+ * based on code by Uli Köhler: https://techoverflow.net/2013/03/29/reading-tar-files-in-c/
+ */
+
+#define __PHYSICSFS_INTERNAL__
+#include "physfs_internal.h"
+
+#if PHYSFS_SUPPORTS_TAR
+#include "physfs_tar.h"
+
+#define ASCII_TO_NUMBER(num) ((num)-48) //Converts an ascii digit to the corresponding number (assuming it is an ASCII digit)
+
+static PHYSFS_uint64 TAR_decodeOctal(char* data, size_t size) {
+    unsigned char* currentPtr = (unsigned char*) data + size;
+    PHYSFS_uint64 sum = 0;
+    PHYSFS_uint64 currentMultiplier = 1;
+    unsigned char* checkPtr = currentPtr;
+
+    for (; checkPtr >= (unsigned char*) data; checkPtr--) {
+        if ((*checkPtr) == 0 || (*checkPtr) == ' ') {
+            currentPtr = checkPtr - 1;
+        }
+    }
+    for (; currentPtr >= (unsigned char*) data; currentPtr--) {
+        sum += ASCII_TO_NUMBER(*currentPtr) * currentMultiplier;
+        currentMultiplier *= 8;
+    }
+    return sum;
+}
+
+static bool TAR_magic(union block* block) {
+	return (memcmp("ustar", block->header.magic, 5) == 0);
+}
+
+static size_t TAR_fileSize(union block* block) {
+	return TAR_decodeOctal(block->header.size, sizeof(block->header.size));
+}
+
+static bool TAR_checksum(union block* block) {
+	PHYSFS_sint64 unsigned_sum = 0;
+	PHYSFS_sint64 signed_sum = 0;
+	PHYSFS_uint64 reference_chksum = 0;
+	char orig_chksum[8];
+	int i = 0;
+
+	memcpy(orig_chksum, block->header.chksum, 8);
+	memset(block->header.chksum, ' ', 8);
+
+	for(; i < BLOCKSIZE; i++) {
+		unsigned_sum += ((unsigned char*) block->buffer)[i];
+		signed_sum += ((signed char*) block->buffer)[i];
+	}
+	memcpy(block->header.chksum, orig_chksum, 8);
+	reference_chksum = TAR_decodeOctal(orig_chksum, 12);
+	return (reference_chksum == unsigned_sum || reference_chksum == signed_sum);
+}
+
+static bool TAR_loadEntries(PHYSFS_Io *io, void *arc)
+{
+	union block zero_block = { 0 };
+	union block current_block = { 0 };
+	bool long_name = false;
+	const PHYSFS_Allocator* allocator = PHYSFS_getAllocator();
+        char name[PATH_MAX] = { 0 };
+	PHYSFS_uint64 size  = 0;
+	PHYSFS_uint64 pos   = 0;
+	PHYSFS_uint64 pad   = 0;
+	PHYSFS_uint64 count = 0;
+	char*         buf   = NULL;
+
+	/* read header block until zero-only terminated block */
+	for(; __PHYSFS_readAll(io, current_block.buffer, BLOCKSIZE); count++)
+	{
+		if((memcmp(current_block.buffer, zero_block.buffer, BLOCKSIZE) == 0))
+			return true;
+
+		/* verify magic */
+		if(!TAR_magic(&current_block))
+			return false;
+
+		/* verify checksum */
+		if(!TAR_checksum(&current_block))
+			return false;
+
+		memset(name, '\0', PATH_MAX);
+		/* support prefix */
+		if(strlen(current_block.header.prefix) > 0)
+		{
+			strcpy(name, current_block.header.prefix);
+			name[strlen(name)] = '/';
+		}
+		strcpy(&name[strlen(name)], current_block.header.name);
+
+		/* add file type entry */
+		if (current_block.header.typeflag == REGTYPE || current_block.header.typeflag == 0) {
+			/* support long file names */
+			if(long_name) {
+				strcpy(&name[0], current_block.header.name);
+				BAIL_IF_ERRPASS(!__PHYSFS_readAll(io, current_block.buffer, BLOCKSIZE), 0);
+				long_name = false;
+				count++;
+			}
+			size = TAR_fileSize(&current_block);
+			pos  = (count + 1) * BLOCKSIZE;
+			pad  = (BLOCKSIZE - (size % BLOCKSIZE)) % BLOCKSIZE;
+			buf  = allocator->Malloc(size + pad);
+			/* add entry to arc */
+			BAIL_IF_ERRPASS(!__PHYSFS_readAll(io, buf, size + pad), 0);
+			count += ((size + pad) / BLOCKSIZE );
+			allocator->Free(buf);
+			BAIL_IF_ERRPASS(!UNPK_addEntry(arc, name, 0, -1, -1, pos, size), 0);
+		}
+		/* add directory type entry */
+		else if(current_block.header.typeflag == DIRTYPE)
+		{
+			BAIL_IF_ERRPASS(!UNPK_addEntry(arc, name, 1, -1, -1, 0, 0), 0);
+		}
+		/* long name mode */
+		else if(current_block.header.typeflag == GNUTYPE_LONGNAME)
+		{
+			long_name = true;
+		}
+		else
+		{
+			// UNHANDLED
+		}
+	}
+	return false;
+}
+
+static void *TAR_openArchive(PHYSFS_Io *io, const char *name,
+                              int forWriting, int *claimed)
+{
+    void *unpkarc = NULL;
+
+    assert(io != NULL);  /* shouldn't ever happen. */
+
+    BAIL_IF(forWriting, PHYSFS_ERR_READ_ONLY, NULL);
+
+    unpkarc = UNPK_openArchive(io, 0, 1);
+    BAIL_IF_ERRPASS(!unpkarc, NULL);
+
+    if (!TAR_loadEntries(io, unpkarc))
+    {
+        UNPK_abandonArchive(unpkarc);
+        return NULL;
+    } /* if */
+
+    *claimed = 1;
+
+    return unpkarc;
+} /* TAR_openArchive */
+
+
+const PHYSFS_Archiver __PHYSFS_Archiver_TAR =
+{
+    CURRENT_PHYSFS_ARCHIVER_API_VERSION,
+    {
+        "TAR",
+        "POSIX GNU tar archives / Chasm: the Rift Remastered",
+        "Jon Daniel <jondaniel879@gmail.com>",
+        "https://www.gnu.org/software/tar",
+        0,
+    },
+    TAR_openArchive,
+    UNPK_enumerate,
+    UNPK_openRead,
+    UNPK_openWrite,
+    UNPK_openAppend,
+    UNPK_remove,
+    UNPK_mkdir,
+    UNPK_stat,
+    UNPK_closeArchive
+};
+
+#endif  /* defined PHYSFS_SUPPORTS_TAR */
+
+/* end of physfs_archiver_tar.c ... */
+

src/physfs_internal.h

@@ -89,6 +89,7 @@ extern const PHYSFS_Archiver __PHYSFS_Archiver_HOG;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_MVL;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_WAD;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_CSM;
+extern const PHYSFS_Archiver __PHYSFS_Archiver_TAR;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_SLB;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_ISO9660;
 extern const PHYSFS_Archiver __PHYSFS_Archiver_VDF;
@@ -204,6 +205,9 @@ void __PHYSFS_smallFree(void *ptr);
 #ifndef PHYSFS_SUPPORTS_CSM
 #define PHYSFS_SUPPORTS_CSM PHYSFS_SUPPORTS_DEFAULT
 #endif
+#ifndef PHYSFS_SUPPORTS_TAR
+#define PHYSFS_SUPPORTS_TAR PHYSFS_SUPPORTS_DEFAULT
+#endif
 #ifndef PHYSFS_SUPPORTS_QPAK
 #define PHYSFS_SUPPORTS_QPAK PHYSFS_SUPPORTS_DEFAULT
 #endif