#define _POSIX_C_SOURCE 200809L
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <stddef.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <sys/mman.h>

#define ARENA_IMPL
#include "str.h"
#include "arena.h"
#define ARGS_IMPL
#include "args.h"

#define ARENA(n, sz) Arena n; { static char arena_backarr[sz];\
	n.beg = arena_backarr; n.end = arena_backarr + sizeof(arena_backarr);\
	__asm("":"+r"(n.beg)); __asm("":"+r"(n.end)); }

int read_all(FILE *f, Str *buf, Arena *a) {
	if (!f) return -1;
	buf->s = a->beg;
	buf->n = fread(a->beg, 1, (uintptr_t)(a->end - a->beg) - 1, f);
	a->beg += buf->n;
	return ferror(f) ? -1 : 0;
}

int next_line(Str *src, Str *line) {
	if (src->n < 1) return 0;
	line->s = src->s;
	char *newln = memchr(src->s, '\n', src->n);
	line->n = newln ? newln - src->s : src->n;
	src->s += line->n + 1;
	src->n -= line->n + 1;
	return 1;
}

void str_putf(Str s, FILE *f) {
	fwrite(s.s, 1, s.n, f);
}

void str_put(Str s) {
	str_putf(s, stdout);
}

char to_xdigit(int x) {
	if (x > 9) {
		return 'A' + (x - 10);
	} else {
		return '0' + x;
	}
}

void str_cat_uri(Str *s, Str uri, Arena *a) {
	str_catc(s, '\'', a);
	for (isize i = 0; i < uri.n; i++) {
		char c = uri.s[i];
		if (c == '\'' || c == '%') {
			str_catc(s, '%', a);
			str_catc(s, to_xdigit((c & 0xff) >> 4), a);
			str_catc(s, to_xdigit(c & 0xf), a);
		} else {
			str_catc(s, c, a);
		}
	}
	str_catc(s, '\'', a);
}

void str_cat_html(Str *s, Str uri, Arena *a) {
	for (isize i = 0; i < uri.n; i++) {
		char c = uri.s[i];
		switch (c) {
		case '&': str_cat(s, S("&amp;"), a); break;
		case '<': str_cat(s, S("&lt;"), a); break;
		case '>': str_cat(s, S("&gt;"), a); break;
		default: str_catc(s, c, a); break;
		}
	}
}

int is_ol_item(Str s) {
	Str h = str_cut(s, '.').head;
	if (h.n < 1) return 0;
	for (isize i = 0; i < h.n; i++) {
		if (!(h.s[i] >= '0' && h.s[i] <= '9')) return 0;
	}
	return 1;
}

typedef enum {
	LINE_BLANK, LINE_PARA,
	LINE_LINK, LINE_FIGURE,
	LINE_UL, LINE_OL,
	LINE_HDR1, LINE_HDR2, LINE_HDR3, LINE_CODE,
	LINE_BQUOT,
} LineMode;

LineMode lm_chg(LineMode from, LineMode to, Str *out, Arena *a) {
#undef S
#define S(s) {s,sizeof(s)-1}
	static Str op[] = {
		[LINE_BLANK] = S(""),
		[LINE_PARA] = S("<p>"),
		[LINE_LINK] = S("<ul>\n<li>"),
		[LINE_FIGURE] = S("<figure>"),
		[LINE_UL] = S("<ul>\n<li>"),
		[LINE_OL] = S("<ol>\n<li>"),
		[LINE_HDR1] = S("<h1>"),
		[LINE_HDR2] = S("<h2>"),
		[LINE_HDR3] = S("<h3>"),
		[LINE_CODE] = S("<pre><code>"),
		[LINE_BQUOT] = S("<blockquote>"),
	};
	static Str cl[] = {
		[LINE_BLANK] = S(""),
		[LINE_PARA] = S("</p>"),
		[LINE_LINK] = S("</li>\n</ul>"),
		[LINE_FIGURE] = S("</figure>"),
		[LINE_UL] = S("</li>\n</ul>"),
		[LINE_OL] = S("</li>\n</ol>"),
		[LINE_HDR1] = S("</h1>"),
		[LINE_HDR2] = S("</h2>"),
		[LINE_HDR3] = S("</h3>"),
		[LINE_CODE] = S("</code></pre>"),
		[LINE_BQUOT] = S("</blockquote>"),
	};
	static Str cont[] = {
		[LINE_BLANK] = S(""),
		[LINE_PARA] = S("<br>\n"),
		[LINE_FIGURE] = S("</figure>\n<figure>"),
		[LINE_LINK] = S("</li>\n<li>"),
		[LINE_UL] = S("</li>\n<li>"),
		[LINE_OL] = S("</li>\n<li>"),
		[LINE_HDR1] = S("</h1>\n<h1>"),
		[LINE_HDR2] = S("</h2>\n<h2>"),
		[LINE_HDR3] = S("</h3>\n<h3>"),
		[LINE_CODE] = S("\n"),
		[LINE_BQUOT] = S("<br>\n"),
	};
#undef S
#define S(s) (Str){s,sizeof(s)-1}
	if (from == to) {
		str_cat(out, cont[from], a);
	} else {
		str_cat(out, cl[from], a);
		str_catc(out, '\n', a);
		str_cat(out, op[to], a);
	}
	return to;
}

typedef struct Doc Doc;
struct Doc {
	Str html;
	Str title;
	Doc *prev, *next;
};

int has_image_ext(Str url) {
	return str_ends(url, S(".png"))
		|| str_ends(url, S(".jpg"))
		|| str_ends(url, S(".jpeg"))
		|| str_ends(url, S(".webp"));
}

Str str_replace_end(Str s, Str a, Str b, Arena *m) {
	if (!str_ends(s, a)) return s;
	char *p = new_arr(m, char, s.n + b.n - a.n);
	memcpy(p, s.s, s.n - a.n);
	memcpy(p + s.n - a.n, b.s, b.n);
	return (Str) { p, s.n + b.n - a.n };
}

int wdoc(FILE *f, Doc **dp, Arena *a, Arena *scratch) {
	Str buf, line, out = {0}, title = {0};
	if (read_all(f, &buf, scratch)) return -1;
	LineMode lm = LINE_BLANK;
	while (next_line(&buf, &line)) {
		if (str_starts(line, S("```"))) {
			lm = lm_chg(lm, lm == LINE_CODE ? LINE_BLANK : LINE_CODE, &out, a);
			continue;
		} else if (lm == LINE_CODE) {
			lm = lm_chg(lm, LINE_CODE, &out, a);
			str_cat(&out, line, a);
			continue;
		} else if (line.n == 0) {
			lm = lm_chg(lm, LINE_BLANK, &out, a);
		} else if (str_starts(line, S("=>"))) {
			line = str_trim(str_skip(line, 2));
			isize i = 0;
			while (i < line.n && !is_space(line.s[i])) i++;
			Str url = { line.s, i };
			line = str_trim(str_skip(line, i));
			if (!str_starts(url, S("gemini://"))) {
				url = str_replace_end(url, S(".gmi"), S(".html"), scratch);
			}
			if (has_image_ext(url)) {
				lm = lm_chg(lm, LINE_FIGURE, &out, a);
				str_cat(&out, S("<img src="), a);
				str_cat_uri(&out, url, a);
				str_catc(&out, '>', a);
				if (line.n > 0) {
					str_cat(&out, S("<figcaption>"), a);
					str_cat_html(&out, line, a);
					str_cat(&out, S("</figcaption>"), a);
				}
			} else {
				Str display = line.n > 0 ? line : url;
				lm = lm_chg(lm, LINE_LINK, &out, a);
				str_cat(&out, S("<a href="), a);
				str_cat_uri(&out, url, a);
				str_catc(&out, '>', a);
				str_cat_html(&out, display, a);
				str_cat(&out, S("</a>"), a);
			}
		} else if (str_starts(line, S("*"))) {
			lm = lm_chg(lm, LINE_UL, &out, a);
			str_cat_html(&out, str_trim(str_skip(line, 1)), a);
		} else if (is_ol_item(line)) {
			lm = lm_chg(lm, LINE_OL, &out, a);
			str_cat_html(&out, str_trim(str_cut(line, '.').tail), a);
		} else if (str_starts(line, S("###"))) {
			lm = lm_chg(lm, LINE_HDR3, &out, a);
			str_cat_html(&out, str_trim(str_skip(line, 3)), a);
		} else if (str_starts(line, S("##"))) {
			lm = lm_chg(lm, LINE_HDR2, &out, a);
			str_cat_html(&out, str_trim(str_skip(line, 2)), a);
		} else if (str_starts(line, S("#"))) {
			lm = lm_chg(lm, LINE_HDR1, &out, a);
			title = str_trim(str_skip(line, 1));
			str_cat_html(&out, title, a);
		} else if (str_starts(line, S(">"))) {
			lm = lm_chg(lm, LINE_BQUOT, &out, a);
			str_cat_html(&out, str_trim(str_skip(line, 1)), a);
		} else {
			lm = lm_chg(lm, LINE_PARA, &out, a);
			str_cat_html(&out, line, a);
		}
	}
	lm = lm_chg(lm, LINE_BLANK, &out, a);
	Doc *d = new(a, Doc);
	if (title.s) d->title = str_dup(title, a);
	d->html = out;
	d->prev = (*dp);
	if (*dp) (*dp)->next = d;
	*dp = d;
	return 0;
}

uint64_t str_hash(Str s) {
	uint64_t h = 14695981039346656037LU;
	for (isize i = 0; i < s.n; i++) h = (h ^ (s.s[i] & 0xff)) * 1099511628211LU;
	return h;
}

/* --hvar bgcolor:'#fcc,#cfc,#ccf,#cff,#ffc,#fcf' */
int hvar_calc(Str param, Str *name, Str *val, Str filename) {
	Cut c = str_cut(param, ':');
	if (!c.head.n || !c.tail.n) return -1;
	*name = c.head;
	usize n = 0;
	for (Str h = c.tail; h.n > 0; h = str_cut(h, ',').tail) n++;
	if (!n) return -1;
	srand(str_hash(filename));
	usize j = rand() % n;
	usize i = 0;
	for (Str h = c.tail; h.n > 0; h = str_cut(h, ',').tail) {
		if (i == j) {
			*val = str_cut(h, ',').head;
			return 0;
		}
		i++;
	}
	return -1;
}

void usage(const char *cmd) {
	fprintf(stderr, "usage: %s -?\n"
			"       %s [-s] [-c FILE] [-t TITLE] [-h NAME:ARG1[,ARG2,ARG3...]] [FILES...]\n"
			"\n"
			" -? --help         show this help text\n"
			" -s --standalone   prefix with html metadata\n"
			" -h --hvar         define a css variable (--name) with a random value,\n"
			"                   selected by a hash of the document's title\n"
			" -c --css          embed the given file within a <style> element\n"
			" -t --title        set the document title (does nothing without --standalone)\n",
			cmd, cmd);
}

typedef struct {
	int standalone;
	int from_stdin;
	Str stylesheet;
	Str hvarv[1024];
	Str title;
	int hvarc;
} Options;

Str html_head(Options *o, Arena *m, Arena *l) {
	Str h = S("<!DOCTYPE html>\n"
			"<meta charset=utf-8>\n"
			"<meta name=viewport content='width=device-width,initial-scale=1'>\n");
	if (o->title.s) {
		str_cat(&h, S("<title>"), m);
		str_cat_html(&h, o->title, m);
		str_cat(&h, S("</title>\n"), m);
	}
	if (o->stylesheet.s) {
		FILE *f = fopen(str_to_cstr(o->stylesheet, m), "r/o");
		if (!f) {
			str_putf(o->stylesheet, stderr);
			fprintf(stderr, ": %s\n", strerror(errno));
			exit(1);
		}
		str_cat(&h, S("<style>\n"), m);
		if (o->hvarc > 0) {
			str_cat(&h, S(":root{"), m);
			for (int i = 0; i < o->hvarc; i++) {
				Str name, val;
				if (hvar_calc(o->hvarv[i], &name, &val, o->title)) {
					fprintf(stderr, "invalid argument given to --hvar\n");
					exit(1);
				}
				str_cat(&h, S("--"), m);
				str_cat(&h, name, m);
				str_catc(&h, ':', m);
				str_cat(&h, val, m);
				str_catc(&h, ';', m);
			}
			str_catc(&h, '}', m);
		}
		Str css;
		if (read_all(f, &css, l)) {
			fprintf(stderr, "failed to read stylesheet: %s\n", strerror(errno));
			exit(1);
		}
		str_cat(&h, css, m);
		str_cat(&h, S("</style>\n"), m);
	}
	return h;
}

#define countof(x) (sizeof(x) / sizeof(*x))
int main(int argc, const char **argv) {
	(void)argc;

	ARENA(perm, 1 << 20)
	ARENA(scratch, 1 << 20)

	Doc *doc = 0;
	Options opts = { 0 };
	int r;

	ArgsState a = args_begin(argv);
	Str param = { 0 };
	opts.from_stdin = 1;

	while ((r = arg_get(&a, "?sc:h:t:", &param,
					"help", '?',
					"standalone", 's',
					":title", 't',
					":css", 'c',
					":hvar", 'h')) >= ARG_OK) {
		Arena reset = scratch;
		FILE *f;
		switch (r) {
		case '?':
			usage(argv[0]);
			return 0;
		case 's':
			opts.standalone = 1;
			break;
		case 'c':
			opts.stylesheet = param;
			break;
		case 't':
			opts.title = param;
			break;
		case 'h':
			if (opts.hvarc == countof(opts.hvarv)) {
				fprintf(stderr, "too many hash variables!\n");
				return 1;
			}
			opts.hvarv[opts.hvarc++] = param;
			break;
		default:
			if (str_eql(param, S("-"))) {
				if (wdoc(stdin, &doc, &perm, &scratch)) {
					fprintf(stderr, "error reading document"
							" from stdin: %s\n",
							strerror(errno));
					return 1;
				}
			} else {
				opts.from_stdin = 0;
				f = fopen(str_to_cstr(param, &scratch), "r/o");
				if (wdoc(f, &doc, &perm, &scratch)) {
					fwrite(param.s, 1, param.n, stderr);
					fprintf(stderr, ": %s\n", strerror(errno));
					return 1;
				}
				fclose(f);
				break;
			}
		}
		scratch = reset;
	}

	switch (r) {
	case ARG_BAD:
		fprintf(stderr, "unknown option '");
		fwrite(param.s, 1, param.n, stderr);
		fprintf(stderr, "' (see %s --help)\n", argv[0]);
		return 1;
	case ARG_EMPTY:
		fprintf(stderr, "'");
		fwrite(param.s, 1, param.n, stderr);
		fprintf(stderr, "' option expected an argument (see %s --help)\n", argv[0]);
		return 1;
	}

	if (opts.from_stdin) {
		if (wdoc(stdin, &doc, &perm, &scratch)) {
			fprintf(stderr, "error reading document from stdin: "
					"%s\n", strerror(errno));
			return 1;

		}
	}

	if (doc && !opts.title.s) opts.title = doc->title;
	while (doc && doc->prev) {
		doc = doc->prev;
		if (doc->title.s && !opts.title.s) opts.title = doc->title;
	}

	if (opts.standalone) {
		str_put(html_head(&opts, &perm, &scratch));
	}

	while (doc) {
		str_put(doc->html);
		doc = doc->next;
	}

	return 0;
}