adds support for named subpatterns to mb_ereg_replace

ju1ius · ju1ius · commit af8b54916050 · 2016-07-31T02:25:50.000+02:00
Named subpatterns are now passed to `mb_ereg_replace_callback`.

This commit also adds a subset of the oniguruma back-reference syntax
for replacements:
* `\k&lt;name&gt;` and `\k'name'` for named subpatterns.
* `\k&lt;n&gt;` and `\k'n'` for numbered subpatterns
These last two notations allow referencing numbered groups where n &gt; 9.
diff --git a/ext/mbstring/php_mbregex.c b/ext/mbstring/php_mbregex.c
@@ -693,6 +693,129 @@ mb_regex_groups_iter(const OnigUChar* name, const OnigUChar* name_end, int ngrou
 }
 /* }}} */
 
+/*
+ * Helper for _php_mb_regex_ereg_replace_exec
+ */
+/* {{{ mb_regex_substitute */
+static inline void mb_regex_substitute(
+	smart_str *pbuf,
+	char *subject,
+	size_t subject_len,
+	char *replace,
+	size_t replace_len,
+	php_mb_regex_t *regexp,
+	OnigRegion *regs,
+	const mbfl_encoding *enc
+) {
+	char *p, *sp, *eos;
+	int no; /* bakreference group number */
+	int clen; /* byte-length of the current character */
+
+	p = replace;
+	eos = replace + replace_len;
+
+	while (p < eos) {
+		clen = (int) php_mb_mbchar_bytes_ex(p, enc);
+		if (clen != 1 || p == eos || p[0] != '\\') {
+			/* skip anything that's not an ascii backslash */
+			smart_str_appendl(pbuf, p, clen);
+			p += clen;
+			continue;
+		}
+		sp = p; /* save position */
+		clen = (int) php_mb_mbchar_bytes_ex(++p, enc);
+		if (clen != 1 || p == eos) {
+			/* skip escaped multibyte char */
+			p += clen;
+			smart_str_appendl(pbuf, sp, p - sp);
+			continue;
+		}
+		no = -1;
+		switch (p[0]) {
+			case '0':
+				no = 0;
+				p++;
+				break;
+			case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				if (!onig_noname_group_capture_is_active(regexp)) {
+					/*
+					 * FIXME:
+					 * Oniguruma throws a compile error if numbered backrefs are used with named groups in the pattern.
+					 * For now we just ignore them, but in the future we might want to raise a warning
+					 * and abort the whole replace operation.
+					 */
+					p++;
+					smart_str_appendl(pbuf, sp, p - sp);
+					continue;
+				}
+				no = p[0] - '0';
+				p++;
+				break;
+			case 'k':
+				clen = (int) php_mb_mbchar_bytes_ex(++p, enc);
+				if (clen != 1 || p == eos || (p[0] != '<' && p[0] != '\'')) {
+					/* not a backref delimiter */
+					p += clen;
+					smart_str_appendl(pbuf, sp, p - sp);
+					continue;
+				}
+				/* try to consume everything until next delimiter */
+				char delim = p[0] == '<' ? '>' : '\'';
+				char *name, *name_end;
+				int maybe_num = 1;
+				name_end = name = p + 1;
+				while (name_end < eos) {
+					clen = (int) php_mb_mbchar_bytes_ex(name_end, enc);
+					if (clen != 1) {
+						name_end += clen;
+						maybe_num = 0;
+						continue;
+					}
+					if (name_end[0] == delim) break;
+					if (maybe_num && !isdigit(name_end[0])) {
+						maybe_num = 0;
+					}
+					name_end++;
+				}
+				if (name_end < eos) {
+					p = name_end + 1;
+					/* we have either a name or a number */
+					if (maybe_num) {
+						/* check for NaN */
+						if (name[0] == '0' && name_end - name > 1) {
+							break;
+						}
+						no = (int) strtoul(name, NULL, 10);
+						break;
+					}
+					no = onig_name_to_backref_number(regexp, (OnigUChar *)name, (OnigUChar *)name_end, regs);
+					break;
+				}
+				/* we failed to find a '>' */
+				smart_str_appendl(pbuf, sp, p - sp);
+				continue;
+			default:
+				p += clen;
+				smart_str_appendl(pbuf, sp, p - sp);
+				continue;
+		}
+		if (no < 0 || no >= regs->num_regs) {
+			/* invalid group number reference, keep the escape sequence in the output */
+			smart_str_appendl(pbuf, sp, p - sp);
+			continue;
+		}
+		if (regs->beg[no] >= 0 && regs->beg[no] < regs->end[no] && (size_t)regs->end[no] <= subject_len) {
+			smart_str_appendl(pbuf, subject + regs->beg[no], regs->end[no] - regs->beg[no]);
+		}
+	}
+
+	if (p < eos) {
+		smart_str_appendl(pbuf, p, eos - p);
+	}
+}
+/* }}} */
+
 /*
  * php functions
  */
@@ -859,14 +982,12 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 	char *string;
 	size_t string_len;
 
-	char *p;
 	php_mb_regex_t *re;
 	OnigSyntaxType *syntax;
 	OnigRegion *regs = NULL;
 	smart_str out_buf = {0};
 	smart_str eval_buf = {0};
 	smart_str *pbuf;
-	size_t i;
 	int err, eval, n;
 	OnigUChar *pos;
 	OnigUChar *string_lim;
@@ -976,38 +1097,11 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 			break;
 		}
 		if (err >= 0) {
-#if moriyoshi_0
-			if (regs->beg[0] == regs->end[0]) {
-				php_error_docref(NULL, E_WARNING, "Empty regular expression");
-				break;
-			}
-#endif
 			/* copy the part of the string before the match */
 			smart_str_appendl(&out_buf, (char *)pos, (size_t)((OnigUChar *)(string + regs->beg[0]) - pos));
 
 			if (!is_callable) {
-				/* copy replacement and backrefs */
-				i = 0;
-				p = replace;
-				while (i < replace_len) {
-					int fwd = (int) php_mb_mbchar_bytes_ex(p, enc);
-					n = -1;
-					if ((replace_len - i) >= 2 && fwd == 1 &&
-					p[0] == '\\' && p[1] >= '0' && p[1] <= '9') {
-						n = p[1] - '0';
-					}
-					if (n >= 0 && n < regs->num_regs) {
-						if (regs->beg[n] >= 0 && regs->beg[n] < regs->end[n] && (size_t)regs->end[n] <= string_len) {
-							smart_str_appendl(pbuf, string + regs->beg[n], regs->end[n] - regs->beg[n]);
-						}
-						p += 2;
-						i += 2;
-					} else {
-						smart_str_appendl(pbuf, p, fwd);
-						p += fwd;
-						i += fwd;
-					}
-				}
+				mb_regex_substitute(pbuf, string, string_len, replace, replace_len, re, regs, enc);
 			}
 
 			if (eval) {
@@ -1047,6 +1141,10 @@ static void _php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAMETERS, OnigOp
 				for (i = 0; i < regs->num_regs; i++) {
 					add_next_index_stringl(&subpats, string + regs->beg[i], regs->end[i] - regs->beg[i]);
 				}
+				if (onig_number_of_names(re) > 0) {
+					mb_regex_groups_iter_args args = {&subpats, string, string_len, regs};
+					onig_foreach_name(re, mb_regex_groups_iter, &args);
+				}
 
 				ZVAL_COPY_VALUE(&args[0], &subpats);
 				/* null terminate buffer */
diff --git a/ext/mbstring/tests/mb_ereg_replace_callback.phpt b/ext/mbstring/tests/mb_ereg_replace_callback.phpt
@@ -8,8 +8,16 @@ function_exists('mb_ereg_replace_callback') or die("skip mb_ereg_replace_callbac
 --FILE--
 <?php
 $str = 'abc 123 #",; $foo';
-echo mb_ereg_replace_callback('(\S+)', function($m){return $m[1].'('.strlen($m[1]).')';}, $str);
+
+echo mb_ereg_replace_callback('(\S+)', function ($m) {
+    return $m[1].'('.strlen($m[1]).')';
+}, $str), "\n";
+
+echo mb_ereg_replace_callback('(?<word>\w+) (?<digit>\d+).*', function ($m) {
+    return sprintf("%s-%s", $m['digit'], $m['word']);
+}, $str), "\n";
 ?>
 --EXPECT--
 abc(3) 123(3) #",;(4) $foo(4)
+123-abc
 
diff --git a/ext/mbstring/tests/mb_ereg_replace_named_subpatterns.phpt b/ext/mbstring/tests/mb_ereg_replace_named_subpatterns.phpt
@@ -0,0 +1,27 @@
+--TEST--
+mb_ereg_replace() with named subpatterns
+--SKIPIF--
+<?php
+extension_loaded('mbstring') or die('skip mbstring not available');
+function_exists('mb_ereg_replace') or die("skip mb_ereg_replace() is not available in this build");
+?>
+--FILE--
+<?php
+    mb_regex_set_options('');
+    echo mb_ereg_replace('(?<a>\s*)(?<b>\w+)(?<c>\s*)', '\k<a>_\k<b>_\k<c>', 'a b c d e' ), "\n";
+    // \k'word' syntax
+    echo mb_ereg_replace('(?<word>[a-z]+)',"<\k'word'>", 'abc def ghi'), PHP_EOL;
+    // numbered captures with \k<n> syntax
+    echo mb_ereg_replace('(1)(2)(3)(4)(5)(6)(7)(8)(9)(a)(\10)', '\0-\k<10>-', '123456789aa'), PHP_EOL;
+    // backref 0 works, but 01 is ignored
+    echo mb_ereg_replace('a', "\k'0'_\k<01>", 'a'), PHP_EOL;
+    // Numbered backref is ignored if named backrefs are present
+    echo mb_ereg_replace('(?<a>A)\k<a>', '-\1-', 'AA'), PHP_EOL;
+?>
+
+--EXPECT--
+_a_ _b_ _c_ _d_ _e_
+<abc> <def> <ghi>
+123456789aa-a-
+a_\k<01>
+-\1-
