kleingrep/regex/matching.go

package regex

import (
	"fmt"
	"strconv"
	"unicode"
)

// A Match represents a match found by the regex in a given string.
// It is represented as a list of groups, where the nth element contains
// the contents of the nth capturing group. Note that the group may not be valid
// (see [Group.IsValid]). The element at index 0 is known
// as the 0-group, and represents the contents of the entire match.
//
// See [Reg.FindSubmatch] for an example.
type Match []Group

// a Group represents a capturing group. It contains the start and index of the group.
type Group struct {
	StartIdx int
	EndIdx   int
}

func newMatch(size int) Match {
	toRet := make([]Group, size)
	for i := range toRet {
		toRet[i].StartIdx = -1
		toRet[i].EndIdx = -1
	}
	return toRet
}

// Returns a string containing the indices of all (valid) groups in the match
func (m Match) String() string {
	var toRet string
	for i, g := range m {
		if g.IsValid() {
			toRet += fmt.Sprintf("Group %d\n", i)
			toRet += g.String()
			toRet += "\n"
		}
	}
	return toRet
}

// String converts the Group into a string representation.
func (idx Group) String() string {
	return fmt.Sprintf("%d\t%d", idx.StartIdx, idx.EndIdx)
}

// IsValid returns whether a group is valid (ie. whether it matched any text). It
// simply ensures that both indices of the group are >= 0.
func (g Group) IsValid() bool {
	return g.StartIdx >= 0 && g.EndIdx >= 0
}

// Simple function, makes it easier to map this over a list of matches
func getZeroGroup(m Match) Group {
	return m[0]
}

func copyThread(to *nfaState, from nfaState) {
	to.threadGroups = append([]Group{}, from.threadGroups...)
}

// Find returns the 0-group of the leftmost match of the regex in the given string.
// An error value != nil indicates that no match was found.
func (re Reg) Find(str string) (Group, error) {
	match, err := re.FindNthMatch(str, 1)
	if err != nil {
		return Group{}, fmt.Errorf("no matches found")
	}
	return getZeroGroup(match), nil
}

// Match returns a boolean value, indicating whether the regex found a match in the given string.
func (re Reg) Match(str string) bool {
	_, err := re.Find(str)
	return err == nil
}

// CompileMatch compiles expr and returns true if str contains a match of the expression.
// It is equivalent to [regexp.Match].
// An optional list of flags may be provided (see [ReFlag]).
// It returns an error (!= nil) if there was an error compiling the expression.
func CompileMatch(expr string, str string, flags ...ReFlag) (bool, error) {
	re, err := Compile(expr, flags...)
	if err != nil {
		return false, err
	}
	return re.Match(str), nil
}

// FindAll returns a slice containing all the 0-groups of the regex in the given string.
// A 0-group represents the match without any submatches.
func (re Reg) FindAll(str string) []Group {
	indices := re.FindAllSubmatch(str)
	zeroGroups := funcMap(indices, getZeroGroup)
	return zeroGroups
}

// FindString returns the text of the leftmost match of the regex in the given string.
// The return value will be an empty string in two situations:
//  1. No match was found
//  2. The match was an empty string
func (re Reg) FindString(str string) string {
	match, err := re.FindNthMatch(str, 1)
	if err != nil {
		return ""
	}
	zeroGroup := getZeroGroup(match)
	return str[zeroGroup.StartIdx:zeroGroup.EndIdx]
}

// FindSubmatch returns the leftmost match of the regex in the given string, including
// the submatches matched by capturing groups. The returned [Match] will always contain the same
// number of groups. The validity of a group (whether or not it matched anything) can be determined with
// [Group.IsValid], or by checking that both indices of the group are >= 0.
// The second-return value is nil if no match was found.
func (re Reg) FindSubmatch(str string) (Match, error) {
	match, err := re.FindNthMatch(str, 1)
	if err != nil {
		return Match{}, fmt.Errorf("no match found")
	} else {
		return match, nil
	}
}

// FindStringSubmatch is the 'string' version of [FindSubmatch]. It returns a slice of strings,
// where the string at index i contains the text matched by the i-th capturing group.
// The 0-th index represents the entire match.
// An empty string at index n could mean:
// ,
//  1. Group n did not find a match
//  2. Group n found a zero-length match
//
// A return value of nil indicates no match.
func (re Reg) FindStringSubmatch(str string) []string {
	matchStr := make([]string, re.numGroups+1)
	match, err := re.FindSubmatch(str)
	if err != nil {
		return nil
	}
	nonEmptyMatchFound := false
	for i := range match {
		if match[i].IsValid() {
			matchStr[i] = str[match[i].StartIdx:match[i].EndIdx]
			nonEmptyMatchFound = true
		} else {
			matchStr[i] = ""
		}
	}
	if nonEmptyMatchFound == false {
		return nil
	}
	return matchStr
}

// FindAllString is the 'all' version of [FindString].
// It returns a slice of strings containing the text of all matches of
// the regex in the given string.
func (re Reg) FindAllString(str string) []string {
	zerogroups := re.FindAll(str)
	matchStrs := funcMap(zerogroups, func(g Group) string {
		return str[g.StartIdx:g.EndIdx]
	})
	return matchStrs
}

// FindNthMatch return the 'n'th match of the regex in the given string.
// It returns an error (!= nil) if there are fewer than 'n' matches in the string.
func (re Reg) FindNthMatch(str string, n int) (Match, error) {
	idx := 0
	matchNum := 0
	str_runes := []rune(str)
	var matchFound bool
	var matchIdx Match
	for idx <= len(str_runes) {
		matchFound, matchIdx, idx = findAllSubmatchHelper(re.start, str_runes, idx, re.numGroups, re.preferLongest)
		if matchFound {
			matchNum++
		}
		if matchNum == n {
			return matchIdx, nil
		}
	}
	// We haven't found the nth match after scanning the string - Return an error
	return nil, fmt.Errorf("invalid match index - too few matches found")
}

// FindAllSubmatch returns a slice of matches in the given string.
func (re Reg) FindAllSubmatch(str string) []Match {
	idx := 0
	str_runes := []rune(str)
	var matchFound bool
	var matchIdx Match
	indices := make([]Match, 0)
	for idx <= len(str_runes) {
		matchFound, matchIdx, idx = findAllSubmatchHelper(re.start, str_runes, idx, re.numGroups, re.preferLongest)
		if matchFound {
			indices = append(indices, matchIdx)
		}
	}

	return indices
}

func addStateToList(str []rune, idx int, list []nfaState, state nfaState, threadGroups []Group, visited []nfaState, preferLongest bool) []nfaState {
	if stateExists(list, state) || stateExists(visited, state) {
		return list
	}
	visited = append(visited, state)

	if state.isKleene || state.isQuestion {
		copyThread(state.splitState, state)
		list = addStateToList(str, idx, list, *state.splitState, threadGroups, visited, preferLongest)
		copyThread(state.next, state)
		list = addStateToList(str, idx, list, *state.next, threadGroups, visited, preferLongest)
		return list
	}
	if state.isAlternation {
		copyThread(state.next, state)
		list = addStateToList(str, idx, list, *state.next, threadGroups, visited, preferLongest)
		copyThread(state.splitState, state)
		list = addStateToList(str, idx, list, *state.splitState, threadGroups, visited, preferLongest)
		return list
	}
	state.threadGroups = append([]Group{}, threadGroups...)
	if state.assert != noneAssert {
		if state.checkAssertion(str, idx, preferLongest) {
			copyThread(state.next, state)
			return addStateToList(str, idx, list, *state.next, state.threadGroups, visited, preferLongest)
		}
	}
	if state.groupBegin {
		state.threadGroups[state.groupNum].StartIdx = idx
		return addStateToList(str, idx, list, *state.next, state.threadGroups, visited, preferLongest)
	}
	if state.groupEnd {
		state.threadGroups[state.groupNum].EndIdx = idx
		return addStateToList(str, idx, list, *state.next, state.threadGroups, visited, preferLongest)
	}
	return append(list, state)

}

// Helper for FindAllMatches. Returns whether it found a match, the
// first Match it finds, and how far it got into the string ie. where
// the next search should start from.
func findAllSubmatchHelper(start *nfaState, str []rune, offset int, numGroups int, preferLongest bool) (bool, Match, int) {
	// Base case - exit if offset exceeds string's length
	if offset > len(str) {
		// 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
	}
	resetThreads(start)

	currentStates := make([]nfaState, 0)
	nextStates := make([]nfaState, 0)
	i := offset // Index in string

	// If the first state is an assertion, makes sure the assertion
	// is true before we do _anything_ else.
	if start.assert != noneAssert {
		if start.checkAssertion(str, offset, preferLongest) == false {
			i++
			return false, []Group{}, i
		}
	}

	start.threadGroups = newMatch(numGroups + 1)
	start.threadGroups[0].StartIdx = i
	currentStates = addStateToList(str, i, currentStates, *start, start.threadGroups, nil, preferLongest)
	var match Match = nil
	for idx := i; idx <= len(str); idx++ {
		if len(currentStates) == 0 {
			break
		}
		for currentStateIdx := 0; currentStateIdx < len(currentStates); currentStateIdx++ {
			currentState := currentStates[currentStateIdx]

			if currentState.threadGroups == nil {
				currentState.threadGroups = newMatch(numGroups + 1)
				currentState.threadGroups[0].StartIdx = idx
			}

			if currentState.isLast {
				currentState.threadGroups[0].EndIdx = idx
				match = append([]Group{}, currentState.threadGroups...)
				if !preferLongest {
					break
				}
			} else if !currentState.isAlternation && !currentState.isKleene && !currentState.isQuestion && !currentState.groupBegin && !currentState.groupEnd && currentState.assert == noneAssert { // Normal character
				if currentState.contentContains(str, idx, preferLongest) {
					nextStates = addStateToList(str, idx+1, nextStates, *currentState.next, currentState.threadGroups, nil, preferLongest)
				}
			}
		}
		currentStates = append([]nfaState{}, nextStates...)
		nextStates = nil
	}
	if match != nil {
		if offset == match[0].EndIdx {
			return true, match, match[0].EndIdx + 1
		}
		return true, match, match[0].EndIdx
	}
	return false, []Group{}, i + 1
}

// Expand appends template to dst, expanding any variables in template to the relevant capturing group.
//
// A variable is of the form '$n', where 'n' is a number. It will be replaced by the contents of the n-th capturing group.
// To insert a literal $, do not put a number after it. Alternatively, you can use $$.
// src is the input string, and match must be the result of [Reg.FindSubmatch].
func (re Reg) Expand(dst string, template string, src string, match Match) string {
	templateRuneSlc := []rune(template)
	srcRuneSlc := []rune(src)
	i := 0
	for i < len(templateRuneSlc) {
		c := templateRuneSlc[i]
		if c == '$' {
			i += 1
			// The dollar sign is the last character of the string, or it is proceeded by another dollar sign
			if i >= len(templateRuneSlc) || templateRuneSlc[i] == '$' {
				dst += "$"
				i++
			} else {
				numStr := ""
				for unicode.IsDigit(templateRuneSlc[i]) {
					numStr += string(templateRuneSlc[i])
					i++
				}
				if numStr == "" {
					dst += "$"
				} else {
					num, _ := strconv.Atoi(numStr)
					if num < len(match) {
						dst += string(srcRuneSlc[match[num].StartIdx:match[num].EndIdx])
					} else {
						dst += "$" + numStr
					}
				}
			}
		} else {
			dst += string(c)
			i++
		}
	}
	return dst
}

// LiteralPrefix returns a string that must begin any match of the given regular expression.
// The second return value is true if the string comprises the entire expression.
func (re Reg) LiteralPrefix() (prefix string, complete bool) {
	state := re.start
	if state.assert != noneAssert {
		state = state.next
	}
	for !(state.isLast) && (!state.isAlternation) && len(state.content) == 1 && state.assert == noneAssert {
		if state.groupBegin || state.groupEnd {
			state = state.next
			continue
		}
		prefix += string(rune(state.content[0]))
		state = state.next
	}
	if state.isLast {
		complete = true
	} else {
		complete = false
	}
	return prefix, complete
}