#include <ctype.h>
#include <string.h>

#include "parse.h"
#include "err.h"

struct str_slice {
	const char *s;
	size_t n;
};

int isurlch(char c) {
	return isalpha(c) || isdigit(c) || c == '-' || c == '.' || c == '_' || c == '~' || c == '!' || c == '$' || c == '\'' || c == '(' || c == ')' || c == '*' || c == '+' || c == ',' || c == ';' || c == '=' || c == '%' || c == '@' || c == ':' || c == '/';
}

void parse_plain_url(struct doc *d, struct doc_line *l, size_t i) {
	char url[l->len + 1];
	size_t start = i - 1;
	while (start > 0 && isalpha(l->txt[start])) start--;
	if (!isalpha(l->txt[start])) start++;
	size_t end = i + 3;
	while (end < l->len && isurlch(l->txt[end])) end++;
	if (end == i + 3) return;
	size_t urln = end - start;
	memcpy(url, &l->txt[start], urln);
	url[urln] = 0;
	l->link = doc_add_link(d, url);
}

int parse_plain(struct doc *d, const buf_t *b) {
	doc_init(d);
	for (size_t i = 0; i < b->sz; i++) {
		char c = b->buf[i];
		if (c == '\n') {
			struct doc_line *l = doc_line_at(d, d->latest);
			for (size_t i = 1; i + 2 < l->len; i++) {
				if (l->txt[i] == ':' && l->txt[i + 1] == '/' && l->txt[i + 2] == '/') {
					parse_plain_url(d, l, i);
					break;
				}
			}
			doc_new_line(d);
		} else {
			doc_add_textn(d, &c, 1);
		}
	}
	return 0;
}

static struct str_slice gmbit(size_t *i, const char *s, size_t n) {
	struct str_slice ss = {
		&s[*i],
		0
	};
	while (*i < n && s[*i] != '\t') {
		*i += 1;
		ss.n++;
	}
	*i += 1;
	return ss;
}

size_t scatss(char *buf, size_t i, size_t n, struct str_slice ss) {
	size_t si = 0;
	while (i < n && si < ss.n) {
		buf[i++] = ss.s[si++];
	}
	return i;
}

int parse_gophermap_line(struct doc *d, const char *s, size_t n) {
	char url[512] = "gopher://";
	size_t urln = 9;
	struct {
		char item_type;
		struct str_slice dstr;
		struct str_slice sel;
		struct str_slice host;
		struct str_slice port;
	} bits;
	size_t i = 0;
	bits.item_type = s[i++];
	bits.dstr = gmbit(&i, s, n);
	bits.sel = gmbit(&i, s, n);
	bits.host = gmbit(&i, s, n);
	bits.port = gmbit(&i, s, n);
	switch (bits.item_type) {
	case '.':
		if (n == 1) return 1;
	default:
		urln = scatss(url, urln, sizeof url, bits.host);
		if (urln < sizeof url) url[urln++] = ':';
		urln = scatss(url, urln, sizeof url, bits.port);
		if (urln < sizeof url) url[urln++] = '/';
		if (urln < sizeof url) url[urln++] = bits.item_type;
		urln = scatss(url, urln, sizeof url, bits.sel);
		url[urln] = 0;
		doc_set_link(d, doc_add_link(d, url));
	case 'i':
		doc_add_textn(d, bits.dstr.s, bits.dstr.n);
		doc_new_line(d);
		break;
	}
	return 0;
}

int parse_gophermap(struct doc *d, const buf_t *b) {
	doc_init(d);
	size_t ln_start = 0;
	for (size_t i = 0; i < b->sz; i++) {
		if (b->buf[i] == '\r') continue;
		if (b->buf[i] == '\n') {
			char *ln_str = &b->buf[ln_start];
			size_t ln_len = i - ln_start;
			if (i > 0 && b->buf[i - 1] == '\r') ln_len--;
			if (parse_gophermap_line(d, ln_str, ln_len)) {
				break;
			}
			ln_start = i + 1;
		}
	}
	return 0;
}

int parse_doc(enum doc_type type, struct doc *d, const buf_t *b) {
	switch (type) {
	case DOC_PLAIN:
		return parse_plain(d, b);
	case DOC_GOPHERMAP:
		return parse_gophermap(d, b);
	default:
		perr("unsupported doctype");
		return -1;
	}
}