parser.go 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550
  1. package muesli
  2. import (
  3. _ "bufio"
  4. _ "bytes"
  5. _ "errors"
  6. "fmt"
  7. _ "io"
  8. _ "os"
  9. _ "reflect"
  10. _ "runtime"
  11. _ "strings"
  12. _ "unicode"
  13. // "gitlab.com/beoran/woe/graphviz"
  14. // _ "gitlab.com/beoran/woe/monolog"
  15. )
  16. /* Grammar:
  17. Desrired syntax (verified LL(1) on smlweb.cpsc.ucalgary.ca)
  18. PROGRAM -> STATEMENTS.
  19. CLOSED -> BLOCK | LIST | PARENTHESIS .
  20. STATEMENTS -> STATEMENT STATEMENTS | .
  21. STATEMENT -> CLOSED | EXPRESSION eos | eos .
  22. COMMAND -> word PARAMETERS.
  23. PARAMETERS -> PARAMETER PARAMETERS | .
  24. PARAMETER -> WORDVALUE | GETTER | SETTER | CLOSED .
  25. EXPRESSION -> COMMAND | GETTER | SETTER | VALUE.
  26. PARENTHESIS -> closeparen EXPRESSION openparen .
  27. BLOCK -> openblock STATEMENTS closeblock .
  28. LIST -> openlist PARAMETERS closelist .
  29. VALUE -> string | int | float | symbol | type .
  30. WORDVALUE -> word | VALUE.
  31. SETTER -> set TARGET .
  32. TARGET -> word PARAMETER | GETTER PARAMETER .
  33. GETTER -> get ORIGIN .
  34. ORIGIN -> word | SETTER | GETTER .
  35. *
  36. * program -> statements
  37. * statements -> statement+
  38. * statement -> get / set / command
  39. *
  40. */
  41. type Parser struct {
  42. Lexer *Lexer
  43. current Token
  44. LoggerWrapper
  45. Errors []ParserError
  46. }
  47. // panic with this type on errors that would prevent the parser
  48. // from making progress.
  49. type ParserError struct {
  50. *Parser
  51. *Token
  52. Chain error
  53. }
  54. func (pe ParserError) Error() string {
  55. return fmt.Sprintf("%s %s", pe.Token.String(), pe.Chain.Error())
  56. }
  57. func (parser *Parser) SetLogger(logger Logger) {
  58. parser.LoggerWrapper = LoggerWrapper{logger}
  59. }
  60. func (parser *Parser) Errorf(message string, args ...interface{}) ParserError {
  61. err := fmt.Errorf(message, args...)
  62. pe := ParserError { Parser: parser, Token:&parser.current, Chain: err }
  63. parser.Errors = append(parser.Errors, pe)
  64. return pe
  65. }
  66. func (parser *Parser) Panicf(message string, args ...interface{}) {
  67. pe := parser.Errorf(message, args...)
  68. panic(pe)
  69. }
  70. func (parser *Parser) Advance() {
  71. token := parser.Lexer.Lex()
  72. parser.current = token
  73. parser.LogDebug("Next token: %s\n", token.String())
  74. }
  75. /* Skips tokens until a closer, namely ), }, ], EOX or EOF is found,
  76. * to skip parse errors, and to be able to continue parsing despite errors. */
  77. func (parser *Parser) SkipError() Token {
  78. parser.Advance()
  79. for {
  80. parser.Advance()
  81. if parser.NextIsErrorSkipped() {
  82. return parser.current
  83. }
  84. if parser.NextIs(TokenKindError, TokenKindNone) {
  85. parser.Panicf("Cannot recover from parse error. Parse ended.")
  86. }
  87. }
  88. return parser.current
  89. }
  90. /* Looks at the current token and advances the lexer if the token is of any of
  91. the token kinds given in kinds. In this case it will return the accepted
  92. token and advance the parser. Otherwise, it will call parser.Panicf with a
  93. * "Syntax error unexpected <token>".*/
  94. func (parser *Parser) Require(kinds ...TokenKind) Token {
  95. if parser.current.IsNone() {
  96. parser.Advance()
  97. }
  98. expected := ""
  99. sep := ""
  100. for _, kind := range kinds {
  101. if kind == parser.current.TokenKind {
  102. accepted := parser.current
  103. parser.Advance()
  104. parser.LogDebug("Require: Accepted token: %s\n", accepted.String())
  105. return accepted
  106. }
  107. expected = fmt.Sprintf("%s%s%s", expected, sep, kind.String())
  108. }
  109. parser.Panicf("error: expected one of the following: %s", expected)
  110. return Token{}
  111. }
  112. func (parser *Parser) NewAstError(message string, args ...interface{}) *Ast {
  113. sv := StringValue(fmt.Sprintf(message+" at token "+parser.current.String(), args))
  114. pos := parser.current.Position
  115. tok := NewToken(TokenKindError, sv, pos)
  116. parser.Errorf(message, args...)
  117. return NewAst(AstKindError, nil, EmptyAstArray(), tok)
  118. }
  119. // Also handles none or error vales of the token
  120. func (parser *Parser) NewAst(kind AstKind, parent * Ast, children []*Ast, value Token) *Ast{
  121. if value.IsNone() {
  122. return NewAstNone()
  123. }
  124. if value.IsError() {
  125. return NewAst(AstKindError, parent, children, value)
  126. }
  127. return NewAst(kind, parent, children, value)
  128. }
  129. func (parser *Parser) ParseValue() *Ast {
  130. parser.LogDebug("ParseValue: %s\n", parser.current.String())
  131. value := parser.Require(TokenKindInteger, TokenKindString,
  132. TokenKindBoolean, TokenKindNil, TokenKindNil, TokenKindFloat, TokenKindSymbol)
  133. return parser.NewAst(AstKindValue, nil, EmptyAstArray(), value)
  134. }
  135. func (parser *Parser) ParseWordValue() *Ast {
  136. parser.LogDebug("ParseWordValue: %s\n", parser.current.String())
  137. value := parser.Require(TokenKindInteger, TokenKindString,
  138. TokenKindBoolean, TokenKindNil, TokenKindNil, TokenKindFloat, TokenKindSymbol,
  139. TokenKindType, TokenKindWord)
  140. return parser.NewAst(AstKindWordValue, nil, EmptyAstArray(), value)
  141. }
  142. func (parser *Parser) ParseArgument() *Ast {
  143. parser.LogDebug("ParseArgument: %s\n", parser.current.String())
  144. switch {
  145. case parser.NextIsGet(): return parser.ParseGet()
  146. case parser.NextIsSet(): return parser.ParseSet()
  147. case parser.NextIsClosed(): return parser.ParseClosed()
  148. case parser.NextIsWordValue(): return parser.ParseWordValue()
  149. default: parser.Panicf("error: in argument: expected $, =, }, word or value")
  150. return nil
  151. }
  152. }
  153. func (parser *Parser) ParseArguments() *Ast {
  154. parser.LogDebug("ParseArguments: %s\n", parser.current.String())
  155. ast := NewAstWithToken(AstKindArguments, parser.current)
  156. var children []*Ast
  157. for parser.NextIsArgument() {
  158. child := parser.ParseArgument()
  159. children = append(children, child)
  160. }
  161. ast.AppendChildren(children...)
  162. return ast
  163. }
  164. func (parser *Parser) ParseList() *Ast {
  165. parser.LogDebug("ParseList: %s\n", parser.current.String())
  166. op := parser.Require(TokenKindOpenList)
  167. if op.IsError() {
  168. return parser.NewAstError("Unexpected value.")
  169. }
  170. ast := NewAstWithToken(AstKindList, op)
  171. args := parser.ParseArguments()
  172. if AstIsError(args) {
  173. return args
  174. }
  175. if cp := parser.Require(TokenKindCloseList); cp.IsError() {
  176. return parser.NewAstError("expected closing brackets")
  177. }
  178. ast.AppendChild(args)
  179. return ast
  180. }
  181. func (parser *Parser) ParseParenthesis() *Ast {
  182. parser.LogDebug("ParseParenthesis: %s\n", parser.current.String())
  183. op := parser.Require(TokenKindOpenParen)
  184. if op.IsError() {
  185. return parser.NewAstError("Unexpected value.")
  186. }
  187. ast := NewAstWithToken(AstKindParenthesis, op)
  188. expr := parser.ParseExpression()
  189. if expr.IsNone() {
  190. return parser.NewAstError("expected expression")
  191. }
  192. if AstIsError(expr) {
  193. return expr
  194. }
  195. if cp := parser.Require(TokenKindCloseParen); cp.IsError() {
  196. return parser.NewAstError("expected closing parenthesis")
  197. }
  198. ast.AppendChild(expr)
  199. return ast
  200. }
  201. func (parser *Parser) ParseBlock() *Ast {
  202. parser.LogDebug("ParseBlock: %s\n", parser.current.String())
  203. op := parser.Require(TokenKindOpenBlock)
  204. if op.IsError() {
  205. return parser.NewAstError("Unexpected value.")
  206. }
  207. ast := NewAstWithToken(AstKindBlock, op)
  208. stats := parser.ParseStatements()
  209. if stats.IsNone() {
  210. return parser.NewAstError("expected expression")
  211. }
  212. if AstIsError(stats) {
  213. return stats
  214. }
  215. if cp := parser.Require(TokenKindCloseBlock); cp.IsError() {
  216. return parser.NewAstError("expected closing block")
  217. }
  218. ast.AppendChild(stats)
  219. return ast
  220. }
  221. /* Parses the target of a set expression */
  222. func (parser *Parser) ParseTarget() *Ast {
  223. parser.LogDebug("ParseTarget: %s\n", parser.current.String())
  224. var target *Ast
  225. switch {
  226. case parser.NextIs(TokenKindWord, TokenKindType, TokenKindSymbol):
  227. // Direct target to a set
  228. token := parser.Require(TokenKindWord, TokenKindType, TokenKindSymbol)
  229. target = NewAstWithToken(AstKindTarget, token)
  230. case parser.NextIsGet():
  231. // Indirect target
  232. target = NewAstWithToken(AstKindTarget, parser.current)
  233. target.AppendChild(parser.ParseGet())
  234. case parser.NextIsClosed():
  235. // Indirect target
  236. target = NewAstWithToken(AstKindTarget, parser.current)
  237. target.AppendChild(parser.ParseClosed())
  238. default:
  239. parser.Panicf("Malformed setter.")
  240. return nil
  241. }
  242. argument := parser.ParseArgument()
  243. if argument.IsNone() {
  244. return parser.NewAstError("Expected argument to set")
  245. }
  246. target.AppendChild(argument)
  247. return target
  248. }
  249. /* Parses the origin of a get expression */
  250. func (parser *Parser) ParseOrigin() *Ast {
  251. parser.LogDebug("ParseOrigin: %s\n", parser.current.String())
  252. var target *Ast
  253. switch {
  254. case parser.NextIs(TokenKindWord, TokenKindType, TokenKindSymbol):
  255. // Direct target to a set
  256. token := parser.Require(TokenKindWord, TokenKindType, TokenKindSymbol)
  257. target = NewAstWithToken(AstKindTarget, token)
  258. case parser.NextIsGet():
  259. // Indirect target
  260. target = NewAstWithToken(AstKindTarget, parser.current)
  261. target.AppendChild(parser.ParseGet())
  262. case parser.NextIsClosed():
  263. // Indirect target
  264. target = NewAstWithToken(AstKindTarget, parser.current)
  265. target.AppendChild(parser.ParseClosed())
  266. default:
  267. parser.Panicf("Malformed getter")
  268. return nil
  269. }
  270. return target
  271. }
  272. func (parser *Parser) ParseSet() *Ast {
  273. parser.LogDebug("ParseSet: %s\n", parser.current.String())
  274. set := parser.Require(TokenKindSet)
  275. ast := NewAstWithToken(AstKindSet, set)
  276. target := parser.ParseTarget()
  277. ast.AppendChild(target)
  278. return ast
  279. }
  280. func (parser *Parser) ParseGet() *Ast {
  281. parser.LogDebug("ParseGet: %s\n", parser.current.String())
  282. get := parser.Require(TokenKindGet)
  283. ast := NewAstWithToken(AstKindGet, get)
  284. target := parser.ParseOrigin()
  285. ast.AppendChild(target)
  286. return ast
  287. }
  288. func (parser *Parser) ParseCommand() *Ast {
  289. parser.LogDebug("ParseCommand: %s\n", parser.current.String())
  290. word := parser.Require(TokenKindWord, TokenKindType)
  291. arguments := parser.ParseArguments()
  292. command := NewAstWithToken(AstKindCommand, word)
  293. command.AppendChild(arguments)
  294. return command
  295. }
  296. func (parser *Parser) ParseClosed() *Ast {
  297. parser.LogDebug("ParseClosed: %s\n", parser.current.String())
  298. switch {
  299. case parser.NextIs(TokenKindOpenBlock): return parser.ParseBlock()
  300. case parser.NextIs(TokenKindOpenParen): return parser.ParseParenthesis()
  301. case parser.NextIs(TokenKindOpenList): return parser.ParseList()
  302. default:
  303. parser.Panicf("Syntax error in closed, expected {, (, [")
  304. return nil
  305. }
  306. }
  307. func (parser *Parser) ParseExpression() *Ast {
  308. parser.LogDebug("ParseExpression: %s\n", parser.current.String())
  309. switch {
  310. case parser.NextIsWord(): return parser.ParseCommand()
  311. case parser.NextIsSet(): return parser.ParseSet()
  312. case parser.NextIsGet(): return parser.ParseGet()
  313. case parser.NextIsValue(): return parser.ParseValue()
  314. default:
  315. parser.Panicf("Syntax error in expression, expected word, $, =, value")
  316. return nil
  317. }
  318. }
  319. func (parser *Parser) ParseExpressionStatement() *Ast {
  320. parser.LogDebug("ParseExpressionStatement: %s\n", parser.current.String())
  321. expr := parser.ParseExpression()
  322. if expr.IsNone() {
  323. return NewAstNone()
  324. }
  325. // Expression statements must end on EOX or EOF
  326. if eox := parser.Require(TokenKindEOX, TokenKindEOF); eox.IsError() {
  327. return parser.NewAstError("expected end of statement")
  328. }
  329. return expr
  330. }
  331. func (parser *Parser) ParseEmptyStatement() *Ast {
  332. parser.LogDebug("ParseEmptyStatement: %s\n", parser.current.String())
  333. if eox := parser.Require(TokenKindEOX, TokenKindEOF); eox.IsError() {
  334. return parser.NewAstError("expected end of statement")
  335. }
  336. return NewAstWithToken(AstKindStatement, parser.current)
  337. }
  338. func (parser Parser) NextIs(kinds ...TokenKind) bool {
  339. if (parser.current.TokenKind == TokenKindNone) {
  340. parser.Advance()
  341. }
  342. for _, kind := range kinds {
  343. if kind == parser.current.TokenKind {
  344. return true
  345. }
  346. }
  347. return false
  348. }
  349. func (parser Parser) NextIsClosed() bool {
  350. return parser.NextIs(TokenKindOpenBlock, TokenKindOpenList, TokenKindOpenParen)
  351. }
  352. func (parser Parser) NextIsWord() bool {
  353. return parser.NextIs(TokenKindWord)
  354. }
  355. func (parser Parser) NextIsGet() bool {
  356. return parser.NextIs(TokenKindGet)
  357. }
  358. func (parser Parser) NextIsSet() bool {
  359. return parser.NextIs(TokenKindSet)
  360. }
  361. func (parser Parser) NextIsValue() bool {
  362. return parser.NextIs(TokenKindString, TokenKindType,
  363. TokenKindInteger, TokenKindFloat, TokenKindBoolean, TokenKindNil,
  364. TokenKindSymbol)
  365. }
  366. func (parser Parser) NextIsWordValue() bool {
  367. return parser.NextIs(TokenKindWord, TokenKindString, TokenKindType,
  368. TokenKindInteger, TokenKindFloat, TokenKindBoolean, TokenKindNil,
  369. TokenKindSymbol)
  370. }
  371. func (parser Parser) NextIsArgument() bool {
  372. return parser.NextIs(TokenKindOpenBlock, TokenKindOpenList, TokenKindOpenParen,
  373. TokenKindSet, TokenKindGet, TokenKindWord, TokenKindString, TokenKindType,
  374. TokenKindInteger, TokenKindFloat, TokenKindBoolean, TokenKindNil,
  375. TokenKindSymbol)
  376. }
  377. func (parser Parser) NextIsErrorSkipped() bool {
  378. return parser.NextIs(TokenKindCloseBlock, TokenKindCloseList, TokenKindCloseParen,
  379. TokenKindEOX, TokenKindEOF)
  380. }
  381. func (parser Parser) NextIsEOX() bool {
  382. return parser.NextIs(TokenKindEOX)
  383. }
  384. func (parser Parser) NextIsEOF() bool {
  385. return parser.NextIs(TokenKindEOF)
  386. }
  387. func (parser Parser) NextIsExpression() bool {
  388. return parser.NextIs(TokenKindWord,
  389. TokenKindGet,
  390. TokenKindSet,
  391. TokenKindString,
  392. TokenKindType,
  393. TokenKindInteger,
  394. TokenKindFloat,
  395. TokenKindBoolean, TokenKindNil,
  396. TokenKindSymbol)
  397. }
  398. func (parser Parser) NextIsStatement() bool {
  399. return parser.NextIsExpression() || parser.NextIsClosed() || parser.NextIsEOX()
  400. }
  401. func (parser *Parser) ParseStatement() *Ast {
  402. parser.LogDebug("ParseStatement: %s\n", parser.current.String())
  403. switch {
  404. case parser.NextIsClosed(): return parser.ParseClosed()
  405. case parser.NextIsExpression(): return parser.ParseExpressionStatement()
  406. case parser.NextIsEOX(): return parser.ParseEmptyStatement()
  407. default:
  408. parser.Panicf("Expected closed, expression, or EOX in statement")
  409. return parser.NewAstError("Expected closed, expression, or EOX in statement")
  410. }
  411. }
  412. func (parser *Parser) ParseStatements() *Ast {
  413. parser.LogDebug("ParseStatements: %s\n", parser.current.String())
  414. ast := NewAstWithToken(AstKindStatements, parser.current)
  415. var children []*Ast
  416. for parser.NextIsStatement() {
  417. child := parser.ParseStatement()
  418. children = append(children, child)
  419. }
  420. ast.AppendChildren(children...)
  421. return ast
  422. }
  423. func (parser *Parser) ParseProgram() *Ast {
  424. parser.LogDebug("ParseProgram: %s\n", parser.current.String())
  425. stats := parser.ParseStatements()
  426. eof := parser.Require(TokenKindEOF)
  427. aeof := NewAstWithToken(AstKindEnd, eof)
  428. if eof.IsNone() {
  429. aeof = parser.NewAstError("Expected EOF, have: %s", parser.current.String())
  430. }
  431. children := []*Ast{stats, aeof}
  432. return NewAst(AstKindProgram, nil, children, NoToken())
  433. }
  434. func (parser *Parser) Parse() (ast *Ast) {
  435. defer func() {
  436. if err := recover() ; err != nil {
  437. if perr, ok := err.(ParserError) ; ok{
  438. ast = parser.NewAstError("Parse error, recovered: %s\n", perr)
  439. } else {
  440. panic(err)
  441. }
  442. }
  443. } ()
  444. parser.Advance()
  445. ast = parser.ParseProgram()
  446. return ast
  447. }
  448. func NewParser(lexer *Lexer) *Parser {
  449. parser := &Parser{Lexer: lexer, current: NoToken(), LoggerWrapper: LoggerWrapper{nil}}
  450. return parser
  451. }
  452. func NewParserFromString(input string) *Parser {
  453. lexer := NewLexerFromString(input)
  454. return NewParser(lexer)
  455. }
  456. func NewParserFromFilename(filename string) (*Parser, error) {
  457. lexer, err := NewLexerFromFilename(filename)
  458. if err != nil {
  459. return nil, err
  460. }
  461. return NewParser(lexer), nil
  462. }