From a382841dc7d554b2ba3920d52f73e4cec428743e Mon Sep 17 00:00:00 2001
From: Marvin Borner
Date: Mon, 12 Aug 2019 16:01:51 +0200
Subject: Basic parser functions

---
 src/Lexical.kt | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 64 insertions(+)
 create mode 100644 src/Lexical.kt

(limited to 'src/Lexical.kt')

diff --git a/src/Lexical.kt b/src/Lexical.kt
new file mode 100644
index 0000000..bd8bc3a
--- /dev/null
+++ b/src/Lexical.kt
@@ -0,0 +1,64 @@
+class Lexical {
+    fun analyze(source: String): MutableList<Pair<String, TokenType>> {
+        var buffer = ""
+        val tokens = mutableListOf<Pair<String, TokenType>>()
+        for (i in source.indices) {
+            if (source.length > i + 1 && getType(source[i]) == getType(source[i + 1])) {
+                buffer += source[i]
+            } else {
+                buffer += source[i]
+                if (getType(source[i]) != Type.Empty)
+                    tokens += buffer to getTokenType(buffer to getType(source[i]))
+                buffer = ""
+            }
+        }
+        return tokens
+    }
+
+    /**
+     * Matches the characters to groups ([Type])
+     */
+    private fun getType(char: Char): Type {
+        return when (char) {
+            in 'a'..'z', in 'A'..'Z' -> Type.Alphabetic
+            in '1'..'9' -> Type.Numeric
+            ' ' -> Type.Empty
+            else -> Type.Unknown
+        }
+    }
+
+    /**
+     * Matches the tokens to a [TokenType]
+     */
+    private fun getTokenType(token: Pair<String, Type>): TokenType {
+        var type = when (token.first) {
+            "print" -> TokenType.Keyword // TODO: DataType matching
+            "=", "+", "-", "*", "/", "==", ">", "<" -> TokenType.Operator
+            ";", ":" -> TokenType.Symbol
+            else -> TokenType.Unknown
+        }
+        if (type == TokenType.Unknown)
+            type = when (token.second) {
+                Type.Numeric -> TokenType.Constant
+                Type.Alphabetic -> TokenType.Identifier
+                else -> TokenType.Unknown
+            }
+        return type
+    }
+}
+
+private enum class Type {
+    Alphabetic,
+    Numeric,
+    Empty,
+    Unknown
+}
+
+enum class TokenType {
+    Keyword,
+    Identifier,
+    Operator,
+    Constant,
+    Symbol,
+    Unknown
+}
\ No newline at end of file
-- 
cgit v1.2.3