implement calendar parsing

1 files changed, 877 insertions, 0 deletions

diff --git a/src/calendar.c b/src/calendar.c
index 46fd68c..1240547 100644
--- a/src/calendar.c
+++ b/src/calendar.c
@@ -1 +1,878 @@
 #include "calendar.h"
+
+#include "err.h"
+#include "opt.h"
+
+#include <errno.h>
+#include <ctype.h>
+#include <limits.h>
+#include <stdlib.h>
+
+#define ERRP(...) \
+	do { \
+		fprintf(stderr, "err: %zu:%zu: ", s->last_line, s->last_col); \
+		fprintf(stderr, __VA_ARGS__); \
+		fprintf(stderr, "\n"); \
+	} while (0)
+
+struct state_s {
+	bool err_flag;
+	size_t line, col;
+	size_t last_line, last_col;
+	struct entry_interval_s cur_warn;
+	bool cur_urgent;
+	bool cur_local;
+	FILE *f;
+	struct calendar_s *cal;
+};
+
+enum cmd_type_e {
+	CMD_TYPE_WARN,
+	CMD_TYPE_URGENT,
+	CMD_TYPE_LOCAL,
+};
+
+struct cmd_s {
+	enum cmd_type_e type;
+	struct entry_interval_s iv;
+	bool b;
+};
+
+struct pos_s {
+	size_t line, col;
+};
+
+static struct pos_s sub_get_pos(struct state_s *s)
+{
+	struct pos_s p = {
+		.line = s->line,
+		.col = s->col
+	};
+
+	return p;
+}
+
+static void sub_set_pos(struct state_s *s, struct pos_s p)
+{
+	s->last_line = p.line;
+	s->last_col = p.col;
+}
+
+static bool sub_eat_spaces(struct state_s *s)
+{
+	int c;
+	bool seen = false;
+
+	while (true) {
+		c = getc(s->f);
+
+		if (c == '\n') {
+			s->col = 1;
+			s->line++;
+			continue;
+		}
+
+		if (!isspace(c)) {
+			ungetc(c, s->f);
+			break;
+		}
+
+		s->col++;
+
+		seen = true;
+	}
+
+	return seen;
+}
+
+static struct entry_interval_s sub_get_interval(struct state_s *s)
+{
+	struct entry_interval_s iv = { 0 };
+	int r, count;
+	long l;
+	bool ready = false;
+
+	sub_set_pos(s, sub_get_pos(s));
+
+	while (true) {
+		errno = 0;
+		r = fscanf(s->f, "%ld%n", &l, &count);
+
+		/* read err */
+		if (errno) {
+			ERRP("invalid time interval");
+			s->err_flag = true;
+			return iv;
+		}
+
+		/* no int found */
+		if (r == 0) {
+			/* err if digit */
+			r = getc(s->f);
+			if (isdigit(r)) {
+				ERRP("expected time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			ungetc(r, s->f);
+
+			/* err if empty so far */
+			if (ready == false) {
+				ERRP("expected time interval");
+				s->err_flag = true;
+			}
+
+			/* done */
+			return iv;
+		}
+
+		/* range err interval */
+		if (l <= 0) {
+			ERRP("invalid time interval");
+			s->err_flag = true;
+			return iv;
+		}
+
+		if (l > INT_MAX) {
+			ERRP("time interval too large");
+			s->err_flag = true;
+			return iv;
+		}
+
+		s->col += count;
+
+		switch (getc(s->f)) {
+		case 'y':
+			if (iv.year != 0) {
+				ERRP("duplicate year entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.year = l;
+			break;
+
+		case 'm':
+			if (iv.month != 0) {
+				ERRP("duplicate month entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.month = l;
+			break;
+
+		case 'd':
+			if (iv.day != 0) {
+				ERRP("duplicate day entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.day = l;
+			break;
+
+		case 'H':
+			if (iv.hour != 0) {
+				ERRP("duplicate hour entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.hour = l;
+			break;
+
+		case 'M':
+			if (iv.minute != 0) {
+				ERRP("duplicate minute entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.minute = l;
+			break;
+
+		case 'S':
+			if (iv.second != 0) {
+				ERRP("duplicate second entry in time interval");
+				s->err_flag = true;
+				return iv;
+			}
+			iv.second = l;
+			break;
+
+		default:
+			ERRP("invalid time interval");
+			s->err_flag = true;
+			return iv;
+		}
+
+		s->col++;
+		ready = true;
+
+		sub_eat_spaces(s);
+	}
+}
+
+static time_t sub_get_date(struct state_s *s)
+{
+	int r, count;
+	long l;
+	int i;
+	char c;
+	time_t rt = 0;
+	struct tm tmp;
+	struct tm t = {
+		.tm_year = 1,
+		.tm_mday = 1,
+		.tm_isdst = -1,
+	};
+
+	sub_set_pos(s, sub_get_pos(s));
+
+	for (i = 0; i < 6; i++) {
+		c = getc(s->f);
+		if (c == '_') {
+			count = 1;
+			switch (i) {
+			/* year */
+			case 0:
+			case 1:
+			case 2:
+				l = 1;
+				break;
+			default:
+				l = 0;
+				break;
+			}
+			goto skip_num;
+		}
+		ungetc(c, s->f);
+
+		errno = 0;
+		r = fscanf(s->f, "%ld%n", &l, &count);
+
+		/* read err */
+		if (errno) {
+			ERRP("invalid date");
+			s->err_flag = true;
+			return rt;
+		}
+
+		/* no int found */
+		if (r == 0) {
+			ERRP("expected date");
+			s->err_flag = true;
+			return rt;
+		}
+
+		/* range err interval */
+		if (l < 0) {
+			ERRP("invalid date");
+			s->err_flag = true;
+			return rt;
+		}
+
+		if (l > INT_MAX) {
+			ERRP("invalid date");
+			s->err_flag = true;
+			return rt;
+		}
+
+skip_num:
+
+		switch (i) {
+		case 0:
+			t.tm_year = l - 1900;
+			break;
+		case 1:
+			t.tm_mon = l - 1;
+			break;
+		case 2:
+			t.tm_mday = l;
+			break;
+		case 3:
+			t.tm_hour = l;
+			break;
+		case 4:
+			t.tm_min = l;
+			break;
+		case 5:
+			t.tm_sec = l;
+			break;
+		}
+
+		s->col += count;
+
+		if (i < 2) {
+			c = getc(s->f);
+
+			if (c != '-') {
+				ungetc(c, s->f);
+				goto validate;
+			}
+		s->col++;
+
+		} else if (i == 2) {
+			sub_eat_spaces(s);
+			c = getc(s->f);
+			ungetc(c, s->f);
+			if (!isdigit(c) && c != '_')
+				goto validate;
+		} else if (i < 5) {
+			c = getc(s->f);
+
+			if (c != ':') {
+				ungetc(c, s->f);
+				goto validate;
+			}
+		}
+	}
+
+validate:
+	tmp = t;
+
+	/* NOTE: timegm is non-POSIX, but commonplace */
+	if (s->cur_local)
+		rt = mktime(&t);
+	else
+		rt = timegm(&t);
+
+	if (rt == -1) {
+		ERRP("invalid date");
+		s->err_flag = true;
+	}
+
+	if (
+		tmp.tm_year != t.tm_year ||
+		tmp.tm_mon  != t.tm_mon ||
+		tmp.tm_mday != t.tm_mday ||
+		tmp.tm_hour != t.tm_hour ||
+		tmp.tm_min  != t.tm_min ||
+		tmp.tm_sec  != t.tm_sec
+	) {
+		ERRP("invalid date!");
+		s->err_flag = true;
+	}
+
+	return rt;
+}
+
+static inline struct cmd_s sub_get_cmd_warn(struct state_s *s)
+{
+	struct cmd_s cmd = {
+		.type = CMD_TYPE_WARN
+	};
+
+	cmd.iv = sub_get_interval(s);
+
+	return cmd;
+}
+
+static bool sub_get_bool(struct state_s *s)
+{
+	bool r = false;
+
+	sub_set_pos(s, sub_get_pos(s));
+
+	switch (tolower(getc(s->f))) {
+	case '1':
+		r = true;
+		s->col += 1;
+		break;
+
+	case '0':
+		s->col += 1;
+		break;
+
+	case 't':
+		if (
+			tolower(getc(s->f)) == 'r' &&
+			tolower(getc(s->f)) == 'u' &&
+			tolower(getc(s->f)) == 'e'
+		) {
+			r = true;
+			s->col += 4;
+		} else {
+			s->err_flag = true;
+		}
+		break;
+
+	case 'y':
+		if (
+			tolower(getc(s->f)) == 'e' &&
+			tolower(getc(s->f)) == 's'
+		) {
+			r = true;
+			s->col += 3;
+		} else {
+			s->err_flag = true;
+		}
+		break;
+
+	case 'n':
+		if (
+			tolower(getc(s->f)) == 'o'
+		) {
+			s->col += 2;
+		} else {
+			s->err_flag = true;
+		}
+		break;
+
+	case 'f':
+		if (
+			tolower(getc(s->f)) == 'a' &&
+			tolower(getc(s->f)) == 'l' &&
+			tolower(getc(s->f)) == 's' &&
+			tolower(getc(s->f)) == 'e'
+		) {
+			s->col += 5;
+		} else {
+			s->err_flag = true;
+		}
+		break;
+
+	case 'o':
+			switch (tolower(getc(s->f))) {
+			case 'n':
+				r = true;
+				s->col += 2;
+				break;
+			case 'f':
+				if (tolower(getc(s->f)) == 'f') {
+					s->col += 3;
+				} else {
+					s->err_flag = true;
+				}
+				break;
+			default:
+				s->err_flag = true;
+				break;
+			}
+			break;
+
+	default:
+			s->err_flag = true;
+			break;
+	}
+
+	if (s->err_flag)
+		ERRP("invalid boolean");
+
+	return r;
+}
+
+static inline struct cmd_s sub_get_cmd_urgent(struct state_s *s)
+{
+	struct cmd_s cmd = {
+		.type = CMD_TYPE_URGENT
+	};
+
+	cmd.b = sub_get_bool(s);
+
+	return cmd;
+}
+
+static inline struct cmd_s sub_get_cmd_local(struct state_s *s)
+{
+	struct cmd_s cmd = {
+		.type = CMD_TYPE_LOCAL
+	};
+
+	cmd.b = sub_get_bool(s);
+
+	return cmd;
+}
+
+static inline struct cmd_s sub_get_cmd(struct state_s *s)
+{
+	struct cmd_s cmd = { 0 };
+	struct pos_s p = {
+		.line = s->line,
+		.col = s->col - 1,
+	};
+
+	sub_set_pos(s, p);
+
+	switch (getc(s->f)) {
+	case 'w':
+		if (
+			getc(s->f) == 'a' &&
+			getc(s->f) == 'r' &&
+			getc(s->f) == 'n'
+		) {
+			s->col += 4;
+			sub_eat_spaces(s);
+			return sub_get_cmd_warn(s);
+		}
+		break;
+
+	case 'u':
+		if (
+			getc(s->f) == 'r' &&
+			getc(s->f) == 'g' &&
+			getc(s->f) == 'e' &&
+			getc(s->f) == 'n' &&
+			getc(s->f) == 't'
+		) {
+			s->col += 6;
+			sub_eat_spaces(s);
+			return sub_get_cmd_urgent(s);
+		}
+		break;
+
+	case 'l':
+		if (
+			getc(s->f) == 'o' &&
+			getc(s->f) == 'c' &&
+			getc(s->f) == 'a' &&
+			getc(s->f) == 'l'
+		) {
+			s->col += 5;
+			sub_eat_spaces(s);
+			return sub_get_cmd_urgent(s);
+		}
+		break;
+	}
+
+	ERRP("unknown command");
+	s->err_flag = true;
+	return cmd;
+}
+
+static char* sub_get_msg(struct state_s *s)
+{
+	const size_t step_size = 64;
+	size_t count = 0, width = 0;
+	char *msg = NULL;
+	char c;
+
+	while ( (c = getc(s->f)) != '\n' && c != EOF) {
+		s->col++;
+		if (count == width) {
+			width += step_size;
+			msg = realloc(msg, width * sizeof(*msg));
+
+			if (msg == NULL) {
+				ERRM("failed to allocate memory");
+				s->err_flag = true;
+				return NULL;
+			}
+		}
+
+		if (c == opt_line_delim() || c == opt_col_delim()) {
+			sub_set_pos(s, sub_get_pos(s));
+			ERRP("message may not contain line or column delimiter characters");
+			s->err_flag = true;
+			free(msg);
+			return NULL;
+		}
+
+		msg[count] = c;
+		count++;
+	}
+
+	s->col = 0;
+	s->line++;
+
+	return msg;
+}
+
+static inline struct entry_s sub_get_entry_on(struct state_s *s)
+{
+	struct entry_s e = {
+		.type = ENTRY_TYPE_ON
+	};
+	time_t rt;
+	int c;
+
+	rt = sub_get_date(s);
+	if (s->err_flag)
+		return e;
+
+	e.start = rt;
+
+	sub_eat_spaces(s);
+
+	c = getc(s->f);
+
+	if (c != ',') {
+		if (c == 'w'
+			&& getc(s->f) == 'a'
+			&& getc(s->f) == 'r'
+			&& getc(s->f) == 'n'
+		) {
+			s->col += 4;
+			e.warn = sub_get_interval(s);
+			if (s->err_flag)
+				return e;
+		} else {
+			ungetc(c, s->f);
+			return e;
+		}
+	}
+
+	s->col++;
+
+	sub_eat_spaces(s);
+	e.msg = sub_get_msg(s);
+	return e;
+}
+
+static inline struct entry_s sub_get_entry_every(struct state_s *s)
+{
+	struct entry_s e = {
+		.type = ENTRY_TYPE_EVERY,
+		.urgent = s->cur_urgent,
+		.local = s->cur_local,
+		.warn = s->cur_warn,
+	};
+	bool from_seen = false, to_seen = false, warn_seen = false;
+	char c;
+	struct pos_s p = {
+		.line = s->last_line,
+		.col = s->last_col,
+	};
+
+	e.every = sub_get_interval(s);
+
+	if (s->err_flag)
+		return e;
+
+	while (true) {
+		sub_eat_spaces(s);
+		sub_set_pos(s, p);
+
+		switch (c = getc(s->f)) {
+		case 'f':
+			if (
+				getc(s->f) == 'r' &&
+				getc(s->f) == 'o' &&
+				getc(s->f) == 'm'
+			) {
+				if (from_seen) {
+					ERRP("invalid calendar entry");
+					s->err_flag = true;
+					return e;
+				}
+				from_seen = true;
+
+				s->col += 4;
+				sub_eat_spaces(s);
+				e.start = sub_get_date(s);
+
+				if (s->err_flag)
+					return e;
+
+				break;
+			}
+
+		case 't':
+			if (
+				getc(s->f) == 'o'
+			) {
+				if (to_seen) {
+					ERRP("invalid calendar entry");
+					s->err_flag = true;
+					return e;
+				}
+				to_seen = true;
+
+				s->col += 2;
+				sub_eat_spaces(s);
+				e.end = sub_get_date(s);
+				e.has_end = true;
+
+				if (s->err_flag)
+					return e;
+
+				break;
+			}
+
+		case 'w':
+			if (
+				getc(s->f) == 'a' &&
+				getc(s->f) == 'r' &&
+				getc(s->f) == 'n'
+			) {
+				if (warn_seen) {
+					ERRP("invalid calendar entry");
+					s->err_flag = true;
+					return e;
+				}
+				warn_seen = true;
+
+				s->col += 4;
+				sub_eat_spaces(s);
+				e.warn = sub_get_interval(s);
+
+				if (s->err_flag)
+					return e;
+
+				break;
+			}
+
+		default:
+			goto done;
+		}
+	}
+
+done:
+	if (!from_seen) {
+		ERRP("invalid calendar entry");
+		s->err_flag = true;
+	}
+
+	ungetc(c, s->f);
+
+	sub_eat_spaces(s);
+
+	c = getc(s->f);
+
+	if (c != ',') {
+		ungetc(c, s->f);
+		return e;
+	}
+
+	s->col++;
+
+	sub_eat_spaces(s);
+	e.msg = sub_get_msg(s);
+	return e;
+}
+
+static inline struct entry_s sub_get_entry(struct state_s *s)
+{
+	struct entry_s e = { 0 };
+
+	switch (getc(s->f)) {
+	case 'o':
+		if (
+			getc(s->f) == 'n'
+		) {
+			s->col += 2;
+			sub_eat_spaces(s);
+			return sub_get_entry_on(s);
+		}
+		break;
+
+	case 'e':
+		if (
+			getc(s->f) == 'v' &&
+			getc(s->f) == 'e' &&
+			getc(s->f) == 'r' &&
+			getc(s->f) == 'y'
+		) {
+			s->col += 5;
+			sub_eat_spaces(s);
+			return sub_get_entry_every(s);
+		}
+		break;
+	}
+
+	ERRP("invalid calendar entry");
+	s->err_flag = true;
+
+	return e;
+}
+
+static inline void sub_cal_push(struct state_s *s, struct calendar_s *cal,
+	struct entry_s *e)
+{
+	const size_t step_size = 64;
+
+	if (cal->count == cal->width) {
+		cal->width += step_size;
+		cal->entries = realloc(cal->entries, cal->width * sizeof(*e));
+
+		if (cal->entries == NULL) {
+			ERRM("failed to allocate memory");
+			s->err_flag = true;
+			return;
+		}
+	}
+
+	cal->entries[cal->count] = *e;
+	cal->count++;
+}
+
+static void sub_cal_free(struct calendar_s *cal)
+{
+	size_t i;
+
+	if (cal->entries == NULL)
+		return;
+
+	for (i = 0; i < cal->count; i++) {
+		if (cal->entries[i].msg != NULL)
+			free(cal->entries[i].msg);
+	}
+
+	free(cal->entries);
+	cal->entries = NULL;
+}
+
+struct calendar_s calendar_parse(FILE *f)
+{
+	struct calendar_s cal = {
+		.count = 0,
+		.err_flag = false,
+	};
+	struct state_s s = {
+		.err_flag = false,
+		.line = 1,
+		.col = 1,
+		.cur_warn = {
+			.day = 1,
+		},
+		.cur_urgent = false,
+		.cur_local = true,
+		.f = f,
+		.cal = &cal,
+	};
+	int c;
+	struct cmd_s cmd;
+	struct entry_s e;
+
+	while (true) {
+		sub_eat_spaces(&s);
+		sub_set_pos(&s, sub_get_pos(&s));
+
+		c = getc(f);
+		if (c == EOF)
+			break;
+
+		if (c == '.') {
+			s.col++;
+			cmd = sub_get_cmd(&s);
+
+			if (s.err_flag) {
+				sub_cal_free(&cal);
+				cal.err_flag = true;
+				return cal;
+			}
+
+			switch (cmd.type) {
+			case CMD_TYPE_WARN:
+				s.cur_warn = cmd.iv;
+				break;
+			case CMD_TYPE_URGENT:
+				s.cur_urgent = cmd.b;
+				break;
+			case CMD_TYPE_LOCAL:
+				s.cur_local = cmd.b;
+				break;
+			}
+
+			continue;
+		}
+
+		ungetc(c, f);
+
+		e = sub_get_entry(&s);
+		if (s.err_flag) {
+			sub_cal_free(&cal);
+			cal.err_flag = true;
+			return cal;
+		}
+		sub_cal_push(&s, &cal, &e);
+	}
+
+	return cal;
+}