X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=parsers.c;h=9fe2da77f6b1c5126753f1717647a08975955689;hp=c5f7a08708f20d703137e8b02fd54a7162ae0493;hb=d26129baa47eb9e558d53f463ad18269dfc99607;hpb=686f8b0f33b78b0fe456fd53852550f73e15eb38 diff --git a/parsers.c b/parsers.c index c5f7a087..9fe2da77 100644 --- a/parsers.c +++ b/parsers.c @@ -1,4 +1,4 @@ -const char parsers_rcs[] = "$Id: parsers.c,v 1.289 2014/07/25 11:55:47 fabiankeil Exp $"; +const char parsers_rcs[] = "$Id: parsers.c,v 1.302 2015/12/27 12:54:12 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/parsers.c,v $ @@ -96,6 +96,7 @@ static char *get_header_line(struct iob *iob); static jb_err scan_headers(struct client_state *csp); static jb_err header_tagger(struct client_state *csp, char *header); static jb_err parse_header_time(const char *header_time, time_t *result); +static jb_err parse_time_header(const char *header, time_t *result); static jb_err crumble (struct client_state *csp, char **header); static jb_err filter_header (struct client_state *csp, char **header); @@ -435,7 +436,9 @@ jb_err decompress_iob(struct client_state *csp) * This is to protect the parsing of gzipped data, * but it should(?) be valid for deflated data also. */ - log_error(LOG_LEVEL_ERROR, "Buffer too small decompressing iob"); + log_error(LOG_LEVEL_ERROR, + "Insufficient data to start decompression. Bytes in buffer: %d", + csp->iob->eod - csp->iob->cur); return JB_ERR_COMPRESS; } @@ -1115,7 +1118,8 @@ static void enforce_header_order(struct list *headers, const struct list *ordere * server and header filtering. * * Returns : JB_ERR_OK in case off success, or - * JB_ERR_MEMORY on out-of-memory error. + * JB_ERR_MEMORY on some out-of-memory errors, or + * JB_ERR_PARSE in case of fatal parse errors. * *********************************************************************/ jb_err sed(struct client_state *csp, int filter_server_headers) @@ -1141,9 +1145,9 @@ jb_err sed(struct client_state *csp, int filter_server_headers) check_negative_tag_patterns(csp, PATTERN_SPEC_NO_REQUEST_TAG_PATTERN); } - while ((err == JB_ERR_OK) && (v->str != NULL)) + while (v->str != NULL) { - for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL); p = p->next) + for (p = csp->headers->first; p != NULL; p = p->next) { /* Header crunch()ed in previous run? -> ignore */ if (p->str == NULL) continue; @@ -1153,6 +1157,10 @@ jb_err sed(struct client_state *csp, int filter_server_headers) (v->len == CHECK_EVERY_HEADER_REMAINING)) { err = v->parser(csp, &(p->str)); + if (err != JB_ERR_OK) + { + return err; + } } } v++; @@ -1657,6 +1665,8 @@ static jb_err server_keep_alive(struct client_state *csp, char **header) csp->flags |= CSP_FLAG_SERVER_KEEP_ALIVE_TIMEOUT_SET; } + freez(*header); + return JB_ERR_OK; } @@ -2190,11 +2200,12 @@ static jb_err server_content_type(struct client_state *csp, char **header) */ if ((csp->content_type & CT_TEXT) || (csp->action->flags & ACTION_FORCE_TEXT_MODE)) { + jb_err err; freez(*header); *header = strdup_or_die("Content-Type: "); - string_append(header, csp->action->string[ACTION_STRING_CONTENT_TYPE]); - if (header == NULL) + err = string_append(header, csp->action->string[ACTION_STRING_CONTENT_TYPE]); + if (JB_ERR_OK != err) { log_error(LOG_LEVEL_HEADER, "Insufficient memory to replace Content-Type!"); return JB_ERR_MEMORY; @@ -2316,8 +2327,7 @@ static jb_err server_content_encoding(struct client_state *csp, char **header) /* * Log a warning if the user expects the content to be filtered. */ - if ((csp->rlist != NULL) && - (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) + if (content_filters_enabled(csp->action)) { log_error(LOG_LEVEL_INFO, "SDCH-compressed content detected, content filtering disabled. " @@ -2661,13 +2671,12 @@ static jb_err server_last_modified(struct client_state *csp, char **header) } else if (0 == strcmpic(newval, "randomize")) { - const char *header_time = *header + sizeof("Last-Modified:"); - log_error(LOG_LEVEL_HEADER, "Randomizing: %s", *header); - if (JB_ERR_OK != parse_header_time(header_time, &last_modified)) + if (JB_ERR_OK != parse_time_header(*header, &last_modified)) { - log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s in %s (crunching!)", header_time, *header); + log_error(LOG_LEVEL_HEADER, + "Couldn't parse time in %s (crunching!)", *header); freez(*header); } else @@ -3278,6 +3287,13 @@ static jb_err client_host(struct client_state *csp, char **header) { char *p, *q; + if (strlen(*header) < 7) + { + log_error(LOG_LEVEL_HEADER, "Removing empty Host header"); + freez(*header); + return JB_ERR_OK; + } + if (!csp->http->hostport || (*csp->http->hostport == '*') || *csp->http->hostport == ' ' || *csp->http->hostport == '\0') { @@ -3364,11 +3380,10 @@ static jb_err client_if_modified_since(struct client_state *csp, char **header) } else /* add random value */ { - const char *header_time = *header + sizeof("If-Modified-Since:"); - - if (JB_ERR_OK != parse_header_time(header_time, &tm)) + if (JB_ERR_OK != parse_time_header(*header, &tm)) { - log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s in %s (crunching!)", header_time, *header); + log_error(LOG_LEVEL_HEADER, + "Couldn't parse time in %s (crunching!)", *header); freez(*header); } else @@ -3592,9 +3607,8 @@ static jb_err client_host_adder(struct client_state *csp) if (!csp->http->hostport || !*(csp->http->hostport)) { - /* XXX: When does this happen and why is it OK? */ - log_error(LOG_LEVEL_INFO, "Weirdness in client_host_adder detected and ignored."); - return JB_ERR_OK; + log_error(LOG_LEVEL_ERROR, "Destination host unknown."); + return JB_ERR_PARSE; } /* @@ -3832,6 +3846,7 @@ static jb_err client_connection_header_adder(struct client_state *csp) * is a partial range (HTTP status 206) * - Rewrite HTTP/1.1 answers to HTTP/1.0 if +downgrade * action applies. + * - Normalize the HTTP-version. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) @@ -3841,36 +3856,84 @@ static jb_err client_connection_header_adder(struct client_state *csp) * original string if necessary. * * Returns : JB_ERR_OK on success, or - * JB_ERR_MEMORY on out-of-memory error. + * JB_ERR_PARSE on fatal parse errors. * *********************************************************************/ static jb_err server_http(struct client_state *csp, char **header) { - sscanf(*header, "HTTP/%*d.%*d %d", &(csp->http->status)); + char *reason_phrase = NULL; + char *new_response_line; + char *p; + size_t length; + unsigned int major_version; + unsigned int minor_version; + + /* Get the reason phrase which start after the second whitespace */ + p = strchr(*header, ' '); + if (NULL != p) + { + p++; + reason_phrase = strchr(p, ' '); + } + + if (reason_phrase != NULL) + { + reason_phrase++; + } + else + { + log_error(LOG_LEVEL_ERROR, + "Response line lacks reason phrase: %s", *header); + reason_phrase=""; + } + + if (3 != sscanf(*header, "HTTP/%u.%u %d", &major_version, + &minor_version, &(csp->http->status))) + { + log_error(LOG_LEVEL_ERROR, + "Failed to parse the response line: %s", *header); + return JB_ERR_PARSE; + } + if (csp->http->status == 206) { csp->content_type = CT_TABOO; } - if ((csp->action->flags & ACTION_DOWNGRADE) != 0) + if (major_version != 1 || (minor_version != 0 && minor_version != 1)) { - /* XXX: Should we do a real validity check here? */ - if (strlen(*header) > 8) - { - (*header)[7] = '0'; - log_error(LOG_LEVEL_HEADER, "Downgraded answer to HTTP/1.0"); - } - else - { - /* - * XXX: Should we block the request or - * enlist a valid status code line here? - */ - log_error(LOG_LEVEL_INFO, "Malformed server response detected. " - "Downgrading to HTTP/1.0 impossible."); - } + /* + * According to RFC 7230 2.6 intermediaries MUST send + * their own HTTP-version in forwarded messages. + */ + log_error(LOG_LEVEL_ERROR, + "Unsupported HTTP version. Downgrading to 1.1."); + major_version = 1; + minor_version = 1; + } + + if (((csp->action->flags & ACTION_DOWNGRADE) != 0) && (minor_version == 1)) + { + log_error(LOG_LEVEL_HEADER, "Downgrading answer to HTTP/1.0"); + minor_version = 0; + } + + /* Rebuild response line. */ + length = sizeof("HTTP/1.1 200 ") + strlen(reason_phrase) + 1; + new_response_line = malloc_or_die(length); + + snprintf(new_response_line, length, "HTTP/%u.%u %d %s", + major_version, minor_version, csp->http->status, reason_phrase); + + if (0 != strcmp(*header, new_response_line)) + { + log_error(LOG_LEVEL_HEADER, "Response line '%s' changed to '%s'", + *header, new_response_line); } + freez(*header); + *header = new_response_line; + return JB_ERR_OK; } @@ -4279,6 +4342,44 @@ static jb_err parse_header_time(const char *header_time, time_t *result) } +/********************************************************************* + * + * Function : parse_time_header + * + * Description : Parses the time in an HTTP time header to get + * the numerical respresentation. + * + * Parameters : + * 1 : header = HTTP header with a time value + * 2 : result = storage for header_time in seconds + * + * Returns : JB_ERR_OK if the time format was recognized, or + * JB_ERR_PARSE otherwise. + * + *********************************************************************/ +static jb_err parse_time_header(const char *header, time_t *result) +{ + const char *header_time; + + header_time = strchr(header, ':'); + + /* + * Currently this can't happen as all callers are called + * through sed() which requires a header name followed by + * a colon. + */ + assert(header_time != NULL); + + header_time++; + if (*header_time == ' ') + { + header_time++; + } + + return parse_header_time(header_time, result); + +} + /********************************************************************* * @@ -4307,6 +4408,8 @@ jb_err get_destination_from_headers(const struct list *headers, struct http_requ char *p; char *host; + assert(!http->ssl); + host = get_header_value(headers, "Host:"); if (NULL == host) @@ -4332,12 +4435,12 @@ jb_err get_destination_from_headers(const struct list *headers, struct http_requ } else { - http->port = http->ssl ? 443 : 80; + http->port = 80; } /* Rebuild request URL */ freez(http->url); - http->url = strdup(http->ssl ? "https://" : "http://"); + http->url = strdup("http://"); string_append(&http->url, http->hostport); string_append(&http->url, http->path); if (http->url == NULL) @@ -4348,6 +4451,26 @@ jb_err get_destination_from_headers(const struct list *headers, struct http_requ log_error(LOG_LEVEL_HEADER, "Destination extracted from \"Host:\" header. New request URL: %s", http->url); + /* + * Regenerate request line in "proxy format" + * to make rewrites more convenient. + * XXX: Code duplication. + */ + assert(http->cmd != NULL); + freez(http->cmd); + http->cmd = strdup_or_die(http->gpc); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->url); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->ver); + if (http->cmd == NULL) + { + return JB_ERR_MEMORY; + } + + log_error(LOG_LEVEL_HEADER, "Faked request-Line: %s", + http->cmd); + return JB_ERR_OK; }