1 files changed, 82 insertions, 37 deletions

diff --git a/xdelta3/xdelta3.h b/xdelta3/xdelta3.h
index 11bbdb8..1b8f083 100644
--- a/xdelta3/xdelta3.h
+++ b/xdelta3/xdelta3.h
@@ -17,7 +17,7 @@
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
-/* To know more about Xdelta, start by reading xdelta3.c.  If you are
+/* To learn more about Xdelta, start by reading xdelta3.c.  If you are
  * ready to use the API, continue reading here.  There are two
  * interfaces -- xd3_encode_input and xd3_decode_input -- plus a dozen
  * or so related calls.  This interface is styled after Zlib. */
@@ -25,7 +25,7 @@
 #ifndef _XDELTA3_H_
 #define _XDELTA3_H_
 
-#define _POSIX_SOURCE
+#define _POSIX_SOURCE 200112L
 #define _ISOC99_SOURCE
 #define _C99_SOURCE
 
@@ -71,7 +71,7 @@
  *
  * 8-16MB is reasonable, probably don't need to go larger. */
 #ifndef XD3_HARDMAXWINSIZE
-#define XD3_HARDMAXWINSIZE (1U<<24)
+#define XD3_HARDMAXWINSIZE (1U<<26)
 #endif
 /* The IOPT_SIZE value sets the size of a buffer used to batch
  * overlapping copy instructions before they are optimized by picking
@@ -87,8 +87,7 @@
 #define XD3_DEFAULT_SPREVSZ (1U<<18)
 #endif
 
-/* The default compression level
- */
+/* The default compression level */
 #ifndef XD3_DEFAULT_LEVEL
 #define XD3_DEFAULT_LEVEL 3
 #endif
@@ -101,6 +100,12 @@
 #define XD3_USE_LARGEFILE64 1
 #endif
 
+/* The source window size is limited to 2GB unless
+ * XD3_USE_LARGEWINDOW64 is defined to 1. */
+#ifndef XD3_USE_LARGEWINDOW64
+#define XD3_USE_LARGEWINDOW64 0
+#endif
+
 /* Sizes and addresses within VCDIFF windows are represented as usize_t
  *
  * For source-file offsets and total file sizes, total input and
@@ -109,6 +114,8 @@
  * the 32bit boundary [xdelta3-test.h]).
  */
 #ifndef _WIN32
+#define __STDC_FORMAT_MACROS
+#include <inttypes.h>
 #include <stdint.h>
 #else /* WIN32 case */
 #define WIN32_LEAN_AND_MEAN
@@ -148,8 +155,7 @@ typedef ULONGLONG      uint64_t;
 #endif /* _MSC_VER defined */
 #endif /* _WIN32 defined */
 
-typedef uint32_t usize_t;
-
+/* Settings based on the size of xoff_t (32 vs 64 file offsets) */
 #if XD3_USE_LARGEFILE64
 /* xoff_t is a 64-bit type */
 #define __USE_FILE_OFFSET64 1 /* GLIBC: for 64bit fileops, ... ? */
@@ -162,6 +168,7 @@ typedef uint32_t usize_t;
 #define _FILE_OFFSET_BITS 64
 #endif
 
+/* Set a xoff_t typedef and the "Q" printf insert. */
 #if defined(_WIN32)
 typedef uint64_t xoff_t;
 /* Note: The following generates benign warnings in a mingw
@@ -176,7 +183,7 @@ typedef size_t xoff_t;
 #elif SIZEOF_UNSIGNED_LONG_LONG == 8
 typedef unsigned long long xoff_t;
 #define Q "ll"
-#endif /* #define Q */
+#endif /* typedef and #define Q */
 
 #define SIZEOF_XOFF_T 8
 
@@ -194,9 +201,45 @@ typedef uint32_t xoff_t;
 #define Q
 #endif /* 64 vs 32 bit xoff_t */
 
-/* Note: This gets modified in the 64bithash branch. */
+/* Settings based on the size of usize_t (32 and 64 bit window size) */
+#if XD3_USE_LARGESIZET
+
+/* Set a usize_ttypedef and the "W" printf insert. */
+#if defined(_WIN32)
+typedef uint64_t usize_t;
+/* Note: The following generates benign warnings in a mingw
+ * cross-compiler */
+#define W "I64"
+#elif SIZEOF_UNSIGNED_LONG == 8
+typedef unsigned long usize_t;
+#define W "l"
+#elif SIZEOF_SIZE_T == 8
+typedef size_t usize_t;
+#define W "z"
+#elif SIZEOF_UNSIGNED_LONG_LONG == 8
+typedef unsigned long long usize_t;
+#define W "ll"
+#endif /* typedef and #define W */
+
+#define SIZEOF_USIZE_T 8
+
+#else /* XD3_USE_LARGESIZET == 0 */
+
+#if SIZEOF_UNSIGNED_INT == 4
+typedef unsigned int usize_t;
+#elif SIZEOF_UNSIGNED_LONG == 4
+typedef unsigned long usize_t;
+#else
+typedef uint32_t usize_t;
+#endif /* usize_t is 32 bits */
+
 #define SIZEOF_USIZE_T 4
+#define W
 
+#endif /* 64 vs 32 bit usize_t */
+
+/* Settings based on the size of size_t (the system-provided,
+ * usually-but-maybe-not an unsigned type) */
 #if SIZEOF_SIZE_T == 4
 #define Z "z"
 #elif SIZEOF_SIZE_T == 8
@@ -339,9 +382,12 @@ typedef int              (xd3_comp_table_func) (xd3_stream *stream,
 
 
 #if XD3_DEBUG
-#define XD3_ASSERT(x) \
-    do { if (! (x)) { DP(RINT "%s:%d: XD3 assertion failed: %s\n", __FILE__, __LINE__, #x); \
-    abort (); } } while (0)
+#define XD3_ASSERT(x)                                \
+  do {                                               \
+    if (! (x)) {                                     \
+      DP(RINT "%s:%d: XD3 assertion failed: %s\n",   \
+         __FILE__, __LINE__, #x);                    \
+      abort (); } } while (0)
 #else
 #define XD3_ASSERT(x) (void)0
 #endif  /* XD3_DEBUG */
@@ -589,9 +635,9 @@ struct _xd3_dinst
 /* the decoded form of a single (half) instruction. */
 struct _xd3_hinst
 {
-  uint8_t     type;
-  uint32_t    size;  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t    addr;  /* TODO: why decode breaks if this is usize_t? */
+  uint8_t    type;
+  usize_t    size;
+  usize_t    addr;
 };
 
 /* the form of a whole-file instruction */
@@ -618,7 +664,7 @@ struct _xd3_desect
 {
   const uint8_t *buf;
   const uint8_t *buf_max;
-  uint32_t       size;  /* TODO: why decode breaks if this is usize_t? */
+  usize_t        size;
   usize_t        pos;
 
   /* used in xdelta3-decode.h */
@@ -666,9 +712,13 @@ struct _xd3_smatcher
 /* hash table size & power-of-two hash function. */
 struct _xd3_hash_cfg
 {
-  usize_t           size;
-  usize_t           shift;
-  usize_t           mask;
+  usize_t  size;       // Number of buckets
+  usize_t  shift;
+  usize_t  mask;
+  usize_t  look;       // How wide is this checksum
+  usize_t  multiplier; // K * powers[0]
+  usize_t *powers;     // Array of [0,look) where powers[look-1] == 1
+                       // and powers[N] = powers[N+1]*K (Rabin-Karp)
 };
 
 /* the sprev list */
@@ -728,7 +778,7 @@ struct _xd3_config
   xd3_alloc_func    *alloc;
   xd3_free_func     *freef;
   void              *opaque;        /* Not used. */
-  int                flags;         /* stream->flags are initialized
+  uint32_t           flags;         /* stream->flags are initialized
                                      * from xd3_config & never
                                      * modified by the library.  Use
                                      * xd3_set_flags to modify flags
@@ -771,8 +821,8 @@ struct _xd3_source
   usize_t             srclen;        /* length of this source window */
   xoff_t              srcbase;       /* offset of this source window
                                         in the source itself */
-  int                 shiftby;       /* for power-of-two blocksizes */
-  int                 maskby;        /* for power-of-two blocksizes */
+  usize_t             shiftby;       /* for power-of-two blocksizes */
+  usize_t             maskby;        /* for power-of-two blocksizes */
   xoff_t              cpyoff_blocks; /* offset of dec_cpyoff in blocks */
   usize_t             cpyoff_blkoff; /* offset of copy window in
                                         blocks, remainder */
@@ -833,7 +883,7 @@ struct _xd3_stream
   xd3_free_func    *free;             /* free function */
   void*             opaque;           /* private data object passed to
                                          alloc, free, and getblk */
-  int               flags;            /* various options */
+  uint32_t          flags;            /* various options */
 
   /* secondary compressor configuration */
   xd3_sec_cfg       sec_data;         /* Secondary compressor config: data */
@@ -931,13 +981,11 @@ struct _xd3_stream
 
   usize_t           dec_secondid;     /* Optional secondary compressor ID. */
 
-  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t          dec_codetblsz;    /* Optional code table: length. */
+  usize_t           dec_codetblsz;    /* Optional code table: length. */
   uint8_t          *dec_codetbl;      /* Optional code table: storage. */
   usize_t           dec_codetblbytes; /* Optional code table: position. */
 
-  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t          dec_appheadsz;    /* Optional application header:
+  usize_t           dec_appheadsz;    /* Optional application header:
                                          size. */
   uint8_t          *dec_appheader;    /* Optional application header:
                                          storage */
@@ -948,15 +996,12 @@ struct _xd3_stream
   uint8_t           dec_cksum[4];     /* Optional checksum: storage. */
   uint32_t          dec_adler32;      /* Optional checksum: value. */
 
-  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t           dec_cpylen;       /* length of copy window
+  usize_t            dec_cpylen;       /* length of copy window
                                           (VCD_SOURCE or VCD_TARGET) */
   xoff_t             dec_cpyoff;       /* offset of copy window
                                           (VCD_SOURCE or VCD_TARGET) */
-  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t           dec_enclen;       /* length of delta encoding */
-  /* TODO: why decode breaks if this is usize_t? */
-  uint32_t           dec_tgtlen;       /* length of target window */
+  usize_t            dec_enclen;       /* length of delta encoding */
+  usize_t            dec_tgtlen;       /* length of target window */
 
 #if USE_UINT64
   uint64_t          dec_64part;       /* part of a decoded uint64_t */
@@ -1262,7 +1307,7 @@ const char* xd3_strerror (int ret);
    specified flags. */
 static inline
 void    xd3_init_config (xd3_config *config,
-                         int         flags)
+                         uint32_t    flags)
 {
   memset (config, 0, sizeof (*config));
   config->flags = flags;
@@ -1325,7 +1370,7 @@ usize_t xd3_encoder_srclen (xd3_stream *stream) {
 
 /* Checks for legal flag changes. */
 static inline
-void xd3_set_flags (xd3_stream *stream, int flags)
+void xd3_set_flags (xd3_stream *stream, uint32_t flags)
 {
   /* The bitwise difference should contain only XD3_FLUSH or
      XD3_SKIP_WINDOW */
@@ -1350,8 +1395,8 @@ void xd3_blksize_div (const xoff_t offset,
                       const xd3_source *source,
                       xoff_t *blkno,
                       usize_t *blkoff) {
-  *blkno = (xoff_t) (offset >> source->shiftby);
-  *blkoff = (usize_t) (offset & source->maskby);
+  *blkno = offset >> source->shiftby;
+  *blkoff = offset & source->maskby;
   XD3_ASSERT (*blkoff < source->blksize);
 }