V: use POSIX regexs, and read one symbol at a time

v/syms.v

@@ -2,45 +2,77 @@ module main
 
 import os
 import flag
-import regex // FIXME: use POSIX regexs
 
 import v.vmod
 const manifest = vmod.from_file('v.mod') or { panic(err) }
 
+#include "regex.h"
+
+const reg_extended = 1
+[typedef]
+struct C.regex_t {}
+[typedef]
+struct C.regmatch_t {
+	rm_so isize
+	rm_eo isize
+}
+fn C.regcomp(&C.regex_t, &char, int) int
+fn C.regexec(&C.regex_t, &char, usize, []C.regmatch_t, int) int
+fn C.regerror(int, &C.regex_t, &char, usize) usize
+
 [noreturn]
 fn error_exit(code int, msg string) {
 	eprintln(msg)
 	exit(code)
 }
 
+fn get_symbol(re &C.regex_t, s string, start isize) (isize, isize)
+{
+	matches := []C.regmatch_t{len: 2}
+	if C.regexec(re, unsafe {s.str + start}, 1, matches.data, 0) != 0 {
+		return -1, -1
+	}
+	return start + matches[0].rm_so, start + matches[0].rm_eo
+}
+
 fn main() {
 	mut fp := flag.new_flag_parser(os.args)
 	fp.application(manifest.name)
 	fp.version(manifest.version)
 	fp.description(manifest.description)
-	default_symbol := '[\\a\\A]+'
+	default_symbol := '[[:alpha:]]+'
 	symbol := fp.string('symbol', `s`, default_symbol, 'symbols are given by REGEXP')
-	mut re := regex.regex_opt(symbol) or {
-		error_exit(1, 'invalid regex $symbol')
-	}
 	fp.footer('
 The default symbol type is words (-s "$default_symbol"); other useful settings include:
 
-  non-white-space characters: -s "[\\S]+"
-  alphanumerics and underscores: -s "\\w+"
-  XML tags: -s "<([\\a\\a_:][\\w:.-]*)[\\s>]"')
+  non-white-space characters: -s "[^[:space:]]+"
+  alphanumerics and underscores: -s "[[:alnum:]_]+"
+  XML tags: -s "<([a-zA-Z_:][a-zA-Z:.0-9-]*)[[:space:]>]"')
 	fp.skip_executable()
 
+	// Compile regex
+	re := &C.regex_t(0)
+	err := C.regcomp(&re, symbol.str, reg_extended)
+	if err != 0 {
+		errlen := C.regerror(err, &re, &char(0), 0)
+		errbuf := []char{len: int(errlen)}
+		C.regerror(err, &re, &errbuf[0], errlen)
+		error_exit(1, unsafe { (&errbuf[0]).vstring() })
+	}
+
 	additional_args := fp.finalize() ?
 	for file in additional_args {
 		// FIXME: read one line at a time
 		lines := os.read_lines(file) or {
 			error_exit(1, 'cannot open \'$file\'')
 		}
 		for l in lines {
-			// FIXME: read one symbol at a time
-			for sym in re.find_all_str(l) {
-				println(sym)
+			for start, end := isize(0), isize(0); true; start = end {
+				start, end = get_symbol(&re, l, start)
+				if start == -1 {
+					break
+				}
+				println('${l[start..end]}')
 			}
 		}
 	}