src/logsmanagement/parser.c
// SPDX-License-Identifier: GPL-3.0-or-later
/** @file parser.c
* @brief API to parse and search logs
*/
#if !defined(_XOPEN_SOURCE) && !defined(__DARWIN__) && !defined(__APPLE__) && !defined(__FreeBSD__)
/* _XOPEN_SOURCE 700 required by strptime (POSIX 2004) and strndup (POSIX 2008)
* Will need to find a cleaner way of doing this, as currently defining
* _XOPEN_SOURCE 700 can cause issues on Centos 7, MacOS and FreeBSD too. */
#define _XOPEN_SOURCE 700
/* _BSD_SOURCE (glibc <= 2.19) and _DEFAULT_SOURCE (glibc >= 2.20) are required
* to silence "warning: implicit declaration of function ‘strsep’;" that is
* included through libnetdata/inlined.h. */
#define _BSD_SOURCE
#define _DEFAULT_SOURCE
#include <time.h>
#endif
#include "parser.h"
#include "helper.h"
#include <stdio.h>
#include <sys/resource.h>
#include <math.h>
#include <string.h>
static regex_t vhost_regex, req_client_regex, cipher_suite_regex;
const char* const csv_auto_format_guess_matrix[] = {
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - - $request_length $request_time $upstream_response_time", // csvVhostCustom4
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - - $request_length $request_time", // csvVhostCustom3
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - -", // csvVhostCombined
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent $request_length $request_time $upstream_response_time", // csvVhostCustom2
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent $request_length $request_time", // csvVhostCustom1
"$host:$server_port $remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent", // csvVhostCommon
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - - $request_length $request_time $upstream_response_time", // csvCustom4
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - - $request_length $request_time", // csvCustom3
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent - -", // csvCombined
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent $request_length $request_time $upstream_response_time", // csvCustom2
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent $request_length $request_time", // csvCustom1
"$remote_addr - - [$time_local] \"$request\" $status $body_bytes_sent", // csvCommon
NULL}
;
UNIT_STATIC int count_fields(const char *line, const char delimiter){
const char *ptr;
int cnt, fQuote;
for (cnt = 1, fQuote = 0, ptr = line; *ptr != '\n' && *ptr != '\r' && *ptr != '\0'; ptr++ ){
if (fQuote) {
if (*ptr == '\"') {
if ( ptr[1] == '\"' ) {
ptr++;
continue;
}
fQuote = 0;
}
continue;
}
if(*ptr == '\"'){
fQuote = 1;
continue;
}
if(*ptr == delimiter){
cnt++;
while(*(ptr+1) == delimiter) ptr++;
continue;
}
}
if (fQuote) {
return -1;
}
return cnt;
}
/**
* @brief Parse a delimited string into an array of strings.
* @details Given a string containing no linebreaks, or containing line breaks
* which are escaped by "double quotes", extract a NULL-terminated
* array of strings, one for every delimiter-separated value in the row.
* @param[in] line The input string to be parsed.
* @param[in] delimiter The delimiter to be used to split the string.
* @param[in] num_fields The expected number of fields in \p line. If a negative
* number is provided, they will be counted.
* @return A NULL-terminated array of strings with the delimited values in \p line,
* or NULL in any other case.
* @todo This function has not been benchmarked or optimised.
*/
static inline char **parse_csv( const char *line, const char delimiter, int num_fields) {
char **buf, **bptr, *tmp, *tptr;
const char *ptr;
int fQuote, fEnd;
if(num_fields < 0){
num_fields = count_fields(line, delimiter);
if ( num_fields == -1 ) {
return NULL;
}
}
buf = mallocz( sizeof(char*) * (num_fields+1) );
tmp = mallocz( strlen(line) + 1 );
bptr = buf;
for ( ptr = line, fQuote = 0, *tmp = '\0', tptr = tmp, fEnd = 0; ; ptr++ ) {
if ( fQuote ) {
if ( !*ptr ) {
break;
}
if ( *ptr == '\"' ) {
if ( ptr[1] == '\"' ) {
*tptr++ = '\"';
ptr++;
continue;
}
fQuote = 0;
}
else {
*tptr++ = *ptr;
}
continue;
}
if(*ptr == '\"'){
fQuote = 1;
continue;
}
else if(*ptr == '\0'){
fEnd = 1;
*tptr = '\0';
*bptr = strdupz( tmp );
if ( !*bptr ) {
for ( bptr--; bptr >= buf; bptr-- ) {
freez( *bptr );
}
freez( buf );
freez( tmp );
return NULL;
}
bptr++;
tptr = tmp;
break;
}
else if(*ptr == delimiter){
*tptr = '\0';
*bptr = strdupz( tmp );
if ( !*bptr ) {
for ( bptr--; bptr >= buf; bptr-- ) {
freez( *bptr );
}
freez( buf );
freez( tmp );
return NULL;
}
bptr++;
tptr = tmp;
continue;
}
else{
*tptr++ = *ptr;
continue;
}
if ( fEnd ) {
break;
}
}
*bptr = NULL;
freez( tmp );
return buf;
}
/**
* @brief Search a buffer for a keyword (or regular expression)
* @details Search the source buffer for a keyword (or regular expression) and
* copy matches to the destination buffer.
* @param[in] src The source buffer to be searched
* @param[in] src_sz Size of \p src
* @param[in, out] dest The destination buffer where the results will be
* written out to. If NULL, the results will just be discarded.
* @param[out] dest_sz Size of \p dest
* @param[in] keyword The keyword or pattern to be searched in the src buffer
* @param[in] regex The precompiled regular expression to be search in the
* src buffer. If NULL, \p keyword will be used instead.
* @param[in] ignore_case Perform case insensitive search if 1.
* @return Number of matches, or -1 in case of error
*/
int search_keyword( char *src, size_t src_sz __maybe_unused,
char *dest, size_t *dest_sz,
const char *keyword, regex_t *regex,
const int ignore_case){
m_assert(src[src_sz - 1] == '\0', "src[src_sz - 1] should be '\0' but it's not");
m_assert((dest && dest_sz) || (!dest && !dest_sz), "either both dest and dest_sz exist, or none does");
if(unlikely(dest && !dest_sz))
return -1;
regex_t regex_compiled;
if(regex)
regex_compiled = *regex;
else{
char regexString[MAX_REGEX_SIZE];
const int regex_flags = ignore_case ? REG_EXTENDED | REG_NEWLINE | REG_ICASE : REG_EXTENDED | REG_NEWLINE;
snprintf(regexString, MAX_REGEX_SIZE, ".*(%s).*", keyword);
int rc;
if (unlikely((rc = regcomp(®ex_compiled, regexString, regex_flags)))){
size_t regcomp_err_str_size = regerror(rc, ®ex_compiled, 0, 0);
char *regcomp_err_str = mallocz(regcomp_err_str_size);
regerror(rc, ®ex_compiled, regcomp_err_str, regcomp_err_str_size);
fatal("Could not compile regular expression:%.*s, error: %s", (int) MAX_REGEX_SIZE, regexString, regcomp_err_str);
}
}
regmatch_t groupArray[1];
int matches = 0;
char *cursor = src;
if(dest_sz)
*dest_sz = 0;
for ( ; ; matches++){
if (regexec(®ex_compiled, cursor, 1, groupArray, REG_NOTBOL | REG_NOTEOL))
break; // No more matches
if (groupArray[0].rm_so == -1)
break; // No more groups
size_t match_len = (size_t) (groupArray[0].rm_eo - groupArray[0].rm_so);
// debug_log( "Match %d [%2d-%2d]:%.*s\n", matches, groupArray[0].rm_so,
// groupArray[0].rm_eo, (int) match_len, cursor + groupArray[0].rm_so);
if(dest && dest_sz){
memcpy( &dest[*dest_sz], cursor + groupArray[0].rm_so, match_len);
*dest_sz += match_len + 1;
dest[*dest_sz - 1] = '\n';
}
cursor += groupArray[0].rm_eo;
}
if(!regex)
regfree(®ex_compiled);
return matches;
}
/**
* @brief Extract web log parser configuration from string
* @param[in] log_format String that describes the log format
* @param[in] delimiter Delimiter to be used when parsing a CSV log format
* @return Pointer to struct that contains the extracted log format
* configuration or NULL if no fields found in log_format.
*/
Web_log_parser_config_t *read_web_log_parser_config(const char *log_format, const char delimiter){
int num_fields = count_fields(log_format, delimiter);
if(num_fields <= 0) return NULL;
/* If first execution of this function, initialise regexs */
static int regexs_initialised = 0;
// TODO: Tests needed for following regexs.
if(!regexs_initialised){
assert(regcomp(&vhost_regex, "^[a-zA-Z0-9:.-]+$", REG_NOSUB | REG_EXTENDED) == 0);
assert(regcomp(&req_client_regex, "^([0-9a-f:.]+|localhost)$", REG_NOSUB | REG_EXTENDED) == 0);
assert(regcomp(&cipher_suite_regex, "^[A-Z0-9_-]+$", REG_NOSUB | REG_EXTENDED) == 0);
regexs_initialised = 1;
}
Web_log_parser_config_t *wblp_config = callocz(1, sizeof(Web_log_parser_config_t));
wblp_config->num_fields = num_fields;
wblp_config->delimiter = delimiter;
char **parsed_format = parse_csv(log_format, delimiter, num_fields); // parsed_format is NULL-terminated
wblp_config->fields = callocz(num_fields, sizeof(web_log_line_field_t));
unsigned int fields_off = 0;
for(int i = 0; i < num_fields; i++ ){
if(strcmp(parsed_format[i], "$host:$server_port") == 0 ||
strcmp(parsed_format[i], "%v:%p") == 0) {
wblp_config->fields[fields_off++] = VHOST_WITH_PORT;
continue;
}
if(strcmp(parsed_format[i], "$host") == 0 ||
strcmp(parsed_format[i], "$http_host") == 0 ||
strcmp(parsed_format[i], "%v") == 0) {
wblp_config->fields[fields_off++] = VHOST;
continue;
}
if(strcmp(parsed_format[i], "$server_port") == 0 ||
strcmp(parsed_format[i], "%p") == 0) {
wblp_config->fields[fields_off++] = PORT;
continue;
}
if(strcmp(parsed_format[i], "$scheme") == 0) {
wblp_config->fields[fields_off++] = REQ_SCHEME;
continue;
}
if(strcmp(parsed_format[i], "$remote_addr") == 0 ||
strcmp(parsed_format[i], "%a") == 0 ||
strcmp(parsed_format[i], "%h") == 0) {
wblp_config->fields[fields_off++] = REQ_CLIENT;
continue;
}
if(strcmp(parsed_format[i], "$request") == 0 ||
strcmp(parsed_format[i], "%r") == 0) {
wblp_config->fields[fields_off++] = REQ;
continue;
}
if(strcmp(parsed_format[i], "$request_method") == 0 ||
strcmp(parsed_format[i], "%m") == 0) {
wblp_config->fields[fields_off++] = REQ_METHOD;
continue;
}
if(strcmp(parsed_format[i], "$request_uri") == 0 ||
strcmp(parsed_format[i], "%U") == 0) {
wblp_config->fields[fields_off++] = REQ_URL;
continue;
}
if(strcmp(parsed_format[i], "$server_protocol") == 0 ||
strcmp(parsed_format[i], "%H") == 0) {
wblp_config->fields[fields_off++] = REQ_PROTO;
continue;
}
if(strcmp(parsed_format[i], "$request_length") == 0 ||
strcmp(parsed_format[i], "%I") == 0) {
wblp_config->fields[fields_off++] = REQ_SIZE;
continue;
}
if(strcmp(parsed_format[i], "$request_time") == 0 ||
strcmp(parsed_format[i], "%D") == 0) {
wblp_config->fields[fields_off++] = REQ_PROC_TIME;
continue;
}
if(strcmp(parsed_format[i], "$status") == 0 ||
strcmp(parsed_format[i], "%>s") == 0 ||
strcmp(parsed_format[i], "%s") == 0) {
wblp_config->fields[fields_off++] = RESP_CODE;
continue;
}
if(strcmp(parsed_format[i], "$bytes_sent") == 0 ||
strcmp(parsed_format[i], "$body_bytes_sent") == 0 ||
strcmp(parsed_format[i], "%b") == 0 ||
strcmp(parsed_format[i], "%O") == 0 ||
strcmp(parsed_format[i], "%B") == 0) {
wblp_config->fields[fields_off++] = RESP_SIZE;
continue;
}
if(strcmp(parsed_format[i], "$upstream_response_time") == 0) {
wblp_config->fields[fields_off++] = UPS_RESP_TIME;
continue;
}
if(strcmp(parsed_format[i], "$ssl_protocol") == 0) {
wblp_config->fields[fields_off++] = SSL_PROTO;
continue;
}
if(strcmp(parsed_format[i], "$ssl_cipher") == 0) {
wblp_config->fields[fields_off++] = SSL_CIPHER_SUITE;
continue;
}
if(strcmp(parsed_format[i], "$time_local") == 0 || strcmp(parsed_format[i], "[$time_local]") == 0 ||
strcmp(parsed_format[i], "%t") == 0 || strcmp(parsed_format[i], "[%t]") == 0) {
wblp_config->fields = reallocz(wblp_config->fields, (num_fields + 1) * sizeof(web_log_line_field_t));
wblp_config->fields[fields_off++] = TIME;
wblp_config->fields[fields_off++] = TIME; // TIME takes 2 fields
wblp_config->num_fields++; // TIME takes 2 fields
continue;
}
wblp_config->fields[fields_off++] = CUSTOM;
}
for(int i = 0; parsed_format[i] != NULL; i++)
freez(parsed_format[i]);
freez(parsed_format);
return wblp_config;
}
/**
* @brief Parse a web log line to extract individual fields.
* @param[in] wblp_config Configuration that specifies how to parse the line.
* @param[in] line Web log record to be parsed. '\n', '\r' or '\0' terminated.
* @param[out] log_line_parsed Struct that stores the results of parsing.
*/
void parse_web_log_line(const Web_log_parser_config_t *wblp_config,
char *line, size_t line_len,
Log_line_parsed_t *log_line_parsed){
/* Read parsing configuration */
web_log_line_field_t *fields_format = wblp_config->fields;
const int num_fields_config = wblp_config->num_fields;
const char delimiter = wblp_config->delimiter;
const int verify = wblp_config->verify_parsed_logs;
/* Consume new lines and spaces at end of line */
for(; line[line_len-1] == '\n' || line[line_len-1] == '\r' || line[line_len-1] == ' '; line_len--);
char *field = line;
char *offset = line;
size_t field_size = 0;
for(int i = 0; i < num_fields_config; i++ ){
/* Consume double quotes and extra delimiters at beginning of field */
while(*field == '"' || *field == delimiter) field++, offset++;
/* Find offset boundaries of next field in line */
while(((size_t)(offset - line) < line_len) && *offset != delimiter) offset++;
if(unlikely(*(offset - 1) == '"')) offset--;
field_size = (size_t) (offset - field);
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Field[%d]:%.*s", i, (int)field_size, field);
#endif
if(fields_format[i] == CUSTOM){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: CUSTOM or UNKNOWN):%.*s", i, (int)field_size, field);
#endif
goto next_item;
}
char *port = field;
size_t port_size = 0;
size_t vhost_size = 0;
if(fields_format[i] == VHOST_WITH_PORT){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: VHOST_WITH_PORT):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->vhost[0] = '\0';
log_line_parsed->port = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
goto next_item;
}
while(*port != ':' && vhost_size < field_size) { port++; vhost_size++; }
if(likely(vhost_size < field_size)) {
/* ':' detected in string */
port++;
port_size = field_size - vhost_size - 1;
field_size = vhost_size; // now field represents vhost and port is separate
}
else {
/* no ':' detected in string - invalid */
log_line_parsed->vhost[0] = '\0';
log_line_parsed->port = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
goto next_item;
}
}
if(fields_format[i] == VHOST_WITH_PORT || fields_format[i] == VHOST){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: VHOST):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->vhost[0] = '\0';
log_line_parsed->parsing_errors++;
goto next_item;
}
// TODO: Add below case in code!!!
// nginx $host and $http_host return ipv6 in [], apache doesn't
// TODO: TEST! This case hasn't been tested!
// char *pch = strchr(parsed[i], ']');
// if(pch){
// *pch = '\0';
// memmove(parsed[i], parsed[i]+1, strlen(parsed[i]));
// }
snprintfz(log_line_parsed->vhost, VHOST_MAX_LEN, "%.*s", (int) field_size, field);
if(verify){
// if(field_size >= VHOST_MAX_LEN){
// #if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
// collector_error("VHOST is invalid");
// #endif
// log_line_parsed->vhost[0] = '\0';
// log_line_parsed->parsing_errors++;
// goto next_item; // TODO: Not entirely right, as it will also skip PORT parsing in case of VHOST_WITH_PORT
// }
if(unlikely(regexec(&vhost_regex, log_line_parsed->vhost, 0, NULL, 0) == REG_NOMATCH)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("VHOST is invalid");
#endif
// log_line_parsed->vhost[0] = 'invalid';
snprintf(log_line_parsed->vhost, sizeof(WEB_LOG_INVALID_HOST_STR), WEB_LOG_INVALID_HOST_STR);
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted VHOST:%s", log_line_parsed->vhost);
#endif
if(fields_format[i] == VHOST) goto next_item;
}
if(fields_format[i] == VHOST_WITH_PORT || fields_format[i] == PORT){
if(fields_format[i] != VHOST_WITH_PORT){
port = field;
port_size = field_size;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: PORT):%.*s", i, (int) port_size, port);
#endif
if(unlikely(port[0] == '-' && port_size == 1)){
log_line_parsed->port = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
goto next_item;
}
char port_d[PORT_MAX_LEN];
snprintfz( port_d, PORT_MAX_LEN, "%.*s", (int) port_size, port);
if(likely(str2int(&log_line_parsed->port, port_d, 10) == STR2XX_SUCCESS)){
if(verify){
if(unlikely(log_line_parsed->port < 80 || log_line_parsed->port > 49151)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("PORT is invalid (<80 or >49151)");
#endif
log_line_parsed->port = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
}
}
}
else{
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting PORT from string");
#endif
log_line_parsed->port = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted PORT:%d", log_line_parsed->port);
#endif
goto next_item;
}
if(fields_format[i] == REQ_SCHEME){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ_SCHEME):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->req_scheme[0] = '\0';
log_line_parsed->parsing_errors++;
goto next_item;
}
snprintfz(log_line_parsed->req_scheme, REQ_SCHEME_MAX_LEN, "%.*s", (int) field_size, field);
if(verify){
if(unlikely( strcmp(log_line_parsed->req_scheme, "http") &&
strcmp(log_line_parsed->req_scheme, "https"))){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_SCHEME is invalid (must be either 'http' or 'https')");
#endif
log_line_parsed->req_scheme[0] = '\0';
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_SCHEME:%s", log_line_parsed->req_scheme);
#endif
goto next_item;
}
if(fields_format[i] == REQ_CLIENT){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ_CLIENT):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->req_client[0] = '\0';
log_line_parsed->parsing_errors++;
goto next_item;
}
snprintfz(log_line_parsed->req_client, REQ_CLIENT_MAX_LEN, "%.*s", (int)field_size, field);
if(verify){
int regex_rc = regexec(&req_client_regex, log_line_parsed->req_client, 0, NULL, 0);
if (likely(regex_rc == 0)) {/* do nothing */}
else if (unlikely(regex_rc == REG_NOMATCH)) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_CLIENT is invalid");
#endif
snprintf(log_line_parsed->req_client, REQ_CLIENT_MAX_LEN, "%s", WEB_LOG_INVALID_CLIENT_IP_STR);
log_line_parsed->parsing_errors++;
}
else {
size_t err_msg_size = regerror(regex_rc, &req_client_regex, NULL, 0);
char *err_msg = mallocz(err_msg_size);
regerror(regex_rc, &req_client_regex, err_msg, err_msg_size);
collector_error("req_client_regex error:%s", err_msg);
freez(err_msg);
m_assert(0, "req_client_regex has failed");
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_CLIENT:%s", log_line_parsed->req_client);
#endif
goto next_item;
}
if(fields_format[i] == REQ || fields_format[i] == REQ_METHOD){
/* If fields_format[i] == REQ, then field is filled in with request in the previous code */
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ or REQ_METHOD):%.*s", i, (int)field_size, field);
#endif
snprintfz( log_line_parsed->req_method, REQ_METHOD_MAX_LEN, "%.*s", (int)field_size, field);
if(verify){
if( unlikely(
/* GET and POST are the most common requests, so check them first */
strcmp(log_line_parsed->req_method, "GET") &&
strcmp(log_line_parsed->req_method, "POST") &&
strcmp(log_line_parsed->req_method, "ACL") &&
strcmp(log_line_parsed->req_method, "BASELINE-CONTROL") &&
strcmp(log_line_parsed->req_method, "BIND") &&
strcmp(log_line_parsed->req_method, "CHECKIN") &&
strcmp(log_line_parsed->req_method, "CHECKOUT") &&
strcmp(log_line_parsed->req_method, "CONNECT") &&
strcmp(log_line_parsed->req_method, "COPY") &&
strcmp(log_line_parsed->req_method, "DELETE") &&
strcmp(log_line_parsed->req_method, "HEAD") &&
strcmp(log_line_parsed->req_method, "LABEL") &&
strcmp(log_line_parsed->req_method, "LINK") &&
strcmp(log_line_parsed->req_method, "LOCK") &&
strcmp(log_line_parsed->req_method, "MERGE") &&
strcmp(log_line_parsed->req_method, "MKACTIVITY") &&
strcmp(log_line_parsed->req_method, "MKCALENDAR") &&
strcmp(log_line_parsed->req_method, "MKCOL") &&
strcmp(log_line_parsed->req_method, "MKREDIRECTREF") &&
strcmp(log_line_parsed->req_method, "MKWORKSPACE") &&
strcmp(log_line_parsed->req_method, "MOVE") &&
strcmp(log_line_parsed->req_method, "OPTIONS") &&
strcmp(log_line_parsed->req_method, "ORDERPATCH") &&
strcmp(log_line_parsed->req_method, "PATCH") &&
strcmp(log_line_parsed->req_method, "PRI") &&
strcmp(log_line_parsed->req_method, "PROPFIND") &&
strcmp(log_line_parsed->req_method, "PROPPATCH") &&
strcmp(log_line_parsed->req_method, "PUT") &&
strcmp(log_line_parsed->req_method, "REBIND") &&
strcmp(log_line_parsed->req_method, "REPORT") &&
strcmp(log_line_parsed->req_method, "SEARCH") &&
strcmp(log_line_parsed->req_method, "TRACE") &&
strcmp(log_line_parsed->req_method, "UNBIND") &&
strcmp(log_line_parsed->req_method, "UNCHECKOUT") &&
strcmp(log_line_parsed->req_method, "UNLINK") &&
strcmp(log_line_parsed->req_method, "UNLOCK") &&
strcmp(log_line_parsed->req_method, "UPDATE") &&
strcmp(log_line_parsed->req_method, "UPDATEREDIRECTREF") &&
strcmp(log_line_parsed->req_method, "-"))) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_METHOD is invalid");
#endif
log_line_parsed->req_method[0] = '\0';
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_METHOD:%s", log_line_parsed->req_method);
#endif
if(fields_format[i] == REQ && field[0] != '-') {
while(*(offset + 1) == delimiter) offset++; // Consume extra whitespace characters
field = ++offset;
while(*offset != delimiter && ((size_t)(offset - line) < line_len)) offset++;
field_size = (size_t) (offset - field);
}
else goto next_item;
}
if(fields_format[i] == REQ || fields_format[i] == REQ_URL){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ or REQ_URL):%.*s", i, (int)field_size, field);
#endif
snprintfz( log_line_parsed->req_URL, REQ_URL_MAX_LEN, "%.*s", (int)field_size, field);
// if(unlikely(field[0] == '-' && field_size == 1)){
// log_line_parsed->req_method[0] = '\0';
// log_line_parsed->parsing_errors++;
// }
//if(verify){} ??
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_URL:%s", log_line_parsed->req_URL ? log_line_parsed->req_URL : "NULL!");
#endif
if(fields_format[i] == REQ) {
while(*(offset + 1) == delimiter) offset++; // Consume extra whitespace characters
field = ++offset;
while(*offset != delimiter && ((size_t)(offset - line) < line_len)) offset++;
field_size = (size_t) (offset - field);
}
else goto next_item;
}
if(fields_format[i] == REQ || fields_format[i] == REQ_PROTO){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ or REQ_PROTO):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->req_proto[0] = '\0';
log_line_parsed->parsing_errors++;
goto next_item;
}
if(unlikely( field_size > REQ_PROTO_PREF_SIZE + REQ_PROTO_MAX_LEN - 1)){
field_size = REQ_PROTO_PREF_SIZE + REQ_PROTO_MAX_LEN - 1;
}
size_t req_proto_num_size = field_size - REQ_PROTO_PREF_SIZE;
if(verify){
if(unlikely(field_size < 6 ||
req_proto_num_size == 0 ||
strncmp(field, "HTTP/", REQ_PROTO_PREF_SIZE) ||
( strncmp(&field[REQ_PROTO_PREF_SIZE], "1", req_proto_num_size) &&
strncmp(&field[REQ_PROTO_PREF_SIZE], "1.0", req_proto_num_size) &&
strncmp(&field[REQ_PROTO_PREF_SIZE], "1.1", req_proto_num_size) &&
strncmp(&field[REQ_PROTO_PREF_SIZE], "2", req_proto_num_size) &&
strncmp(&field[REQ_PROTO_PREF_SIZE], "2.0", req_proto_num_size)))) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_PROTO is invalid");
#endif
log_line_parsed->req_proto[0] = '\0';
log_line_parsed->parsing_errors++;
}
else snprintfz( log_line_parsed->req_proto, req_proto_num_size + 1,
"%.*s", (int)req_proto_num_size, &field[REQ_PROTO_PREF_SIZE]);
}
else snprintfz( log_line_parsed->req_proto, req_proto_num_size + 1,
"%.*s", (int)req_proto_num_size, &field[REQ_PROTO_PREF_SIZE]);
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_PROTO:%s", log_line_parsed->req_proto);
#endif
goto next_item;
}
if(fields_format[i] == REQ_SIZE){
/* TODO: Differentiate between '-' or 0 and an invalid request size.
* right now, all these will set req_size == 0 */
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ_SIZE):%.*s", i, (int)field_size, field);
#endif
char req_size_d[REQ_SIZE_MAX_LEN];
snprintfz( req_size_d, REQ_SIZE_MAX_LEN, "%.*s", (int) field_size, field);
if(field[0] == '-' && field_size == 1) {
log_line_parsed->req_size = 0; // Request size can be '-'
}
else if(likely(str2int(&log_line_parsed->req_size, req_size_d, 10) == STR2XX_SUCCESS)){
if(verify){
if(unlikely(log_line_parsed->req_size < 0)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_SIZE is invalid (<0)");
#endif
log_line_parsed->req_size = 0;
log_line_parsed->parsing_errors++;
}
}
}
else{
collector_error("Error while extracting REQ_SIZE from string");
log_line_parsed->req_size = 0;
log_line_parsed->parsing_errors++;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_SIZE:%d", log_line_parsed->req_size);
#endif
goto next_item;
}
if(fields_format[i] == REQ_PROC_TIME){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: REQ_PROC_TIME):%.*s", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->req_proc_time = WEB_LOG_INVALID_PORT;
log_line_parsed->parsing_errors++;
goto next_item;
}
float f = 0;
char req_proc_time_d[REQ_PROC_TIME_MAX_LEN];
snprintfz( req_proc_time_d, REQ_PROC_TIME_MAX_LEN, "%.*s", (int) field_size, field);
if(memchr(field, '.', field_size)){ // nginx time is in seconds with a milliseconds resolution.
if(likely(str2float(&f, req_proc_time_d) == STR2XX_SUCCESS)){
log_line_parsed->req_proc_time = (int) (f * 1.0E6);
}
else {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting REQ_PROC_TIME from string");
#endif
log_line_parsed->req_proc_time = 0;
log_line_parsed->parsing_errors++;
}
}
else{ // apache time is in microseconds
if(unlikely(str2int(&log_line_parsed->req_proc_time, req_proc_time_d, 10) != STR2XX_SUCCESS)) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting REQ_PROC_TIME from string");
#endif
log_line_parsed->req_proc_time = 0;
log_line_parsed->parsing_errors++;
}
}
if(verify){
if(unlikely(log_line_parsed->req_proc_time < 0)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("REQ_PROC_TIME is invalid (<0)");
#endif
log_line_parsed->req_proc_time = 0;
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted REQ_PROC_TIME:%d", log_line_parsed->req_proc_time);
#endif
goto next_item;
}
if(fields_format[i] == RESP_CODE){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: RESP_CODE):%.*s\n", i, (int)field_size, field);
#endif
if(unlikely(field[0] == '-' && field_size == 1)){
log_line_parsed->resp_code = 0;
log_line_parsed->parsing_errors++;
goto next_item;
}
char resp_code_d[REQ_RESP_CODE_MAX_LEN];
snprintfz( resp_code_d, REQ_RESP_CODE_MAX_LEN, "%.*s", (int)field_size, field);
if(likely(str2int(&log_line_parsed->resp_code, resp_code_d, 10) == STR2XX_SUCCESS)){
if(verify){
/* rfc7231
* Informational responses (100–199),
* Successful responses (200–299),
* Redirects (300–399),
* Client errors (400–499),
* Server errors (500–599). */
if(unlikely(log_line_parsed->resp_code < 100 || log_line_parsed->resp_code > 599)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("RESP_CODE is invalid (<100 or >599)");
#endif
log_line_parsed->resp_code = 0;
log_line_parsed->parsing_errors++;
}
}
}
else{
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting RESP_CODE from string");
#endif
log_line_parsed->resp_code = 0;
log_line_parsed->parsing_errors++;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted RESP_CODE:%d", log_line_parsed->resp_code);
#endif
goto next_item;
}
if(fields_format[i] == RESP_SIZE){
/* TODO: Differentiate between '-' or 0 and an invalid response size.
* right now, all these will set resp_size == 0 */
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: RESP_SIZE):%.*s", i, (int)field_size, field);
#endif
char resp_size_d[REQ_RESP_SIZE_MAX_LEN];
snprintfz( resp_size_d, REQ_RESP_SIZE_MAX_LEN, "%.*s", (int)field_size, field);
if(field[0] == '-' && field_size == 1) {
log_line_parsed->resp_size = 0; // Response size can be '-'
}
else if(likely(str2int(&log_line_parsed->resp_size, resp_size_d, 10) == STR2XX_SUCCESS)){
if(verify){
if(unlikely(log_line_parsed->resp_size < 0)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("RESP_SIZE is invalid (<0)");
#endif
log_line_parsed->resp_size = 0;
log_line_parsed->parsing_errors++;
}
}
}
else {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting RESP_SIZE from string");
#endif
log_line_parsed->resp_size = 0;
log_line_parsed->parsing_errors++;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted RESP_SIZE:%d", log_line_parsed->resp_size);
#endif
goto next_item;
}
if(fields_format[i] == UPS_RESP_TIME){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: UPS_RESP_TIME):%.*s", i, (int)field_size, field);
#endif
if(field[0] == '-' && field_size == 1) {
log_line_parsed->ups_resp_time = 0;
log_line_parsed->parsing_errors++;
goto next_item;
}
/* Times of several responses are separated by commas and colons. Following the
* Go parser implementation, where only the first one is kept, the others are
* discarded. Also, there must be no space in between them. Needs testing... */
char *pch = memchr(field, ',', field_size);
if(pch) field_size = pch - field;
float f = 0;
char ups_resp_time_d[UPS_RESP_TIME_MAX_LEN];
snprintfz( ups_resp_time_d, UPS_RESP_TIME_MAX_LEN, "%.*s", (int)field_size, field);
if(memchr(field, '.', field_size)){ // nginx time is in seconds with a milliseconds resolution.
if(likely(str2float(&f, ups_resp_time_d) == STR2XX_SUCCESS)){
log_line_parsed->ups_resp_time = (int) (f * 1.0E6);
}
else {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting UPS_RESP_TIME from string");
#endif
log_line_parsed->ups_resp_time = 0;
log_line_parsed->parsing_errors++;
}
}
else{ // unlike in the REQ_PROC_TIME case, apache doesn't have an equivalent here
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("Error while extracting UPS_RESP_TIME from string");
#endif
log_line_parsed->ups_resp_time = 0;
log_line_parsed->parsing_errors++;
}
if(verify){
if(unlikely(log_line_parsed->ups_resp_time < 0)){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("UPS_RESP_TIME is invalid (<0)");
#endif
log_line_parsed->ups_resp_time = 0;
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted UPS_RESP_TIME:%d", log_line_parsed->ups_resp_time);
#endif
goto next_item;
}
if(fields_format[i] == SSL_PROTO){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: SSL_PROTO):%.*s", i, (int)field_size, field);
#endif
if(field[0] == '-' && field_size == 1) {
log_line_parsed->ssl_proto[0] = '\0';
log_line_parsed->parsing_errors++;
goto next_item;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "SSL_PROTO field size:%zu", field_size);
#endif
snprintfz( log_line_parsed->ssl_proto, SSL_PROTO_MAX_LEN, "%.*s", (int)field_size, field);
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "log_line_parsed->ssl_proto:%s", log_line_parsed->ssl_proto);
#endif
if(verify){
if(unlikely(strcmp(log_line_parsed->ssl_proto, "TLSv1") &&
strcmp(log_line_parsed->ssl_proto, "TLSv1.1") &&
strcmp(log_line_parsed->ssl_proto, "TLSv1.2") &&
strcmp(log_line_parsed->ssl_proto, "TLSv1.3") &&
strcmp(log_line_parsed->ssl_proto, "SSLv2") &&
strcmp(log_line_parsed->ssl_proto, "SSLv3"))) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("SSL_PROTO is invalid");
#endif
log_line_parsed->ssl_proto[0] = '\0';
log_line_parsed->parsing_errors++;
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted SSL_PROTO:%s", log_line_parsed->ssl_proto);
#endif
goto next_item;
}
if(fields_format[i] == SSL_CIPHER_SUITE){
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: SSL_CIPHER_SUITE):%.*s", i, (int)field_size, field);
#endif
if(field[0] == '-' && field_size == 1) {
log_line_parsed->ssl_cipher[0] = '\0';
log_line_parsed->parsing_errors++;
}
snprintfz( log_line_parsed->ssl_cipher, SSL_CIPHER_SUITE_MAX_LEN, "%.*s", (int)field_size, field);
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "before: SSL_CIPHER_SUITE:%s", log_line_parsed->ssl_cipher);
#endif
if(verify){
int regex_rc = regexec(&cipher_suite_regex, log_line_parsed->ssl_cipher, 0, NULL, 0);
if (likely(regex_rc == 0)){/* do nothing */}
else if (unlikely(regex_rc == REG_NOMATCH)) {
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
collector_error("SSL_CIPHER_SUITE is invalid");
#endif
log_line_parsed->ssl_cipher[0] = '\0';
log_line_parsed->parsing_errors++;
}
else {
size_t err_msg_size = regerror(regex_rc, &cipher_suite_regex, NULL, 0);
char *err_msg = mallocz(err_msg_size);
regerror(regex_rc, &cipher_suite_regex, err_msg, err_msg_size);
collector_error("cipher_suite_regex error:%s", err_msg);
freez(err_msg);
m_assert(0, "cipher_suite_regex has failed");
}
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Extracted SSL_CIPHER_SUITE:%s", log_line_parsed->ssl_cipher);
#endif
goto next_item;
}
if(fields_format[i] == TIME){
if(wblp_config->skip_timestamp_parsing){
while(*offset != ']') offset++;
i++;
offset++;
goto next_item;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Item %d (type: TIME - 1st of 2 fields):%.*s", i, (int)field_size, field);
#endif
// TODO: What if TIME is invalid?
// if(field[0] == '-' && field_size == 1) {
// log_line_parsed->timestamp = 0;
// log_line_parsed->parsing_errors++;
// ++i;
// goto next_item;
// }
char *datetime = field;
if(memchr(datetime, '[', field_size)) {
datetime++;
field_size--;
}
struct tm ltm = {0};
char *tz_str = strptime(datetime, "%d/%b/%Y:%H:%M:%S", <m);
if(unlikely(tz_str == NULL)){
collector_error("TIME datetime parsing failed");
log_line_parsed->timestamp = 0;
log_line_parsed->parsing_errors++;
goto next_item;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "strptime() result: year:%d mon:%d day:%d hour:%d min:%d sec:%d",
ltm.tm_year, ltm.tm_mon, ltm.tm_mday,
ltm.tm_hour, ltm.tm_min, ltm.tm_sec);
#endif
/* Deal with 2nd part of datetime i.e. timezone */
m_assert(*tz_str == ' ', "Invalid TIME timezone");
++tz_str;
m_assert(*tz_str == '+' || *tz_str == '-', "Invalid TIME timezone");
char tz_sign = *tz_str;
char *tz_str_end = ++tz_str;
while(*tz_str_end != ']') tz_str_end++;
m_assert(tz_str_end - tz_str == 4, "Invalid TIME timezone string length");
char tz_num[4];
memcpy(tz_num, tz_str, tz_str_end - tz_str);
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "TIME 2nd part: %.*s", (int)(tz_str_end - tz_str), tz_str);
#endif
long int tz = strtol(tz_str, NULL, 10);
long int tz_h = tz / 100;
long int tz_m = tz % 100;
int64_t tz_adj = (int64_t) tz_h * 3600 + (int64_t) tz_m * 60;
if(tz_sign == '+') tz_adj *= -1; // if timezone is positive, we need to subtract it to get GMT
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
debug_log( "Timezone: int:%ld, hrs:%ld, mins:%ld", tz, tz_h, tz_m);
#endif
if(-1 == (log_line_parsed->timestamp = timegm(<m) + tz_adj)){
collector_error("TIME datetime parsing failed");
log_line_parsed->timestamp = 0;
log_line_parsed->parsing_errors++;
}
#if ENABLE_PARSE_WEB_LOG_LINE_DEBUG
char tb[80];
strftime(tb, sizeof(tb), "%c", <m );
debug_log( "Extracted TIME:%ld", log_line_parsed->timestamp);
debug_log( "Extracted TIME string:%s", tb);
#endif
offset = tz_str_end + 1; // WARNING! this modifies the offset but it is required in the TIME case.
++i; // TIME takes up 2 fields_format[] spaces, so skip the next one
goto next_item;
}
next_item:
/* If offset is located beyond the end of the line, terminate parsing */
if(unlikely((size_t) (offset - line) >= line_len)) break;
field = ++offset;
}
}
/**
* @brief Extract web log metrics from a group of web log fields.
* @param[in] parser_config Configuration specifying how and what web log
* metrics to extract.
* @param[in] line_parsed Web logs fields extracted from a web log line.
* @param[out] metrics Web logs metrics exctracted from the \p line_parsed
* web log fields, using the \p parser_config configuration.
*/
void extract_web_log_metrics(Log_parser_config_t *parser_config,
Log_line_parsed_t *line_parsed,
Web_log_metrics_t *metrics){
/* Extract number of parsed lines */
/* NOTE: Commented out as it is done in flb_collect_logs_cb() now. */
// metrics->num_lines++;
/* Extract vhost */
// TODO: Reduce number of reallocs
if((parser_config->chart_config & CHART_VHOST) && *line_parsed->vhost){
int i;
for(i = 0; i < metrics->vhost_arr.size; i++){
if(!strcmp(metrics->vhost_arr.vhosts[i].name, line_parsed->vhost)){
metrics->vhost_arr.vhosts[i].count++;
break;
}
}
if(metrics->vhost_arr.size == i){ // Vhost not found in array - need to append
metrics->vhost_arr.size++;
if(metrics->vhost_arr.size >= metrics->vhost_arr.size_max){
metrics->vhost_arr.size_max = metrics->vhost_arr.size * VHOST_BUFFS_SCALE_FACTOR + 1;
metrics->vhost_arr.vhosts = reallocz( metrics->vhost_arr.vhosts,
metrics->vhost_arr.size_max * sizeof(struct log_parser_metrics_vhost));
}
snprintf(metrics->vhost_arr.vhosts[metrics->vhost_arr.size - 1].name, VHOST_MAX_LEN, "%s", line_parsed->vhost);
metrics->vhost_arr.vhosts[metrics->vhost_arr.size - 1].count = 1;
}
}
/* Extract port */
// TODO: Reduce number of reallocs
if((parser_config->chart_config & CHART_PORT) && line_parsed->port){
int i;
for(i = 0; i < metrics->port_arr.size; i++){
if(metrics->port_arr.ports[i].port == line_parsed->port){
metrics->port_arr.ports[i].count++;
break;
}
}
if(metrics->port_arr.size == i){ // Port not found in array - need to append
metrics->port_arr.size++;
if(metrics->port_arr.size >= metrics->port_arr.size_max){
metrics->port_arr.size_max = metrics->port_arr.size * PORT_BUFFS_SCALE_FACTOR + 1;
metrics->port_arr.ports = reallocz( metrics->port_arr.ports,
metrics->port_arr.size_max * sizeof(struct log_parser_metrics_port));
}
if(line_parsed->port == WEB_LOG_INVALID_PORT)
snprintfz(metrics->port_arr.ports[metrics->port_arr.size - 1].name, PORT_MAX_LEN, WEB_LOG_INVALID_PORT_STR);
else
snprintfz(metrics->port_arr.ports[metrics->port_arr.size - 1].name, PORT_MAX_LEN, "%d", line_parsed->port);
metrics->port_arr.ports[metrics->port_arr.size - 1].port = line_parsed->port;
metrics->port_arr.ports[metrics->port_arr.size - 1].count = 1;
}
}
/* Extract client metrics */
if(( parser_config->chart_config & ( CHART_IP_VERSION | CHART_REQ_CLIENT_CURRENT | CHART_REQ_CLIENT_ALL_TIME)) && *line_parsed->req_client) {
/* Invalid IP version */
if(unlikely(!strcmp(line_parsed->req_client, WEB_LOG_INVALID_CLIENT_IP_STR))){
if(parser_config->chart_config & CHART_IP_VERSION) metrics->ip_ver.invalid++;
}
else if(strchr(line_parsed->req_client, ':')){
/* IPv6 version */
if(parser_config->chart_config & CHART_IP_VERSION) metrics->ip_ver.v6++;
/* Unique Client IPv6 Address current poll */
if(parser_config->chart_config & CHART_REQ_CLIENT_CURRENT){
int i;
for(i = 0; i < metrics->req_clients_current_arr.ipv6_size; i++){
if(!strcmp(metrics->req_clients_current_arr.ipv6_req_clients[i], line_parsed->req_client)) break;
}
if(metrics->req_clients_current_arr.ipv6_size == i){ // Req client not found in array - need to append
metrics->req_clients_current_arr.ipv6_size++;
metrics->req_clients_current_arr.ipv6_req_clients = reallocz(metrics->req_clients_current_arr.ipv6_req_clients,
metrics->req_clients_current_arr.ipv6_size * sizeof(*metrics->req_clients_current_arr.ipv6_req_clients));
snprintf(metrics->req_clients_current_arr.ipv6_req_clients[metrics->req_clients_current_arr.ipv6_size - 1],
REQ_CLIENT_MAX_LEN, "%s", line_parsed->req_client);
}
}
/* Unique Client IPv6 Address all-time */
if(parser_config->chart_config & CHART_REQ_CLIENT_ALL_TIME){
int i;
for(i = 0; i < metrics->req_clients_alltime_arr.ipv6_size; i++){
if(!strcmp(metrics->req_clients_alltime_arr.ipv6_req_clients[i], line_parsed->req_client)) break;
}
if(metrics->req_clients_alltime_arr.ipv6_size == i){ // Req client not found in array - need to append
metrics->req_clients_alltime_arr.ipv6_size++;
metrics->req_clients_alltime_arr.ipv6_req_clients = reallocz(metrics->req_clients_alltime_arr.ipv6_req_clients,
metrics->req_clients_alltime_arr.ipv6_size * sizeof(*metrics->req_clients_alltime_arr.ipv6_req_clients));
snprintf(metrics->req_clients_alltime_arr.ipv6_req_clients[metrics->req_clients_alltime_arr.ipv6_size - 1],
REQ_CLIENT_MAX_LEN, "%s", line_parsed->req_client);
}
}
}
else{
/* IPv4 version */
if(parser_config->chart_config & CHART_IP_VERSION) metrics->ip_ver.v4++;
/* Unique Client IPv4 Address current poll */
if(parser_config->chart_config & CHART_REQ_CLIENT_CURRENT){
int i;
for(i = 0; i < metrics->req_clients_current_arr.ipv4_size; i++){
if(!strcmp(metrics->req_clients_current_arr.ipv4_req_clients[i], line_parsed->req_client)) break;
}
if(metrics->req_clients_current_arr.ipv4_size == i){ // Req client not found in array - need to append
metrics->req_clients_current_arr.ipv4_size++;
metrics->req_clients_current_arr.ipv4_req_clients = reallocz(metrics->req_clients_current_arr.ipv4_req_clients,
metrics->req_clients_current_arr.ipv4_size * sizeof(*metrics->req_clients_current_arr.ipv4_req_clients));
snprintf(metrics->req_clients_current_arr.ipv4_req_clients[metrics->req_clients_current_arr.ipv4_size - 1],
REQ_CLIENT_MAX_LEN, "%s", line_parsed->req_client);
}
}
/* Unique Client IPv4 Address all-time */
if(parser_config->chart_config & CHART_REQ_CLIENT_ALL_TIME){
int i;
for(i = 0; i < metrics->req_clients_alltime_arr.ipv4_size; i++){
if(!strcmp(metrics->req_clients_alltime_arr.ipv4_req_clients[i], line_parsed->req_client)) break;
}
if(metrics->req_clients_alltime_arr.ipv4_size == i){ // Req client not found in array - need to append
metrics->req_clients_alltime_arr.ipv4_size++;
metrics->req_clients_alltime_arr.ipv4_req_clients = reallocz(metrics->req_clients_alltime_arr.ipv4_req_clients,
metrics->req_clients_alltime_arr.ipv4_size * sizeof(*metrics->req_clients_alltime_arr.ipv4_req_clients));
snprintf(metrics->req_clients_alltime_arr.ipv4_req_clients[metrics->req_clients_alltime_arr.ipv4_size - 1],
REQ_CLIENT_MAX_LEN, "%s", line_parsed->req_client);
}
}
}
}
/* Extract request method */
if(parser_config->chart_config & CHART_REQ_METHODS){
for(int i = 0; i < REQ_METHOD_ARR_SIZE; i++){
if(!strcmp(line_parsed->req_method, req_method_str[i])){
metrics->req_method[i]++;
break;
}
}
}
/* Extract request protocol */
if(parser_config->chart_config & CHART_REQ_PROTO){
if(!strcmp(line_parsed->req_proto, "1") || !strcmp(line_parsed->req_proto, "1.0")) metrics->req_proto.http_1++;
else if(!strcmp(line_parsed->req_proto, "1.1")) metrics->req_proto.http_1_1++;
else if(!strcmp(line_parsed->req_proto, "2") || !strcmp(line_parsed->req_proto, "2.0")) metrics->req_proto.http_2++;
else metrics->req_proto.other++;
}
/* Extract bytes received and sent */
if(parser_config->chart_config & CHART_BANDWIDTH){
metrics->bandwidth.req_size += line_parsed->req_size;
metrics->bandwidth.resp_size += line_parsed->resp_size;
}
/* Extract request processing time */
if((parser_config->chart_config & CHART_REQ_PROC_TIME) && line_parsed->req_proc_time){
if(line_parsed->req_proc_time < metrics->req_proc_time.min || metrics->req_proc_time.min == 0){
metrics->req_proc_time.min = line_parsed->req_proc_time;
}
if(line_parsed->req_proc_time > metrics->req_proc_time.max || metrics->req_proc_time.max == 0){
metrics->req_proc_time.max = line_parsed->req_proc_time;
}
metrics->req_proc_time.sum += line_parsed->req_proc_time;
metrics->req_proc_time.count++;
}
/* Extract response code family, response code & response code type */
if(parser_config->chart_config & (CHART_RESP_CODE_FAMILY | CHART_RESP_CODE | CHART_RESP_CODE_TYPE)){
switch(line_parsed->resp_code / 100){
/* Note: 304 and 401 should be treated as resp_success */
case 1:
metrics->resp_code_family.resp_1xx++;
metrics->resp_code[line_parsed->resp_code - 100]++;
metrics->resp_code_type.resp_success++;
break;
case 2:
metrics->resp_code_family.resp_2xx++;
metrics->resp_code[line_parsed->resp_code - 100]++;
metrics->resp_code_type.resp_success++;
break;
case 3:
metrics->resp_code_family.resp_3xx++;
metrics->resp_code[line_parsed->resp_code - 100]++;
if(line_parsed->resp_code == 304) metrics->resp_code_type.resp_success++;
else metrics->resp_code_type.resp_redirect++;
break;
case 4:
metrics->resp_code_family.resp_4xx++;
metrics->resp_code[line_parsed->resp_code - 100]++;
if(line_parsed->resp_code == 401) metrics->resp_code_type.resp_success++;
else metrics->resp_code_type.resp_bad++;
break;
case 5:
metrics->resp_code_family.resp_5xx++;
metrics->resp_code[line_parsed->resp_code - 100]++;
metrics->resp_code_type.resp_error++;
break;
default:
metrics->resp_code_family.other++;
metrics->resp_code[RESP_CODE_ARR_SIZE - 1]++;
metrics->resp_code_type.other++;
break;
}
}
/* Extract SSL protocol */
if(parser_config->chart_config & CHART_SSL_PROTO){
if(!strcmp(line_parsed->ssl_proto, "TLSv1")) metrics->ssl_proto.tlsv1++;
else if(!strcmp(line_parsed->ssl_proto, "TLSv1.1")) metrics->ssl_proto.tlsv1_1++;
else if(!strcmp(line_parsed->ssl_proto, "TLSv1.2")) metrics->ssl_proto.tlsv1_2++;
else if(!strcmp(line_parsed->ssl_proto, "TLSv1.3")) metrics->ssl_proto.tlsv1_3++;
else if(!strcmp(line_parsed->ssl_proto, "SSLv2")) metrics->ssl_proto.sslv2++;
else if(!strcmp(line_parsed->ssl_proto, "SSLv3")) metrics->ssl_proto.sslv3++;
else metrics->ssl_proto.other++;
}
/* Extract SSL cipher suite */
// TODO: Reduce number of reallocs
if((parser_config->chart_config & CHART_SSL_CIPHER) && *line_parsed->ssl_cipher){
int i;
for(i = 0; i < metrics->ssl_cipher_arr.size; i++){
if(!strcmp(metrics->ssl_cipher_arr.ssl_ciphers[i].name, line_parsed->ssl_cipher)){
metrics->ssl_cipher_arr.ssl_ciphers[i].count++;
break;
}
}
if(metrics->ssl_cipher_arr.size == i){ // SSL cipher suite not found in array - need to append
metrics->ssl_cipher_arr.size++;
metrics->ssl_cipher_arr.ssl_ciphers = reallocz(metrics->ssl_cipher_arr.ssl_ciphers,
metrics->ssl_cipher_arr.size * sizeof(struct log_parser_metrics_ssl_cipher));
snprintf( metrics->ssl_cipher_arr.ssl_ciphers[metrics->ssl_cipher_arr.size - 1].name,
SSL_CIPHER_SUITE_MAX_LEN, "%s", line_parsed->ssl_cipher);
metrics->ssl_cipher_arr.ssl_ciphers[metrics->ssl_cipher_arr.size - 1].count = 1;
}
}
metrics->timestamp = line_parsed->timestamp;
}
/**
* @brief Try to automatically detect the configuration for a web log parser.
* @details It tries to automatically detect the configuration to be used for
* a web log parser, by parsing a single web log line record and trying to pick
* a matching configuration (from a static list of predefined ones.)
* @param[in] line Null-terminated web log line to use in guessing the configuration.
* @param[in] delimiter Delimiter used to break down \p line in separate fields.
* @returns Pointer to the web log parser configuration if automatic detection
* was sucessful, otherwise NULL.
*/
Web_log_parser_config_t *auto_detect_web_log_parser_config(char *line, const char delimiter){
for(int i = 0; csv_auto_format_guess_matrix[i] != NULL; i++){
Web_log_parser_config_t *wblp_config = read_web_log_parser_config(csv_auto_format_guess_matrix[i], delimiter);
if(count_fields(line, delimiter) == wblp_config->num_fields){
wblp_config->verify_parsed_logs = 1; // Verification must be turned on to be able to pick up parsing_errors
Log_line_parsed_t line_parsed = (Log_line_parsed_t) {0};
parse_web_log_line(wblp_config, line, strlen(line), &line_parsed);
if(line_parsed.parsing_errors == 0){
return wblp_config;
}
}
freez(wblp_config->fields);
freez(wblp_config);
}
return NULL;
}