Rewrite header parsing.

1 files changed, 52 insertions, 47 deletions

diff --git a/libressl.patch b/libressl.patch
index bcc7860..564d620 100644
--- a/libressl.patch
+++ b/libressl.patch
@@ -135,7 +135,7 @@ index 9f1dd67ec..7e8da6df2 100644
  static void tls_xwrite(tls_state_t *tls, int len)
  {
 diff --git a/networking/wget.c b/networking/wget.c
-index 6a64836fb..0a8e282cc 100644
+index 6a64836fb..918ad6812 100644
 --- a/networking/wget.c
 +++ b/networking/wget.c
 @@ -169,7 +169,11 @@
@@ -191,7 +191,7 @@ index 6a64836fb..0a8e282cc 100644
  /* IPv6 knows scoped address types i.e. link and site local addresses. Link
   * local addresses can have a scope identifier to specify the
   * interface/link an address is valid on (e.g. fe80::1%eth0). This scope
-@@ -482,32 +519,63 @@ static char* sanitize_string(char *s)
+@@ -482,18 +519,45 @@ static char* sanitize_string(char *s)
         return s;
  }
  
@@ -239,12 +239,7 @@ index 6a64836fb..0a8e282cc 100644
         c = *buf_ptr;
  #if 1
         /* Disallow any control chars: trim at first char < 0x20 */
-+       printf("wget fgets_trim_sanitize 5\n");
-        sanitize_string(G.wget_buf);
-+       printf("wget fgets_trim_sanitize 6\n");
- #else
-        *buf_ptr = '\0';
-        buf_ptr = strchrnul(G.wget_buf, '\r');
+@@ -504,10 +568,12 @@ static char fgets_trim_sanitize(FILE *fp, const char *fmt)
         *buf_ptr = '\0';
  #endif
  
@@ -257,26 +252,40 @@ index 6a64836fb..0a8e282cc 100644
  
         return c;
  }
-@@ -636,6 +704,32 @@ static void parse_url(const char *src_url, struct host_info *h)
+@@ -636,6 +702,46 @@ static void parse_url(const char *src_url, struct host_info *h)
          */
  }
  
-+static char *tls_sanitized_hdr(struct tls *ctx)
++static int tls_sanitized_hdr(struct tls *ctx, char* start, char **header, char **value)
 +{
 +       printf("wget tls_sanitized_hdr 1\n");
 +       printf("wget_buf was: %s\n",G.wget_buf);
-+       char *s, *hdrval;
 +       char c;
-+       c = tlsgets_trim_sanitize(ctx, "  %s\n");
++       char *s;
++       //c = tlsgets_trim_sanitize(ctx, "  %s\n");
++       if(*start =='\n'){
++               printf("Found end of headers\n");
++               return 0;
++       }
 +       if(G.wget_buf[0] == '\0')
-+               return  NULL;
++               return  -1;
 +       printf("While getting header, buf was: %s\n", G.wget_buf);
++       *header = G.wget_buf;
++       /* Convert to lowercase */
 +       for(s = G.wget_buf; isalnum(*s) || *s == '-' || *s == '.' || *s == '_'; ++s)
 +               *s |= 0x20;
++       /* Find seperator */
 +       if(*s != ':')
 +               bb_error_msg_and_die("Bad header line: %s", G.wget_buf);
 +       *s++ = '\0';
-+       hdrval = skip_whitespace(s);
++       s = skip_whitespace(s);
++       *value = s;
++       s = strchr(s,'\n');
++       if(*s == NULL) {
++               printf("Buffer too small!\n");
++               while (tls_read(ctx, &c, sizeof(char)) == 1 && c != '\n')
++                       /*Consume the rest of the header */;
++       }
 +       if (c != '\n'){
 +               printf("buffer too small!\n");
 +               while (tls_read(ctx,&c,sizeof(char)) == 1 && c != EOF && c != '\n')
@@ -290,7 +299,7 @@ index 6a64836fb..0a8e282cc 100644
  static char *get_sanitized_hdr(FILE *fp)
  {
         char *s, *hdrval;
-@@ -689,6 +783,7 @@ static void reset_beg_range_to_zero(void)
+@@ -689,6 +795,7 @@ static void reset_beg_range_to_zero(void)
  }
  
  #if ENABLE_FEATURE_WGET_OPENSSL
@@ -298,7 +307,7 @@ index 6a64836fb..0a8e282cc 100644
  static int spawn_https_helper_openssl(const char *host, unsigned port)
  {
         char *allocated = NULL;
-@@ -777,6 +872,38 @@ static int spawn_https_helper_openssl(const char *host, unsigned port)
+@@ -777,6 +884,38 @@ static int spawn_https_helper_openssl(const char *host, unsigned port)
  # endif
         return sp[0];
  }
@@ -337,7 +346,7 @@ index 6a64836fb..0a8e282cc 100644
  #endif
  
  #if ENABLE_FEATURE_WGET_HTTPS
-@@ -1151,6 +1278,7 @@ static void download_one_url(const char *url)
+@@ -1151,6 +1290,7 @@ static void download_one_url(const char *url)
         server.user = NULL;
         target.user = NULL;
  
@@ -345,7 +354,7 @@ index 6a64836fb..0a8e282cc 100644
         parse_url(url, &target);
  
         /* Use the proxy if necessary */
-@@ -1172,6 +1300,7 @@ static void download_one_url(const char *url)
+@@ -1172,6 +1312,7 @@ static void download_one_url(const char *url)
                         server.host = target.host;
                 }
         }
@@ -353,7 +362,7 @@ index 6a64836fb..0a8e282cc 100644
  
         if (ENABLE_FEATURE_IPV6)
                 strip_ipv6_scope_id(target.host);
-@@ -1191,6 +1320,7 @@ static void download_one_url(const char *url)
+@@ -1191,6 +1332,7 @@ static void download_one_url(const char *url)
                         G.fname_out = fname_out_alloc = xstrdup(G.fname_out);
                 }
         }
@@ -361,7 +370,7 @@ index 6a64836fb..0a8e282cc 100644
  #if ENABLE_FEATURE_WGET_STATUSBAR
         G.curfile = bb_get_last_path_component_nostrip(G.fname_out);
  #endif
-@@ -1206,15 +1336,19 @@ static void download_one_url(const char *url)
+@@ -1206,15 +1348,19 @@ static void download_one_url(const char *url)
                  * We are not sure it exists on remote side */
         }
  
@@ -381,7 +390,7 @@ index 6a64836fb..0a8e282cc 100644
         /*G.content_len = 0; - redundant, got_clen = 0 is enough */
         G.got_clen = 0;
         G.chunked = 0;
-@@ -1229,37 +1363,64 @@ static void download_one_url(const char *url)
+@@ -1229,37 +1375,64 @@ static void download_one_url(const char *url)
  
                 /* Open socket to http(s) server */
  #if ENABLE_FEATURE_WGET_OPENSSL
@@ -459,7 +468,7 @@ index 6a64836fb..0a8e282cc 100644
                 /* Send HTTP request */
                 if (use_proxy) {
                         SENDFMT(sfp, "GET %s://%s/%s HTTP/1.1\r\n",
-@@ -1270,6 +1431,7 @@ static void download_one_url(const char *url)
+@@ -1270,6 +1443,7 @@ static void download_one_url(const char *url)
                                 (option_mask32 & WGET_OPT_POST) ? "POST" : "GET",
                                 target.path);
                 }
@@ -467,7 +476,7 @@ index 6a64836fb..0a8e282cc 100644
                 if (!USR_HEADER_HOST)
                         SENDFMT(sfp, "Host: %s\r\n", target.host);
                 if (!USR_HEADER_USER_AGENT)
-@@ -1280,6 +1442,7 @@ static void download_one_url(const char *url)
+@@ -1280,6 +1454,7 @@ static void download_one_url(const char *url)
                  */
                 SENDFMT(sfp, "Connection: close\r\n");
  
@@ -475,7 +484,7 @@ index 6a64836fb..0a8e282cc 100644
  #if ENABLE_FEATURE_WGET_AUTHENTICATION
                 if (target.user && !USR_HEADER_AUTH) {
                         SENDFMT(sfp, "Proxy-Authorization: Basic %s\r\n"+6,
-@@ -1291,6 +1454,7 @@ static void download_one_url(const char *url)
+@@ -1291,6 +1466,7 @@ static void download_one_url(const char *url)
                 }
  #endif
  
@@ -483,7 +492,7 @@ index 6a64836fb..0a8e282cc 100644
                 if (G.beg_range != 0 && !USR_HEADER_RANGE)
                         SENDFMT(sfp, "Range: bytes=%"OFF_FMT"u-\r\n", G.beg_range);
  
-@@ -1300,6 +1464,7 @@ static void download_one_url(const char *url)
+@@ -1300,6 +1476,7 @@ static void download_one_url(const char *url)
                         fputs(G.extra_headers, sfp);
                 }
  
@@ -491,7 +500,7 @@ index 6a64836fb..0a8e282cc 100644
                 if (option_mask32 & WGET_OPT_POST_FILE) {
                         int fd = xopen_stdin(G.post_file);
                         G.post_data = xmalloc_read(fd, NULL);
-@@ -1314,15 +1479,18 @@ static void download_one_url(const char *url)
+@@ -1314,15 +1491,18 @@ static void download_one_url(const char *url)
                                 );
                         }
  # if ENABLE_PLATFORM_MINGW32
@@ -510,7 +519,7 @@ index 6a64836fb..0a8e282cc 100644
                 } else
  # else
                         SENDFMT(sfp,
-@@ -1338,6 +1506,7 @@ static void download_one_url(const char *url)
+@@ -1338,6 +1518,7 @@ static void download_one_url(const char *url)
                         SENDFMT(sfp, "\r\n");
                 }
  
@@ -518,7 +527,7 @@ index 6a64836fb..0a8e282cc 100644
                 fflush(sfp);
  
  /* Tried doing this unconditionally.
-@@ -1345,31 +1514,61 @@ static void download_one_url(const char *url)
+@@ -1345,31 +1526,55 @@ static void download_one_url(const char *url)
   */
  #if SSL_SUPPORTED
                 if (target.protocol == P_HTTPS) {
@@ -542,7 +551,6 @@ index 6a64836fb..0a8e282cc 100644
 +                       bb_error_msg_and_die("Failed to read tmpfile: %s\n", strerror(errno));
 +               if(tls_write(ctx, outbuf, wlen) < wlen)
 +                       bb_error_msg_and_die("Failed to write:%s\n",tls_error(ctx));
-+       
 +               
 +               printf("About to seek\n");
 +               fseek(dfp,0,SEEK_END);
@@ -564,16 +572,11 @@ index 6a64836fb..0a8e282cc 100644
  
                 str = G.wget_buf;
 +               printf("wget_buf was: %s\n",G.wget_buf);
-+               printf("wget download_one_url 27-2\n");
                 str = skip_non_whitespace(str);
-+               printf("wget download_one_url 27-3\n");
                 str = skip_whitespace(str);
-+               printf("wget download_one_url 27-4\n");
                 // FIXME: no error check
                 // xatou wouldn't work: "200 OK"
-+               printf("wget download_one_url 27-5\n");
                 status = atoi(str);
-+               printf("wget download_one_url 27-6\n");
 +               printf("Status was: %d\n",status);
                 switch (status) {
                 case 0:
@@ -583,20 +586,22 @@ index 6a64836fb..0a8e282cc 100644
                                 /* eat all remaining headers */;
                         goto read_response;
  
-@@ -1441,7 +1640,11 @@ However, in real world it was observed that some web servers
+@@ -1441,7 +1646,13 @@ However, in real world it was observed that some web servers
                 /*
                  * Retrieve HTTP headers.
                  */
 -               while ((str = get_sanitized_hdr(sfp)) != NULL) {
 +               printf("wget download_one_url 28\n");
 +               printf("At this point wget_buf is: %s\n",G.wget_buf);
-+               while ((str = tls_sanitized_hdr(ctx)) != NULL) {
++               str = strchr(str,'\n');
++               char *header, *value;
++               while (tls_sanitized_hdr(ctx, str, &header, &value) > 0) {
 +                       printf("wget Retreiving the next part of the headers\n");
 +                       printf("wget str was: %s\n", str);
                         static const char keywords[] ALIGN1 =
                                 "content-length\0""transfer-encoding\0""location\0";
                         enum {
-@@ -1453,6 +1656,7 @@ However, in real world it was observed that some web servers
+@@ -1453,6 +1664,7 @@ However, in real world it was observed that some web servers
  
                         /* strip trailing whitespace */
                         char *s = strchrnul(str, '\0') - 1;
@@ -604,7 +609,7 @@ index 6a64836fb..0a8e282cc 100644
                         while (s >= str && (*s == ' ' || *s == '\t')) {
                                 *s = '\0';
                                 s--;
-@@ -1460,6 +1664,7 @@ However, in real world it was observed that some web servers
+@@ -1460,6 +1672,7 @@ However, in real world it was observed that some web servers
                         key = index_in_strings(keywords, G.wget_buf) + 1;
                         if (key == KEY_content_length) {
                                 G.content_len = BB_STRTOOFF(str, NULL, 10);
@@ -612,7 +617,7 @@ index 6a64836fb..0a8e282cc 100644
                                 if (G.content_len < 0 || errno) {
                                         bb_error_msg_and_die("content-length %s is garbage", str);
                                 }
-@@ -1469,6 +1674,7 @@ However, in real world it was observed that some web servers
+@@ -1469,6 +1682,7 @@ However, in real world it was observed that some web servers
                         if (key == KEY_transfer_encoding) {
                                 if (strcmp(str_tolower(str), "chunked") != 0)
                                         bb_error_msg_and_die("transfer encoding '%s' is not supported", str);
@@ -620,7 +625,7 @@ index 6a64836fb..0a8e282cc 100644
                                 G.chunked = 1;
                         }
                         if (key == KEY_location && status >= 300) {
-@@ -1494,9 +1700,11 @@ However, in real world it was observed that some web servers
+@@ -1494,9 +1708,11 @@ However, in real world it was observed that some web servers
                                                 goto resolve_lsa;
                                         } /* else: lsa stays the same: we use proxy */
                                 }
@@ -632,7 +637,7 @@ index 6a64836fb..0a8e282cc 100644
  //             if (status >= 300)
  //                     bb_error_msg_and_die("bad redirection (no Location: header from server)");
  
-@@ -1514,6 +1722,7 @@ However, in real world it was observed that some web servers
+@@ -1514,6 +1730,7 @@ However, in real world it was observed that some web servers
  
         free(lsa);
  
@@ -640,7 +645,7 @@ index 6a64836fb..0a8e282cc 100644
         if (!(option_mask32 & WGET_OPT_SPIDER)) {
                 if (G.output_fd < 0)
                         G.output_fd = xopen(G.fname_out, G.o_flags);
-@@ -1536,6 +1745,7 @@ However, in real world it was observed that some web servers
+@@ -1536,6 +1753,7 @@ However, in real world it was observed that some web servers
                 /* ftpcmd("QUIT", NULL, sfp); - why bother? */
         }
  #endif
@@ -648,7 +653,7 @@ index 6a64836fb..0a8e282cc 100644
         fclose(sfp);
  
         free(server.allocated);
-@@ -1544,11 +1754,13 @@ However, in real world it was observed that some web servers
+@@ -1544,11 +1762,13 @@ However, in real world it was observed that some web servers
         free(target.user);
         free(fname_out_alloc);
         free(redirected_path);
@@ -662,7 +667,7 @@ index 6a64836fb..0a8e282cc 100644
  #if ENABLE_FEATURE_WGET_LONG_OPTIONS
         static const char wget_longopts[] ALIGN1 =
                 /* name, has_arg, val */
-@@ -1590,6 +1802,7 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
+@@ -1590,6 +1810,7 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
  #endif
  
         INIT_G();
@@ -670,7 +675,7 @@ index 6a64836fb..0a8e282cc 100644
  
  #if ENABLE_FEATURE_WGET_TIMEOUT
         G.timeout_seconds = 900;
-@@ -1626,6 +1839,8 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
+@@ -1626,6 +1847,8 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
                 IF_FEATURE_WGET_LONG_OPTIONS(, &G.post_data)
                 IF_FEATURE_WGET_LONG_OPTIONS(, &G.post_file)
         );
@@ -679,7 +684,7 @@ index 6a64836fb..0a8e282cc 100644
  #if 0 /* option bits debug */
         if (option_mask32 & WGET_OPT_RETRIES) bb_error_msg("-t NUM");
         if (option_mask32 & WGET_OPT_nsomething) bb_error_msg("-nsomething");
-@@ -1670,6 +1885,7 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
+@@ -1670,6 +1893,7 @@ IF_DESKTOP(       "no-parent\0"        No_argument       "\xf0")
         }
  #endif
  
@@ -687,7 +692,7 @@ index 6a64836fb..0a8e282cc 100644
         G.output_fd = -1;
         G.o_flags = O_WRONLY | O_CREAT | O_TRUNC | O_EXCL;
         if (G.fname_out) { /* -O FILE ? */
-@@ -1691,8 +1907,11 @@ IF_DESKTOP(      "no-parent\0"        No_argument       "\xf0")
+@@ -1691,8 +1915,11 @@ IF_DESKTOP(      "no-parent\0"        No_argument       "\xf0")
                 }
         }