lighttpd1.4/src/request.c

#include "first.h"

#include "request.h"
#include "keyvalue.h"
#include "log.h"
#include "inet_ntop_cache.h"

#include <sys/stat.h>

#include <limits.h>
#include <stdlib.h>
#include <string.h>

static int request_check_hostname(buffer *host) {
	enum { DOMAINLABEL, TOPLABEL } stage = TOPLABEL;
	size_t i;
	int label_len = 0;
	size_t host_len, hostport_len;
	char *colon;
	int is_ip = -1; /* -1 don't know yet, 0 no, 1 yes */
	int level = 0;

	/*
	 *       hostport      = host [ ":" port ]
	 *       host          = hostname | IPv4address | IPv6address
	 *       hostname      = *( domainlabel "." ) toplabel [ "." ]
	 *       domainlabel   = alphanum | alphanum *( alphanum | "-" ) alphanum
	 *       toplabel      = alpha | alpha *( alphanum | "-" ) alphanum
	 *       IPv4address   = 1*digit "." 1*digit "." 1*digit "." 1*digit
	 *       IPv6address   = "[" ... "]"
	 *       port          = *digit
	 */

	/* IPv6 adress */
	if (host->ptr[0] == '[') {
		char *c = host->ptr + 1;
		int colon_cnt = 0;

		/* check the address inside [...] */
		for (; *c && *c != ']'; c++) {
			if (*c == ':') {
				if (++colon_cnt > 7) {
					return -1;
				}
			} else if (!light_isxdigit(*c) && '.' != *c) {
				return -1;
			}
		}

		/* missing ] */
		if (!*c) {
			return -1;
		}

		/* check port */
		if (*(c+1) == ':') {
			for (c += 2; *c; c++) {
				if (!light_isdigit(*c)) {
					return -1;
				}
			}
		}
		else if ('\0' != *(c+1)) {
			/* only a port is allowed to follow [...] */
			return -1;
		}
		return 0;
	}

	hostport_len = host_len = buffer_string_length(host);

	if (NULL != (colon = memchr(host->ptr, ':', host_len))) {
		char *c = colon + 1;

		/* check portnumber */
		for (; *c; c++) {
			if (!light_isdigit(*c)) return -1;
		}

		/* remove the port from the host-len */
		host_len = colon - host->ptr;
	}

	/* Host is empty */
	if (host_len == 0) return -1;

	/* if the hostname ends in a "." strip it */
	if (host->ptr[host_len-1] == '.') {
		/* shift port info one left */
		if (NULL != colon) memmove(colon-1, colon, hostport_len - host_len);
		buffer_string_set_length(host, --hostport_len);
		if (--host_len == 0) return -1;
	}


	/* scan from the right and skip the \0 */
	for (i = host_len; i-- > 0; ) {
		const char c = host->ptr[i];

		switch (stage) {
		case TOPLABEL:
			if (c == '.') {
				/* only switch stage, if this is not the last character */
				if (i != host_len - 1) {
					if (label_len == 0) {
						return -1;
					}

					/* check the first character at right of the dot */
					if (is_ip == 0) {
						if (!light_isalnum(host->ptr[i+1])) {
							return -1;
						}
					} else if (!light_isdigit(host->ptr[i+1])) {
						is_ip = 0;
					} else if ('-' == host->ptr[i+1]) {
						return -1;
					} else {
						/* just digits */
						is_ip = 1;
					}

					stage = DOMAINLABEL;

					label_len = 0;
					level++;
				} else if (i == 0) {
					/* just a dot and nothing else is evil */
					return -1;
				}
			} else if (i == 0) {
				/* the first character of the hostname */
				if (!light_isalnum(c)) {
					return -1;
				}
				label_len++;
			} else {
				if (c != '-' && !light_isalnum(c)) {
					return -1;
				}
				if (is_ip == -1) {
					if (!light_isdigit(c)) is_ip = 0;
				}
				label_len++;
			}

			break;
		case DOMAINLABEL:
			if (is_ip == 1) {
				if (c == '.') {
					if (label_len == 0) {
						return -1;
					}

					label_len = 0;
					level++;
				} else if (!light_isdigit(c)) {
					return -1;
				} else {
					label_len++;
				}
			} else {
				if (c == '.') {
					if (label_len == 0) {
						return -1;
					}

					/* c is either - or alphanum here */
					if ('-' == host->ptr[i+1]) {
						return -1;
					}

					label_len = 0;
					level++;
				} else if (i == 0) {
					if (!light_isalnum(c)) {
						return -1;
					}
					label_len++;
				} else {
					if (c != '-' && !light_isalnum(c)) {
						return -1;
					}
					label_len++;
				}
			}

			break;
		}
	}

	/* a IP has to consist of 4 parts */
	if (is_ip == 1 && level != 3) {
		return -1;
	}

	if (label_len == 0) {
		return -1;
	}

	return 0;
}

int http_request_host_normalize(buffer *b, int scheme_port) {
    /*
     * check for and canonicalize numeric IP address and portnum (optional)
     * (IP address may be followed by ":portnum" (optional))
     * - IPv6: "[...]"
     * - IPv4: "x.x.x.x"
     * - IPv4: 12345678   (32-bit decimal number)
     * - IPv4: 012345678  (32-bit octal number)
     * - IPv4: 0x12345678 (32-bit hex number)
     *
     * allow any chars (except ':' and '\0' and stray '[' or ']')
     *   (other code may check chars more strictly or more pedantically)
     * ':'  delimits (optional) port at end of string
     * "[]" wraps IPv6 address literal
     * '\0' should have been rejected earlier were it present
     *
     * any chars includes, but is not limited to:
     * - allow '-' any where, even at beginning of word
     *     (security caution: might be confused for cmd flag if passed to shell)
     * - allow all-digit TLDs
     *     (might be mistaken for IPv4 addr by inet_aton()
     *      unless non-digits appear in subdomain)
     */

    /* Note: not using getaddrinfo() since it does not support "[]" around IPv6
     * and is not as lenient as inet_aton() and inet_addr() for IPv4 strings.
     * Not using inet_pton() (when available) on IPv4 for similar reasons. */

    const char * const p = b->ptr;
    const size_t blen = buffer_string_length(b);
    long port = 0;

    if (*p != '[') {
        char * const colon = (char *)memchr(p, ':', blen);
        if (colon) {
            if (*p == ':') return -1; /*(empty host then port, or naked IPv6)*/
            if (colon[1] != '\0') {
                char *e;
                port = strtol(colon+1, &e, 0); /*(allow decimal, octal, hex)*/
                if (0 < port && port <= USHRT_MAX && *e == '\0') {
                    /* valid port */
                } else {
                    return -1;
                }
            } /*(else ignore stray colon at string end)*/
            buffer_string_set_length(b, (size_t)(colon - p)); /*(remove port str)*/
        }

        if (light_isdigit(*p)) {
            /* (IPv4 address literal or domain starting w/ digit (e.g. 3com))*/
            sock_addr addr;
            if (1 == sock_addr_inet_pton(&addr, p, AF_INET, 0)) {
                sock_addr_inet_ntop_copy_buffer(b, &addr);
            }
        }
    } else { /* IPv6 addr */
      #if defined(HAVE_IPV6) && defined(HAVE_INET_PTON)

        sock_addr addr;
        char *bracket = b->ptr+blen-1;
        char *percent = strchr(b->ptr+1, '%');
        size_t len;
        int rc;
        char buf[INET6_ADDRSTRLEN+16]; /*(+16 for potential %interface name)*/
        if (blen <= 2) return -1; /*(invalid "[]")*/
        if (*bracket != ']') {
            bracket = (char *)memchr(b->ptr+1, ']', blen-1);
            if (NULL == bracket || bracket[1] != ':'  || bracket - b->ptr == 1){
               return -1;
            }
            if (bracket[2] != '\0') { /*(ignore stray colon at string end)*/
                char *e;
                port = strtol(bracket+2, &e, 0); /*(allow decimal, octal, hex)*/
                if (0 < port && port <= USHRT_MAX && *e == '\0') {
                    /* valid port */
                } else {
                    return -1;
                }
            }
        }

        *bracket = '\0';/*(terminate IPv6 string)*/
        if (percent) *percent = '\0'; /*(remove %interface from address)*/
        rc = sock_addr_inet_pton(&addr, b->ptr+1, AF_INET6, 0);
        if (percent) *percent = '%'; /*(restore %interface)*/
        *bracket = ']'; /*(restore bracket)*/
        if (1 != rc) return -1;

        sock_addr_inet_ntop(&addr, buf, sizeof(buf));
        len = strlen(buf);
        if (percent) {
            if (percent > bracket) return -1;
            if (len + (size_t)(bracket - percent) >= sizeof(buf)) return -1;
            memcpy(buf+len, percent, (size_t)(bracket - percent));
            len += (size_t)(bracket - percent);
        }
        buffer_string_set_length(b, 1); /* truncate after '[' */
        buffer_append_string_len(b, buf, len);
        buffer_append_string_len(b, CONST_STR_LEN("]"));

      #else

        return -1;

      #endif
    }

    if (0 != port && port != scheme_port) {
        buffer_append_string_len(b, CONST_STR_LEN(":"));
        buffer_append_int(b, (int)port);
    }

    return 0;
}

static int scheme_port (const buffer *scheme)
{
    return buffer_is_equal_string(scheme, CONST_STR_LEN("https")) ? 443 : 80;
}

int http_request_host_policy (connection *con, buffer *b, const buffer *scheme) {
    return (((con->conf.http_parseopts & HTTP_PARSEOPT_HOST_STRICT)
             && 0 != request_check_hostname(b))
            || ((con->conf.http_parseopts & HTTP_PARSEOPT_HOST_NORMALIZE)
                && 0 != http_request_host_normalize(b, scheme_port(scheme))));
}

#if 0
#define DUMP_HEADER
#endif

static int http_request_split_value(array *vals, buffer *b) {
	size_t i, len;
	int state = 0;

	const char *current;
	const char *token_start = NULL, *token_end = NULL;
	/*
	 * parse
	 *
	 * val1, val2, val3, val4
	 *
	 * into a array (more or less a explode() incl. striping of whitespaces
	 */

	if (buffer_string_is_empty(b)) return 0;

	current = b->ptr;
	len = buffer_string_length(b);
	for (i =  0; i <= len; ++i, ++current) {
		data_string *ds;

		switch (state) {
		case 0: /* find start of a token */
			switch (*current) {
			case ' ':
			case '\t': /* skip white space */
			case ',': /* skip empty token */
				break;
			case '\0': /* end of string */
				return 0;
			default:
				/* found real data, switch to state 1 to find the end of the token */
				token_start = token_end = current;
				state = 1;
				break;
			}
			break;
		case 1: /* find end of token and last non white space character */
			switch (*current) {
			case ' ':
			case '\t':
				/* space - don't update token_end */
				break;
			case ',':
			case '\0': /* end of string also marks the end of a token */
				if (NULL == (ds = (data_string *)array_get_unused_element(vals, TYPE_STRING))) {
					ds = data_string_init();
				}

				buffer_copy_string_len(ds->value, token_start, token_end-token_start+1);
				array_insert_unique(vals, (data_unset *)ds);

				state = 0;
				break;
			default:
				/* no white space, update token_end to include current character */
				token_end = current;
				break;
			}
			break;
		}
	}

	return 0;
}

static int request_uri_is_valid_char(unsigned char c) {
	if (c <= 32) return 0;
	if (c == 127) return 0;
	if (c == 255) return 0;

	return 1;
}

int http_request_parse(server *srv, connection *con) {
	char *uri = NULL, *proto = NULL, *method = NULL, con_length_set;
	int is_key = 1, key_len = 0, is_ws_after_key = 0, in_folding;
	char *value = NULL, *key = NULL;
	char *reqline_host = NULL;
	int reqline_hostlen = 0;

	enum { HTTP_CONNECTION_UNSET, HTTP_CONNECTION_KEEPALIVE, HTTP_CONNECTION_CLOSE } keep_alive_set = HTTP_CONNECTION_UNSET;

	int line = 0;

	int request_line_stage = 0;
	size_t i, first, ilen;

	int done = 0;
	const unsigned int http_header_strict = (con->conf.http_parseopts & HTTP_PARSEOPT_HEADER_STRICT);

	/*
	 * Request: "^(GET|POST|HEAD) ([^ ]+(\\?[^ ]+|)) (HTTP/1\\.[01])$"
	 * Option : "^([-a-zA-Z]+): (.+)$"
	 * End    : "^$"
	 */

	if (con->conf.log_request_header) {
		log_error_write(srv, __FILE__, __LINE__, "sdsdSb",
				"fd:", con->fd,
				"request-len:", buffer_string_length(con->request.request),
				"\n", con->request.request);
	}

	if (con->request_count > 1 &&
	    con->request.request->ptr[0] == '\r' &&
	    con->request.request->ptr[1] == '\n') {
		/* we are in keep-alive and might get \r\n after a previous POST request.*/

	      #ifdef __COVERITY__
		if (buffer_string_length(con->request.request) < 2) {
			con->keep_alive = 0;
			con->http_status = 400;
			return 0;
		}
	      #endif
		/* coverity[overflow_sink : FALSE] */
		buffer_copy_string_len(con->parse_request, con->request.request->ptr + 2, buffer_string_length(con->request.request) - 2);
	} else {
		/* fill the local request buffer */
		buffer_copy_buffer(con->parse_request, con->request.request);
	}

	keep_alive_set = 0;
	con_length_set = 0;

	/* parse the first line of the request
	 *
	 * should be:
	 *
	 * <method> <uri> <protocol>\r\n
	 * */
	ilen = buffer_string_length(con->parse_request);
	for (i = 0, first = 0; i < ilen && line == 0; i++) {
		switch(con->parse_request->ptr[i]) {
		case '\r':
			if (con->parse_request->ptr[i+1] == '\n') {
				http_method_t r;
				char *nuri = NULL;
				size_t j, jlen;

				/* \r\n -> \0\0 */
				con->parse_request->ptr[i] = '\0';
				con->parse_request->ptr[i+1] = '\0';

				buffer_copy_string_len(con->request.request_line, con->parse_request->ptr, i);

				if (request_line_stage != 2) {
					con->http_status = 400;
					con->response.keep_alive = 0;
					con->keep_alive = 0;

					if (srv->srvconf.log_request_header_on_error) {
						log_error_write(srv, __FILE__, __LINE__, "s", "incomplete request line -> 400");
						log_error_write(srv, __FILE__, __LINE__, "Sb",
								"request-header:\n",
								con->request.request);
					}
					return 0;
				}

				proto = con->parse_request->ptr + first;

				*(uri - 1) = '\0';
				*(proto - 1) = '\0';

				/* we got the first one :) */
				if (HTTP_METHOD_UNSET == (r = get_http_method_key(method))) {
					con->http_status = 501;
					con->response.keep_alive = 0;
					con->keep_alive = 0;

					if (srv->srvconf.log_request_header_on_error) {
						log_error_write(srv, __FILE__, __LINE__, "s", "unknown http-method -> 501");
						log_error_write(srv, __FILE__, __LINE__, "Sb",
								"request-header:\n",
								con->request.request);
					}

					return 0;
				}

				con->request.http_method = r;

				/*
				 * RFC2616 says:
				 *
				 * HTTP-Version   = "HTTP" "/" 1*DIGIT "." 1*DIGIT
				 *
				 * */
				if (0 == strncmp(proto, "HTTP/", sizeof("HTTP/") - 1)) {
					char * major = proto + sizeof("HTTP/") - 1;
					char * minor = strchr(major, '.');
					char *err = NULL;
					int major_num = 0, minor_num = 0;

					int invalid_version = 0;

					if (NULL == minor || /* no dot */
					    minor == major || /* no major */
					    *(minor + 1) == '\0' /* no minor */) {
						invalid_version = 1;
					} else {
						*minor = '\0';
						major_num = strtol(major, &err, 10);

						if (*err != '\0') invalid_version = 1;

						*minor++ = '.';
						minor_num = strtol(minor, &err, 10);

						if (*err != '\0') invalid_version = 1;
					}

					if (invalid_version) {
						con->http_status = 400;
						con->keep_alive = 0;

						if (srv->srvconf.log_request_header_on_error) {
							log_error_write(srv, __FILE__, __LINE__, "s", "unknown protocol -> 400");
							log_error_write(srv, __FILE__, __LINE__, "Sb",
									"request-header:\n",
									con->request.request);
						}
						return 0;
					}

					if (major_num == 1 && minor_num == 1) {
						con->request.http_version = con->conf.allow_http11 ? HTTP_VERSION_1_1 : HTTP_VERSION_1_0;
					} else if (major_num == 1 && minor_num == 0) {
						con->request.http_version = HTTP_VERSION_1_0;
					} else {
						con->http_status = 505;

						if (srv->srvconf.log_request_header_on_error) {
							log_error_write(srv, __FILE__, __LINE__, "s", "unknown HTTP version -> 505");
							log_error_write(srv, __FILE__, __LINE__, "Sb",
									"request-header:\n",
									con->request.request);
						}
						return 0;
					}
				} else {
					con->http_status = 400;
					con->keep_alive = 0;

					if (srv->srvconf.log_request_header_on_error) {
						log_error_write(srv, __FILE__, __LINE__, "s", "unknown protocol -> 400");
						log_error_write(srv, __FILE__, __LINE__, "Sb",
								"request-header:\n",
								con->request.request);
					}
					return 0;
				}

				if (0 == strncmp(uri, "http://", 7) &&
				    NULL != (nuri = strchr(uri + 7, '/'))) {
					reqline_host = uri + 7;
					reqline_hostlen = nuri - reqline_host;

					buffer_copy_string_len(con->request.uri, nuri, proto - nuri - 1);
				} else if (0 == strncmp(uri, "https://", 8) &&
				    NULL != (nuri = strchr(uri + 8, '/'))) {
					reqline_host = uri + 8;
					reqline_hostlen = nuri - reqline_host;

					buffer_copy_string_len(con->request.uri, nuri, proto - nuri - 1);
				} else {
					/* everything looks good so far */
					buffer_copy_string_len(con->request.uri, uri, proto - uri - 1);
				}

				/* check uri for invalid characters */
				jlen = buffer_string_length(con->request.uri);
				if (http_header_strict) {
					for (j = 0; j < jlen && request_uri_is_valid_char(con->request.uri->ptr[j]); j++) ;
				} else {
					char *z = memchr(con->request.uri->ptr, '\0', jlen);
					j = (NULL == z) ? jlen : (size_t)(z - con->request.uri->ptr);
				}
				if (j < jlen) {
						con->http_status = 400;
						con->keep_alive = 0;

						if (srv->srvconf.log_request_header_on_error) {
							unsigned char buf[2];
							buf[0] = con->request.uri->ptr[j];
							buf[1] = '\0';

							if (con->request.uri->ptr[j] > 32 &&
							    con->request.uri->ptr[j] != 127) {
								/* the character is printable -> print it */
								log_error_write(srv, __FILE__, __LINE__, "ss",
										"invalid character in URI -> 400",
										buf);
							} else {
								/* a control-character, print ascii-code */
								log_error_write(srv, __FILE__, __LINE__, "sd",
										"invalid character in URI -> 400",
										con->request.uri->ptr[j]);
							}

							log_error_write(srv, __FILE__, __LINE__, "Sb",
									"request-header:\n",
									con->request.request);
						}

						return 0;
				}

				buffer_copy_buffer(con->request.orig_uri, con->request.uri);

				con->http_status = 0;

				i++;
				line++;
				first = i+1;
			}
			break;
		case ' ':
			switch(request_line_stage) {
			case 0:
				/* GET|POST|... */
				method = con->parse_request->ptr + first;
				first = i + 1;
				break;
			case 1:
				/* /foobar/... */
				uri = con->parse_request->ptr + first;
				first = i + 1;
				break;
			default:
				/* ERROR, one space to much */
				con->http_status = 400;
				con->response.keep_alive = 0;
				con->keep_alive = 0;

				if (srv->srvconf.log_request_header_on_error) {
					log_error_write(srv, __FILE__, __LINE__, "s", "overlong request line -> 400");
					log_error_write(srv, __FILE__, __LINE__, "Sb",
							"request-header:\n",
							con->request.request);
				}
				return 0;
			}

			request_line_stage++;
			break;
		}
	}

	in_folding = 0;

	if (buffer_string_is_empty(con->request.uri)) {
		con->http_status = 400;
		con->response.keep_alive = 0;
		con->keep_alive = 0;

		if (srv->srvconf.log_request_header_on_error) {