Started working on '-m num' flag : print the <num>th match

Instead, I tweaked the rest of the matching function, so that a special
check isn't necessary. If we are trying to match at the end of a string,
we skip any of the actual matching and proceed straight to finding
0-length matches.

This change was made because, with the special case, capturing groups
weren't getting updated if we had an end-of-string match.

main.go

@@ -51,6 +51,9 @@ func shuntingYard(re string) []postfixNode {
 	// Eventually, I might be able to add it into the main parsing loop, to reduce the time
 	// complexity.
 	// A numeric range has the syntax: <num1-num2>. Ir matches all numbers in this range.
+	//
+	// Also check for non-capturing groups. The LPAREN of a non-capturing group looks like this: '(?:'
+	// I take this out, and put in a special character - NONCAPLPAREN_CHAR.
 	for i := 0; i < len(re_runes_orig); i++ {
 		c := re_runes_orig[i]
 		if c == '<' && (i == 0 || (re_runes_orig[i-1] != '\\' && re_runes_orig[i-1] != '?')) {
@@ -82,7 +85,9 @@ func shuntingYard(re string) []postfixNode {
 			fmt.Sscanf(tmpStr, "%d-%d", &rangeStart, &rangeEnd)
 			regex := range2regex(rangeStart, rangeEnd)
 			re_runes = append(re_runes, []rune(regex)...)
-
+		} else if c == '(' && i < len(re_runes_orig)-2 && re_runes_orig[i+1] == '?' && re_runes_orig[i+2] == ':' {
+			re_runes = append(re_runes, NONCAPLPAREN_CHAR)
+			i += 2
 		} else {
 			re_runes = append(re_runes, c)
 		}
@@ -148,7 +153,11 @@ func shuntingYard(re string) []postfixNode {
 			}
 			re_postfix = append(re_postfix, re_runes[i]) // Append closing brace
 		}
-		if i < len(re_runes) && re_runes[i] == '(' && (i == 0 || re_runes[i-1] != '\\') && (i < len(re_runes)-1 && re_runes[i+1] == '?') { // Unescaped open parentheses followed by question mark = lokaround. Don't mess with it.
+		if i < len(re_runes)-3 && string(re_runes[i+1:i+4]) == "(?:" { // Non-capturing lparen
+			re_postfix = append(re_postfix, NONCAPLPAREN_CHAR)
+			i += 3
+		}
+		if i < len(re_runes) && re_runes[i] == '(' && (i == 0 || re_runes[i-1] != '\\') && (i < len(re_runes)-2 && re_runes[i+1] == '?' && slices.Contains([]rune{'=', '!', '<'}, re_runes[i+2])) { // Unescaped open parentheses followed by question mark then '<', '!' or '=' => lokaround. Don't mess with it.
 			i++ // Step inside
 			if i == len(re_runes)-1 || (re_runes[i+1] != '=' && re_runes[i+1] != '!' && re_runes[i+1] != '<') {
 				panic("Invalid regex. Lookaround intended?")
@@ -174,7 +183,7 @@ func shuntingYard(re string) []postfixNode {
 			}
 			continue
 		}
-		if i < len(re_runes) && (re_runes[i] != '(' && re_runes[i] != '|' && re_runes[i] != '\\') || (i > 0 && re_runes[i-1] == '\\') { // Every character should be concatenated if it is escaped
+		if i < len(re_runes) && (re_runes[i] != '(' && re_runes[i] != NONCAPLPAREN_CHAR && re_runes[i] != '|' && re_runes[i] != '\\') || (i > 0 && re_runes[i-1] == '\\') { // Every character should be concatenated if it is escaped
 			if i < len(re_runes)-1 {
 				if re_runes[i+1] != '|' && re_runes[i+1] != '*' && re_runes[i+1] != '+' && re_runes[i+1] != '?' && re_runes[i+1] != ')' && re_runes[i+1] != '{' {
 					re_postfix = append(re_postfix, CONCAT)
@@ -197,7 +206,7 @@ func shuntingYard(re string) []postfixNode {
 			b. If not, keep popping from opStack (and appending to outQueue) until:
 				i. opStack is empty, OR
 				ii. current character has greater priority than top of opStack
-		3. If current character is '(', push to opStack
+		3. If current character is '(' or NONCAPLPAREN_CHAR, push to opStack
 		4. If current character is ')', pop from opStack (and append to outQueue) until '(' is found. Discard parantheses.
 		5. If current character is '[', find all the characters until ']', then create a postfixNode containing all these contents. Add this node to outQueue.
 		6. If current character is '{', find the appropriate numeric specifier (range start, range end). Apply the range to the postfixNode at the end of outQueue.
@@ -382,32 +391,35 @@ func shuntingYard(re string) []postfixNode {
 			}
 
 			idx := len(outQueue) - 1
-			// Get the most recently added non-paren node
+			// Get the last added node
-			for node := outQueue[idx]; idx >= 0 && (node.nodetype == RPAREN || node.nodetype == LPAREN); node = outQueue[idx] {
+			if idx < 0 || outQueue[idx].nodetype == LPAREN {
-				idx--
-			}
-			if idx < 0 {
 				panic("Numeric specifier with no content.")
 			}
 			outQueue[idx].startReps = startRangeNum
 			outQueue[idx].endReps = endRangeNum
 		}
-		if c == '(' {
+		if c == '(' || c == NONCAPLPAREN_CHAR {
 			opStack = append(opStack, c)
-			outQueue = append(outQueue, newPostfixNode(c))
+			if c == '(' { // We only push _capturing_ group parentheses to outQueue
+				outQueue = append(outQueue, newPostfixNode(c))
+			}
 			numOpenParens++
 		}
 		if c == ')' {
-			// Keep popping from opStack until we encounter an opening parantheses. Panic if we reach the end of the stack.
+			// Keep popping from opStack until we encounter an opening parantheses or a NONCAPLPAREN_CHAR. Panic if we reach the end of the stack.
-			for val, err := peek(opStack); val != '('; val, err = peek(opStack) {
+			var val rune
+			var err error
+			for val, err = peek(opStack); val != '(' && val != NONCAPLPAREN_CHAR; val, err = peek(opStack) {
 				if err != nil {
 					panic("ERROR: Imbalanced parantheses.")
 				}
 				to_append := mustPop(&opStack)
 				outQueue = append(outQueue, newPostfixNode(to_append))
 			}
-			_ = mustPop(&opStack)                            // Get rid of opening parentheses
+			_ = mustPop(&opStack) // Get rid of opening parentheses
-			outQueue = append(outQueue, newPostfixNode(')')) // Add closing parentheses
+			if val == '(' {       // Whatever was inside the parentheses was a _capturing_ group, so we append the closing parentheses as well
+				outQueue = append(outQueue, newPostfixNode(')')) // Add closing parentheses
+			}
 			numOpenParens--
 		}
 	}
@@ -594,7 +606,8 @@ func main() {
 	multiLineFlag := flag.Bool("t", false, "Multi-line mode. Treats newline just like any character.")
 	printMatchesFlag := flag.Bool("p", false, "Prints start and end index of each match. Can only be used with '-t' for multi-line mode.")
 	caseInsensitiveFlag = flag.Bool("i", false, "Case-insensitive. Disregard the case of all characters.")
-	substituteText := flag.String("s", "", "Substitute the contents  of each match with the given string. Overrides -o and -v")
+	matchNum := flag.Int("m", 0, "Print the match with the given index. Eg. -m 3 prints the third match.")
+	substituteText := flag.String("s", "", "Substitute the contents of each match with the given string. Overrides -o and -v")
 	flag.Parse()
 
 	// In multi-line mode, 'dot' metacharacter also matches newline
@@ -607,14 +620,23 @@ func main() {
 	if *onlyFlag {
 		*lineFlag = false
 	}
-	// Check if substitute text has been enabled
+	// Check if substitute and matchNum flags have been enabled
 	substituteFlagEnabled := false
+	matchNumFlagEnabled := false
 	flag.Visit(func(f *flag.Flag) {
 		if f.Name == "s" {
 			substituteFlagEnabled = true
 		}
+		if f.Name == "m" {
+			matchNumFlagEnabled = true
+		}
 	})
 
+	// Validate matchNumFlag - must be positive integer
+	if matchNumFlagEnabled && *matchNum < 1 {
+		panic("Invalid match number to print.")
+	}
+
 	// Process:
 	// 1. Convert regex into postfix notation (Shunting-Yard algorithm)
 	// 		a. Add explicit concatenation operators to facilitate this

matching.go

@@ -60,8 +60,7 @@ func (g Group) isValid() bool {
 // takeZeroState takes the 0-state (if such a transition exists) for all states in the
 // given slice. It returns the resulting states. If any of the resulting states is a 0-state,
 // the second ret val is true.
-// The third ret val is a list of all the group numbers of all the opening parentheses we crossed,
+// If a state begins or ends a capturing group, its 'thread' is updated to contain the correct index.
-// and the fourth is a list of all the closing parentheses we passed
 func takeZeroState(states []*State, numGroups int, idx int) (rtv []*State, isZero bool) {
 	for _, state := range states {
 		if len(state.transitions[EPSILON]) > 0 {
@@ -94,11 +93,7 @@ func takeZeroState(states []*State, numGroups int, idx int) (rtv []*State, isZer
 // from any of the given states, given the string and our position in it.
 // It uses the same algorithm to find zero-states as the one inside the loop,
 // so I should probably put it in a function.
-// It also returns all the capturing groups that both begin and end at the current index.
+func zeroMatchPossible(str []rune, idx int, numGroups int, states ...*State) bool {
-// This is because, by definition, zero-states don't move forward in the string.
-func zeroMatchPossible(str []rune, idx int, numGroups int, states ...*State) (bool, []int, []int) {
-	allOpenParenGroups := make([]int, 0)
-	allCloseParenGroups := make([]int, 0)
 	zeroStates, isZero := takeZeroState(states, numGroups, idx)
 	tempstates := make([]*State, 0, len(zeroStates)+len(states))
 	tempstates = append(tempstates, states...)
@@ -113,10 +108,10 @@ func zeroMatchPossible(str []rune, idx int, numGroups int, states ...*State) (bo
 	}
 	for _, state := range tempstates {
 		if state.isEmpty && (state.assert == NONE || state.checkAssertion(str, idx)) && state.isLast {
-			return true, allOpenParenGroups, allCloseParenGroups
+			return true
 		}
 	}
-	return false, allOpenParenGroups, allCloseParenGroups
+	return false
 }
 
 // Prunes the slice by removing overlapping indices.
@@ -173,28 +168,6 @@ func findAllMatchesHelper(start *State, str []rune, offset int, numGroups int) (
 		// The second value here shouldn't be used, because we should exit when the third return value is > than len(str)
 		return false, []Group{}, offset
 	}
-	// 'Base case' - if we are at the end of the string, check if we can add a zero-length match
-	if offset == len(str) {
-		// Get all zero-state matches. If we can get to a zero-state without matching anything, we
-		// can add a zero-length match. This is all true only if the start state itself matches nothing.
-		if start.isEmpty {
-			to_return := newMatch(numGroups + 1)
-			if start.groupBegin {
-				to_return[start.groupNum].startIdx = offset
-			}
-			if ok, openGrps, closeGrps := zeroMatchPossible(str, offset, numGroups, start); ok {
-				for _, gIdx := range openGrps {
-					to_return[gIdx].startIdx = offset
-				}
-				for _, gIdx := range closeGrps {
-					to_return[gIdx].endIdx = offset
-				}
-				to_return[0] = Group{offset, offset}
-				return true, to_return, offset + 1
-			}
-		}
-		return false, []Group{}, offset + 1
-	}
 
 	// Hold a list of match indices for the current run. When we
 	// can no longer find a match, the match with the largest range is
@@ -329,7 +302,7 @@ func findAllMatchesHelper(start *State, str []rune, offset int, numGroups int) (
 
 		// Check if we can find a zero-length match
 		if foundPath == false {
-			if ok, _, _ := zeroMatchPossible(str, i, numGroups, currentStates...); ok {
+			if ok := zeroMatchPossible(str, i, numGroups, currentStates...); ok {
 				if tempIndices[0].isValid() == false {
 					tempIndices[0] = Group{startIdx, startIdx}
 				}
@@ -375,7 +348,7 @@ func findAllMatchesHelper(start *State, str []rune, offset int, numGroups int) (
 	for _, state := range currentStates {
 		// Only add the match if the start index is in bounds. If the state has an assertion,
 		// make sure the assertion checks out.
-		if state.isLast && startIdx < len(str) {
+		if state.isLast && i <= len(str) {
 			if state.assert == NONE || state.checkAssertion(str, i) {
 				for j := 1; j < numGroups+1; j++ {
 					tempIndices[j] = state.threadGroups[j]

misc.go

@@ -13,6 +13,7 @@ var RBRACKET rune = 0xF0001
 var ANY_CHAR rune = 0xF0002    // Represents any character - used for states where the allChars flag is on.
 var LPAREN_CHAR rune = 0xF0003 // Parentheses in regex are concatenated with this - it acts as a pseudio-parentheses
 var RPAREN_CHAR rune = 0xF0004
+var NONCAPLPAREN_CHAR rune = 0xF0005 // Represents a non-capturing group's LPAREN
 
 // Returns true if str[idx] and str[idx-1] are separated by a word boundary.
 func isWordBoundary(str []rune, idx int) bool {
@@ -26,7 +27,7 @@ func isWordBoundary(str []rune, idx int) bool {
 
 func isNormalChar(c rune) bool {
 	specialChars := []rune(`?*\^${}()+|[].~<>`)
-	specialChars = append(specialChars, LBRACKET, RBRACKET)
+	specialChars = append(specialChars, LBRACKET, RBRACKET, NONCAPLPAREN_CHAR)
 	return !slices.Contains(specialChars, c)
 }
 

nfa.go

@@ -28,7 +28,6 @@ type State struct {
 	transitions                map[int][]*State // Transitions to different states (maps a character (int representation) to a _list of states. This is useful if one character can lead multiple states eg. ab|aa)
 	isKleene                   bool             // Identifies whether current node is a 0-state representing Kleene star
 	assert                     assertType       // Type of assertion of current node - NONE means that the node doesn't assert anything
-	zeroMatchFound             bool             // Whether or not the state has been used for a zero-length match - only relevant for zero states
 	allChars                   bool             // Whether or not the state represents all characters (eg. a 'dot' metacharacter). A 'dot' node doesn't store any contents directly, as it would take up too much space
 	except                     []rune           // Only valid if allChars is true - match all characters _except_ the ones in this block. Useful for inverting character classes.
 	lookaroundRegex            string           // Only for lookaround states - Contents of the regex that the lookaround state holds
@@ -37,7 +36,9 @@ type State struct {
 	groupBegin                 bool             // Whether or not the node starts a capturing group
 	groupEnd                   bool             // Whether or not the node ends a capturing group
 	groupNum                   int              // Which capturing group the node starts / ends
-	threadGroups               []Group          // Assuming that a state is part of a 'thread' in the matching process, this array stores the indices of capturing groups in the current thread. As matches are found for this state, its groups will be copied over.
+	// The following properties depend on the current match - I should think about resetting them for every match.
+	zeroMatchFound bool    // Whether or not the state has been used for a zero-length match - only relevant for zero states
+	threadGroups   []Group // Assuming that a state is part of a 'thread' in the matching process, this array stores the indices of capturing groups in the current thread. As matches are found for this state, its groups will be copied over.
 }
 
 // Clones the NFA starting from the given state.
@@ -118,15 +119,21 @@ func (s State) checkAssertion(str []rune, idx int) bool {
 	if s.isLookaround() {
 		// The process here is simple:
 		// 		1. Compile the regex stored in the state's contents.
-		// 		2. Run it on the test string.
+		// 		2. Run it on a subset of the test string, that ends after the current index in the string
 		// 		3. Based on the kind of lookaround (and the indices we get), determine what action to take.
 		startState := s.lookaroundNFA
-		matchIndices := findAllMatches(startState, str, startState.lookaroundNumCaptureGroups)
+		var strToMatch []rune
+		if s.assert == PLA || s.assert == NLA {
+			strToMatch = str[idx:]
+		} else {
+			strToMatch = str[:idx]
+		}
+		matchIndices := findAllMatches(startState, strToMatch, startState.lookaroundNumCaptureGroups)
 
 		numMatchesFound := 0
 		for _, matchIdx := range matchIndices {
-			if s.assert == PLA || s.assert == NLA { // Lookahead - return true (or false) if at least one match starts at the current index
+			if s.assert == PLA || s.assert == NLA { // Lookahead - return true (or false) if at least one match starts at 0. Zero is used because the test-string _starts_ from idx.
-				if matchIdx[0].startIdx == idx {
+				if matchIdx[0].startIdx == 0 {
 					numMatchesFound++
 				}
 			}

re_test.go

@@ -149,14 +149,13 @@ var reTests = []struct {
 	{"^((3[7-9])|([4-9][0-9])|([1-9][0-9][0-9])|(1000))$", "400", []Group{{0, 3}}},
 	{"^((3[7-9])|([4-9][0-9])|([1-9][0-9][0-9])|(1000))$", "4000", []Group{}},
 
-	// Todo - add lookaround tests
+	// Lookaround tests
 	{"(?<=bo)y", "boy", []Group{{2, 3}}},
 	{"bo(?=y)", "boy", []Group{{0, 2}}},
-
-	// Todo - add numeric range tests
-	// Todo - add capturing group tests
 	{"(?<=f)f+(?=f)", "fffff", []Group{{1, 4}}},
 	{"(?<=f)f+(?=f)", "fffffa", []Group{{1, 4}}},
+
+	// Todo - add numeric range tests
 }
 
 var groupTests = []struct {
@@ -179,6 +178,10 @@ var groupTests = []struct {
 	{"(aaa)|(aaaa)", "aaaa", []Match{[]Group{{0, 4}, {-1, -1}, {0, 4}}}},
 	{"(aaaa)|(aaa)", "aaaa", []Match{[]Group{{0, 4}, {0, 4}, {-1, -1}}}},
 	{"(a)|(aa)", "aa", []Match{[]Group{{0, 2}, {-1, -1}, {0, 2}}}},
+	{"(a?)a?", "b", []Match{[]Group{{0, 0}, {0, 0}}, []Group{{1, 1}, {1, 1}}}},
+	{"(a?)a?", "ab", []Match{[]Group{{0, 1}, {0, 1}}, []Group{{1, 1}, {1, 1}}, []Group{{2, 2}, {2, 2}}}},
+	{"(a?)a?", "aa", []Match{[]Group{{0, 2}, {0, 1}}, []Group{{2, 2}, {2, 2}}}},
+	{"a((b.d){3})", "abfdbhdbid", []Match{[]Group{{0, 10}, {1, 10}, {7, 10}}}},
 }
 
 func TestFindAllMatches(t *testing.T) {

todo.txt

@@ -3,3 +3,4 @@
 3. Fix adding concatenation operators in shunting-yard function (very janky, compares against operators individually)
 Ideas for flags:
     -m <num> : Print <num>th match (-m 1 = first match, -m 2 = second match)
+    -g <num> : Print the <num>th group