218 lines
9.3 KiB
Go
218 lines
9.3 KiB
Go
// Copyright (c) 2024-2026 Celestino Amoroso (celestino.amoroso@gmail.com).
|
|
// All rights reserved.
|
|
|
|
// Symbol.go
|
|
package scan
|
|
|
|
import (
|
|
"strings"
|
|
)
|
|
|
|
var symbolMap map[Symbol]symbolSpec
|
|
|
|
type SymbolClass int16
|
|
|
|
const (
|
|
SymClassOperator SymbolClass = iota
|
|
SymClassCommand
|
|
SymClassIdentifier
|
|
SymClassDelimiter
|
|
SymClassParenthesis
|
|
SymClassDeclaration
|
|
SymClassValue
|
|
SymClassOther
|
|
)
|
|
|
|
type symbolSpec struct {
|
|
repr string
|
|
kind SymbolClass
|
|
opType TermPosition
|
|
}
|
|
|
|
func init() {
|
|
|
|
symbolMap = map[Symbol]symbolSpec{
|
|
SymUnknown: {"<unknown>", SymClassOther, PosLeaf}, // -1: Unknown symbol
|
|
SymNone: {"<null>", SymClassOther, PosLeaf}, // 0: Null value for variable of type symbol
|
|
SymError: {"<error>", SymClassOther, PosLeaf}, // 1: Error reading from stream
|
|
SymEos: {"<eos>", SymClassOther, PosLeaf}, // 2: End of stream
|
|
SymMinus: {"-", SymClassOperator, PosInfix}, // 3: '-'
|
|
SymMinusEqual: {"-=", SymClassOperator, PosInfix}, // 4: '-='
|
|
SymDoubleMinus: {"--", SymClassOperator, PosPostfix}, // 5: '--'
|
|
SymPlus: {"+", SymClassOperator, PosInfix}, // 6: '+'
|
|
SymPlusEqual: {"+=", SymClassOperator, PosInfix}, // 7: '+='
|
|
SymDoublePlus: {"++", SymClassOperator, PosPostfix}, // 8: '++'
|
|
SymStar: {"*", SymClassOperator, PosInfix}, // 9: '*'
|
|
SymDoubleStar: {"**", SymClassOperator, PosInfix}, // 10: '**'
|
|
SymSlash: {"/", SymClassOperator, PosInfix}, // 11: '/'
|
|
SymBackSlash: {"\\", SymClassOperator, PosLeaf}, // 12: '\'
|
|
SymVertBar: {"|", SymClassOperator, PosInfix}, // 13: '|'
|
|
SymDoubleVertBar: {"||", SymClassOperator, PosInfix}, // 14: '||'
|
|
SymComma: {",", SymClassOperator, PosInfix}, // 15: ','
|
|
SymColon: {":", SymClassOperator, PosInfix}, // 16: ':'
|
|
SymSemiColon: {";", SymClassOperator, PosInfix}, // 17: ';'
|
|
SymDot: {".", SymClassOperator, PosInfix}, // 18: '.'
|
|
SymDotSlash: {"./", SymClassOperator, PosInfix}, // 19: './'
|
|
SymQuote: {"'", SymClassDelimiter, PosLeaf}, // 20: '\''
|
|
SymDoubleQuote: {"\"", SymClassDelimiter, PosLeaf}, // 21: '"'
|
|
SymBackTick: {"`", SymClassDelimiter, PosLeaf}, // 22: '`'
|
|
SymExclamation: {"!", SymClassOperator, PosPostfix}, // 23: '!'
|
|
SymQuestion: {"?", SymClassOperator, PosInfix}, // 24: '?'
|
|
SymAmpersand: {"&", SymClassOperator, PosInfix}, // 25: '&'
|
|
SymDoubleAmpersand: {"&&", SymClassOperator, PosInfix}, // 26: '&&'
|
|
SymPercent: {"%", SymClassOperator, PosInfix}, // 27: '%'
|
|
SymAt: {"@", SymClassOperator, PosPrefix}, // 28: '@'
|
|
SymUndescore: {"_", SymClassIdentifier, PosLeaf}, // 29: '_'
|
|
SymEqual: {"=", SymClassOperator, PosInfix}, // 30: '='
|
|
SymColonEqual: {":=", SymClassOperator, PosInfix}, // 31: ':='
|
|
SymDoubleEqual: {"==", SymClassOperator, PosInfix}, // 32: '=='
|
|
SymLess: {"<", SymClassOperator, PosInfix}, // 33: '<'
|
|
SymLessOrEqual: {"<=", SymClassOperator, PosInfix}, // 34: '<='
|
|
SymGreater: {">", SymClassOperator, PosInfix}, // 35: '>'
|
|
SymGreaterOrEqual: {">=", SymClassOperator, PosInfix}, // 36: '>='
|
|
SymLessGreater: {"<>", SymClassOperator, PosInfix}, // 37: '<>'
|
|
SymNotEqual: {"!=", SymClassOperator, PosInfix}, // 38: '!='
|
|
SymDollar: {"$", SymClassOperator, PosPrefix}, // 39: '$'
|
|
SymHash: {"#", SymClassOperator, PosPrefix}, // 40: '#'
|
|
SymOpenRound: {"(", SymClassParenthesis, PosPrefix}, // 41: '('
|
|
SymClosedRound: {")", SymClassParenthesis, PosPostfix}, // 42: ')'
|
|
SymOpenSquare: {"[", SymClassParenthesis, PosPrefix}, // 43: '['
|
|
SymClosedSquare: {"]", SymClassParenthesis, PosPostfix}, // 44: ']'
|
|
SymOpenBrace: {"{", SymClassParenthesis, PosPrefix}, // 45: '{'
|
|
SymClosedBrace: {"}", SymClassParenthesis, PosPostfix}, // 46: '}'
|
|
SymTilde: {"~", SymClassOperator, PosPrefix}, // 47: '~'
|
|
SymDoubleQuestion: {"??", SymClassOperator, PosInfix}, // 48: '??'
|
|
SymQuestionEqual: {"?=", SymClassOperator, PosInfix}, // 49: '?='
|
|
SymQuestionExclam: {"?!", SymClassOperator, PosInfix}, // 50: '?!'
|
|
SymDoubleAt: {"@@", SymClassCommand, PosLeaf}, // 51: '@@'
|
|
SymDoubleColon: {"::", SymClassOperator, PosInfix}, // 52: '::'
|
|
SymDoubleGreater: {">>", SymClassOperator, PosInfix}, // 53: '>>'
|
|
SymDoubleLess: {"<<", SymClassOperator, PosInfix}, // 54: '<<'
|
|
SymCaret: {"^", SymClassOperator, PosInfix}, // 55: '^'
|
|
SymDollarRound: {"$(", SymClassOperator, PosPrefix}, // 56: '$('
|
|
SymOpenClosedRound: {"()", SymClassOperator, PosPostfix}, // 57: '()'
|
|
SymDoubleDollar: {"$$", SymClassCommand, PosLeaf}, // 58: '$$'
|
|
SymDoubleDot: {"..", SymClassOperator, PosInfix}, // 59: '..'
|
|
SymTripleDot: {"...", SymClassOperator, PosPostfix}, // 60: '...'
|
|
SymStarEqual: {"*=", SymClassOperator, PosInfix}, // 61: '*='
|
|
SymSlashEqual: {"/=", SymClassOperator, PosInfix}, // 62: '/='
|
|
SymPercEqual: {"%=", SymClassOperator, PosInfix}, // 63: '%='
|
|
SymDoubleLessEqual: {"<<=", SymClassOperator, PosInfix}, // 64: '<<='
|
|
SymDoubleGreaterEqual: {">>=", SymClassOperator, PosInfix}, // 65: '>>='
|
|
SymAmpersandEqual: {"&=", SymClassOperator, PosInfix}, // 66: '&='
|
|
SymVertBarEqual: {"|=", SymClassOperator, PosInfix}, // 67: '|='
|
|
SymCaretEqual: {"^=", SymClassOperator, PosInfix}, // 68: '^='
|
|
SymPlusGreater: {"+>", SymClassOperator, PosInfix}, // 69: '+>'
|
|
SymLessPlus: {"<+", SymClassOperator, PosInfix}, // 70: '<+'
|
|
SymPreInc: {"++", SymClassOperator, PosPrefix}, // 71: '++'
|
|
SymPreDec: {"--", SymClassOperator, PosPrefix}, // 72: '--'
|
|
SymOpenSquareLess: {"[<", SymClassOperator, PosPrefix}, // 97: '[<'
|
|
SymGreaterClosedSquare: {">]", SymClassOperator, PosPostfix}, // 98: '>]'
|
|
// SymChangeSign
|
|
// SymUnchangeSign
|
|
// SymIdentifier
|
|
// SymBool
|
|
// SymInteger
|
|
// SymVariable
|
|
// SymFloat
|
|
// SymFraction
|
|
// SymString
|
|
// SymIterator
|
|
// SymOr: "or",
|
|
// SymAnd: "and",
|
|
// SymNot: "not",
|
|
// SymComment
|
|
// SymFuncCall
|
|
// SymFuncDef
|
|
// SymList
|
|
// SymLinkedList
|
|
// SymDict
|
|
// SymIndex
|
|
// SymExpression
|
|
// SymSelector // <selector> ::= <expr> "?" <selector-case> {":" <selector-case>} ["::" <default-selector-case>]
|
|
// SymSelectorCase // <selector-case> ::= [<list>] "{" <multi-expr> "}"
|
|
// // SymOpenComment // 0: '/*'
|
|
// // SymClosedComment // 0: '*/'
|
|
// // SymOneLineComment // 0: '//'
|
|
// keywordBase
|
|
SymKwAnd: {"and", SymClassOperator, PosInfix},
|
|
SymKwNot: {"not", SymClassOperator, PosInfix},
|
|
SymKwOr: {"or", SymClassOperator, PosInfix},
|
|
SymKwBut: {"but", SymClassOperator, PosInfix},
|
|
SymKwMap: {"map", SymClassOperator, PosInfix},
|
|
SymKwFilter: {"filter", SymClassOperator, PosInfix},
|
|
SymKwDigest: {"digest", SymClassOperator, PosInfix},
|
|
SymKwCat: {"cat", SymClassOperator, PosInfix},
|
|
SymKwJoin: {"join", SymClassOperator, PosInfix},
|
|
SymKwGroupBy: {"groupby", SymClassOperator, PosInfix},
|
|
SymKwFunc: {"func(", SymClassDeclaration, PosPrefix},
|
|
SymKwBuiltin: {"builtin", SymClassOperator, PosPrefix},
|
|
SymKwPlugin: {"plugin", SymClassOperator, PosPrefix},
|
|
SymKwIn: {"in", SymClassOperator, PosInfix},
|
|
SymKwInclude: {"include", SymClassOperator, PosPrefix},
|
|
SymKwNil: {"nil", SymClassValue, PosLeaf},
|
|
SymKwUnset: {"unset", SymClassOperator, PosPrefix},
|
|
}
|
|
}
|
|
|
|
func SymToString(sym Symbol) string {
|
|
if s, ok := symbolMap[sym]; ok {
|
|
return s.repr
|
|
}
|
|
return ""
|
|
}
|
|
|
|
func SymListToString(symList []Symbol, quote bool) string {
|
|
var sb strings.Builder
|
|
if len(symList) == 0 {
|
|
sb.WriteString("<nothing>")
|
|
} else {
|
|
for _, sym := range symList {
|
|
if sb.Len() > 0 {
|
|
sb.WriteByte(',')
|
|
sb.WriteByte(' ')
|
|
}
|
|
if quote {
|
|
sb.WriteByte('`')
|
|
}
|
|
sb.WriteString(SymToString(sym))
|
|
if quote {
|
|
sb.WriteByte('`')
|
|
}
|
|
}
|
|
}
|
|
return sb.String()
|
|
}
|
|
|
|
func StringEndsWithOperator(s string) bool {
|
|
return endingOperator(s) != SymNone
|
|
}
|
|
|
|
func endingOperator(s string) (sym Symbol) {
|
|
var matchLength = 0
|
|
var repr string
|
|
sym = SymNone
|
|
lower := strings.TrimRight(strings.ToLower(s), " \t")
|
|
for symbol, spec := range symbolMap {
|
|
if len(spec.repr) > matchLength || repr == spec.repr {
|
|
if strings.HasSuffix(lower, spec.repr) {
|
|
if isNotEndingSymbol(spec) && repr != spec.repr {
|
|
repr = spec.repr
|
|
matchLength = len(spec.repr)
|
|
sym = symbol
|
|
} else {
|
|
sym = SymNone
|
|
break
|
|
// matchLength = 0
|
|
}
|
|
}
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
func isNotEndingSymbol(spec symbolSpec) bool {
|
|
return (spec.kind == SymClassOperator && (spec.opType == PosInfix || spec.opType == PosPrefix)) ||
|
|
(spec.kind == SymClassParenthesis && spec.opType == PosPrefix)
|
|
}
|