-/*
- Copyright (C) Andrew Tridgell 1996
- Copyright (C) Paul Mackerras 1996
-
- This program is free software; you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation; either version 2 of the License, or
- (at your option) any later version.
-
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with this program; if not, write to the Free Software
- Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
-*/
+/*
+ * Routines used by the file-transfer code.
+ *
+ * Copyright (C) 1996 Andrew Tridgell
+ * Copyright (C) 1996 Paul Mackerras
+ * Copyright (C) 2003-2009 Wayne Davison
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, visit the http://fsf.org website.
+ */
#include "rsync.h"
+#include "itypes.h"
#include "zlib/zlib.h"
extern int do_compression;
-static int compression_level = Z_DEFAULT_COMPRESSION;
+extern int module_id;
+extern int def_compress_level;
+extern char *skip_compress;
-/* determine the compression level based on a wildcard filename list */
-void set_compression(char *fname)
+static int compression_level, per_file_default_level;
+
+struct suffix_tree {
+ struct suffix_tree *sibling;
+ struct suffix_tree *child;
+ char letter, word_end;
+};
+
+static char *match_list;
+static struct suffix_tree *suftree;
+
+static void add_suffix(struct suffix_tree **prior, char ltr, const char *str)
{
- extern int module_id;
- char *dont;
- char *tok;
+ struct suffix_tree *node, *newnode;
+
+ if (ltr == '[') {
+ const char *after = strchr(str, ']');
+ /* Treat "[foo" and "[]" as having a literal '['. */
+ if (after && after++ != str+1) {
+ while ((ltr = *str++) != ']')
+ add_suffix(prior, ltr, after);
+ return;
+ }
+ }
- if (!do_compression) return;
+ for (node = *prior; node; prior = &node->sibling, node = node->sibling) {
+ if (node->letter == ltr) {
+ if (*str)
+ add_suffix(&node->child, *str, str+1);
+ else
+ node->word_end = 1;
+ return;
+ }
+ if (node->letter > ltr)
+ break;
+ }
+ if (!(newnode = new(struct suffix_tree)))
+ out_of_memory("add_suffix");
+ newnode->sibling = node;
+ newnode->child = NULL;
+ newnode->letter = ltr;
+ *prior = newnode;
+ if (*str) {
+ add_suffix(&newnode->child, *str, str+1);
+ newnode->word_end = 0;
+ } else
+ newnode->word_end = 1;
+}
- compression_level = Z_DEFAULT_COMPRESSION;
- dont = lp_dont_compress(module_id);
+static void add_nocompress_suffixes(const char *str)
+{
+ char *buf, *t;
+ const char *f = str;
- if (!dont || !*dont) return;
+ if (!(buf = new_array(char, strlen(f) + 1)))
+ out_of_memory("add_nocompress_suffixes");
- if ((dont[0] == '*') && (!dont[1])) {
- /* an optimization to skip the rest of this routine */
- compression_level = 0;
- return;
+ while (*f) {
+ if (*f == '/') {
+ f++;
+ continue;
+ }
+
+ t = buf;
+ do {
+ if (isUpper(f))
+ *t++ = toLower(f);
+ else
+ *t++ = *f;
+ } while (*++f != '/' && *f);
+ *t++ = '\0';
+
+ add_suffix(&suftree, *buf, buf+1);
}
- dont = strdup(dont);
- fname = strdup(fname);
- if (!dont || !fname) return;
+ free(buf);
+}
- strlower(dont);
- strlower(fname);
+static void init_set_compression(void)
+{
+ const char *f;
+ char *t, *start;
- for (tok=strtok(dont," ");tok;tok=strtok(NULL," ")) {
- if (fnmatch(tok, fname, 0) == 0) {
- compression_level = 0;
+ if (skip_compress)
+ add_nocompress_suffixes(skip_compress);
+
+ /* A non-daemon transfer skips the default suffix list if the
+ * user specified --skip-compress. */
+ if (skip_compress && module_id < 0)
+ f = "";
+ else
+ f = lp_dont_compress(module_id);
+
+ if (!(match_list = t = new_array(char, strlen(f) + 2)))
+ out_of_memory("set_compression");
+
+ per_file_default_level = def_compress_level;
+
+ while (*f) {
+ if (*f == ' ') {
+ f++;
+ continue;
+ }
+
+ start = t;
+ do {
+ if (isUpper(f))
+ *t++ = toLower(f);
+ else
+ *t++ = *f;
+ } while (*++f != ' ' && *f);
+ *t++ = '\0';
+
+ if (t - start == 1+1 && *start == '*') {
+ /* Optimize a match-string of "*". */
+ *match_list = '\0';
+ suftree = NULL;
+ per_file_default_level = 0;
break;
}
+
+ /* Move *.foo items into the stuffix tree. */
+ if (*start == '*' && start[1] == '.' && start[2]
+ && !strpbrk(start+2, ".?*")) {
+ add_suffix(&suftree, start[2], start+3);
+ t = start;
+ }
+ }
+ *t++ = '\0';
+}
+
+/* determine the compression level based on a wildcard filename list */
+void set_compression(const char *fname)
+{
+ const struct suffix_tree *node;
+ const char *s;
+ char ltr;
+
+ if (!do_compression)
+ return;
+
+ if (!match_list)
+ init_set_compression();
+
+ compression_level = per_file_default_level;
+
+ if (!*match_list && !suftree)
+ return;
+
+ if ((s = strrchr(fname, '/')) != NULL)
+ fname = s + 1;
+
+ for (s = match_list; *s; s += strlen(s) + 1) {
+ if (iwildmatch(s, fname)) {
+ compression_level = 0;
+ return;
+ }
+ }
+
+ if (!(node = suftree) || !(s = strrchr(fname, '.'))
+ || s == fname || !(ltr = *++s))
+ return;
+
+ while (1) {
+ if (isUpper(<r))
+ ltr = toLower(<r);
+ while (node->letter != ltr) {
+ if (node->letter > ltr)
+ return;
+ if (!(node = node->sibling))
+ return;
+ }
+ if ((ltr = *++s) == '\0') {
+ if (node->word_end)
+ compression_level = 0;
+ return;
+ }
+ if (!(node = node->child))
+ return;
}
- free(dont);
- free(fname);
}
/* non-compressing recv token */
-static int simple_recv_token(int f,char **data)
+static int32 simple_recv_token(int f, char **data)
{
- static int residue;
+ static int32 residue;
static char *buf;
- int n;
+ int32 n;
if (!buf) {
- buf = (char *)malloc(CHUNK_SIZE);
- if (!buf) out_of_memory("simple_recv_token");
+ buf = new_array(char, CHUNK_SIZE);
+ if (!buf)
+ out_of_memory("simple_recv_token");
}
if (residue == 0) {
- int i = read_int(f);
- if (i <= 0) return i;
+ int32 i = read_int(f);
+ if (i <= 0)
+ return i;
residue = i;
}
return n;
}
-
/* non-compressing send token */
-static void simple_send_token(int f,int token,
- struct map_struct *buf,OFF_T offset,int n)
+static void simple_send_token(int f, int32 token, struct map_struct *buf,
+ OFF_T offset, int32 n)
{
- extern int write_batch; /* dw */
- int hold_int; /* dw */
-
if (n > 0) {
- int l = 0;
- while (l < n) {
- int n1 = MIN(CHUNK_SIZE,n-l);
- write_int(f,n1);
- write_buf(f,map_ptr(buf,offset+l,n1),n1);
- if (write_batch) {
- write_batch_delta_file( (char *) &n1, sizeof(int) );
- write_batch_delta_file(map_ptr(buf,offset+l,n1),n1);
- }
- l += n1;
+ int32 len = 0;
+ while (len < n) {
+ int32 n1 = MIN(CHUNK_SIZE, n-len);
+ write_int(f, n1);
+ write_buf(f, map_ptr(buf, offset+len, n1), n1);
+ len += n1;
}
}
/* a -2 token means to send data only and no token */
- if (token != -2) {
- write_int(f,-(token+1));
- if (write_batch) {
- hold_int = -(token+1);
- write_batch_delta_file( (char *) &hold_int, sizeof(int) );
- }
- }
+ if (token != -2)
+ write_int(f, -(token+1));
}
-
/* Flag bytes in compressed stream are encoded as follows: */
#define END_FLAG 0 /* that's all folks */
#define TOKEN_LONG 0x20 /* followed by 32-bit token number */
#define MAX_DATA_COUNT 16383 /* fit 14 bit count into 2 bytes with flags */
+/* zlib.h says that if we want to be able to compress something in a single
+ * call, avail_out must be at least 0.1% larger than avail_in plus 12 bytes.
+ * We'll add in 0.1%+16, just to be safe (and we'll avoid floating point,
+ * to ensure that this is a compile-time value). */
+#define AVAIL_OUT_SIZE(avail_in_size) ((avail_in_size)*1001/1000+16)
+
/* For coding runs of tokens */
-static int last_token = -1;
-static int run_start;
-static int last_run_end;
+static int32 last_token = -1;
+static int32 run_start;
+static int32 last_run_end;
/* Deflation state */
static z_stream tx_strm;
/* Output buffer */
static char *obuf;
-static int obuf_size;
+
+/* We want obuf to be able to hold both MAX_DATA_COUNT+2 bytes as well as
+ * AVAIL_OUT_SIZE(CHUNK_SIZE) bytes, so make sure that it's large enough. */
+#if MAX_DATA_COUNT+2 > AVAIL_OUT_SIZE(CHUNK_SIZE)
+#define OBUF_SIZE (MAX_DATA_COUNT+2)
+#else
+#define OBUF_SIZE AVAIL_OUT_SIZE(CHUNK_SIZE)
+#endif
/* Send a deflated token */
static void
-send_deflated_token(int f, int token,
- struct map_struct *buf, OFF_T offset, int nb, int toklen)
+send_deflated_token(int f, int32 token, struct map_struct *buf, OFF_T offset,
+ int32 nb, int32 toklen)
{
- int n, r;
+ int32 n, r;
static int init_done, flush_pending;
- extern int write_batch; /* dw */
- char temp_byte; /* dw */
if (last_token == -1) {
/* initialization */
Z_DEFLATED, -15, 8,
Z_DEFAULT_STRATEGY) != Z_OK) {
rprintf(FERROR, "compression init failed\n");
- exit_cleanup(RERR_STREAMIO);
+ exit_cleanup(RERR_PROTOCOL);
}
-#if MAX_DATA_COUNT+2 > CHUNK_SIZE+128 /* this shouldn't ever happen... */
- obuf_size = MAX_DATA_COUNT+2;
-#else
- obuf_size = CHUNK_SIZE+128;
-#endif
- if ((obuf = malloc(obuf_size)) == NULL)
+ if ((obuf = new_array(char, OBUF_SIZE)) == NULL)
out_of_memory("send_deflated_token");
init_done = 1;
} else
last_run_end = 0;
run_start = token;
flush_pending = 0;
-
} else if (last_token == -2) {
run_start = token;
-
} else if (nb != 0 || token != last_token + 1
|| token >= run_start + 65536) {
/* output previous run */
n = last_token - run_start;
if (r >= 0 && r <= 63) {
write_byte(f, (n==0? TOKEN_REL: TOKENRUN_REL) + r);
- if (write_batch) { /* dw */
- temp_byte = (char)( (n==0? TOKEN_REL: TOKENRUN_REL) + r);
- write_batch_delta_file(&temp_byte,sizeof(char));
- }
} else {
write_byte(f, (n==0? TOKEN_LONG: TOKENRUN_LONG));
write_int(f, run_start);
- if (write_batch) { /* dw */
- temp_byte = (char)(n==0? TOKEN_LONG: TOKENRUN_LONG);
- write_batch_delta_file(&temp_byte,sizeof(temp_byte));
- write_batch_delta_file((char *)&run_start,sizeof(run_start));
- }
}
if (n != 0) {
write_byte(f, n);
write_byte(f, n >> 8);
- if (write_batch) { /* dw */
- write_batch_delta_file((char *)&n,sizeof(char));
- temp_byte = (char) n >> 8;
- write_batch_delta_file(&temp_byte,sizeof(temp_byte));
- }
}
last_run_end = last_token;
run_start = token;
obuf[0] = DEFLATED_DATA + (n >> 8);
obuf[1] = n;
write_buf(f, obuf, n+2);
- if (write_batch) /* dw */
- write_batch_delta_file(obuf,n+2);
}
}
} while (nb != 0 || tx_strm.avail_out == 0);
if (token == -1) {
/* end of file - clean up */
write_byte(f, END_FLAG);
- if (write_batch) { /* dw */
- temp_byte = END_FLAG;
- write_batch_delta_file((char *)&temp_byte,sizeof(temp_byte));
- }
-
} else if (token != -2) {
- /* add the data in the current block to the compressor's
- history and hash table */
- tx_strm.next_in = (Bytef *) map_ptr(buf, offset, toklen);
- tx_strm.avail_in = toklen;
- tx_strm.next_out = (Bytef *) obuf;
- tx_strm.avail_out = obuf_size;
- r = deflate(&tx_strm, Z_INSERT_ONLY);
- if (r != Z_OK || tx_strm.avail_in != 0) {
- rprintf(FERROR, "deflate on token returned %d (%d bytes left)\n",
- r, tx_strm.avail_in);
- exit_cleanup(RERR_STREAMIO);
- }
+ /* Add the data in the current block to the compressor's
+ * history and hash table. */
+ do {
+ /* Break up long sections in the same way that
+ * see_deflate_token() does. */
+ int32 n1 = toklen > 0xffff ? 0xffff : toklen;
+ toklen -= n1;
+ tx_strm.next_in = (Bytef *)map_ptr(buf, offset, n1);
+ tx_strm.avail_in = n1;
+ tx_strm.next_out = (Bytef *) obuf;
+ tx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
+ r = deflate(&tx_strm, Z_INSERT_ONLY);
+ if (r != Z_OK || tx_strm.avail_in != 0) {
+ rprintf(FERROR, "deflate on token returned %d (%d bytes left)\n",
+ r, tx_strm.avail_in);
+ exit_cleanup(RERR_STREAMIO);
+ }
+ } while (toklen > 0);
}
}
-
/* tells us what the receiver is in the middle of doing */
static enum { r_init, r_idle, r_running, r_inflating, r_inflated } recv_state;
static char *dbuf;
/* for decoding runs of tokens */
-static int rx_token;
-static int rx_run;
+static int32 rx_token;
+static int32 rx_run;
/* Receive a deflated token and inflate it */
-static int
-recv_deflated_token(int f, char **data)
+static int32 recv_deflated_token(int f, char **data)
{
- int n, r, flag;
static int init_done;
- static int saved_flag;
+ static int32 saved_flag;
+ int32 n, flag;
+ int r;
for (;;) {
switch (recv_state) {
rx_strm.zfree = NULL;
if (inflateInit2(&rx_strm, -15) != Z_OK) {
rprintf(FERROR, "inflate init failed\n");
- exit_cleanup(RERR_STREAMIO);
+ exit_cleanup(RERR_PROTOCOL);
}
- if ((cbuf = malloc(MAX_DATA_COUNT)) == NULL
- || (dbuf = malloc(CHUNK_SIZE)) == NULL)
+ if (!(cbuf = new_array(char, MAX_DATA_COUNT))
+ || !(dbuf = new_array(char, AVAIL_OUT_SIZE(CHUNK_SIZE))))
out_of_memory("recv_deflated_token");
init_done = 1;
} else {
/* check previous inflated stuff ended correctly */
rx_strm.avail_in = 0;
rx_strm.next_out = (Bytef *)dbuf;
- rx_strm.avail_out = CHUNK_SIZE;
+ rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
r = inflate(&rx_strm, Z_SYNC_FLUSH);
- n = CHUNK_SIZE - rx_strm.avail_out;
+ n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
/*
* Z_BUF_ERROR just means no progress was
* made, i.e. the decompressor didn't have
case r_inflating:
rx_strm.next_out = (Bytef *)dbuf;
- rx_strm.avail_out = CHUNK_SIZE;
+ rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
r = inflate(&rx_strm, Z_NO_FLUSH);
- n = CHUNK_SIZE - rx_strm.avail_out;
+ n = AVAIL_OUT_SIZE(CHUNK_SIZE) - rx_strm.avail_out;
if (r != Z_OK) {
rprintf(FERROR, "inflate returned %d (%d bytes)\n", r, n);
exit_cleanup(RERR_STREAMIO);
* put the data corresponding to a token that we've just returned
* from recv_deflated_token into the decompressor's history buffer.
*/
-static void see_deflate_token(char *buf, int len)
+static void see_deflate_token(char *buf, int32 len)
{
- int r, blklen;
+ int r;
+ int32 blklen;
unsigned char hdr[5];
rx_strm.avail_in = 0;
}
}
rx_strm.next_out = (Bytef *)dbuf;
- rx_strm.avail_out = CHUNK_SIZE;
+ rx_strm.avail_out = AVAIL_OUT_SIZE(CHUNK_SIZE);
r = inflate(&rx_strm, Z_SYNC_FLUSH);
- if (r != Z_OK) {
+ if (r != Z_OK && r != Z_BUF_ERROR) {
rprintf(FERROR, "inflate (token) returned %d\n", r);
exit_cleanup(RERR_STREAMIO);
}
/**
* Transmit a verbatim buffer of length @p n followed by a token.
- * If token == -1 then we have reached EOF
+ * If token == -1 then we have reached EOF
* If n == 0 then don't send a buffer
*/
-void send_token(int f,int token,struct map_struct *buf,OFF_T offset,
- int n,int toklen)
+void send_token(int f, int32 token, struct map_struct *buf, OFF_T offset,
+ int32 n, int32 toklen)
{
- if (!do_compression) {
- simple_send_token(f,token,buf,offset,n);
- } else {
+ if (!do_compression)
+ simple_send_token(f, token, buf, offset, n);
+ else
send_deflated_token(f, token, buf, offset, n, toklen);
- }
}
-
/*
* receive a token or buffer from the other end. If the reurn value is >0 then
* it is a data buffer of that length, and *data will point at the data.
* if the return value is -i then it represents token i-1
* if the return value is 0 then the end has been reached
*/
-int recv_token(int f,char **data)
+int32 recv_token(int f, char **data)
{
int tok;
/*
* look at the data corresponding to a token, if necessary
*/
-void see_token(char *data, int toklen)
+void see_token(char *data, int32 toklen)
{
if (do_compression)
see_deflate_token(data, toklen);