Wrote more backreference tests

regex/doc.go

@@ -18,7 +18,7 @@ Single characters:
 	[^abc]			Negated character class - match any character except a, b and c
 	[^a-z]			Negated character range - do not match any character from a to z
 	\[				Match a literal '['. Backslashes can escape any character with special meaning, including another backslash.
-	\452			Match the character with the octal value 452 (up to 3 digits)
+	\0452			Match the character with the octal value 452 (up to 4 digits, first digit must be 0)
 	\xFF			Match the character with the hex value FF (exactly 2 characters)
 	\x{0000FF}		Match the character with the hex value 0000FF (exactly 6 characters)
 	\n				Newline
@@ -93,6 +93,10 @@ Lookarounds:
 	(?<=x)y			Positive lookbehind - Match y if preceded by x
 	(?<!x)y			Negative lookbehind - Match y if NOT preceded by x
 
+Backreferences:
+
+	(xy)\1			Match 'xy' followed by the text most recently captured by group 1 (in this case, 'xy')
+
 Numeric ranges:
 
 	<x-y>			Match any number from x to y (inclusive) (x and y must be positive numbers)
@@ -156,6 +160,7 @@ The following features from [regexp] are (currently) NOT supported:
 The following features are not available in [regexp], but are supported in my engine:
  1. Lookarounds
  2. Numeric ranges
+ 3. Backreferences
 
 I hope to shorten the first list, and expand the second.
 */

regex/matching.go

@@ -319,7 +319,7 @@ func findAllSubmatchHelper(start *nfaState, str []rune, offset int, numGroups in
 				if currentState.contentContains(str, idx, preferLongest) {
 					nextStates = addStateToList(str, idx+1, nextStates, *currentState.next, currentState.threadGroups, nil, preferLongest)
 				}
-			} else if currentState.isBackreference {
+			} else if currentState.isBackreference && currentState.threadGroups[currentState.referredGroup].IsValid() {
 				groupLength := currentState.threadGroups[currentState.referredGroup].EndIdx - currentState.threadGroups[currentState.referredGroup].StartIdx
 				if currentState.threadBackref == groupLength {
 					currentState.threadBackref = 0

regex/re_test.go

@@ -314,10 +314,6 @@ var reTests = []struct {
 	{`\0009`, nil, "\x009", []Group{{0, 2}}},
 	{`\0141`, nil, "a", []Group{{0, 1}}},
 
-	// At this point, the python test suite has a bunch
-	// of backreference tests. Since my engine doesn't
-	// implement backreferences, I've skipped those tests.
-
 	{`*a`, nil, ``, nil},
 	{`(*)b`, nil, ``, nil},
 	{`a**`, nil, ``, nil},
@@ -585,9 +581,33 @@ var groupTests = []struct {
 	{`(.*)c(.*)`, nil, `abcde`, []Match{[]Group{{0, 5}, {0, 2}, {3, 5}}}},
 	{`\((.*), (.*)\)`, nil, `(a, b)`, []Match{[]Group{{0, 6}, {1, 2}, {4, 5}}}},
 
-	// At this point, the python test suite has a bunch
+	// Backreference tests
-	// of backreference tests. Since my engine doesn't
+	{`(abc)\1`, nil, `abcabc`, []Match{[]Group{{0, 6}, {0, 3}}}},
-	// implement backreferences, I've skipped those tests.
+	{`([a-c]+)\1`, nil, `abcabc`, []Match{[]Group{{0, 6}, {0, 3}}}},
+	{`([a-c]*)\1`, nil, `abcabc`, []Match{[]Group{{0, 6}, {0, 3}}, []Group{{6, 6}, {6, 6}}}},
+	{`^(.+)?B`, nil, `AB`, []Match{[]Group{{0, 2}, {0, 1}}}},
+	{`(a+).\1$`, nil, `aaaaa`, []Match{[]Group{{0, 5}, {0, 2}}}},
+	{`^(a+).\1$`, nil, `aaaa`, []Match{}},
+	{`(a)\1`, nil, `aa`, []Match{[]Group{{0, 2}, {0, 1}}}},
+	{`(a+)\1`, nil, `aa`, []Match{[]Group{{0, 2}, {0, 1}}}},
+	{`(a+)+\1`, nil, `aa`, []Match{[]Group{{0, 2}, {0, 1}}}},
+	{`(a).+\1`, nil, `aba`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`(a)ba*\1`, nil, `aba`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`(aa|a)a\1$`, nil, `aaa`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`(a|aa)a\1$`, nil, `aaa`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`(a+)a\1$`, nil, `aaa`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`([abc]*)\1`, nil, `abcabc`, []Match{[]Group{{0, 6}, {0, 3}}, []Group{{6, 6}, {6, 6}}}},
+	{`(a)(?:b)\1`, nil, `aba`, []Match{[]Group{{0, 3}, {0, 1}}}},
+	{`(a)(?:b)\1`, nil, `abb`, []Match{}},
+	{`(?:a)(b)\1`, nil, `aba`, []Match{}},
+	{`(?:a)(b)\1`, nil, `abb`, []Match{[]Group{{0, 3}, {1, 2}}}},
+	{`(?:(cat)|(dog))\2`, nil, `catdog`, []Match{}},
+	{`(?:a)\1`, nil, `aa`, nil},
+	{`((cat)|(dog)|(cow)|(bat))\4`, nil, `cowcow`, []Match{[]Group{{0, 6}, {0, 3}, {-1, -1}, {-1, -1}, {0, 3}, {-1, -1}}}},
+	{`(a|b)*\1`, nil, `abb`, []Match{[]Group{{0, 3}, {1, 2}}}},
+	{`(a|b)*\1`, nil, `aba`, []Match{}},
+	{`(a|b)*\1`, nil, `bab`, []Match{}},
+	{`(a|b)*\1`, nil, `baa`, []Match{[]Group{{0, 3}, {1, 2}}}},
 
 	{`(a)(b)c|ab`, nil, `ab`, []Match{[]Group{{0, 2}}}},
 	{`(a)+x`, nil, `aaax`, []Match{[]Group{{0, 4}, {2, 3}}}},
@@ -792,23 +812,24 @@ func TestFindSubmatch(t *testing.T) {
 				if test.result != nil {
 					panic(err)
 				}
-			}
+			} else {
-			match, err := regComp.FindSubmatch(test.str)
+				match, err := regComp.FindSubmatch(test.str)
-			if err != nil {
+				if err != nil {
-				if len(test.result) != 0 {
+					if len(test.result) != 0 {
-					t.Errorf("Wanted %v got no match\n", test.result[0])
+						t.Errorf("Wanted %v got no match\n", test.result[0])
-				}
-			} else if len(test.result) == 0 {
-				t.Errorf("Wanted no match got %v\n", match)
-			}
-			for i := range match {
-				if match[i].IsValid() {
-					if test.result[0][i] != match[i] {
-						t.Errorf("Wanted %v	Got %v\n", test.result[0], match)
 					}
-				} else {
+				} else if len(test.result) == 0 {
-					if i < len(test.result) && test.result[0][i].IsValid() {
+					t.Errorf("Wanted no match got %v\n", match)
-						t.Errorf("Wanted %v	Got %v\n", test.result[0], match)
+				}
+				for i := range match {
+					if match[i].IsValid() {
+						if test.result[0][i] != match[i] {
+							t.Errorf("Wanted %v	Got %v\n", test.result[0], match)
+						}
+					} else {
+						if i < len(test.result) && test.result[0][i].IsValid() {
+							t.Errorf("Wanted %v	Got %v\n", test.result[0], match)
+						}
 					}
 				}
 			}
@@ -823,10 +844,22 @@ func TestFindStringSubmatch(t *testing.T) {
 				if test.result != nil {
 					panic(err)
 				}
-			}
+			} else {
-			matchStr := regComp.FindStringSubmatch(test.str)
+				matchStr := regComp.FindStringSubmatch(test.str)
-			if matchStr == nil {
+				if matchStr == nil {
-				if len(test.result) != 0 {
+					if len(test.result) != 0 {
+						expectedStr := funcMap(test.result[0], func(g Group) string {
+							if g.IsValid() {
+								return test.str[g.StartIdx:g.EndIdx]
+							} else {
+								return ""
+							}
+						})
+						t.Errorf("Wanted %v got no match\n", expectedStr)
+					}
+				} else if len(test.result) == 0 {
+					t.Errorf("Wanted no match got %v\n", matchStr)
+				} else {
 					expectedStr := funcMap(test.result[0], func(g Group) string {
 						if g.IsValid() {
 							return test.str[g.StartIdx:g.EndIdx]
@@ -834,26 +867,15 @@ func TestFindStringSubmatch(t *testing.T) {
 							return ""
 						}
 					})
-					t.Errorf("Wanted %v got no match\n", expectedStr)
+					for i, groupStr := range matchStr {
-				}
+						if groupStr == "" {
-			} else if len(test.result) == 0 {
+							if i < len(expectedStr) && expectedStr[i] != "" {
-				t.Errorf("Wanted no match got %v\n", matchStr)
+								t.Errorf("Wanted %v	Got %v\n", expectedStr, matchStr)
-			} else {
+							}
-				expectedStr := funcMap(test.result[0], func(g Group) string {
+						} else {
-					if g.IsValid() {
+							if expectedStr[i] != groupStr {
-						return test.str[g.StartIdx:g.EndIdx]
+								t.Errorf("Wanted %v	Got %v\n", expectedStr, matchStr)
-					} else {
+							}
-						return ""
-					}
-				})
-				for i, groupStr := range matchStr {
-					if groupStr == "" {
-						if i < len(expectedStr) && expectedStr[i] != "" {
-							t.Errorf("Wanted %v	Got %v\n", expectedStr, matchStr)
-						}
-					} else {
-						if expectedStr[i] != groupStr {
-							t.Errorf("Wanted %v	Got %v\n", expectedStr, matchStr)
 						}
 					}
 				}
@@ -870,10 +892,24 @@ func TestFindAllStringSubmatch(t *testing.T) {
 				if test.result != nil {
 					panic(err)
 				}
-			}
+			} else {
-			matchStrs := regComp.FindAllStringSubmatch(test.str)
+				matchStrs := regComp.FindAllStringSubmatch(test.str)
-			if matchStrs == nil {
+				if matchStrs == nil {
-				if len(test.result) != 0 {
+					if len(test.result) != 0 {
+						expectedStrs := funcMap(test.result, func(m Match) []string {
+							return funcMap(m, func(g Group) string {
+								if g.IsValid() {
+									return test.str[g.StartIdx:g.EndIdx]
+								} else {
+									return ""
+								}
+							})
+						})
+						t.Errorf("Wanted %v got no match\n", expectedStrs)
+					}
+				} else if len(test.result) == 0 {
+					t.Errorf("Wanted no match got %v\n", matchStrs)
+				} else {
 					expectedStrs := funcMap(test.result, func(m Match) []string {
 						return funcMap(m, func(g Group) string {
 							if g.IsValid() {
@@ -883,29 +919,16 @@ func TestFindAllStringSubmatch(t *testing.T) {
 							}
 						})
 					})
-					t.Errorf("Wanted %v got no match\n", expectedStrs)
+					for i, matchStr := range matchStrs {
-				}
+						for j, groupStr := range matchStr {
-			} else if len(test.result) == 0 {
+							if groupStr == "" {
-				t.Errorf("Wanted no match got %v\n", matchStrs)
+								if j < len(expectedStrs[i]) && expectedStrs[i][j] != "" {
-			} else {
+									t.Errorf("Wanted %v	Got %v\n", expectedStrs, matchStrs)
-				expectedStrs := funcMap(test.result, func(m Match) []string {
+								}
-					return funcMap(m, func(g Group) string {
+							} else {
-						if g.IsValid() {
+								if expectedStrs[i][j] != groupStr {
-							return test.str[g.StartIdx:g.EndIdx]
+									t.Errorf("Wanted %v	Got %v\n", expectedStrs, matchStrs)
-						} else {
+								}
-							return ""
-						}
-					})
-				})
-				for i, matchStr := range matchStrs {
-					for j, groupStr := range matchStr {
-						if groupStr == "" {
-							if j < len(expectedStrs[i]) && expectedStrs[i][j] != "" {
-								t.Errorf("Wanted %v	Got %v\n", expectedStrs, matchStrs)
-							}
-						} else {
-							if expectedStrs[i][j] != groupStr {
-								t.Errorf("Wanted %v	Got %v\n", expectedStrs, matchStrs)
 							}
 						}
 					}
@@ -923,17 +946,18 @@ func TestFindAllSubmatch(t *testing.T) {
 				if test.result != nil {
 					panic(err)
 				}
-			}
+			} else {
-			matchIndices := regComp.FindAllSubmatch(test.str)
+				matchIndices := regComp.FindAllSubmatch(test.str)
-			for i := range matchIndices {
+				for i := range matchIndices {
-				for j := range matchIndices[i] {
+					for j := range matchIndices[i] {
-					if matchIndices[i][j].IsValid() {
+						if matchIndices[i][j].IsValid() {
-						if test.result[i][j] != matchIndices[i][j] {
+							if test.result[i][j] != matchIndices[i][j] {
-							t.Errorf("Wanted %v	Got %v\n", test.result, matchIndices)
+								t.Errorf("Wanted %v	Got %v\n", test.result, matchIndices)
-						}
+							}
-					} else {
+						} else {
-						if i < len(test.result) && j < len(test.result[i]) && test.result[i][j].IsValid() {
+							if i < len(test.result) && j < len(test.result[i]) && test.result[i][j].IsValid() {
-							t.Errorf("Wanted %v	Got %v\n", test.result, matchIndices)
+								t.Errorf("Wanted %v	Got %v\n", test.result, matchIndices)
+							}
 						}
 					}
 				}