Add next_entry

format/desktop_entry/desktop_entry.ha

@@ -1,198 +1,44 @@
 use locale;
 use strings;
 
-// Represents a file of the generic key/value format used by desktop entries.
-// Specification: §3
-export type file = struct {
-	preceeding_lines: []line,
-	groups:           []group,
-};
-
-// A named group of key/value entries.
-// Specification: §3.2
-export type group = struct {
-	name:  str,
-	lines: []line,
-};
-
 // A line in the file, which can be a comment (or a blank line), an entry, or
 // a localized entry.
-export type line = (comment | entry | localized_entry);
+export type line = (blank | comment | group_header | entry);
+
+// A blank line.
+// Specification: §3.1
+export type blank = void;
 
 // A comment.
 // Specification: §3.1
 export type comment = str;
 
-// A key/value pair.
-// Specification: §3.3
-export type entry = struct {
-	key:   str,
-	value: value,
-};
+// A group header.
+// Specification: §3.2
+export type group_header = str;
 
-// A localized entry.
-// Specification: §5
-export type localized_entry = struct {
+// An entry in a desktop file. Entries without an explicitly stated locale are
+// assigned [[locale::c]].
+// Specification: §3.3, §5
+export type entry = struct {
+	group:  str,
 	key:    str,
-	value:  value,
+	value:  str,
 	locale: locale::locale,
 };
 
-// An entry value. Values that reference memory (such as [[str]]) are borrowed
-// from the [[file]]. These may be free'd or overwritten when other functions in
-// this module are called, so [[value_dup]] is required to extend its lifetime.
-export type value = (str | bool | f32);
-
-// Gets a non-localized value from a file. If the group does not exist, or the
-// group exists but the key isn't in it, it returns void.
-export fn file_get(fil: *file, group_name: str, key: str) (value | void) = {
-	let grou = match (file_find_group(fil, group_name)) {
-	case let index: size => yield &fil.groups[index];
-	case void            => return void;
-	};
-	let lin = match (group_find_entry(grou, key)) {
-	case let index: size => yield grou.lines[index];
-	case void            => return void;
-	};
-	match (lin) {
-	case let entr: entry => return entr.value;
-	case                 => abort();
-	};
+// Duplicates an [[entry]]. Use [[entry_finish]] to get rid of it.
+export fn entry_dup(entr: entry) entry = entry {
+	group  = strings::dup(entr.group),
+	key    = strings::dup(entr.key),
+	value  = strings::dup(entr.value),
+	locale = locale::dup(entr.locale),
 };
 
-// Gets a localized value from a file. If the group does not exist, or the group
-// exists but the key isnt in it, it returns void. If the key is in the group
-// but is not localized to the specified locale, the non-localized value is
-// returned.
-export fn file_get_localized(
-	fil:        *file,
-	group_name: str,
-	key:        str,
-	local:      locale::locale,
-) (value | void) = {
-	let grou = match (file_find_group(fil, group_name)) {
-	case let index: size => yield &fil.groups[index];
-	case void            => return void;
-	};
-	let lin = match (group_find_localized_entry(grou, key, local)) {
-	case let index: size => yield grou.lines[index];
-	case void =>
-		yield match(group_find_entry(grou, key)) {
-		case let index: size => yield index;
-		case void            => return void;
-		};
-	};
-	match (lin) {
-	case let entr: localized_entry => return entr.value;
-	case let entr: entry           => return entr.value;
-	case                           => abort();
-	};
-};
-
-//export fn file_add
-
-//export fn file_add_localized
-
-//export fn file_remove
-
-//export fn file_remove_localized
-
-//export fn file_encode
-
-//export fn file_finish
-
-fn file_find_group(fil: *file, group_name: str) (size | void) = {
-	let index = 0z;
-	for (let grou .. fil.groups) {
-		if (grou.name == group_name) {
-			return index;
-		};
-		index += 1;
-	};
-};
-
-fn group_find_entry(grou: *group, key: str) (size | void) = {
-	let index = 0z;
-	for (let lin .. grou.lines) {
-		match (lin) {
-		case let entr: entry =>
-			if (entr.key == key) {
-				return index;
-			};
-		case => void;
-		};
-		index += 1;
-	};
-};
-
-fn group_find_localized_entry(grou: *group, key: str, local: locale::locale) (size | void) = {
-	// The matching is done as follows. If LC_MESSAGES is of the form
-	// lang_COUNTRY.ENCODING@MODIFIER, then it will match a key of the form
-	// lang_COUNTRY@MODIFIER. If such a key does not exist, it will attempt
-	// to match lang_COUNTRY followed by lang@MODIFIER. Then, a match
-	// against lang by itself will be attempted. Finally, if no matching
-	// key is found the required key without a locale specified is used.
-	// The encoding from the LC_MESSAGES value is ignored when matching.
-	// 
-	// If LC_MESSAGES does not have a MODIFIER field, then no key with a
-	// modifier will be matched. Similarly, if LC_MESSAGES does not have a
-	// COUNTRY field, then no key with a country specified will be matched.
-	// If LC_MESSAGES just has a lang field, then it will do a straight
-	// match to a key with a similar value.
-
-	let lang_country_modifier = local;
-	lang_country_modifier.encoding = "";
-	match(group_find_localized_entry_exact(grou, key, lang_country_modifier)) {
-	case let index: size => return index;
-	case                 =>	void;
-	};
-	
-	let lang_country = lang_country_modifier;
-	lang_country.modifier = "";
-	match(group_find_localized_entry_exact(grou, key, lang_country)) {
-	case let index: size => return index;
-	case                 =>	void;
-	};
-	
-	let lang_modifier = lang_country_modifier;
-	lang_modifier.country = "";
-	match(group_find_localized_entry_exact(grou, key, lang_modifier)) {
-	case let index: size => return index;
-	case                 =>	void;
-	};
-
-	let lang = lang_modifier;
-	lang.modifier = "";
-	match(group_find_localized_entry_exact(grou, key, lang)) {
-	case let index: size => return index;
-	case                 =>	void;
-	};
-	
-	return void;
-};
-
-fn group_find_localized_entry_exact(grou: *group, key: str, local: locale::locale) (size | void) = {
-	let index = 0z;
-	for (let lin .. grou.lines) {
-		match (lin) {
-		case let entr: localized_entry =>
-			if (entr.key == key && locale::equal(entr.locale, local)) {
-				return index;
-			};
-		case => void;
-		};
-		index += 1;
-	};
-};
-
-// Duplicates a [[value]]. Use [[value_finish]] to get rid of it.
-export fn value_dup(valu: value) value = match (valu) {
-case let valu: str => yield strings::dup(valu);
-case               => yield valu;
-};
-
-// Frees an [[entry]] previously duplicated with [[entry_dup]].
-export fn value_finish(valu: value) void = match (valu) {
-case let valu: str => free(valu);
-case               => void;
+// Frees memory associated with an [[entry]].
+export fn entry_finish(entr: entry) void = {
+	free(entr.group);
+	free(entr.key);
+	free(entr.value);
+	locale::finish(entr.locale);
 };

format/desktop_entry/error.ha

@@ -0,0 +1,35 @@
+use encoding::utf8;
+use io;
+
+// Any error that may occur during parsing.
+export type error = !(syntaxerr | io::error | utf8::invalid);
+
+// All syntax errors defined in this module.
+export type syntaxerr = !(
+	invalid_group_header |
+	invalid_entry |
+	invalid_ascii |
+	invalid_escape);
+
+// Returned when a malformed group header is encountered while parsing.
+export type invalid_group_header = !void;
+
+// Returned when a malformed entry is encountered while parsing.
+export type invalid_entry = !void;
+
+// Returned when ASCII text was expected while parsing, but something else was
+// given.
+export type invalid_ascii = !void;
+
+// Returned when an invalid escape sequence was enctountered while parsing.
+export type invalid_escape = !void;
+
+// Converts a desktop entry [[error]] into a user-friendly string.
+export fn strerror(err: error) str = match (err) {
+case invalid_group_header   => yield "invalid group header";
+case invalid_entry          => yield "invalid entry";
+case invalid_ascii          => yield "invalid ascii";
+case invalid_escape         => yield "invalid escape";
+case let err: io::error     => yield io::strerror(err);
+case let err: utf8::invalid => yield utf8::strerror(err);
+};

format/desktop_entry/scan.ha

@@ -0,0 +1,122 @@
+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 [[line_dup]] to retain a copy. If all you
+// want is the file's data, use next_entry.
+// FIXME: there is no line_dup
+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 let err: (io::error | utf8::invalid) =>
+		return err;
+	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 = 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));
+	if (!validate_entry_key(key)) return invalid_entry;
+
+	let (key, local_string) = strings::cut(key, "[");
+	let local = if (local_string != "") {
+		yield locale::c;
+	} else {
+		local_string = strings::rtrim(local_string, ']');
+		if (!validate_entry_locale(local_string)) return invalid_entry;
+		
+		yield match(locale::parse(local_string)) {
+		case let local: locale::locale => yield local;
+		case locale::invalid           => return invalid_entry;
+		};
+	};
+
+	return entry {
+		key    = key,
+		value  = valu,
+		locale = local,
+		...
+	};
+};
+
+fn validate_entry_key(key: str) bool = {
+	return strings::contains(key, '[', ']', '=');
+};
+
+fn validate_entry_locale(key: str) bool = {
+	return strings::contains(key, '[', ']', '=');
+};

locale/+linux.ha

@@ -52,29 +52,29 @@ fn get_locale(var: str) locale =
 		};
 	};
 
-fn get_locale_no_fallback(var: str) (locale | errors::invalid) =
+fn get_locale_no_fallback(var: str) (locale | invalid) =
 	match (get_env_locale(var)) {
 	case let local: locale => yield local;
 	case                   => yield (get_locale_conf_entry(var));
 	};
 
-fn get_env_locale(var: str) (locale | errors::invalid) =
+fn get_env_locale(var: str) (locale | invalid) =
 	match (os::getenv(var)) {
 	case let env: str => return parse(env);
-	case              => return errors::invalid;
+	case              => return invalid;
 	};
 
 let locale_conf: []str = [];
 
 @fini fn locale_conf() void = strings::freeall(locale_conf);
 
-fn get_locale_conf_entry(var: str) (locale | errors::invalid) = {
+fn get_locale_conf_entry(var: str) (locale | invalid) = {
 	get_locale_conf();
 	for (let entry .. locale_conf) {
 		let (key, value) = strings::cut(entry, "=");
 		if (key == var) return parse(value);
 	};
-	return errors::invalid;
+	return invalid;
 };
 
 fn get_locale_conf() []str = {

locale/-linux.ha

@@ -1,5 +1,3 @@
-use errors;
-
 // Returns the locale to use for character classification and case conversion.
 // The memory is statically allocated and must not be free'd. It may be
 // overwritten later, so use [[dup]] to extend its lifetime.

locale/error.ha

@@ -0,0 +1,7 @@
+// All errors defined in this module.
+export type error = !invalid;
+
+// Returned when a malformed locale is encountered.
+export type invalid = !void;
+
+export fn strerror(err: error) str = "invalid locale";

locale/locale.ha

@@ -21,16 +21,16 @@ export def c = locale {
 //   lang_COUNTRY.ENCODING@MODIFIER
 // 
 // Where _COUNTRY, .ENCODING, and @MODIFIER may be omitted. The function
-// returns a [[locale]], or [[errors::invalid]] if the input cannot be parsed.
-// All memory is borrowed from the input, so [[finish]] should not be used to
-// free it.
-export fn parse(in: str) (locale | errors::invalid) = {
+// returns a [[locale]], or [[invalid]] if the input cannot be parsed. All
+// memory is borrowed from the input, so [[finish]] should not be used to free
+// it.
+export fn parse(in: str) (locale | invalid) = {
 	let (in, modifier) = strings::rcut(in, "@");
-	if (strings::compare(in, "") == 0) return void: errors::invalid;
+	if (strings::compare(in, "") == 0) return void: invalid;
 	let (in, encoding) = strings::rcut(in, ".");
-	if (strings::compare(in, "") == 0) return void: errors::invalid;
+	if (strings::compare(in, "") == 0) return void: invalid;
 	let (in, country) = strings::rcut(in, "_");
-	if (strings::compare(in, "") == 0) return void: errors::invalid;
+	if (strings::compare(in, "") == 0) return void: invalid;
 	return locale {
 		lang     = in,
 		country  = country,

locale/locale_test.ha

@@ -1,5 +1,4 @@
 use fmt;
-use errors;
 use strings;
 
 @test fn parse_full() void = {
@@ -64,8 +63,8 @@ use strings;
 
 @test fn parse_error() void = {
 	let local = match(parse("_COUNTRY.ENCODING@MODIFIER")) {
-	case errors::invalid => void;
-	case                 => abort("error");
+	case invalid => void;
+	case         => abort("no error");
 	};
 };