fixes #15939, fixes #15464, fixes #16169, fixes #16226 VM now supports addr(mystring[ind]) (index + index assignment) (#15987)

* fix #15939, fix #15464 VM now supports `addr(mystring[ind])` (index + index assignment), var char return etc
* cleanups
* cstring tests
* add test for bug #15464
* improve test coverage

changelog.md

@@ -69,6 +69,7 @@
 
 - Added `nim --eval:cmd` to evaluate a command directly, see `nim --help`.
 
+- VM now supports `addr(mystring[ind])` (index + index assignment)
 
 
 ## Tool changes

compiler/vm.nim

@@ -20,6 +20,7 @@ import
 
 from semfold import leValueConv, ordinalValToString
 from evaltempl import evalTemplate
+from magicsys import getSysType
 
 const
   traceCode = defined(nimVMDebug)
@@ -123,7 +124,7 @@ proc derefPtrToReg(address: BiggestInt, typ: PType, r: var TFullReg, isAssign: b
     else:
       r.ensureKind(rkind)
       let val = cast[ptr T](address)[]
-      when T is SomeInteger:
+      when T is SomeInteger | char:
         r.field = BiggestInt(val)
       else:
         r.field = val
@@ -131,6 +132,7 @@ proc derefPtrToReg(address: BiggestInt, typ: PType, r: var TFullReg, isAssign: b
 
   ## see also typeinfo.getBiggestInt
   case typ.kind
+  of tyChar: fun(intVal, char, rkInt)
   of tyInt: fun(intVal, int, rkInt)
   of tyInt8: fun(intVal, int8, rkInt)
   of tyInt16: fun(intVal, int16, rkInt)
@@ -677,6 +679,23 @@ proc rawExecute(c: PCtx, start: int, tos: PStackFrame): TFullReg =
         regs[ra].intVal = s[idx].ord
       else:
         stackTrace(c, tos, pc, formatErrorIndexBound(idx, s.len-1))
+    of opcLdStrIdxAddr:
+      # a = addr(b[c]); similar to opcLdArrAddr
+      decodeBC(rkNode)
+      if regs[rc].intVal > high(int):
+        stackTrace(c, tos, pc, formatErrorIndexBound(regs[rc].intVal, high(int)))
+      let idx = regs[rc].intVal.int
+      let s = regs[rb].node.strVal.addr # or `byaddr`
+      if idx <% s[].len:
+         # `makePtrType` not accessible from vm.nim
+        let typ = newType(tyPtr, nextId c.idgen, c.module.owner)
+        typ.add getSysType(c.graph, c.debug[pc], tyChar)
+        let node = newNodeIT(nkIntLit, c.debug[pc], typ) # xxx nkPtrLit
+        node.intVal = cast[int](s[][idx].addr)
+        node.flags.incl nfIsPtr
+        regs[ra].node = node
+      else:
+        stackTrace(c, tos, pc, formatErrorIndexBound(idx, s[].len-1))
     of opcWrArr:
       # a[b] = c
       decodeBC(rkNode)

compiler/vmdef.nim

@@ -78,6 +78,7 @@ type
     opcWrDeref,
     opcWrStrIdx,
     opcLdStrIdx, # a = b[c]
+    opcLdStrIdxAddr,  # a = addr(b[c])
 
     opcAddInt,
     opcAddImmInt,

compiler/vmgen.nim

@@ -1654,8 +1654,8 @@ proc genArrAccessOpcode(c: PCtx; n: PNode; dest: var TDest; opc: TOpcode;
   let a = c.genx(n[0], flags)
   let b = c.genIndex(n[1], n[0].typ)
   if dest < 0: dest = c.getTemp(n.typ)
-  if opc == opcLdArr and {gfNodeAddr} * flags != {}:
-    c.gABC(n, opcLdArrAddr, dest, a, b)
+  if opc in {opcLdArrAddr, opcLdStrIdxAddr} and gfNodeAddr in flags:
+    c.gABC(n, opc, dest, a, b)
   elif needsRegLoad():
     var cc = c.getTemp(n.typ)
     c.gABC(n, opc, cc, a, b)
@@ -1749,11 +1749,13 @@ proc genCheckedObjAccess(c: PCtx; n: PNode; dest: var TDest; flags: TGenFlags) =
 proc genArrAccess(c: PCtx; n: PNode; dest: var TDest; flags: TGenFlags) =
   let arrayType = n[0].typ.skipTypes(abstractVarRange-{tyTypeDesc}).kind
   if arrayType in {tyString, tyCString}:
-    genArrAccessOpcode(c, n, dest, opcLdStrIdx, {})
+    let opc = if gfNodeAddr in flags: opcLdStrIdxAddr else: opcLdStrIdx
+    genArrAccessOpcode(c, n, dest, opc, flags)
   elif arrayType == tyTypeDesc:
     c.genTypeLit(n.typ, dest)
   else:
-    genArrAccessOpcode(c, n, dest, opcLdArr, flags)
+    let opc = if gfNodeAddr in flags: opcLdArrAddr else: opcLdArr
+    genArrAccessOpcode(c, n, dest, opc, flags)
 
 proc getNullValueAux(t: PType; obj: PNode, result: PNode; conf: ConfigRef; currPosition: var int) =
   if t != nil and t.len > 0 and t[0] != nil:

tests/misc/taddr.nim

@@ -196,10 +196,76 @@ template test14339() = # bug #14339
       when not defined(js): # pending bug #16003
         doAssert a.val == 5
 
+template testStatic15464() = # bug #15464
+  proc access(s: var seq[char], i: int): var char = s[i]
+  proc access(s: var string, i: int): var char = s[i]
+  static:
+    var s = @['a', 'b', 'c']
+    access(s, 2) = 'C'
+    doAssert access(s, 2) == 'C'
+  static:
+    var s = "abc"
+    access(s, 2) = 'C'
+    doAssert access(s, 2) == 'C'
+
+proc test15464() = # bug #15464 (v2)
+  proc access(s: var seq[char], i: int): var char = s[i]
+  proc access(s: var string, i: int): var char = s[i]
+  block:
+    var s = @['a', 'b', 'c']
+    access(s, 2) = 'C'
+    doAssert access(s, 2) == 'C'
+  block:
+    var s = "abc"
+    access(s, 2) = 'C'
+    doAssert access(s, 2) == 'C'
+
+block: # bug #15939
+  block:
+    const foo = "foo"
+    proc proc1(s: var string) =
+      if s[^1] notin {'a'..'z'}:
+        s = ""
+    proc proc2(f: string): string =
+      result = f
+      proc1(result)
+    const bar = proc2(foo)
+    doAssert bar == "foo"
+
+proc test15939() = # bug #15939 (v2)
+  template fn(a) =
+    let pa = a[0].addr
+    doAssert pa != nil
+    doAssert pa[] == 'a'
+    pa[] = 'x'
+    doAssert pa[] == 'x'
+    doAssert a == "xbc"
+    when not defined js: # otherwise overflows
+      let pa2 = cast[ptr char](cast[int](pa) + 1)
+      doAssert pa2[] == 'b'
+      pa2[] = 'B'
+      doAssert a == "xBc"
+
+  # mystring[ind].addr
+  var a = "abc"
+  fn(a)
+
+  # mycstring[ind].addr
+  template cstringTest =
+    var a2 = "abc"
+    var b2 = a2.cstring
+    fn(b2)
+  when nimvm: cstringTest()
+  else: # can't take address of cstring element in js
+    when not defined(js): cstringTest()
+
 template main =
   # xxx wrap all other tests here like that so they're also tested in VM
   test14420()
   test14339()
+  test15464()
+  test15939()
 
+testStatic15464()
 static: main()
 main()