Make router/directory parsing nondestructive and more const-friendly

svn:r890

src/common/crypto.c

@@ -441,7 +441,7 @@ int crypto_pk_write_public_key_to_string(crypto_pk_env_t *env, char **dest, int
   return 0;
 }
 
-int crypto_pk_read_public_key_from_string(crypto_pk_env_t *env, char *src, int len) {
+int crypto_pk_read_public_key_from_string(crypto_pk_env_t *env, const char *src, int len) {
   BIO *b; 
 
   assert(env && src);
@@ -820,7 +820,7 @@ crypto_cipher_advance(crypto_cipher_env_t *env, long delta)
 
 
 /* SHA-1 */
-int crypto_SHA_digest(unsigned char *m, int len, unsigned char *digest)
+int crypto_SHA_digest(const unsigned char *m, int len, unsigned char *digest)
 {
   assert(m && digest);
   return (SHA1(m,len,digest) == NULL);
@@ -1034,7 +1034,7 @@ char *crypto_perror()
 }
 
 int 
-base64_encode(char *dest, int destlen, char *src, int srclen)
+base64_encode(char *dest, int destlen, const char *src, int srclen)
 {
   EVP_ENCODE_CTX ctx;
   int len, ret;
@@ -1046,13 +1046,13 @@ base64_encode(char *dest, int destlen, char *src, int srclen)
     return -1;
 
   EVP_EncodeInit(&ctx);
-  EVP_EncodeUpdate(&ctx, dest, &len, src, srclen);
+  EVP_EncodeUpdate(&ctx, dest, &len, (char*) src, srclen);
   EVP_EncodeFinal(&ctx, dest+len, &ret);
   ret += len;
   return ret;
 }
 int 
-base64_decode(char *dest, int destlen, char *src, int srclen)
+base64_decode(char *dest, int destlen, const char *src, int srclen)
 {
   EVP_ENCODE_CTX ctx;
   int len, ret;
@@ -1063,7 +1063,7 @@ base64_decode(char *dest, int destlen, char *src, int srclen)
     return -1;
 
   EVP_DecodeInit(&ctx);
-  EVP_DecodeUpdate(&ctx, dest, &len, src, srclen);
+  EVP_DecodeUpdate(&ctx, dest, &len, (char*) src, srclen);
   EVP_DecodeFinal(&ctx, dest, &ret);
   ret += len;
   return ret;

src/common/crypto.h

@@ -38,7 +38,7 @@ int crypto_pk_generate_key(crypto_pk_env_t *env);
 int crypto_pk_read_private_key_from_file(crypto_pk_env_t *env, FILE *src);
 int crypto_pk_read_public_key_from_file(crypto_pk_env_t *env, FILE *src);
 int crypto_pk_write_public_key_to_string(crypto_pk_env_t *env, char **dest, int *len);
-int crypto_pk_read_public_key_from_string(crypto_pk_env_t *env, char *src, int len);
+int crypto_pk_read_public_key_from_string(crypto_pk_env_t *env, const char *src, int len);
 int crypto_pk_write_private_key_to_file(crypto_pk_env_t *env, FILE *dest);
 int crypto_pk_write_private_key_to_filename(crypto_pk_env_t *env, const char *fname);
 int crypto_pk_write_public_key_to_file(crypto_pk_env_t *env, FILE *dest);
@@ -58,8 +58,8 @@ int crypto_pk_public_checksig(crypto_pk_env_t *env, unsigned char *from, int fro
 int crypto_pk_get_fingerprint(crypto_pk_env_t *pk, char *fp_out);
 int crypto_pk_check_fingerprint_syntax(const char *s);
 
-int base64_encode(char *dest, int destlen, char *src, int srclen);
-int base64_decode(char *dest, int destlen, char *src, int srclen);
+int base64_encode(char *dest, int destlen, const char *src, int srclen);
+int base64_decode(char *dest, int destlen, const char *src, int srclen);
 
 /* Key negotiation */
 typedef struct crypto_dh_env_st {
@@ -95,7 +95,7 @@ int crypto_cipher_advance(crypto_cipher_env_t *env, long delta);
 crypto_cipher_env_t *crypto_create_init_cipher(int cipher_type, char *key, char *iv, int encrypt_mode);
 
 /* SHA-1 */
-int crypto_SHA_digest(unsigned char *m, int len, unsigned char *digest);
+int crypto_SHA_digest(const unsigned char *m, int len, unsigned char *digest);
 
 /* random numbers */
 int crypto_seed_rng();

src/common/util.c

@@ -54,12 +54,21 @@ char *tor_strdup(const char *s) {
   return dup;
 }
 
+char *tor_strndup(const char *s, size_t n) {
+  char *dup;
+  assert(s);
+  dup = tor_malloc(n+1);
+  strncpy(dup, s, n);
+  dup[n] = 0;
+  return dup;
+}
+
 /*
  *    String manipulation
  */
 
 /* return the first char of s that is not whitespace and not a comment */
-char *eat_whitespace(char *s) {
+const char *eat_whitespace(const char *s) {
   assert(s);
 
   while(isspace(*s) || *s == '#') {
@@ -75,14 +84,14 @@ char *eat_whitespace(char *s) {
   return s;
 }
 
-char *eat_whitespace_no_nl(char *s) {
+const char *eat_whitespace_no_nl(const char *s) {
   while(*s == ' ' || *s == '\t') 
     ++s;
   return s;
 }
 
 /* return the first char of s that is whitespace or '#' or '\0 */
-char *find_whitespace(char *s) {
+const char *find_whitespace(const char *s) {
   assert(s);
 
   while(*s && !isspace(*s) && *s != '#')

src/common/util.h

@@ -36,11 +36,12 @@ void *tor_malloc(size_t size);
 void *tor_malloc_zero(size_t size);
 void *tor_realloc(void *ptr, size_t size);
 char *tor_strdup(const char *s);
+char *tor_strndup(const char *s, size_t n);
 #define tor_free(p) do {if(p) {free(p); (p)=NULL;}} while(0)
 
-char *eat_whitespace(char *s);
-char *eat_whitespace_no_nl(char *s);
-char *find_whitespace(char *s);
+const char *eat_whitespace(const char *s);
+const char *eat_whitespace_no_nl(const char *s);
+const char *find_whitespace(const char *s);
 
 void tor_gettimeofday(struct timeval *timeval);
 long tv_udiff(struct timeval *start, struct timeval *end);

src/or/dirserv.c

@@ -202,7 +202,8 @@ dirserv_add_descriptor(const char **desc)
   routerinfo_t *ri = NULL;
   int i, r;
   char *start, *end;
-  char *desc_tmp = NULL, *cp;
+  char *desc_tmp = NULL;
+  const char *cp;
   size_t desc_len;
 
   start = strstr(*desc, "router ");
@@ -218,9 +219,7 @@ dirserv_add_descriptor(const char **desc)
     end = start+strlen(start);
   }
   desc_len = end-start;
-  cp = desc_tmp = tor_malloc(desc_len+1);
-  strncpy(desc_tmp, start, desc_len);
-  desc_tmp[desc_len]='\0';
+  cp = desc_tmp = tor_strndup(start, desc_len);
 
   /* Check: is the descriptor syntactically valid? */
   ri = router_get_entry_from_string(&cp);

src/or/or.h

@@ -749,11 +749,11 @@ void router_get_routerlist(routerlist_t **prouterlist);
 void routerinfo_free(routerinfo_t *router);
 void router_mark_as_down(char *nickname);
 int router_set_routerlist_from_file(char *routerfile);
-int router_get_dir_hash(char *s, char *digest);
-int router_get_router_hash(char *s, char *digest);
-int router_set_routerlist_from_directory(char *s, crypto_pk_env_t *pkey);
-routerinfo_t *router_get_entry_from_string(char**s);
-int router_add_exit_policy_from_string(routerinfo_t *router, char *s);
+int router_get_dir_hash(const char *s, char *digest);
+int router_get_router_hash(const char *s, char *digest);
+int router_set_routerlist_from_directory(const char *s, crypto_pk_env_t *pkey);
+routerinfo_t *router_get_entry_from_string(const char **s);
+int router_add_exit_policy_from_string(routerinfo_t *router, const char *s);
 int router_supports_exit_address(uint32_t addr, uint16_t port,
                                  routerinfo_t *router);
 int router_compare_addr_to_exit_policy(uint32_t addr, uint16_t port,

src/or/router.c

@@ -378,7 +378,8 @@ int router_dump_router_to_string(char *s, int maxlen, routerinfo_t *router,
   int result=0;
   struct exit_policy_t *tmpe;
 #ifdef DEBUG_ROUTER_DUMP_ROUTER_TO_STRING
-  char *s_tmp, *s_dup;
+  char *s_tmp, *s_dup; 
+  const char *cp;
   routerinfo_t *ri_tmp;
 #endif
   
@@ -494,8 +495,8 @@ int router_dump_router_to_string(char *s, int maxlen, routerinfo_t *router,
   s[written+1] = 0;
 
 #ifdef DEBUG_ROUTER_DUMP_ROUTER_TO_STRING
-  s_tmp = s_dup = tor_strdup(s);
-  ri_tmp = router_get_entry_from_string(&s_tmp);
+  cp = s_tmp = s_dup = tor_strdup(s);
+  ri_tmp = router_get_entry_from_string(&cp);
   if (!ri_tmp) {
     log_fn(LOG_ERR, "We just generated a router descriptor we can't parse: <<%s>>", 
            s);

src/or/routerlist.c

@@ -21,17 +21,17 @@ extern or_options_t options; /* command-line and config-file options */
 
 /****************************************************************************/
 
-struct directory_token;
-typedef struct directory_token directory_token_t;
+struct directory_token_t;
+typedef struct directory_token_t directory_token_t;
 
 /* static function prototypes */
-static int router_set_routerlist_from_string(char *s);
+static int router_set_routerlist_from_string(const char *s);
 static int
-router_get_list_from_string_impl(char **s, routerlist_t **dest,
+router_get_list_from_string_impl(const char **s, routerlist_t **dest,
                                  int n_good_nicknames,
                                  const char **good_nickname_lst);
 static int
-router_get_routerlist_from_directory_impl(char *s, routerlist_t **dest,
+router_get_routerlist_from_directory_impl(const char *s, routerlist_t **dest,
                                           crypto_pk_env_t *pkey);
 static int router_add_exit_policy(routerinfo_t *router, 
                                   directory_token_t *tok);
@@ -245,30 +245,38 @@ typedef enum {
   _SIGNATURE,
   _PUBLIC_KEY,
   _ERR,
-  _EOF
+  _EOF,
+  _NIL
 } directory_keyword;
 
-struct token_table_ent { char *t; int v; };
+typedef enum {
+  NO_ARGS,
+  ARGS,
+  CONCAT_ARGS,
+} arg_syntax;
+ 
+
+struct token_table_ent { char *t; int v; arg_syntax s; };
 
 static struct token_table_ent token_table[] = {
-  { "accept", K_ACCEPT },
-  { "directory-signature", K_DIRECTORY_SIGNATURE },
-  { "reject", K_REJECT },
-  { "router", K_ROUTER },
-  { "recommended-software", K_RECOMMENDED_SOFTWARE },
-  { "signed-directory", K_SIGNED_DIRECTORY },
-  { "signing-key", K_SIGNING_KEY },
-  { "onion-key", K_ONION_KEY },
-  { "link-key", K_LINK_KEY },
-  { "router-signature", K_ROUTER_SIGNATURE },
-  { "published", K_PUBLISHED },
-  { "running-routers", K_RUNNING_ROUTERS },
-  { "platform", K_PLATFORM },
+  { "accept", K_ACCEPT, ARGS },
+  { "directory-signature", K_DIRECTORY_SIGNATURE, NO_ARGS },
+  { "reject", K_REJECT, ARGS },
+  { "router", K_ROUTER, ARGS },
+  { "recommended-software", K_RECOMMENDED_SOFTWARE, ARGS },
+  { "signed-directory", K_SIGNED_DIRECTORY, NO_ARGS },
+  { "signing-key", K_SIGNING_KEY, NO_ARGS },
+  { "onion-key", K_ONION_KEY, NO_ARGS },
+  { "link-key", K_LINK_KEY, NO_ARGS },
+  { "router-signature", K_ROUTER_SIGNATURE, NO_ARGS },
+  { "published", K_PUBLISHED, CONCAT_ARGS },
+  { "running-routers", K_RUNNING_ROUTERS, ARGS },
+  { "platform", K_PLATFORM, ARGS },
   { NULL, -1 }
 };
 
 #define MAX_ARGS 1024
-struct directory_token {
+struct directory_token_t {
   directory_keyword tp;
   union {
     struct {
@@ -287,6 +295,7 @@ struct directory_token {
 static void
 router_release_token(directory_token_t *tok)
 {
+  int i;
   switch (tok->tp) 
     {
     case _SIGNATURE:
@@ -295,14 +304,21 @@ router_release_token(directory_token_t *tok)
     case _PUBLIC_KEY:
       crypto_free_pk_env(tok->val.public_key);
       break;
-    default:
+    case _ERR:
+    case _EOF:
+    case _NIL:
       break;
+    default:
+      for (i = 0; i < tok->val.cmd.n_args; ++i) {
+        tor_free(tok->val.cmd.args[i]);
+      }
     }
+  tok->tp = _NIL;
 }
 
 static int
-_router_get_next_token(char **s, directory_token_t *tok) {
-  char *next;
+_router_get_next_token(const char **s, directory_token_t *tok) {
+  const char *next;
   crypto_pk_env_t *pkey = NULL;
   char *signature = NULL;
   int i, done;
@@ -310,6 +326,8 @@ _router_get_next_token(char **s, directory_token_t *tok) {
   tok->tp = _ERR;
   tok->val.error = "";
 
+  router_release_token(tok);
+  
   *s = eat_whitespace(*s);
   if (!**s) {
     tok->tp = _EOF;
@@ -368,21 +386,42 @@ _router_get_next_token(char **s, directory_token_t *tok) {
     for (i = 0 ; token_table[i].t ; ++i) {
       if (!strncmp(token_table[i].t, *s, next-*s)) {
         tok->tp = token_table[i].v;
-        i = 0;
-        done = (*next == '\n');
-        *s = eat_whitespace_no_nl(next);
-        while (**s != '\n' && i <= MAX_ARGS && !done) {
-          next = find_whitespace(*s);
-          if (*next == '\n')
-            done = 1;
-          *next = 0;
-          tok->val.cmd.args[i++] = *s;
+
+        if (token_table[i].s == ARGS) {
+          i = 0;
+          done = (*next == '\n');
+          *s = eat_whitespace_no_nl(next);
+          while (**s != '\n' && i < MAX_ARGS && !done) {
+            next = find_whitespace(*s);
+            if (*next == '\n')
+              done = 1;
+            tok->val.cmd.args[i++] = tor_strndup(*s,next-*s);
+            *s = eat_whitespace_no_nl(next+1);
+          }
+          tok->val.cmd.n_args = i;
+          if (i >= MAX_ARGS) {
+            /* XXX free args[0..i] */
+            tok->tp = _ERR;
+            tok->val.error = "Too many arguments"; return -1;
+          }
+        } else if (token_table[i].s == CONCAT_ARGS) {
+          *s = eat_whitespace_no_nl(next);
+          next = strchr(*s, '\n');
+          if (!next) {
+            tok->tp = _ERR;
+            tok->val.error = "Unexpected EOF"; return -1;
+          }
+          tok->val.cmd.args[0] = tor_strndup(*s,next-*s);
+          tok->val.cmd.n_args = 1;
           *s = eat_whitespace_no_nl(next+1);
-        };
-        tok->val.cmd.n_args = i;
-        if (i > MAX_ARGS) {
-          tok->tp = _ERR;
-          tok->val.error = "Too many arguments"; return -1;
+        } else {
+          *s = eat_whitespace_no_nl(next);
+          if (**s != '\n') {
+            tok->tp = _ERR;
+            tok->val.error = "Unexpected arguments"; return -1;
+          }
+          tok->val.cmd.n_args = 0;
+          *s = eat_whitespace_no_nl(*s+1);
         }
         return 0;
       }
@@ -444,7 +483,7 @@ router_get_next_token(char **s, directory_token_t *tok) {
 
 /* read routerinfo elements from s, and throw out the ones that
  * don't parse and resolve. */
-static int router_set_routerlist_from_string(char *s)
+static int router_set_routerlist_from_string(const char *s)
 {
   if (router_get_list_from_string_impl(&s, &routerlist, -1, NULL)) {
     log(LOG_WARN, "Error parsing router file");
@@ -457,7 +496,8 @@ static int router_set_routerlist_from_string(char *s)
   return 0;
 }
 
-static int router_get_hash_impl(char *s, char *digest, const char *start_str,
+static int router_get_hash_impl(const char *s, char *digest, 
+                                const char *start_str,
                                 const char *end_str)
 {
   char *start, *end;
@@ -486,22 +526,22 @@ static int router_get_hash_impl(char *s, char *digest, const char *start_str,
   return 0;
 }
 
-int router_get_dir_hash(char *s, char *digest)
+int router_get_dir_hash(const char *s, char *digest)
 {
   return router_get_hash_impl(s,digest,
                               "signed-directory","directory-signature");
 }
-int router_get_router_hash(char *s, char *digest)
+int router_get_router_hash(const char *s, char *digest)
 {
   return router_get_hash_impl(s,digest,
                               "router ","router-signature");
 }
 
 /* return 0 if myversion is in start. Else return -1. */
-int compare_recommended_versions(char *myversion, char *start) {
+int compare_recommended_versions(const char *myversion, const char *start) {
   int len_myversion = strlen(myversion);
   char *comma;
-  char *end = start + strlen(start);
+  const char *end = start + strlen(start);
 
   log_fn(LOG_DEBUG,"checking '%s' in '%s'.", myversion, start);
   
@@ -517,7 +557,7 @@ int compare_recommended_versions(char *myversion, char *start) {
   }
 }
 
-int router_set_routerlist_from_directory(char *s, crypto_pk_env_t *pkey)
+int router_set_routerlist_from_directory(const char *s, crypto_pk_env_t *pkey)
 {
   if (router_get_routerlist_from_directory_impl(s, &routerlist, pkey)) {
     log_fn(LOG_WARN, "Couldn't parse directory.");
@@ -544,18 +584,19 @@ int router_set_routerlist_from_directory(char *s, crypto_pk_env_t *pkey)
 }
 
 static int
-router_get_routerlist_from_directory_impl(char *s, routerlist_t **dest,
+router_get_routerlist_from_directory_impl(const char *s, routerlist_t **dest,
                                           crypto_pk_env_t *pkey)
 {
   directory_token_t tok;
   char digest[20];
   char signed_digest[128];
   routerlist_t *new_dir = NULL;
-  char *versions;
+  char *versions = NULL;
   struct tm published;
   time_t published_on;
-  const char *good_nickname_lst[1024];
-  int n_good_nicknames;
+  char *good_nickname_lst[1024];
+  int n_good_nicknames = 0;
+  int i;
   
 #define NEXT_TOK()                                                      \
   do {                                                                  \