extmod/re1.5: Support escaping within RE classes.

jimmo · dpgeorge · commit ebf8332104c1 · 2019-10-18T12:20:32.000+11:00
Fixes issues #3178 and #5220. Tests are added, including all the cases mentioned in both bugs.
diff --git a/extmod/re1.5/compilecode.c b/extmod/re1.5/compilecode.c
@@ -53,6 +53,9 @@ static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
             PC++; // Skip # of pair byte
             prog->len++;
             for (cnt = 0; *re != ']'; re++, cnt++) {
+                if (*re == '\\') {
+                    ++re;
+                }
                 if (!*re) return NULL;
                 EMIT(PC++, *re);
                 if (re[1] == '-' && re[2] != ']') {
diff --git a/tests/extmod/ure1.py b/tests/extmod/ure1.py
@@ -88,3 +88,23 @@
 
 # bytes objects
 m = re.match(rb'a+?', b'ab');  print(m.group(0))
+print("===")
+
+# escaping
+m = re.match(r'a\.c', 'a.c'); print(m.group(0) if m else '')
+m = re.match(r'a\.b', 'abc'); print(m is None)
+m = re.match(r'a\.b', 'a\\bc'); print(m is None)
+m = re.match(r'[a\-z]', 'abc'); print(m.group(0))
+m = re.match(r'[.\]]*', '.].]a'); print(m.group(0))
+m = re.match(r'[.\]+]*', '.]+.]a'); print(m.group(0))
+m = re.match(r'[a-f0-9x\-yz]*', 'abxcd1-23'); print(m.group(0))
+m = re.match(r'[a\\b]*', 'a\\aa\\bb\\bbab'); print(m.group(0))
+m = re.search(r'[a\-z]', '-'); print(m.group(0))
+m = re.search(r'[a\-z]', 'f'); print(m is None)
+m = re.search(r'[a\]z]', 'a'); print(m.group(0))
+print(re.compile(r'[-a]').split('foo-bar'))
+print(re.compile(r'[a-]').split('foo-bar'))
+print(re.compile(r'[ax\-]').split('foo-bar'))
+print(re.compile(r'[a\-x]').split('foo-bar'))
+print(re.compile(r'[\-ax]').split('foo-bar'))
+print("===")
diff --git a/tests/extmod/ure_error.py b/tests/extmod/ure_error.py
@@ -23,3 +23,4 @@ def test_re(r):
 test_re(r'[')
 test_re(r'([')
 test_re(r'([)')
+test_re(r'[a\]')
