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

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

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, strv_t ln, size_t i) {
	size_t start = i - 1;
	while (start > 0 && isalpha(ln.s[start])) start--;
	if (!isalpha(ln.s[start])) start++;
	size_t end = i + 3;
	while (end < ln.n && isurlch(ln.s[end])) end++;
	if (end == i + 3) return;
	strv_t url = { &ln.s[start], end - start };
	doc_line_at(d, d->latest)->link = doc_add_link(d, url);
}

int parse_plain(struct doc *d, const str_t *b) {
	doc_init(d);
	strv_t ln, buf = strv(b);
	while (strv_split(&buf, '\n', &ln)) {
		if (ln.n > 0 && ln.s[ln.n - 1] == '\r') ln.n--;
		for (size_t j = 1; j + 2 < ln.n; j++) {
			if (ln.s[j] == ':' && ln.s[j + 1] == '/' && ln.s[j + 2] == '/') {
				parse_plain_url(d, ln, j);
			}
		}
		doc_add_line(d, ln);
	}
	return 0;
}

int parse_gophermap_line(struct doc *d, strv_t ln) {
	return 0;
}

int parse_gophermap(struct doc *d, const str_t *b) {
	str_t url;
	struct {
		char item_type;
		strv_t dstr;
		strv_t sel;
		strv_t host;
		strv_t port;
	} bits;
	doc_init(d);
	str_init(&url, 64);
	strv_t ln, buf = { b->buf, b->sz };
	while (strv_split(&buf, '\n', &ln)) {
		if (ln.n > 0 && ln.s[ln.n - 1] == '\r') ln.n--;
		if (ln.n == 1 && ln.s[0] == '.') break;
		bits.item_type = strv_next(&ln);
		strv_split(&ln, '\t', &bits.dstr);
		strv_split(&ln, '\t', &bits.sel);
		strv_split(&ln, '\t', &bits.host);
		strv_split(&ln, '\t', &bits.port);
		switch (bits.item_type) {
		default:
			str_fmt(&url, "%s:%s/%c%s", bits.host, bits.port, bits.item_type, bits.sel);
			doc_set_link(d, doc_add_link(d, strv(url)));
		case 'i':
			doc_add_text(d, bits.dstr);
			doc_new_line(d);
			break;
		}
	}
	str_free(&url);
	return 0;
}

int parse_doc(enum doc_type type, struct doc *d, const str_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;
	}
}