diff options
Diffstat (limited to 'compiler/lexer.nim')
| -rw-r--r-- | compiler/lexer.nim | 25 |
1 files changed, 23 insertions, 2 deletions
diff --git a/compiler/lexer.nim b/compiler/lexer.nim index 4285a090d..217e33675 100644 --- a/compiler/lexer.nim +++ b/compiler/lexer.nim @@ -101,6 +101,7 @@ type base10, # base10 is listed as the first element, # so that it is the correct default value base2, base8, base16 + TToken* = object # a Nimrod token tokType*: TTokType # the type of the token indent*: int # the indentation; != -1 if the token has been @@ -110,6 +111,8 @@ type fNumber*: BiggestFloat # the parsed floating point literal base*: TNumericalBase # the numerical base; only valid for int # or float literals + strongSpaceA*: int8 # leading spaces of an operator + strongSpaceB*: int8 # trailing spaces of an operator literal*: string # the parsed (string) literal; and # documentation comments are here too line*, col*: int @@ -119,7 +122,9 @@ type indentAhead*: int # if > 0 an indendation has already been read # this is needed because scanning comments # needs so much look-ahead - + currLineIndent*: int + strongSpaces*: bool + var gLinesCompiled*: int # all lines that have been compiled @@ -184,6 +189,7 @@ proc initToken*(L: var TToken) = L.tokType = tkInvalid L.iNumber = 0 L.indent = 0 + L.strongSpaceA = 0 L.literal = "" L.fNumber = 0.0 L.base = base10 @@ -193,6 +199,7 @@ proc fillToken(L: var TToken) = L.tokType = tkInvalid L.iNumber = 0 L.indent = 0 + L.strongSpaceA = 0 setLen(L.literal, 0) L.fNumber = 0.0 L.base = base10 @@ -202,6 +209,7 @@ proc openLexer(lex: var TLexer, fileIdx: int32, inputstream: PLLStream) = openBaseLexer(lex, inputstream) lex.fileIdx = fileidx lex.indentAhead = - 1 + lex.currLineIndent = 0 inc(lex.lineNumber, inputstream.lineOffset) proc closeLexer(lex: var TLexer) = @@ -635,6 +643,14 @@ proc getOperator(L: var TLexer, tok: var TToken) = h = h !& ord(c) inc(pos) endOperator(L, tok, pos, h) + # advance pos but don't store it in L.bufpos so the next token (which might + # be an operator too) gets the preceeding spaces: + tok.strongSpaceB = 0 + while buf[pos] == ' ': + inc pos + inc tok.strongSpaceB + if buf[pos] in {CR, LF, nimlexbase.EndOfFile}: + tok.strongSpaceB = -1 proc scanComment(L: var TLexer, tok: var TToken) = var pos = L.bufpos @@ -678,10 +694,12 @@ proc scanComment(L: var TLexer, tok: var TToken) = proc skip(L: var TLexer, tok: var TToken) = var pos = L.bufpos var buf = L.buf + tok.strongSpaceA = 0 while true: case buf[pos] of ' ': inc(pos) + inc(tok.strongSpaceA) of Tabulator: lexMessagePos(L, errTabulatorsAreNotAllowed, pos) inc(pos) @@ -692,8 +710,10 @@ proc skip(L: var TLexer, tok: var TToken) = while buf[pos] == ' ': inc(pos) inc(indent) + tok.strongSpaceA = 0 if buf[pos] > ' ': tok.indent = indent + L.currLineIndent = indent break else: break # EndOfFile also leaves the loop @@ -703,6 +723,7 @@ proc rawGetTok(L: var TLexer, tok: var TToken) = fillToken(tok) if L.indentAhead >= 0: tok.indent = L.indentAhead + L.currLineIndent = L.indentAhead L.indentAhead = -1 else: tok.indent = -1 @@ -812,5 +833,5 @@ proc rawGetTok(L: var TLexer, tok: var TToken) = tok.tokType = tkInvalid lexMessage(L, errInvalidToken, c & " (\\" & $(ord(c)) & ')') inc(L.bufpos) - + dummyIdent = getIdent("") |