/** Is `y` a leap-year? */
static int leap(int y) {
	assert(y >= 1582);
	if(!(y % 400)) return 1;
	if(!(y % 100)) return 0;
	if(!(y % 4)) return 1;
	return 0;
}
/** Convert or narrower type or return zero. */
static union date32 date_to_32(const int y, const int m, const int d) {
	union date32 d32 = { 0 };
	/* Leap year calculations only work at y>=1 and Gregorian Calendar and max
	 23 bits. */
	if(y < 1582 || y > 8388607 || m < 1 || m > 12 || d < 1 || d > 31) goto no;
	switch(m) {
	case 1: case 3: case 5: case 7: case 8: case 10: case 12: break;
	case 4: case 6: case 9: case 11: if(d > 30) goto no; break;
	case 2: if(d > 28 + leap(y)) goto no; break;
	default: assert(0); break;
	}
	d32.year = (unsigned)y, d32.month = (unsigned)m, d32.day = (unsigned)d;
no:
	return d32;
}
/** Tomohiko Sakamoto comp.lang.c 1993-04-10. */
static unsigned weekday(union date32 d) {
	d.year -= d.month < 3;
	return (d.year + d.year / 4 - d.year / 100 + d.year / 400
		+ "-bed=pen+mad."[d.month] + d.day) % 7;
}
/*
### plot with steps
reset session

$Data <<EOD
1,1,0
1,2,0
1,3,0
1,4,2
1,5,1
1,6,3
1,7,3
1,8,1
1,9,3
1,10,8
1,11,1
1,12,0
1,13,3
EOD

set title "Cumulative count" font ",16"
set xlabel "episode"
set ylabel "cumulative count"
set xtics 1
set key bottom right
set grid
unset border

set datafile separator comma
plot $Data u 2:($3) smooth cumulative with steps lw 2 lc "red" ti "cumulative count"
### end of code
*/
static int bible_graph(/*const*/ struct page_tree *const journal) {
	enum { CHILL, BOOK, CHAPTER, WORD, NEXT } state = CHILL;
	struct page_tree_entry entry = { 0, 0 };
	struct lex *lex = 0;
	size_t count = 0;
	for(struct page_tree_iterator p_it = page_tree_begin(journal);
		(entry = page_tree_next(&p_it)).key; ) {
		struct page *const page = entry.value;
		for(struct lex_array_iterator l_it = lex_array_iterator(&page->meaning);
			(lex = lex_array_next(&l_it)); ) {
			switch(lex->symbol) {
			case KJV_BOOK:
				if(state != CHILL && state != WORD) goto catch;
				if(state == WORD) printf("\n");
				fprintf(stderr, "%d-%.2d-%.2d: \"%.*s\", ",
					entry.key->year, entry.key->month, entry.key->day,
					(int)(lex->s1 - lex->s0), lex->s0);
				state = BOOK;
				break;
			case KJV_CHAPTER_VERSE:
				if(state != BOOK) goto catch;
				printf("\"%.*s\", \"", (int)(lex->s1 - lex->s0), lex->s0);
				state = CHAPTER;
				break;
			case KJV_NEXT:
				if(state != WORD) goto catch;
				printf("\"\n");
				break;
			case KJV_TEXT:
				if(state != WORD && state != CHAPTER && state != NEXT)
					goto catch;
				printf("%s%.*s<%d>", state == WORD ? "*" : "",
					(int)(lex->s1 - lex->s0 < 0 ? 10 : lex->s1 - lex->s0),
					lex->s0, (int)(lex->s1 - lex->s0));
				count++;
				state = WORD;
				break;
			default:
				if(state != CHILL && state != WORD) goto catch;
				if(state == WORD) printf("\"\n"), state = CHILL;
				break;
			}
		}
		if(state != CHILL && state != WORD) goto catch;
		if(state == WORD) printf("\n"), state = CHILL;
	}
	printf("Count: %lu.\n", (unsigned long)count);
	return 1;
catch:
	fprintf(stderr, "Bible error.\n");
	if(entry.key) {
		fprintf(stderr, "On date: %d-%.2d-%.2d.\n",
			entry.key->year, entry.key->month, entry.key->day);
		if(lex) fprintf(stderr, "At line %lu.\n", (unsigned long)lex->line);
	}
	errno = EILSEQ;
	return 0;
}

#define C_BLACK "\033[0;30m"
#define C_RED "\033[0;31m"
#define C_GREEN "\033[0;32m"
#define C_YELLOW "\033[0;33m"
#define C_BLUE "\033[0;34m"
#define C_PURPLE "\033[0;35m"
#define C_CYAN "\033[0;36m"
#define C_WHITE "\033[0;37m"

#define CB_BLACK "\033[1;30m"
#define CB_RED "\033[1;31m"
#define CB_GREEN "\033[1;32m"
#define CB_YELLOW "\033[1;33m"
#define CB_BLUE "\033[1;34m"
#define CB_PURPLE "\033[1;35m"
#define CB_CYAN "\033[1;36m"
#define CB_WHITE "\033[1;37m"

#define C_RESET "\033[0m"

int main(int argc, char **argv) {
	int success = EXIT_SUCCESS;
	char *intent = 0;

	/* For reading in files, overwritten. */
	DIR *dir = 0;
	struct dirent *de;
	struct int_array years = int_array(), months = int_array(),
		days = int_array();
	int *y, *y_end;

	struct page_tree journal = page_tree();

	errno = 0;
	if(argc != 2) { intent = "needs journal location, which should"
		" contain <year>/<month>/<day>.txt"; goto catch; }

	/* Get the years list as directories matching a year. */
	if(chdir(argv[1]) == -1 || !(dir = opendir("."))) goto catch;
	while((de = readdir(dir))) {
		struct stat st;
		int year, *p;
		if(!lex_looks_like_year(de->d_name, &year)) continue;
		if(stat(de->d_name, &st)) goto catch;
		if(!S_ISDIR(st.st_mode)) continue;
		if(!(p = int_array_new(&years))) goto catch;
		*p = year;
	}
	closedir(dir), dir = 0;
	/* Sort the years for sensible ordering of parsing. */
	qsort(years.data, years.size, sizeof *years.data, &void_int_cmp);
	fprintf(stderr, "Years in <<%s>>: %s.\n",
		argv[1], int_array_to_string(&years));

	/* Go though each year. */
	for(y = years.data, y_end = y + years.size; y < y_end; y++) {
		char fn[64];
		int *m, *m_end;
		sprintf(fn, "%d", *y);

		/* Get the months as directories. */
		if(chdir(fn) == -1 || !(dir = opendir("."))) goto catch;
		while((de = readdir(dir))) {
			struct stat st;
			int month, *p;
			if(!(month = lex_looks_like_month(de->d_name))) continue;
			if(stat(de->d_name, &st)) goto catch;
			if(!S_ISDIR(st.st_mode)) continue;
			if(!(p = int_array_new(&months))) goto catch;
			*p = month;
		}
		closedir(dir), dir = 0;
		qsort(months.data, months.size, sizeof *months.data, &void_int_cmp);
		fprintf(stderr, "Months in <<%s>>: %s.)\n",
			fn, int_array_to_string(&months));

		/* Go though each month. */
		for(m = months.data, m_end = m + months.size; m < m_end; m++) {
			int *d, *d_end;
			sprintf(fn, "%.2d", *m);

			/* Get the days as files. */
			if(chdir(fn) == -1 || !(dir = opendir("."))) goto catch;
			while((de = readdir(dir))) {
				struct stat st;
				int day, *p;
				/* fixme: Have yyyy-mm-dd to figure out how many days. */
				if(!(day = lex_looks_like_day(de->d_name))) continue;
				if(stat(de->d_name, &st)) goto catch;
				if(S_ISDIR(st.st_mode)) continue;
				if(!(p = int_array_new(&days))) goto catch;
				*p = day;
			}
			closedir(dir), dir = 0;
			qsort(days.data, days.size, sizeof *days.data, &void_int_cmp);
			fprintf(stderr, "Days in <<%s>>: %s.\n",
				fn, int_array_to_string(&days));

			for(d = days.data, d_end = d + days.size; d < d_end; d++) {
				struct lex *lex = 0;
				struct page *page = 0;
				union date32 d32;
				if(!(d32 = date_to_32(*y, *m, *d)).year) { errno = EILSEQ;
					intent = "date parse error"; goto syntax; }
				sprintf(fn, "%.2d.txt", *d);
				if(page_tree_bulk_add(&journal, d32, &page) != TREE_UNIQUE) {
					if(!errno) intent = "not unique", errno = EDOM;
					goto syntax;
				}
				page->entry = char_array();
				page->meaning = lex_array();
				if(!append_file(&page->entry, fn)) goto syntax;
				int first = 1;
				for(lex_reset(page->entry.data); ; ) {
					if(!(lex = lex_array_new(&page->meaning))) goto syntax;
					if(!lex_next(lex)) {
						if(lex->symbol != END) { errno = EILSEQ; goto syntax; }
						break; /* Terminated successfully. */
					}
					switch(lex->symbol) {
					case TEXT: printf("%s%.*s",
						first ? "" : " ", (int)(lex->s1 - lex->s0), lex->s0);
						first = 0; break;
					case PARAGRAPH: printf("\n" C_RESET); break;
					case KJV_BOOK: printf(C_YELLOW "%.*s",
						(int)(lex->s1 - lex->s0), lex->s0); break;
					case KJV_CHAPTER_VERSE: printf(" ch. %.*s",
						(int)(lex->s1 - lex->s0), lex->s0); break;
					case KJV_TEXT: printf("%.*s",
						(int)(lex->s1 - lex->s0), lex->s0); break;
					case KJV_NEXT: printf("(next)\n"); break;
					default:
						fprintf(stderr, "%lu: %s",
							(unsigned long)lex->line, lex_symbols[lex->symbol]);
						if(lex->s0 && lex->s1) {
							if(lex->s0 + INT_MAX < lex->s1)
								intent = "line too long", errno = EILSEQ;
							else
								fprintf(stderr, " <<%.*s>>",
								(int)(lex->s1 - lex->s0), lex->s0);
						}
						fprintf(stderr, ".\n");
						break;
					}
				}
				continue;
syntax:
				fprintf(stderr, "On date: %d-%.2d-%.2d.\n", *y, *m, *d);
				if(!page) goto catch;
				if(!lex) { fprintf(stderr, "While parsing <<<\n%s>>>.\n",
					page->entry.data); goto catch; }
				for(struct lex_array_iterator it
					= lex_array_iterator(&page->meaning);
					lex = lex_array_next(&it); ) {
					fprintf(stderr, "%lu: %s",
						(unsigned long)lex->line, lex_symbols[lex->symbol]);
					if(lex->s0 && lex->s1) {
						if(lex->s0 + INT_MAX < lex->s1)
							intent = "line too long", errno = EILSEQ;
						else
							fprintf(stderr, " <<%.*s>>",
							(int)(lex->s1 - lex->s0), lex->s0);
					}
					fprintf(stderr, ".\n");
				}
				goto catch;
			}

			int_array_clear(&days);
			if(chdir("..") == -1) goto catch;
		}

		int_array_clear(&months);
		if(chdir("..") == -1) goto catch;
		/* fixme: Expand, contact is the next thing that it doesn't get. */
		if(*y == 1993/*1996*/) break;
	}
	page_tree_bulk_finish(&journal);
	int_array_(&years), int_array_(&months), int_array_(&days);
	fprintf(stderr, "Journal has entries: %s\n", page_tree_to_string(&journal));

	/* Do something interesting? */
	if(!bible_graph(&journal)) goto catch;

	goto finally;
catch:
	success = EXIT_FAILURE;
	perror("interpret");
	if(intent) fprintf(stderr, "Further explanation: %s.\n", intent);
finally:
	if(dir && closedir(dir)) success = EXIT_FAILURE, perror("dir");
	int_array_(&years), int_array_(&months), int_array_(&days);
	struct page_tree_entry entry;
	for(struct page_tree_iterator it = page_tree_begin(&journal);
		(entry = page_tree_next(&it)).key; ) {
		struct page *const page = entry.value;
		char z[12];
		date32_to_string(*entry.key, &z);
		lex_array_(&page->meaning);
		char_array_(&page->entry);
	}
	return success;
}

/* Popular KJV 738137 words. */

struct scan scan(const char *const buffer) {
	struct scan scan;
	scan.marker = scan.from = scan.cursor = scan.label = scan.buffer = buffer;
	scan.condition = yycline;
	scan.line = 1;
	return scan;
}

int scan_next(struct scan *const s, struct lex *const x) {
	/*!re2c /**/
	re2c:flags:tags = 1;
	re2c:define:YYCTYPE = char;
	re2c:yyfill:enable = 0;
	re2c:define:YYCURSOR = s->cursor;
	re2c:define:YYMARKER = s->marker;
	re2c:define:YYCONDTYPE = 'condition';
	re2c:define:YYGETCONDITION = 's->condition';
	re2c:define:YYGETCONDITION:naked = 1;
	re2c:define:YYSETCONDITION = 's->condition = @@;';
	re2c:define:YYSETCONDITION:naked = 1;
	sentinel = "\x00";
	newline = "\n";
	unix_control = [\x01-\x08\x0a-\x1f\x7f];
	ws = [ \t];
	glyph = [^] \ (sentinel | unix_control | newline | ws);
	keyword = ([a-zA-Z] | [0-9][0-9_\-]*[a-zA-Z]) [a-zA-Z0-9_\-]*;
	decimal = "-"? ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0];
	natural = [1-9][0-9]*;
	date = "-"? natural "-" [0-1][0-9] "-" [0-1][0-9];
	*/
	const char *s0, *s1;
	/*!stags:re2c format = 'const char *@@;\n'; */
	assert(s && x);
	if(!s->buffer) return 0;
	x->line = s->line;
	x->s0 = x->s1 = 0;
scan:
	/*!re2c /**/
	<*> unix_control { return x->symbol = ILLEGAL, 0; }
	<*> * { return x->symbol = SYNTAX, 0; }
	<*> sentinel /* New line always delimits. */
		{ return x->symbol = s->condition == yycline ? END : ILLEGAL, 0; }
	<expect_line> newline => line { x->line = ++s->line; goto scan; }
	/* Symbols that go at the beginning of a line. */
	<line> newline { x->line = ++s->line; goto scan; }
	<line> "[" :=> edict
	<line> "--" :=> source
	<line> "->" :=> location
	<line> "!" => text { return x->symbol = COMPLETE, 1; }
	<line> "^" => text { return x->symbol = CANCELLED, 1; }
	<line> "#" => text { return x->symbol = HEADING, 1; }
	<line> * :=> text

	<text> newline => line { x->line = ++s->line; goto scan; }
	<text, bible> ws+ { goto scan; }
	<text> @s0 glyph+ @s1
		{ x->s0 = s0, x->s1 = s1; return x->symbol = TEXT, 1; }


	<source> @s0 keyword @s1 => expect_line
		 { x->s0 = s0, x->s1 = s1; return x->symbol = SOURCE_RECALL, 1; }

	<location> "" / "(" :=> map
	<location> "[" ws* @s0 keyword @s1 ws* "]"
		{ x->s0 = s0, x->s1 = s1; return x->symbol = LOCATION_SAVE, 1; }
	<location> @s0 keyword @s1 => expect_line
		{ x->s0 = s0, x->s1 = s1; return x->symbol = LOCATION_RECALL, 1; }

	<map> "(" @s0 decimal "," @s1 decimal ")" => expect_caption
		{ x->s0 = s0, x->s1 = s1; return x->symbol = LOCATION, 1; }

	<edict> "source" :=> source
	<edict> "ed" :=> ed
	<edict> "contact" :=> contact
	<edict> "glider" :=> glider
	<edict> "flight" :=> flight
	<edict> "bible" :=> bible
	<edict> "book" :=> book
	<edict> "movie" :=> movie
	<edict> "tv" :=> tv
	<edict> "medication" :=> medication
	<edict> "idea" :=> idea
	<edict> "vaccine" :=> vaccine
	<edict> "in" :=> in
	<edict> "" / natural :=> significant
	<edict> [0-1][0-9] "-" [0-3][0-9]
		", " [0-2][0-9] ":" [0-5][0-9] "] "
		:=> text /* This is likely WhatsApp conversations. Ignore. */
	/* missed, show, 'First, Second', 'Sounds', 'CSS', ..., 'Swanky', 'Shields' */

	/* How did it get into my journal? */
	<edict> "source"
		{ if(s->is_ws_expected || s->edict.size)
			return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 1, s->is_source = 1;
		s->edict.size = 2;
		s->edict.expect[1] = EXPECT_KEYWORD;
		s->edict.expect[0] = EXPECT_END_TEXT;
		return x->symbol = SOURCE, 1; }
	<edict> "default"
		{ if(s->is_ws_expected || !s->is_source)
			return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 1, s->is_source = 0;
		return x->symbol = DEFAULT, 1; }

	/* Editorializing; looking back. */
	<edict> "ed"
		{ if(s->is_ws_expected || s->edict.size)
			return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 1; /* no idea, just copy; probably should do sth */
		s->edict.size = 1;
		s->edict.expect[0] = EXPECT_END_TEXT; /* Pithy comment. */
		return x->symbol = EDITORIALIZING, 1; }

	/* Score. */
	<edict> "significant"
		{ if(s->is_ws_expected || s->edict.size)
			return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 1;
		s->edict.size = 3;
		s->edict.expect[2] = EXPECT_NATURAL; /* Ordinal. */
		s->edict.expect[1] = EXPECT_RESTRICT_TEXT; /* Name. */
		s->edict.expect[0] = EXPECT_DATE; /* Birthday. */
		return x->symbol = SIGNIFICANT, 1; }
	<edict> @s0 natural @s1
		{ if(s->is_ws_expected || s->edict.size)
			return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 1;
		x->s0 = s0, x->s1 = s1;
		return x->symbol = SIGNIFICANT_RECALL, 1; }

	/* General [edict: whatever]. */
	<edict> ws+ { s->is_ws_expected = 0; goto scan; }
	<edict> ":"
		{ if(!s->edict.size) return x->symbol = SYNTAX, 0;
		s->is_ws_expected = 0, s->is_source = 0;
		expect_pop(); goto scan; }
	<edict_keyword> ws* @s0 keyword @s1 ws* ";"?
		{ x->s0 = s0, x->s1 = s1; expect_pop();
		return x->symbol = ARG_KEYWORD, 1; }
	<edict_date> ws* @s0 date @s1 ws* ";"?
		{ x->s0 = s0, x->s1 = s1; expect_pop();
		return x->symbol = ARG_DATE, 1; }
	<edict_natural> ws* @s0 natural @s1 ws* ";"?
		{ x->s0 = s0, x->s1 = s1; expect_pop();
		return x->symbol = ARG_NATURAL, 1; }
	<edict_restrict_text>
		ws* @s0 (glyph \ [;[\]]) ((glyph \ [;[\]]) | ws)* @s1 ws* ";"?
		{ x->s0 = s0, x->s1 = s1; expect_pop();
		return x->symbol = ARG_RESTRICT_TEXT, 1; }
	<edict_end_text>
		ws* @s0 (glyph \ [[\]]) ((glyph \ [[\]]) | ws)* @s1 ws*
		{ x->s0 = s0, x->s1 = s1; expect_pop();
		return x->symbol = ARG_END_TEXT, 1; }
	<edict, edict_end> "]" => expect_line
		{ if(s->edict.size) return 0; goto scan; }
	*/
}

#define FOR_SYMBOL(X) \
	\
	/* Results. */ \
	X(END, 0), \
	X(SYNTAX, 0), \
	X(ILLEGAL, 0), \
	X(NOT_FOUND, 0), \
	\
	X(ORDERED_LIST_ITEM, &no_vt), \
	X(LIST_ITEM, 0), \
	X(COMPLETE, 0), \
	X(CANCELLED, 0), \
	X(HEADING, 0), \
	/* Text. */ \
	X(PARAGRAPH, 0), \
	X(TEXT, &word_vt), \
	/*This is lazy.*/X(CAPTION, &word_vt), \
	\
	/* Edicts. */ \
	X(SOURCE, &word_vt), \
	X(DEFAULT, 0), \
	X(SOURCE_RECALL, &word_vt), \
	X(LOCATION, 0), \
	X(LOCATION_SAVE, 0), \
	X(LOCATION_RECALL, 0), \
	X(SIGNIFICANT, 0), \
	X(SIGNIFICANT_RECALL, 0), \
	X(EDITORIALIZING, 0), \
	\
	/* Arguments. */ \
	X(ARG_KEYWORD, 0), \
	X(ARG_DATE, &date_vt), \
	X(ARG_NATURAL, &no_vt), \
	X(ARG_RESTRICT_TEXT, &word_vt), \
	X(ARG_END_TEXT, &word_vt), \
	\
	/* Bible */ \
	X(KJV_BOOK, &kjv_book_vt), \
	X(KJV_CHAPTER_VERSE, &word_vt), \
	X(KJV_TEXT, &word_vt), \
	X(KJV_NEXT, 0)

int scan(union date32 date, const char *const buffer);

struct scan {
	const char *marker, *from, *cursor, *limit, *label, *buffer;
	int condition;
	size_t line;
	int is_ws_expected;
};

struct lex {
	size_t line;
#define ARG1(n, m) n
	enum lex_symbol { FOR_SYMBOL(ARG1) } symbol;
#undef ARG1
	const char *s0, *s1;
};
#define STR1(n, m) #n
static const char *const lex_symbols[] = { FOR_SYMBOL(STR1) };
#undef X

struct scan scan(const char *);
int scan_next(struct scan *const s, struct lex *const x);