Add preliminary expression parsing

3 files changed, 253 insertions, 71 deletions

diff --git a/nimterop/ast2.nim b/nimterop/ast2.nim
index 43b6b35..04bb8eb 100644
--- a/nimterop/ast2.nim
+++ b/nimterop/ast2.nim
@@ -1,37 +1,10 @@
 import macros, os, sequtils, sets, strformat, strutils, tables, times
 
-import regex
-
 import compiler/[ast, idents, lineinfos, modulegraphs, msgs, options, parser, renderer]
 
-import "."/treesitter/[api, c, cpp]
-
-import "."/[globals, getters]
-
-proc getCCodeAst*(gState: State, code: string): string =
-  var parser = tsParserNew()
-  var code = code
+import "."/treesitter/api
 
-  defer:
-    parser.tsParserDelete()
-
-
-  doAssert code.nBl, "Empty code"
-  if gState.mode == "c":
-    doAssert parser.tsParserSetLanguage(treeSitterC()), "Failed to load C parser"
-  elif gState.mode == "cpp":
-    doAssert parser.tsParserSetLanguage(treeSitterCpp()), "Failed to load C++ parser"
-  else:
-    doAssert false, &"Invalid parser {gState.mode}"
-
-  var
-    tree = parser.tsParserParseString(nil, code.cstring, code.len.uint32)
-    root = tree.tsTreeRootNode()
-
-  defer:
-    tree.tsTreeDelete()
-
-  return code.printLisp(root)
+import "."/[globals, getters, exprparser]
 
 proc getPtrType*(str: string): string =
   result = case str:
@@ -59,42 +32,8 @@ proc parseString(gState: State, str: string): PNode =
   except:
     decho getCurrentExceptionMsg()
 
-proc getLit*(gState: State, str: string, expression = false): PNode =
-  # Used to convert #define literals into const and expressions
-  # in array sizes
-  #
-  # `expression` is true when `str` should be converted into a Nim expression
-  let
-    str = str.replace(re"/[/*].*?(?:\*/)?$", "").strip()
-
-  if str.contains(re"^[\-]?[\d]+$"):              # decimal
-    result = newIntNode(nkIntLit, parseInt(str))
-
-  elif str.contains(re"^[\-]?[\d]*[.]?[\d]+$"):   # float
-    result = newFloatNode(nkFloatLit, parseFloat(str))
-
-  elif str.contains(re"^0x[\da-fA-F]+$"):         # hexadecimal
-    result = gState.parseString(str)
-
-  elif str.contains(re"^'[[:ascii:]]'$"):         # char
-    result = newNode(nkCharLit)
-    result.intVal = str[1].int64
-
-  elif str.contains(re"""^"[[:ascii:]]+"$"""):    # char *
-    result = newStrNode(nkStrLit, str[1 .. ^2])
-
-  else:
-    decho "Macro AST:"
-    decho str
-    decho nimState.gState.getCCodeAst(str)
-    let
-      str =
-        if expression: gState.getNimExpression(str)
-        else: str
-    result = gState.parseString(str)
-
-  if result.isNil:
-    result = newNode(nkNilLit)
+proc getLit*(nimState: NimState, str: string, expression = false): PNode =
+  result = nimState.codeToNode(str)
 
 proc getOverrideOrSkip(gState: State, node: TSNode, origname: string, kind: NimSymKind): PNode =
   # Check if symbol `origname` of `kind` and `origname` has any cOverride defined
@@ -181,11 +120,9 @@ proc newConstDef(gState: State, node: TSNode, fname = "", fval = ""): PNode =
 
   if name.Bl:
     # Name skipped or overridden since blank
-    result = gState.getOverrideOrSkip(node, origname, nskConst)
-  elif valident.kind in {nkCharLit .. nkStrLit} or
-    (valident.kind == nkStmtList and valident.len > 0 and
-    valident[0].kind in {nkCharLit .. nkStrLit}):
-    if gState.addNewIdentifer(name):
+    result = nimState.getOverrideOrSkip(node, origname, nskConst)
+  elif valident.kind != nkNilLit:
+    if nimState.addNewIdentifer(name):
       # const X* = Y
       #
       # nkConstDef(
diff --git a/nimterop/exprparser.nim b/nimterop/exprparser.nim
new file mode 100644
index 0000000..a4be5d0
--- /dev/null
+++ b/nimterop/exprparser.nim
@@ -0,0 +1,245 @@
+import strformat, strutils, macros
+
+import regex
+
+import compiler/[ast, renderer]
+
+import "."/treesitter/[api, c, cpp]
+
+import "."/[globals, getters]
+
+type
+  ExprParser* = ref object
+    state*: NimState
+    code*: string
+
+proc newExprParser*(state: NimState, code: string): ExprParser =
+  ExprParser(state: state, code: code)
+
+template decho(msg: varargs[string, `$`]) =
+  if exprParser.state.gState.debug:
+    let nimState {.inject.} = exprParser.state
+    necho "# " & join(msg, "")
+
+template val*(node: TSNode): string =
+  exprParser.code.getNodeVal(node)
+
+proc mode*(exprParser: ExprParser): string =
+  exprParser.state.gState.mode
+
+template withCodeAst(exprParser: ExprParser, body: untyped): untyped =
+  var parser = tsParserNew()
+  defer:
+    parser.tsParserDelete()
+
+  doAssert exprParser.code.nBl, "Empty code"
+  if exprParser.mode == "c":
+    doAssert parser.tsParserSetLanguage(treeSitterC()), "Failed to load C parser"
+  elif exprParser.mode == "cpp":
+    doAssert parser.tsParserSetLanguage(treeSitterCpp()), "Failed to load C++ parser"
+  else:
+    doAssert false, &"Invalid parser {exprParser.mode}"
+
+  var
+    tree = parser.tsParserParseString(nil, exprParser.code.cstring, exprParser.code.len.uint32)
+    root {.inject.} = tree.tsTreeRootNode()
+
+  body
+
+  defer:
+    tree.tsTreeDelete()
+
+
+proc getNumNode(number, suffix: string): PNode {.inline.} =
+  result = newNode(nkNilLit)
+  if number.contains("."):
+    let floatSuffix = number[result.len-1]
+    case floatSuffix
+    of 'l', 'L':
+      # TODO: handle long double (128 bits)
+      # result = newNode(nkFloat128Lit)
+      result = newFloatNode(nkFloat64Lit, parseFloat(number[0 ..< number.len - 1]))
+    of 'f', 'F':
+      result = newFloatNode(nkFloat64Lit, parseFloat(number[0 ..< number.len - 1]))
+    else:
+      discard
+    return
+
+  case suffix
+  of "u", "U":
+    result = newNode(nkUintLit)
+  of "l", "L":
+    result = newNode(nkInt32Lit)
+  of "ul", "UL":
+    result = newNode(nkUint32Lit)
+  of "ll", "LL":
+    result = newNode(nkInt64Lit)
+  of "ull", "ULL":
+    result = newNode(nkUint64Lit)
+  else:
+    result = newNode(nkIntLit)
+
+  if number.contains(re"0[xX]"):
+    result.intVal = parseHexInt(number)
+    result.flags = {nfBase16}
+  elif number.contains(re"0[bB]"):
+    result.intVal = parseBinInt(number)
+    result.flags = {nfBase2}
+  elif number.contains(re"0[oO]"):
+    result.intVal = parseOctInt(number)
+    result.flags = {nfBase8}
+  else:
+    result.intVal = parseInt(number)
+
+proc processNumberLiteral*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkNilLit)
+  let nodeVal = node.val
+
+  var match: RegexMatch
+  const reg = re"(\-)?(0\d+|0[xX][0-9a-fA-F]+|0[bB][01]+|\d+\.?\d*[fFlL]?|\d*\.?\d+[fFlL]?|\d+)([ulUL]*)"
+  let found = nodeVal.find(reg, match)
+  if found:
+    let
+      prefix = if match.group(0).len > 0: nodeVal[match.group(0)[0]] else: ""
+      number = nodeVal[match.group(1)[0]]
+      suffix = nodeVal[match.group(2)[0]]
+
+    result = getNumNode(number, suffix)
+
+    if result.kind != nkNilLit and prefix == "-":
+      result = nkPrefix.newTree(
+        exprParser.state.getIdent("-"),
+        result
+      )
+
+proc processCharacterLiteral*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkCharLit)
+  result.intVal = node.val[1].int64
+
+proc processStringLiteral*(exprParser: ExprParser, node: TSNode): PNode =
+  let nodeVal = node.val
+  result = newStrNode(nkStrLit, nodeVal[1 ..< nodeVal.len - 1])
+
+proc processTSNode*(exprParser: ExprParser, node: TSNode): PNode
+
+proc processShiftExpression*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkInfix)
+  let
+    left = node[0]
+    right = node[1]
+  var shiftSym = exprParser.code[left.tsNodeEndByte() ..< right.tsNodeStartByte()].strip()
+
+  case shiftSym
+  of "<<":
+    result.add exprParser.state.getIdent("shl")
+  of ">>":
+    result.add exprParser.state.getIdent("shr")
+  else:
+    discard
+
+  result.add exprParser.processTSNode(left)
+  result.add exprParser.processTSNode(right)
+
+proc processParenthesizedExpr*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkPar)
+  for i in 0 ..< node.len():
+    result.add(exprParser.processTSNode(node[i]))
+
+proc processLogicalExpression*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkPar)
+  let child = node[0]
+  var nimSym = ""
+
+  var binarySym = exprParser.code[node.tsNodeStartByte() ..< child.tsNodeStartByte()].strip()
+  decho "LOG SYM: ", binarySym
+
+  case binarySym
+  of "!":
+    nimSym = "not"
+  else:
+    return newNode(nkNilLit)
+
+  decho "LOG CHILD: ", child.val, ", nim: ", nimSym
+  result.add nkPrefix.newTree(
+    exprParser.state.getIdent(nimSym),
+    exprParser.processTSNode(child)
+  )
+
+proc processBitwiseExpression*(exprParser: ExprParser, node: TSNode): PNode =
+  if node.len() > 1:
+    result = newNode(nkInfix)
+    let left = node[0]
+    let right = node[1]
+    var nimSym = ""
+
+    var binarySym = exprParser.code[left.tsNodeEndByte() ..< right.tsNodeStartByte()].strip()
+    decho "# BIN SYM: ", binarySym
+
+    case binarySym
+    of "|", "||":
+      nimSym = "or"
+    of "&", "&&":
+      nimSym = "and"
+    of "^":
+      nimSym = "xor"
+    else:
+      return newNode(nkNilLit)
+
+    result.add exprParser.state.getIdent(nimSym)
+    result.add exprParser.processTSNode(left)
+    result.add exprParser.processTSNode(right)
+
+  elif node.len() == 1:
+    result = newNode(nkPar)
+    let child = node[0]
+    var nimSym = ""
+
+    var binarySym = exprParser.code[node.tsNodeStartByte() ..< child.tsNodeStartByte()].strip()
+    decho "# BIN SYM: ", binarySym
+
+    case binarySym
+    of "~":
+      nimSym = "not"
+    else:
+      return newNode(nkNilLit)
+
+    result.add nkPrefix.newTree(
+      exprParser.state.getIdent(nimSym),
+      exprParser.processTSNode(child)
+    )
+
+proc processTSNode*(exprParser: ExprParser, node: TSNode): PNode =
+  result = newNode(nkNilLit)
+  decho "# NODE: ", node.getName(), ", VAL: ", node.val
+  case node.getName()
+  of "number_literal":
+    result = exprParser.processNumberLiteral(node)
+  of "string_literal":
+    result = exprParser.processStringLiteral(node)
+  of "char_literal":
+    result = exprParser.processCharacterLiteral(node)
+  of "expression_statement", "ERROR", "translation_unit":
+    # This may be wrong. What can be in an expression?
+    result = exprParser.processTSNode(node[0])
+  of "parenthesized_expression":
+    result = exprParser.processParenthesizedExpr(node)
+  of "bitwise_expression":
+    result = exprParser.processBitwiseExpression(node)
+  of "shift_expression":
+    result = exprParser.processShiftExpression(node)
+  of "logical_expression":
+    result = exprParser.processLogicalExpression(node)
+  of "identifier":
+    var ident = node.val
+    if ident != "_":
+      ident = exprParser.state.getIdentifier(ident, nskConst)
+    result = exprParser.state.getIdent(ident)
+  else:
+    result = newNode(nkNilLit)
+
+  decho "# NODERES: ", result
+
+proc codeToNode*(state: NimState, code: string): PNode =
+  let exprParser = newExprParser(state, code)
+  withCodeAst(exprParser):
+    result = exprParser.processTSNode(root)
\ No newline at end of file
diff --git a/nimterop/getters.nim b/nimterop/getters.nim
index 4ffb5d7..b96ae70 100644
--- a/nimterop/getters.nim
+++ b/nimterop/getters.nim
@@ -221,7 +221,7 @@ proc len*(node: TSNode): int =
     result = node.tsNodeNamedChildCount().int
 
 proc `[]`*(node: TSNode, i: SomeInteger): TSNode =
-  if i < node.len:
+  if i < type(i)(node.len()):
     result = node.tsNodeNamedChild(i.uint32)
 
 proc getName*(node: TSNode): string {.inline.} =