summaryrefslogtreecommitdiff
path: root/parsing.c
diff options
context:
space:
mode:
Diffstat (limited to 'parsing.c')
-rw-r--r--parsing.c260
1 files changed, 208 insertions, 52 deletions
diff --git a/parsing.c b/parsing.c
index 9653eb8..2447363 100644
--- a/parsing.c
+++ b/parsing.c
@@ -2,7 +2,7 @@
#include <string.h>
#include <wctype.h>
#include <wchar.h>
-#include <pthread.h>
+#include <errno.h>
#include "parsing.h"
#include "channel.h"
@@ -15,29 +15,54 @@
#define TEXT_BUFFER_SIZE 0x400
#define LINE_BUFFER_SIZE 0x400
#define TIME_BUFFER_SIZE 0xf
+#define DATE_BUFFER_SIZE 0x20
-static pthread_mutex_t user_mutex, word_mutex, channel_mutex;
+#define max(a, b) ((a) > (b) ? (a) : (b))
static struct user_t *last_user = NULL;
static int in_monolog = 0, monolog_len = 0;
+static struct tm now_global;
+
static inline void add_word(struct user_t *user, wchar_t *word, int len) {
- pthread_mutex_lock(&user_mutex);
user->words++;
- pthread_mutex_unlock(&user_mutex);
word[len] = '\0';
char mbword[TEXT_BUFFER_SIZE];
wcstombs(mbword, word, TEXT_BUFFER_SIZE);
- pthread_mutex_lock(&word_mutex);
struct word_t *word_s = word_get(mbword);
word_s->count++;
- pthread_mutex_unlock(&word_mutex);
+}
+
+static inline char *parse_getline(char *buffer, int bufsize, FILE *f) {
+ char *r;
+
+ r = fgets(buffer, bufsize, f);
+
+ return r;
}
static void process_file(FILE *f, struct channel_t *channel, struct regexset_t *rs) {
char line[LINE_BUFFER_SIZE];
+ const char *log_date_format, *day_date_format;
+ struct tm now;
+
+ if(rs->log_date_format) {
+ log_date_format = rs->log_date_format;
+ } else if(ircstats_config.log_date_format) {
+ log_date_format = ircstats_config.log_date_format;
+ } else {
+ log_date_format = NULL;
+ }
- while(fgets(line, LINE_BUFFER_SIZE, f)) {
+ if(rs->day_date_format) {
+ day_date_format = rs->day_date_format;
+ } else if(ircstats_config.day_date_format) {
+ day_date_format = ircstats_config.day_date_format;
+ } else {
+ day_date_format = NULL;
+ }
+
+ while(parse_getline(line, LINE_BUFFER_SIZE, f)) {
int rc;
int ovector[30];
@@ -48,7 +73,6 @@ static void process_file(FILE *f, struct channel_t *channel, struct regexset_t *
pcre_copy_named_substring(rs->text, line, ovector, rc, "text", text, TEXT_BUFFER_SIZE);
pcre_copy_named_substring(rs->text, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
pcre_copy_named_substring(rs->text, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
- pthread_mutex_lock(&user_mutex);
struct user_t *user = user_get(nick);
if(user == last_user) {
monolog_len++;
@@ -65,7 +89,6 @@ static void process_file(FILE *f, struct channel_t *channel, struct regexset_t *
in_monolog = 0;
monolog_len = 1;
}
- pthread_mutex_unlock(&user_mutex);
/* Calculate array index for lines. */
int hour, min, time_i;
@@ -77,14 +100,21 @@ static void process_file(FILE *f, struct channel_t *channel, struct regexset_t *
wchar_t wtext[TEXT_BUFFER_SIZE];
mbstowcs(wtext, text, TEXT_BUFFER_SIZE);
- pthread_mutex_lock(&user_mutex);
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
user->characters += wcslen(wtext);
user->lines[time_i]++;
- pthread_mutex_unlock(&user_mutex);
- pthread_mutex_lock(&channel_mutex);
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
channel->hours[time_i]++;
- pthread_mutex_unlock(&channel_mutex);
wchar_t word[TEXT_BUFFER_SIZE];
wchar_t *end = wcschr(wtext, '\0');
@@ -112,47 +142,187 @@ static void process_file(FILE *f, struct channel_t *channel, struct regexset_t *
rc = pcre_exec(rs->join, rs->join_e, line, strlen(line), 0, 0, ovector, 30);
if(rc > 0) {
- char nick[NICK_BUFFER_SIZE];
+ char nick[NICK_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
pcre_copy_named_substring(rs->join, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
- pthread_mutex_lock(&user_mutex);
+ pcre_copy_named_substring(rs->join, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->join, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
+
+ int hour, min;
+ hour = atoi(hour_s);
+ min = atoi(min_s);
+
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
struct user_t *user = user_get(nick);
- pthread_mutex_unlock(&user_mutex);
+
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
+ continue;
+ }
+
+ rc = pcre_exec(rs->part, rs->part_e, line, strlen(line), 0, 0, ovector, 30);
+ if(rc > 0) {
+ char nick[NICK_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
+ pcre_copy_named_substring(rs->part, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->part, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->part, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
+
+ int hour, min;
+ hour = atoi(hour_s);
+ min = atoi(min_s);
+
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
+ struct user_t *user = user_get(nick);
+
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
+ continue;
+ }
+
+ rc = pcre_exec(rs->quit, rs->quit_e, line, strlen(line), 0, 0, ovector, 30);
+ if(rc > 0) {
+ char nick[NICK_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
+ pcre_copy_named_substring(rs->quit, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->quit, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->quit, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
+
+ int hour, min;
+ hour = atoi(hour_s);
+ min = atoi(min_s);
+
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
+ struct user_t *user = user_get(nick);
+
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
+ continue;
+ }
+
+ rc = pcre_exec(rs->nick_changed, rs->nick_changed_e, line, strlen(line), 0, 0, ovector, 30);
+ if(rc > 0) {
+ char newnick[NICK_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
+ pcre_copy_named_substring(rs->nick_changed, line, ovector, rc, "new_nick", newnick, NICK_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->nick_changed, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->nick_changed, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
+
+ int hour, min;
+ hour = atoi(hour_s);
+ min = atoi(min_s);
+
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
+ struct user_t *user = user_get(newnick);
+
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
+ continue;
+ }
+
+ rc = pcre_exec(rs->log_opened, rs->log_opened_e, line, strlen(line), 0, 0, ovector, 30);
+ if(rc > 0) {
+ char date[DATE_BUFFER_SIZE];
+
+ if(!log_date_format) {
+ continue;
+ }
+
+ pcre_copy_named_substring(rs->log_opened, line, ovector, rc, "date", date, DATE_BUFFER_SIZE);
+
+ if(!strptime(date, log_date_format, &now)) {
+ fprintf(stderr, "log fail: %s\n", date);
+ continue;
+ }
+
+ now_global = now;
+
+ continue;
+ }
+
+ /* day_changed is optional */
+ rc = rs->day_changed ? pcre_exec(rs->day_changed, rs->day_changed_e, line, strlen(line), 0, 0, ovector, 30) : 0;
+ if(rc > 0) {
+ char date[DATE_BUFFER_SIZE];
+
+ if(!day_date_format) {
+ continue;
+ }
+
+ pcre_copy_named_substring(rs->day_changed, line, ovector, rc, "date", date, DATE_BUFFER_SIZE);
+
+ if(!strptime(date, day_date_format, &now)) {
+ fprintf(stderr, "day fail: %s\n", date);
+ continue;
+ }
+
+ now_global = now;
+
continue;
}
rc = pcre_exec(rs->kick, rs->kick_e, line, strlen(line), 0, 0, ovector, 30);
if(rc > 0) {
- char nick[NICK_BUFFER_SIZE], victim[NICK_BUFFER_SIZE];
+ char nick[NICK_BUFFER_SIZE], victim[NICK_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
+ pcre_copy_named_substring(rs->kick, line, ovector, rc, "hour", hour_s, TIME_BUFFER_SIZE);
+ pcre_copy_named_substring(rs->kick, line, ovector, rc, "minute", min_s, TIME_BUFFER_SIZE);
pcre_copy_named_substring(rs->kick, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
pcre_copy_named_substring(rs->kick, line, ovector, rc, "victim", victim, NICK_BUFFER_SIZE);
- pthread_mutex_lock(&user_mutex);
+
+ int hour, min;
+ hour = atoi(hour_s);
+ min = atoi(min_s);
+
+ now = now_global;
+
+ now.tm_hour = hour;
+ now.tm_min = min;
+ time_t now_ut = mktime(&now);
+
struct user_t *user = user_get(nick),
*victim_user = user_get(victim);
user->kicks++;
victim_user->kicked++;
- pthread_mutex_unlock(&user_mutex);
+
+ if(user->seen_first == 0 || now_ut < user->seen_first) {
+ user->seen_first = now_ut;
+ }
+ user->seen_last = max(now_ut, user->seen_last);
+
continue;
}
}
}
-struct thread_arg_t {
- FILE *f;
- struct channel_t *channel;
- struct regexset_t *rs;
-};
-
-static void *thread_func(void *arg) {
- struct thread_arg_t *ta = arg;
- process_file(ta->f, ta->channel, ta->rs);
-
- return NULL;
-}
-
-void process(int thread_n) {
- pthread_mutex_init(&user_mutex, NULL);
- pthread_mutex_init(&word_mutex, NULL);
- pthread_mutex_init(&channel_mutex, NULL);
+void process() {
/* Parsing stuff goes here. */
for(int chan_i = 0; chan_i < channel_get_count(); chan_i++) {
user_init();
@@ -173,28 +343,14 @@ void process(int thread_n) {
last_user = NULL;
in_monolog = monolog_len = 0;
- pthread_t *threads;
- threads = malloc(sizeof(pthread_t) * thread_n);
- struct thread_arg_t ta;
- ta.f = f;
- ta.channel = channel;
- ta.rs = rs;
- for(int i = 0; i < thread_n; i++) {
- pthread_create(&threads[i], NULL, thread_func, &ta);
- }
- for(int i = 0; i < thread_n; i++) {
- pthread_join(threads[i], NULL);
- }
- free(threads);
+ process_file(f, channel, rs);
fclose(f);
file = file->next;
}
+
export_xml(channel, users);
user_free();
word_free();
}
- pthread_mutex_destroy(&user_mutex);
- pthread_mutex_destroy(&word_mutex);
- pthread_mutex_destroy(&channel_mutex);
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-22 10:17:21 +0100