hare-xdg/format/desktop_entry/scan.ha

use ascii;
use bufio;
use encoding::utf8;
use errors;
use io;
use locale;
use memio;
use strings;

export type scanner = struct {
	scanner: bufio::scanner,
	group:   str,
};

// Creates a desktop entry file scanner. Use [[next]] to read lines. The caller
// must call [[finish]] once they're done with this object.
export fn scan(input: io::handle) scanner = scanner {
	scanner = bufio::newscanner(input),
	...
};

// Returns the next line from a desktop entry file. The return value is
// borrowed from the [[scanner]]. Use [[strings::dup]], [[entry_dup]], etc. to
// retain a copy. If all you want is the file's data, use [[next_entry]].
export fn next(this: *scanner) (line | io::EOF | error) = {
	let text = match (bufio::scan_line(&this.scanner)?) {
	case let text: const str => yield text;
	case io::EOF             => return io::EOF;
	};

	if (text == "") {
		// blank line
		return blank;
		
	} else if (strings::hasprefix(text, '#')) {
		// comment
		return strings::ltrim(text, '#'): comment;
		
	} else if (strings::hasprefix(text, '[')) {
		// group header
		let header = parse_group_header(text)?;
		free(this.group);
		this.group = strings::dup(header: str);
		return header;
		
	} else {
		// key/value pair
		let entry = parse_entry(text)?;
		entry.group = this.group;
		return entry;
	};
};

// Returns the next entry from the desktop file, skipping over blank lines,
// comments, group headers, etc. The return value is borrowed from the
// [[scanner]]. Use [[entry_dup]] to retain a copy.
export fn next_entry(this: *scanner) (entry | io::EOF | error) = {
	for (true) match(next(this)?) {
	case let entr: entry => return entr;
	case io::EOF         => return io::EOF;
	case                 => void;
	};
};

// Frees resources associated with a [[scanner]].
export fn finish(this: *scanner) void = {
	bufio::finish(&this.scanner);
	free(this.group);
};

// memory is borrowed from the input
fn parse_group_header(text: str) (group_header | error) = {
	if (!strings::hasprefix(text, '[') || !strings::hassuffix(text, ']')) {
		return invalid_group_header;
	};

	text = strings::rtrim(strings::ltrim(text, '['), ']');
	if (strings::contains(text, '[', ']')) {
		return invalid_group_header;
	};
	return text: group_header;
};

// memory is borrowed from the input
fn parse_entry(line: str) (entry | error) = {
	if (!strings::contains(line, '=')) return invalid_entry;
	let (key, valu) = strings::cut(line, "=");
	key = strings::ltrim(strings::rtrim(key));

	let (key, local_string) = strings::cut(key, "[");
	let local = if (local_string == "") {
		yield locale::c;
	} else {
		local_string = strings::rtrim(local_string, ']');
		validate_entry_locale(local_string)?;
				
		yield match(locale::parse(local_string)) {
		case let local: locale::locale => yield local;
		case locale::invalid           => return invalid_entry;
		};
	};
	local.encoding = "";

	validate_entry_key(key)?;
	return entry {
		key    = key,
		value  = valu,
		locale = local,
		...
	};
};

fn validate_entry_key(key: str) (void | error) = {
	for (let byte .. strings::toutf8(key)) {
		const ok =
			(byte >= 'A' && byte <= 'Z') ||
			(byte >= 'a' && byte <= 'z') ||
			(byte >= '0' && byte <= '9') ||
			(byte == '-');
		if (!ok) return invalid_entry;
	};
};

fn validate_entry_locale(locale: str) (void | error) = {
	for (let byte .. strings::toutf8(locale)) {
		const bad =
			(byte == '[') ||
			(byte == ']') ||
			(byte == '=') ||
			!ascii::isprint(byte: rune);
		if (bad) return invalid_entry;
	};
};