#include <stdio.h>
#include <string.h>
#include <wctype.h>
#include <wchar.h>
#include <errno.h>

#include "parsing.h"
#include "channel.h"
#include "user.h"
#include "word.h"
#include "config.h"
#include "pg.h"

#define NICK_BUFFER_SIZE 0x100
#define TEXT_BUFFER_SIZE 0x400
#define LINE_BUFFER_SIZE 0x400
#define TIME_BUFFER_SIZE 0xf
#define DATE_BUFFER_SIZE 0x20

#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) {
	user->words++;
	word[len] = '\0';
	char mbword[TEXT_BUFFER_SIZE];
	wcstombs(mbword, word, TEXT_BUFFER_SIZE);
	struct word_t *word_s = word_get(mbword);
	word_s->count++;
}

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;
	}

	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];

		rc = pcre_exec(rs->text, rs->text_e, line, strlen(line), 0, 0, ovector, 30);
		if(rc > 0) {
			char nick[NICK_BUFFER_SIZE], text[TEXT_BUFFER_SIZE], hour_s[TIME_BUFFER_SIZE], min_s[TIME_BUFFER_SIZE];
			pcre_copy_named_substring(rs->text, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
			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);
			struct user_t *user = user_get(nick);
			if(user == last_user) {
				monolog_len++;
				if(!in_monolog && monolog_len >= ircstats_config.monolog_min) {
					in_monolog = 1;
					user->monologs++;
					/* Count first lines. */
					user->monolog_lines += monolog_len;
				} else if(in_monolog) {
					user->monolog_lines++;
				}
			} else {
				last_user = user;
				in_monolog = 0;
				monolog_len = 1;
			}

			/* Calculate array index for lines. */
			int hour, min, time_i;
			hour = atoi(hour_s);
			min = atoi(min_s);
			time_i = hour*4 + min / 15;

			/* Count words. */
			wchar_t wtext[TEXT_BUFFER_SIZE];
			mbstowcs(wtext, text, TEXT_BUFFER_SIZE);

			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]++;

			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]++;

			wchar_t word[TEXT_BUFFER_SIZE];
			wchar_t *end = wcschr(wtext, '\0');
			*word = '\0';
			int len = 0;
			for(wchar_t *pos = wtext; pos < end; pos++) {
				if(iswblank(*pos)) {
					if(len >= ircstats_config.wordlen_min) {
						add_word(user, word, len);
					}
					len = 0;
					*word = '\0';
				} else if(iswalpha(*pos)) {
					word[len++] = towlower(*pos);
				} else {
					len = 0;
					*word = '\0';
				}
			}
			if(len >= ircstats_config.wordlen_min) {
				add_word(user, word, len);
			}
			continue;
		}

		rc = pcre_exec(rs->join, rs->join_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->join, line, ovector, rc, "nick", nick, NICK_BUFFER_SIZE);
			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);

			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], 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);

			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++;

			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;
		}
	}
}

static void save_users(int channel_id) {
	for(int i = 0; i < USERS_MAX; i++) {
		struct user_t *user = &users[i];
		if(!user->nick)
			continue;

		while(user) {
			if(!user->real_user && strcmp(user->nick, "<NULL>"))
				pg_user_set(channel_id, user);
			user = user->next;
		}
	}
}

static void save_words(int channel_id) {
	for(int i = 0; i < WORDS_MAX; i++) {
		struct word_t *word = &words[i];
		if(!word->name)
			continue;

		while(word) {
			pg_word_set(channel_id, word);
			word = word->next;
		}
	}
}

void process() {
	/* Parsing stuff goes here. */
	for(int chan_i = 0; chan_i < channel_get_count(); chan_i++) {
		user_init();
		word_init();
		struct channel_t *channel = channel_get(chan_i);
		int channel_id = pg_channel_get(channel);
		pg_users_get(channel_id);
		pg_words_get(channel_id);
		printf("Channel %s\n", channel->name);
		struct channel_file_t *file = channel->files;
		while(file) {
			struct regexset_t *rs = file->rs;
			FILE *f = fopen(file->path, "r");
			if(!f) {
				fprintf(stderr, "\tFailed to open %s\n", file->path);
				file = file->next;
				continue;
			} else
				printf("\tParsing %s\n", file->path);

			long pos = 0;
			time_t time = 0;
			pg_channel_file_get(channel_id, file, &pos, &time);
			fseek(f, pos, SEEK_SET);
			localtime_r(&time, &now_global);

			last_user = NULL;
			in_monolog = monolog_len = 0;

			process_file(f, channel, rs);

			pg_channel_file_set(channel_id, file, ftell(f), mktime(&now_global));

			fclose(f);
			file = file->next;
		}

		save_users(channel_id);
		save_words(channel_id);

		user_free();
		word_free();
	}
}