Function analyze

index/regexp.go:420–540

analyze returns the regexpInfo for the regexp re.

(re *syntax.Regexp)

// analyze returns the regexpInfo for the regexp re.
func analyze(re *syntax.Regexp) (ret regexpInfo) {
	//println("analyze", re.String())
	//defer func() { println("->", ret.String()) }()
	var info regexpInfo
	switch re.Op {
	case syntax.OpNoMatch:
		return noMatch()

	case syntax.OpEmptyMatch,
		syntax.OpBeginLine, syntax.OpEndLine,
		syntax.OpBeginText, syntax.OpEndText,
		syntax.OpWordBoundary, syntax.OpNoWordBoundary:
		return emptyString()

	case syntax.OpLiteral:
		if re.Flags&syntax.FoldCase != 0 {
			switch len(re.Rune) {
			case 0:
				return emptyString()
			case 1:
				// Single-letter case-folded string:
				// rewrite into char class and analyze.
				re1 := &syntax.Regexp{
					Op: syntax.OpCharClass,
				}
				re1.Rune = re1.Rune0[:0]
				r0 := re.Rune[0]
				re1.Rune = append(re1.Rune, r0, r0)
				for r1 := unicode.SimpleFold(r0); r1 != r0; r1 = unicode.SimpleFold(r1) {
					re1.Rune = append(re1.Rune, r1, r1)
				}
				info = analyze(re1)
				return info
			}
			// Multi-letter case-folded string:
			// treat as concatenation of single-letter case-folded strings.
			re1 := &syntax.Regexp{
				Op:    syntax.OpLiteral,
				Flags: syntax.FoldCase,
			}
			info = emptyString()
			for i := range re.Rune {
				re1.Rune = re.Rune[i : i+1]
				info = concat(info, analyze(re1))
			}
			return info
		}
		info.exact = stringSet{string(re.Rune)}
		info.match = allQuery

	case syntax.OpAnyCharNotNL, syntax.OpAnyChar:
		return anyChar()

	case syntax.OpCapture:
		return analyze(re.Sub[0])

	case syntax.OpConcat:
		return fold(concat, re.Sub, emptyString())