fixes #15688; handle strongSpace overflow issues (#20724)

* fixes #15688; handle `strongSpace` overflow issues

* stop at 1

* change the type of strongSpaceA to bool

compiler/layouter.nim

@@ -510,7 +510,7 @@ proc emitTok*(em: var Emitter; L: Lexer; tok: Token) =
     rememberSplit(splitComma)
     wrSpace em
   of openPars:
-    if tok.strongSpaceA > 0 and not em.endsInWhite and
+    if tok.strongSpaceA and not em.endsInWhite and
         (not em.wasExportMarker or tok.tokType == tkCurlyDotLe):
       wrSpace em
     wr(em, $tok.tokType, ltSomeParLe)
@@ -528,7 +528,7 @@ proc emitTok*(em: var Emitter; L: Lexer; tok: Token) =
     wr(em, $tok.tokType, ltOther)
     if not em.inquote: wrSpace(em)
   of tkOpr, tkDotDot:
-    if em.inquote or ((tok.strongSpaceA == 0 and tok.strongSpaceB == 0) and
+    if em.inquote or (((not tok.strongSpaceA) and tok.strongSpaceB == 0) and
         tok.ident.s notin ["<", ">", "<=", ">=", "==", "!="]):
       # bug #9504: remember to not spacify a keyword:
       lastTokWasTerse = true
@@ -538,7 +538,7 @@ proc emitTok*(em: var Emitter; L: Lexer; tok: Token) =
       if not em.endsInWhite: wrSpace(em)
       wr(em, tok.ident.s, ltOpr)
       template isUnary(tok): bool =
-        tok.strongSpaceB == 0 and tok.strongSpaceA > 0
+        tok.strongSpaceB == 0 and tok.strongSpaceA
 
       if not isUnary(tok):
         rememberSplit(splitBinary)

compiler/lexer.nim

@@ -102,7 +102,7 @@ type
     fNumber*: BiggestFloat    # the parsed floating point literal
     base*: NumericalBase      # the numerical base; only valid for int
                               # or float literals
-    strongSpaceA*: int8       # leading spaces of an operator
+    strongSpaceA*: bool       # leading spaces of an operator
     strongSpaceB*: int8       # trailing spaces of an operator
     literal*: string          # the parsed (string) literal; and
                               # documentation comments are here too
@@ -173,7 +173,7 @@ proc initToken*(L: var Token) =
   L.tokType = tkInvalid
   L.iNumber = 0
   L.indent = 0
-  L.strongSpaceA = 0
+  L.strongSpaceA = false
   L.literal = ""
   L.fNumber = 0.0
   L.base = base10
@@ -186,7 +186,7 @@ proc fillToken(L: var Token) =
   L.tokType = tkInvalid
   L.iNumber = 0
   L.indent = 0
-  L.strongSpaceA = 0
+  L.strongSpaceA = false
   setLen(L.literal, 0)
   L.fNumber = 0.0
   L.base = base10
@@ -958,7 +958,8 @@ proc getOperator(L: var Lexer, tok: var Token) =
   tok.strongSpaceB = 0
   while L.buf[pos] == ' ':
     inc pos
-    inc tok.strongSpaceB
+    if tok.strongSpaceB < 1:
+      inc(tok.strongSpaceB)
   if L.buf[pos] in {CR, LF, nimlexbase.EndOfFile}:
     tok.strongSpaceB = -1
 
@@ -1147,7 +1148,7 @@ proc scanComment(L: var Lexer, tok: var Token) =
 proc skip(L: var Lexer, tok: var Token) =
   var pos = L.bufpos
   tokenBegin(tok, pos)
-  tok.strongSpaceA = 0
+  tok.strongSpaceA = false
   when defined(nimpretty):
     var hasComment = false
     var commentIndent = L.currLineIndent
@@ -1158,7 +1159,8 @@ proc skip(L: var Lexer, tok: var Token) =
     case L.buf[pos]
     of ' ':
       inc(pos)
-      inc(tok.strongSpaceA)
+      if not tok.strongSpaceA:
+        tok.strongSpaceA = true
     of '\t':
       if not L.allowTabs: lexMessagePos(L, errGenerated, pos, "tabs are not allowed, use spaces instead")
       inc(pos)
@@ -1180,7 +1182,7 @@ proc skip(L: var Lexer, tok: var Token) =
           pos = L.bufpos
         else:
           break
-      tok.strongSpaceA = 0
+      tok.strongSpaceA = false
       when defined(nimpretty):
         if L.buf[pos] == '#' and tok.line < 0: commentIndent = indent
       if L.buf[pos] > ' ' and (L.buf[pos] != '#' or L.buf[pos+1] == '#'):

compiler/parser.nim

@@ -301,13 +301,13 @@ proc isUnary(tok: Token): bool =
   ## Check if the given token is a unary operator
   tok.tokType in {tkOpr, tkDotDot} and
   tok.strongSpaceB == 0 and
-  tok.strongSpaceA > 0
+  tok.strongSpaceA
 
 proc checkBinary(p: Parser) {.inline.} =
   ## Check if the current parser token is a binary operator.
   # we don't check '..' here as that's too annoying
   if p.tok.tokType == tkOpr:
-    if p.tok.strongSpaceB > 0 and p.tok.strongSpaceA == 0:
+    if p.tok.strongSpaceB > 0 and not p.tok.strongSpaceA:
       parMessage(p, warnInconsistentSpacing, prettyTok(p.tok))
 
 #| module = stmt ^* (';' / IND{=})
@@ -477,7 +477,7 @@ proc dotExpr(p: var Parser, a: PNode): PNode =
   optInd(p, result)
   result.add(a)
   result.add(parseSymbol(p, smAfterDot))
-  if p.tok.tokType == tkBracketLeColon and p.tok.strongSpaceA <= 0:
+  if p.tok.tokType == tkBracketLeColon and not p.tok.strongSpaceA:
     var x = newNodeI(nkBracketExpr, p.parLineInfo)
     # rewrite 'x.y[:z]()' to 'y[z](x)'
     x.add result[1]
@@ -486,7 +486,7 @@ proc dotExpr(p: var Parser, a: PNode): PNode =
     var y = newNodeI(nkCall, p.parLineInfo)
     y.add x
     y.add result[0]
-    if p.tok.tokType == tkParLe and p.tok.strongSpaceA <= 0:
+    if p.tok.tokType == tkParLe and not p.tok.strongSpaceA:
       exprColonEqExprListAux(p, tkParRi, y)
     result = y
 
@@ -842,7 +842,7 @@ proc primarySuffix(p: var Parser, r: PNode,
     case p.tok.tokType
     of tkParLe:
       # progress guaranteed
-      if p.tok.strongSpaceA > 0:
+      if p.tok.strongSpaceA:
         result = commandExpr(p, result, mode)
         # type sections allow full command syntax
         if mode == pmTypeDef:
@@ -861,13 +861,13 @@ proc primarySuffix(p: var Parser, r: PNode,
       result = parseGStrLit(p, result)
     of tkBracketLe:
       # progress guaranteed
-      if p.tok.strongSpaceA > 0:
+      if p.tok.strongSpaceA:
         result = commandExpr(p, result, mode)
         break
       result = namedParams(p, result, nkBracketExpr, tkBracketRi)
     of tkCurlyLe:
       # progress guaranteed
-      if p.tok.strongSpaceA > 0:
+      if p.tok.strongSpaceA:
         result = commandExpr(p, result, mode)
         break
       result = namedParams(p, result, nkCurlyExpr, tkCurlyRi)
@@ -2386,7 +2386,7 @@ proc parseAll(p: var Parser): PNode =
       parMessage(p, errInvalidIndentation)
 
 proc checkFirstLineIndentation*(p: var Parser) =
-  if p.tok.indent != 0 and p.tok.strongSpaceA > 0:
+  if p.tok.indent != 0 and p.tok.strongSpaceA:
     parMessage(p, errInvalidIndentation)
 
 proc parseTopLevelStmt(p: var Parser): PNode =

nimpretty/tests/exhaustive.nim

@@ -267,7 +267,7 @@ proc emitTok*(em: var Emitter; L: TLexer; tok: TToken) =
     if not em.endsInWhite: wr(" ")
     wr(tok.ident.s)
     template isUnary(tok): bool =
-      tok.strongSpaceB == 0 and tok.strongSpaceA > 0
+      tok.strongSpaceB == 0 and tok.strongSpaceA
 
     if not isUnary(tok) or em.lastTok in {tkOpr, tkDotDot}:
       wr(" ")

nimpretty/tests/expected/exhaustive.nim

@@ -272,7 +272,7 @@ proc emitTok*(em: var Emitter; L: TLexer; tok: TToken) =
     if not em.endsInWhite: wr(" ")
     wr(tok.ident.s)
     template isUnary(tok): bool =
-      tok.strongSpaceB == 0 and tok.strongSpaceA > 0
+      tok.strongSpaceB == 0 and tok.strongSpaceA
 
     if not isUnary(tok) or em.lastTok in {tkOpr, tkDotDot}:
       wr(" ")

tests/lexer/tlexerspaces.nim

@@ -0,0 +1,2 @@
+discard 12 +                                                                                                                                                                                                                                                                           5
+discard 12 + 5