Merge pull request #539 from miloyip/issue538_regexzeromin

Support {0, } and {0, m} in Regex
Tencent · Feb 12, 2016 · 3cb5733 · 3cb5733
2 parents 3c2c162 + 60116cf
commit 3cb5733
Show file tree

Hide file tree

Showing 2 changed files with 67 additions and 6 deletions.
diff --git a/include/rapidjson/internal/regex.h b/include/rapidjson/internal/regex.h
@@ -256,13 +256,13 @@ class GenericRegex {
                 case '{':
                     {
                         unsigned n, m;
-                        if (!ParseUnsigned(ds, &n) || n == 0)
+                        if (!ParseUnsigned(ds, &n))
                             return;
 
                         if (ds.Peek() == ',') {
                             ds.Take();
                             if (ds.Peek() == '}')
-                                m = 0;
+                                m = kInfinityQuantifier;
                             else if (!ParseUnsigned(ds, &m) || m < n)
                                 return;
                         }
@@ -424,15 +424,29 @@ class GenericRegex {
     }
 
     bool EvalQuantifier(Stack<Allocator>& operandStack, unsigned n, unsigned m) {
-        RAPIDJSON_ASSERT(n > 0);
-        RAPIDJSON_ASSERT(m == 0 || n <= m);         // m == 0 means infinity
+        RAPIDJSON_ASSERT(n <= m);
         if (operandStack.GetSize() < sizeof(Frag))
             return false;
 
+        if (n == 0) {
+            if (m == 0)                             // a{0} not support
+                return false;
+            else if (m == kInfinityQuantifier)
+                Eval(operandStack, kZeroOrMore);    // a{0,} -> a*
+            else {
+                Eval(operandStack, kZeroOrOne);         // a{0,5} -> a?
+                for (unsigned i = 0; i < m - 1; i++)
+                    CloneTopOperand(operandStack);      // a{0,5} -> a? a? a? a? a?
+                for (unsigned i = 0; i < m - 1; i++)
+                    Eval(operandStack, kConcatenation); // a{0,5} -> a?a?a?a?a?
+            }
+            return true;
+        }
+
         for (unsigned i = 0; i < n - 1; i++)        // a{3} -> a a a
             CloneTopOperand(operandStack);
 
-        if (m == 0)
+        if (m == kInfinityQuantifier)
             Eval(operandStack, kOneOrMore);         // a{3,} -> a a a+
         else if (m > n) {
             CloneTopOperand(operandStack);          // a{3,5} -> a a a a
@@ -469,6 +483,8 @@ class GenericRegex {
     template <typename InputStream>
     bool ParseUnsigned(DecodedStream<InputStream>& ds, unsigned* u) {
         unsigned r = 0;
+        if (ds.Peek() < '0' || ds.Peek() > '9')
+            return false;
         while (ds.Peek() >= '0' && ds.Peek() <= '9') {
             if (r >= 429496729 && ds.Peek() > '5') // 2^32 - 1 = 4294967295
                 return false; // overflow
@@ -658,6 +674,8 @@ class GenericRegex {
     SizeType stateCount_;
     SizeType rangeCount_;
 
+    static const unsigned kInfinityQuantifier = ~0u;
+
     // For SearchWithAnchoring()
     uint32_t* stateSet_;        // allocated by states_.GetAllocator()
     mutable Stack<Allocator> state0_;

diff --git a/test/unittest/regextest.cpp b/test/unittest/regextest.cpp
@@ -325,6 +325,43 @@ TEST(Regex, QuantifierMinMax3) {
     EXPECT_FALSE(re.Match("abbbbbbd"));
 }
 
+// Issue538
+TEST(Regex, QuantifierMinMax4) {
+    Regex re("a(b|c){0,3}d");
+    ASSERT_TRUE(re.IsValid());
+    EXPECT_TRUE(re.Match("ad"));
+    EXPECT_TRUE(re.Match("abd"));
+    EXPECT_TRUE(re.Match("acd"));
+    EXPECT_TRUE(re.Match("abbd"));
+    EXPECT_TRUE(re.Match("accd"));
+    EXPECT_TRUE(re.Match("abcd"));
+    EXPECT_TRUE(re.Match("abbbd"));
+    EXPECT_TRUE(re.Match("acccd"));
+    EXPECT_FALSE(re.Match("abbbbd"));
+    EXPECT_FALSE(re.Match("add"));
+    EXPECT_FALSE(re.Match("accccd"));
+    EXPECT_FALSE(re.Match("abcbcd"));
+}
+
+// Issue538
+TEST(Regex, QuantifierMinMax5) {
+    Regex re("a(b|c){0,}d");
+    ASSERT_TRUE(re.IsValid());
+    EXPECT_TRUE(re.Match("ad"));
+    EXPECT_TRUE(re.Match("abd"));
+    EXPECT_TRUE(re.Match("acd"));
+    EXPECT_TRUE(re.Match("abbd"));
+    EXPECT_TRUE(re.Match("accd"));
+    EXPECT_TRUE(re.Match("abcd"));
+    EXPECT_TRUE(re.Match("abbbd"));
+    EXPECT_TRUE(re.Match("acccd"));
+    EXPECT_TRUE(re.Match("abbbbd"));
+    EXPECT_TRUE(re.Match("accccd"));
+    EXPECT_TRUE(re.Match("abcbcd"));
+    EXPECT_FALSE(re.Match("add"));
+    EXPECT_FALSE(re.Match("aad"));
+}
+
 #define EURO "\xE2\x82\xAC" // "\xE2\x82\xAC" is UTF-8 sequence of Euro sign U+20AC
 
 TEST(Regex, Unicode) {
@@ -501,6 +538,7 @@ TEST(Regex, Invalid) {
         EXPECT_FALSE(re.IsValid());\
     }
 
+    TEST_INVALID("");
     TEST_INVALID("a|");
     TEST_INVALID("()");
     TEST_INVALID(")");
@@ -517,7 +555,7 @@ TEST(Regex, Invalid) {
     TEST_INVALID("a{0}");
     TEST_INVALID("a{-1}");
     TEST_INVALID("a{}");
-    TEST_INVALID("a{0,}");
+    // TEST_INVALID("a{0,}");   // Support now
     TEST_INVALID("a{,0}");
     TEST_INVALID("a{1,0}");
     TEST_INVALID("a{-1,0}");
@@ -530,4 +568,9 @@ TEST(Regex, Invalid) {
 #undef TEST_INVALID
 }
 
+TEST(Regex, Issue538) {
+    Regex re("^[0-9]+(\\\\.[0-9]+){0,2}");
+    EXPECT_TRUE(re.IsValid());
+}
+
 #undef EURO