Add basic wrappers for zlib/gzip

svn:r2324

4 files changed, 302 insertions, 7 deletions

diff --git a/src/common/Makefile.am b/src/common/Makefile.am
index 205fca43d..cf18c3fec 100644
--- a/src/common/Makefile.am
+++ b/src/common/Makefile.am
@@ -3,6 +3,6 @@ noinst_LIBRARIES = libor.a
 
 #CFLAGS  = -Wall -Wpointer-arith -O2
 
-libor_a_SOURCES = log.c crypto.c fakepoll.c util.c aes.c tortls.c
+libor_a_SOURCES = log.c crypto.c fakepoll.c util.c aes.c tortls.c torgzip.c
 
-noinst_HEADERS = log.h crypto.h fakepoll.h test.h util.h aes.h torint.h tortls.h strlcpy.c strlcat.c
+noinst_HEADERS = log.h crypto.h fakepoll.h test.h util.h aes.h torint.h tortls.h strlcpy.c strlcat.c torgzip.h
diff --git a/src/common/torgzip.c b/src/common/torgzip.c
new file mode 100644
index 000000000..1b3fce2e8
--- /dev/null
+++ b/src/common/torgzip.c
@@ -0,0 +1,221 @@
+/* Copyright 2004 Roger Dingledine */
+/* See LICENSE for licensing information */
+/* $Id$ */
+
+/**
+ * \file torgzip.c
+ *
+ * \brief Simple in-memory gzip implementation.
+ **/
+
+#include "orconfig.h"
+
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <assert.h>
+#include <zlib.h>
+
+#include "util.h"
+#include "log.h"
+#include "torgzip.h"
+
+static int gzip_is_supported = -1;
+
+int
+is_gzip_supported(void)
+{
+  if (gzip_is_supported >= 0)
+    return gzip_is_supported;
+
+  if (!strcmpstart(ZLIB_VERSION, "0.") ||
+      !strcmpstart(ZLIB_VERSION, "1.0") ||
+      !strcmpstart(ZLIB_VERSION, "1.1"))
+    gzip_is_supported = 0;
+  else
+    gzip_is_supported = 1;
+
+  return gzip_is_supported;
+}
+
+static INLINE int
+method_bits(compress_method_t method)
+{
+  /* Bits+16 means "use gzip" in zlib >= 1.2 */
+  return method == GZIP_METHOD ? 15+16 : 15;
+}
+
+int
+tor_gzip_compress(char **out, size_t *out_len,
+		  const char *in, size_t in_len,
+                  compress_method_t method)
+{
+  struct z_stream_s *stream = NULL;
+  size_t out_size;
+  off_t offset;
+
+  tor_assert(out && out_len && in);
+
+  if (method == GZIP_METHOD && !is_gzip_supported()) {
+    /* Old zlib version don't support gzip in deflateInit2 */
+    log_fn(LOG_WARN, "Gzip not supported with zlib %s", ZLIB_VERSION);
+    return -1;
+  }
+
+  *out = NULL;
+
+  stream = tor_malloc_zero(sizeof(struct z_stream_s));
+  stream->zalloc = Z_NULL;
+  stream->zfree = Z_NULL;
+  stream->opaque = NULL;
+  stream->next_in = (unsigned char*) in;
+  stream->avail_in = in_len;
+
+  if (deflateInit2(stream, Z_BEST_COMPRESSION, Z_DEFLATED,
+		   method_bits(method),
+		   8, Z_DEFAULT_STRATEGY) != Z_OK) {
+    printf("Z");
+    log_fn(LOG_WARN, "Error from deflateInit2: %s",
+	   stream->msg?stream->msg:"<no message>");
+    goto err;
+  }
+
+  /* Guess 50% compression. */
+  out_size = in_len / 2;
+  if (out_size < 1024) out_size = 1024;
+  *out = tor_malloc(out_size);
+  stream->next_out = *out;
+  stream->avail_out = out_size;
+
+  while (1) {
+    switch (deflate(stream, Z_FINISH))
+      {
+      case Z_STREAM_END:
+        goto done;
+      case Z_OK:
+        /* In case zlib doesn't work as I think .... */
+        if (stream->avail_out >= stream->avail_in+16)
+          break;
+      case Z_BUF_ERROR:
+	offset = stream->next_out - ((unsigned char*)*out);
+	out_size *= 2;
+	*out = tor_realloc(*out, out_size);
+	stream->next_out = *out + offset;
+	stream->avail_out = out_size - offset;
+	break;
+      default:
+        log_fn(LOG_WARN, "Gzip compression didn't finish: %s",
+               stream->msg ? stream->msg : "<no message>");
+        goto err;
+      }
+  }
+ done:
+  *out_len = stream->total_out;
+  if (deflateEnd(stream)!=Z_OK) {
+    log_fn(LOG_WARN, "Error freeing gzip structures");
+    goto err;
+  }
+  tor_free(stream);
+
+  return 0;
+ err:
+  if (stream) {
+    deflateEnd(stream);
+    tor_free(stream);
+  }
+  if (*out) {
+    tor_free(*out);
+  }
+  return -1;
+}
+
+int
+tor_gzip_uncompress(char **out, size_t *out_len,
+		    const char *in, size_t in_len,
+                    compress_method_t method)
+{
+  struct z_stream_s *stream = NULL;
+  size_t out_size;
+  off_t offset;
+
+  tor_assert(out && out_len && in);
+
+  if (method == GZIP_METHOD && !is_gzip_supported()) {
+    /* Old zlib version don't support gzip in inflateInit2 */
+    log_fn(LOG_WARN, "Gzip not supported with zlib %s", ZLIB_VERSION);
+    return -1;
+  }
+
+  *out = NULL;
+
+  stream = tor_malloc_zero(sizeof(struct z_stream_s));
+  stream->zalloc = Z_NULL;
+  stream->zfree = Z_NULL;
+  stream->opaque = NULL;
+  stream->next_in = (unsigned char*) in;
+  stream->avail_in = in_len;
+
+  if (inflateInit2(stream,
+                   method_bits(method)) != Z_OK) {
+    log_fn(LOG_WARN, "Error from inflateInit2: %s",
+	   stream->msg?stream->msg:"<no message>");
+    goto err;
+  }
+
+  out_size = in_len * 2;  /* guess 50% compression. */
+  if (out_size < 1024) out_size = 1024;
+
+  *out = tor_malloc(out_size);
+  stream->next_out = *out;
+  stream->avail_out = out_size;
+
+  while (1) {
+    switch(inflate(stream, Z_FINISH))
+      {
+      case Z_STREAM_END:
+	goto done;
+      case Z_OK:
+	/* In case zlib doesn't work as I think.... */
+	if (stream->avail_out >= stream->avail_in+16)
+	  break;
+      case Z_BUF_ERROR:
+	offset = stream->next_out - ((unsigned char*)*out);
+	out_size *= 2;
+	*out = tor_realloc(*out, out_size);
+	stream->next_out = *out + offset;
+	stream->avail_out = out_size - offset;
+	break;
+      default:
+	log_fn(LOG_WARN, "Gzip decompression returned an error: %s",
+	       stream->msg ? stream->msg : "<no message>");
+	goto err;
+      }
+
+  }
+ done:
+  *out_len = stream->total_out;
+  if (inflateEnd(stream)!=Z_OK) {
+    log_fn(LOG_WARN, "Error freeing gzip structures");
+    goto err;
+  }
+  tor_free(stream);
+
+  return 0;
+ err:
+  if (stream) {
+    inflateEnd(stream);
+    tor_free(stream);
+  }
+  if (*out) {
+    tor_free(*out);
+  }
+  return -1;
+}
+
+/*
+  Local Variables:
+  mode:c
+  indent-tabs-mode:nil
+  c-basic-offset:2
+  End:
+*/
diff --git a/src/common/torgzip.h b/src/common/torgzip.h
new file mode 100644
index 000000000..61bb6adbc
--- /dev/null
+++ b/src/common/torgzip.h
@@ -0,0 +1,26 @@
+/* Copyright 2003 Roger Dingledine */
+/* See LICENSE for licensing information */
+/* $Id$ */
+
+/**
+ * \file torgzip.h
+ * \brief Headers for torgzip.h
+ **/
+
+#ifndef __TORGZIP_H
+#define __TORGZIP_H
+
+typedef enum { GZIP_METHOD, ZLIB_METHOD } compress_method_t;
+
+int
+tor_gzip_compress(char **out, size_t *out_len,
+		  const char *in, size_t in_len,
+		  compress_method_t method);
+int
+tor_gzip_uncompress(char **out, size_t *out_len,
+		    const char *in, size_t in_len,
+		    compress_method_t method);
+
+int is_gzip_supported(void);
+
+#endif
diff --git a/src/or/test.c b/src/or/test.c
index d9badb746..0ca1ab290 100644
--- a/src/or/test.c
+++ b/src/or/test.c
@@ -14,6 +14,7 @@
 #include <dirent.h>
 #include "or.h"
 #include "../common/test.h"
+#include "../common/torgzip.h"
 
 extern or_options_t options;
 
@@ -552,15 +553,15 @@ test_util() {
   test_eq((void*)555, smartlist_get(sl,5));
 
   smartlist_clear(sl);
-  smartlist_split_string(sl, "abc", ":", 0);
+  smartlist_split_string(sl, "abc", ":", 0, 0);
   test_eq(1, smartlist_len(sl));
   test_streq("abc", smartlist_get(sl, 0));
-  smartlist_split_string(sl, "a::bc::", "::", 0);
+  smartlist_split_string(sl, "a::bc::", "::", 0, 0);
   test_eq(4, smartlist_len(sl));
   test_streq("a", smartlist_get(sl, 1));
   test_streq("bc", smartlist_get(sl, 2));
   test_streq("", smartlist_get(sl, 3));
-  smartlist_split_string(sl, "/def/  /ghijk", "/", 0);
+  smartlist_split_string(sl, "/def/  /ghijk", "/", 0, 0);
   test_eq(8, smartlist_len(sl));
   test_streq("", smartlist_get(sl, 4));
   test_streq("def", smartlist_get(sl, 5));
@@ -569,12 +570,12 @@ test_util() {
   SMARTLIST_FOREACH(sl, char *, cp, tor_free(cp));
   smartlist_clear(sl);
 
-  smartlist_split_string(sl, "a,bbd,cdef", ",", 1);
+  smartlist_split_string(sl, "a,bbd,cdef", ",", SPLIT_SKIP_SPACE, 0);
   test_eq(3, smartlist_len(sl));
   test_streq("a", smartlist_get(sl,0));
   test_streq("bbd", smartlist_get(sl,1));
   test_streq("cdef", smartlist_get(sl,2));
-  smartlist_split_string(sl, " z <> zhasd <>  <> bnud<>   ", "<>", 1);
+  smartlist_split_string(sl, " z <> zhasd <>  <> bnud<>   ", "<>", SPLIT_SKIP_SPACE, 0);
   test_eq(8, smartlist_len(sl));
   test_streq("z", smartlist_get(sl,3));
   test_streq("zhasd", smartlist_get(sl,4));
@@ -583,11 +584,57 @@ test_util() {
   test_streq("", smartlist_get(sl,7));
 
   SMARTLIST_FOREACH(sl, char *, cp, tor_free(cp));
+  smartlist_clear(sl);
+
+  smartlist_split_string(sl, " z <> zhasd <>  <> bnud<>   ", "<>", SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 0);
+  test_eq(3, smartlist_len(sl));
+  test_streq("z", smartlist_get(sl, 0));
+  test_streq("zhasd", smartlist_get(sl, 1));
+  test_streq("bnud", smartlist_get(sl, 2));
+  smartlist_split_string(sl, " z <> zhasd <>  <> bnud<>   ", "<>", SPLIT_SKIP_SPACE|SPLIT_IGNORE_BLANK, 2);
+  test_eq(5, smartlist_len(sl));
+  test_streq("z", smartlist_get(sl, 3));
+  test_streq("zhasd <>  <> bnud<>", smartlist_get(sl, 4));
+
 
   /* XXXX test older functions. */
   smartlist_free(sl);
 }
 
+void
+test_gzip() {
+  char *buf1, *buf2=NULL, *buf3=NULL;
+  size_t len1, len2;
+
+  buf1 = tor_strdup("AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA");
+  if (is_gzip_supported()) {
+    test_assert(!tor_gzip_compress(&buf2, &len1, buf1, strlen(buf1)+1,
+                                   GZIP_METHOD));
+    test_assert(buf2);
+    test_assert(!memcmp(buf2, "\037\213", 2)); /* Gzip magic. */
+
+    test_assert(!tor_gzip_uncompress(&buf3, &len2, buf2, len1, GZIP_METHOD));
+    test_assert(buf3);
+    test_streq(buf1,buf3);
+
+    tor_free(buf2);
+    tor_free(buf3);
+  }
+
+  test_assert(!tor_gzip_compress(&buf2, &len1, buf1, strlen(buf1)+1,
+                                 ZLIB_METHOD));
+  test_assert(buf2);
+  test_assert(!memcmp(buf2, "\x78\xDA", 2)); /* deflate magic. */
+
+  test_assert(!tor_gzip_uncompress(&buf3, &len2, buf2, len1, ZLIB_METHOD));
+  test_assert(buf3);
+  test_streq(buf1,buf3);
+
+  tor_free(buf2);
+  tor_free(buf3);
+  tor_free(buf1);
+}
+
 static void* _squareAndRemoveK4(const char *key, void*val, void *data)
 {
   int *ip = (int*)data;
@@ -1037,6 +1084,7 @@ main(int c, char**v){
   test_crypto();
   test_crypto_dh();
   puts("\n========================= Util ============================");
+  test_gzip();
   test_util();
   test_strmap();
   puts("\n========================= Onion Skins =====================");