diff --git a/core/SqlAstExpression.cpp b/core/SqlAstExpression.cpp
new file mode 100644
index 0000000..618fce3
--- /dev/null
+++ b/core/SqlAstExpression.cpp
@@ -0,0 +1,13 @@
+#include "SqlAstExpression.h"
+
+using namespace SqlAst;
+
+
+std::shared_ptr<SqlAst::Expression> parseExpression(SqlParser &parser)
+{
+	// get token, what is it?
+	// number
+	// symbol
+	// left parenthesis
+
+}
diff --git a/core/SqlAstExpression.h b/core/SqlAstExpression.h
new file mode 100644
index 0000000..6990fd5
--- /dev/null
+++ b/core/SqlAstExpression.h
@@ -0,0 +1,19 @@
+#ifndef SQLASTEXPRESSION_H
+#define SQLASTEXPRESSION_H
+
+#include "SqlAstNode.h"
+
+class SqlParser;
+
+namespace SqlAst {
+
+/// Base class for parts of expressions like calculations, comparisons, function calls etc...
+class Expression: public Node {
+
+};
+
+std::shared_ptr<SqlAst::Expression> parseExpression(SqlParser &parser);
+
+}
+
+#endif // SQLASTEXPRESSION_H
diff --git a/core/SqlAstNode.cpp b/core/SqlAstNode.cpp
new file mode 100644
index 0000000..19c59b4
--- /dev/null
+++ b/core/SqlAstNode.cpp
@@ -0,0 +1,8 @@
+#include "SqlAstNode.h"
+
+using namespace SqlAst;
+
+Node::Node()
+{
+
+}
diff --git a/core/SqlAstNode.h b/core/SqlAstNode.h
new file mode 100644
index 0000000..3c40f41
--- /dev/null
+++ b/core/SqlAstNode.h
@@ -0,0 +1,58 @@
+#ifndef SQLASTNODE_H
+#define SQLASTNODE_H
+
+#include <memory>
+#include <string>
+
+namespace SqlAst {
+
+class Node {
+public:
+	Node();
+	//virtual Node* Clone() const = 0;
+};
+
+class DDLNode: public Node {
+
+};
+
+class CreateTable: public Node {
+
+};
+
+
+// Is there a benefit for having a common base for crud operations???
+class CrudNode: public Node {
+
+};
+
+
+class Insert: public CrudNode {
+
+};
+
+/** Class for representing an identifier.
+ *
+ * This can still be multiple things like:
+ * - name of alias, schema, table and or column
+ * - name of function
+ * - predefined symbol like LOCAL_TIME
+ *
+ * During parsing this cannot always be determined as an alias might be involved that hasn't been parsed yet
+ * so we put a Identifier in the AST a follow up pass could determine what it actually as and act appropriatly
+ *
+ * An identifier can consist of following fields
+ *	- [[schema.]table.]column
+ *	- [alias.]column
+ *  - [schema.]table.function (for function taking a row of type table)
+ *  - alias.function (for function taking a row of type table)
+ *  - schema.function
+ *  - sql symbol like CURRENT_DATE
+ */
+class Identifier: public Node {
+
+};
+
+}
+
+#endif // SQLASTNODE_H
diff --git a/core/SqlAstSelect.cpp b/core/SqlAstSelect.cpp
new file mode 100644
index 0000000..dcf7a5a
--- /dev/null
+++ b/core/SqlAstSelect.cpp
@@ -0,0 +1,17 @@
+#include "SqlAstSelect.h"
+#include "SqlAstSelectList.h"
+#include "SqlParser.h"
+
+using namespace SqlAst;
+
+std::shared_ptr<SqlAst::Select> parseSelect(SqlParser &parser)
+{
+	std::shared_ptr<SqlAst::Select> ast_select = std::make_shared<SqlAst::Select>();
+	// parse select list of expression + aliasses, required
+	auto select_list = parseSelectList(parser);
+	ast_select->setSelectList(select_list);
+
+	// parse optional from list
+
+	return ast_select;
+}
diff --git a/core/SqlAstSelect.h b/core/SqlAstSelect.h
new file mode 100644
index 0000000..d2b7232
--- /dev/null
+++ b/core/SqlAstSelect.h
@@ -0,0 +1,39 @@
+#ifndef SQLASTSELECT_H
+#define SQLASTSELECT_H
+
+#include "SqlAstNode.h"
+//#include "SqlAstSelectList.h"
+#include <memory>
+
+class SqlParser;
+
+namespace SqlAst {
+
+class SelectList;
+class From;
+class Where;
+class GroupBy;
+class Having;
+class OrderBy;
+
+class Select: public CrudNode {
+public:
+
+	void setSelectList(std::shared_ptr<SelectList> list) { select = list; }
+	auto getSelectList() const { return select; }
+	void setFrom(std::shared_ptr<From> f) { from = f; }
+
+private:
+	std::shared_ptr<SelectList> select;
+	std::shared_ptr<From> from;
+	std::shared_ptr<Where> where;
+	std::shared_ptr<GroupBy> groupBy;
+	std::shared_ptr<Having> having;
+	std::shared_ptr<OrderBy> orderBy;
+};
+
+std::shared_ptr<SqlAst::Select> parseSelect(SqlParser &parser);
+
+}
+
+#endif // SQLASTSELECT_H
diff --git a/core/SqlAstSelectList.cpp b/core/SqlAstSelectList.cpp
new file mode 100644
index 0000000..65fec73
--- /dev/null
+++ b/core/SqlAstSelectList.cpp
@@ -0,0 +1,27 @@
+#include "SqlAstSelectList.h"
+#include "SqlAstSelectListEntry.h"
+#include "SqlParser.h"
+
+using namespace SqlAst;
+
+std::shared_ptr<SqlAst::SelectList> parseSelectList(SqlParser &parser)
+{
+	// parse select element
+	// whats next?
+	// comma -> parse an element
+	// something else return and let caller figure it out
+	auto ast_select_list = std::make_shared<SelectList>();
+	while (1) {
+		auto ast_select_elem = parseSelectListEntry(parser);
+		if (ast_select_elem) {
+			ast_select_list->add(ast_select_elem);
+		}
+		else {
+			// todo error
+		}
+		if (!parser.expectToken(BasicTokenType::Comma)) {
+			break;
+		}
+	}
+	return ast_select_list;
+}
diff --git a/core/SqlAstSelectList.h b/core/SqlAstSelectList.h
new file mode 100644
index 0000000..4f092e0
--- /dev/null
+++ b/core/SqlAstSelectList.h
@@ -0,0 +1,33 @@
+#ifndef SQLASTSELECTLIST_H
+#define SQLASTSELECTLIST_H
+
+#include "SqlAstNode.h"
+#include <memory>
+#include <vector>
+
+class SqlParser;
+
+namespace SqlAst {
+
+class SelectListEntry;
+
+class SelectList: public Node {
+public:
+	using EntrySPtr = std::shared_ptr<SelectListEntry>;
+
+	void add(EntrySPtr entry)
+	{
+		entryList.push_back(entry);
+	}
+
+private:
+	using EntryList = std::vector<EntrySPtr>;
+
+	EntryList entryList;
+};
+
+std::shared_ptr<SqlAst::SelectList> parseSelectList(SqlParser &parser);
+
+}
+
+#endif // SQLASTSELECTLIST_H
diff --git a/core/SqlAstSelectListEntry.cpp b/core/SqlAstSelectListEntry.cpp
new file mode 100644
index 0000000..dcf74a9
--- /dev/null
+++ b/core/SqlAstSelectListEntry.cpp
@@ -0,0 +1,34 @@
+#include "SqlAstSelectListEntry.h"
+#include "SqlAstExpression.h"
+#include "SqlParser.h"
+
+using namespace SqlAst;
+
+std::shared_ptr<SelectListEntry> parseSelectListEntry(SqlParser &parser)
+{
+
+	// parse expression
+	auto ast_expr = parseExpression(parser);
+
+	if (parser.consumeOptionalKeyword(Keyword::As)) {
+		// alias required!
+		auto token = parser.expectSymbol();
+
+
+	}
+	else {
+
+		// optional alias
+
+	}
+	// can come three groups of things:
+	// - AS keyword, alias should follow, followed by comma
+	// - a symbol ie an alias followed by comma
+	// - something else ie comma, FROM or INTO or... is something one of the callers should figure out as long as we have valid
+	// selectElem we are happy
+
+
+	auto ast = std::make_shared<SelectListEntry>();
+
+	return ast;
+}
diff --git a/core/SqlAstSelectListEntry.h b/core/SqlAstSelectListEntry.h
new file mode 100644
index 0000000..9ad7476
--- /dev/null
+++ b/core/SqlAstSelectListEntry.h
@@ -0,0 +1,29 @@
+#ifndef SQLASTSELECTLISTENTRY_H
+#define SQLASTSELECTLISTENTRY_H
+
+#include "SqlAstNode.h"
+#include <QString>
+#include <memory>
+#include <vector>
+
+class SqlParser;
+
+namespace SqlAst {
+
+class Expression;
+
+class SelectListEntry: public Node {
+public:
+	void setExpression(std::shared_ptr<Expression> expr) { this->expression = expr; }
+	void setAlias(QString a) { this->alias = alias; }
+private:
+	std::shared_ptr<Expression> expression;
+	QString alias;
+};
+
+std::shared_ptr<SqlAst::SelectListEntry> parseSelectListEntry(SqlParser &parser);
+
+}
+
+
+#endif // SQLASTSELECTLISTENTRY_H
diff --git a/core/SqlLexer.cpp b/core/SqlLexer.cpp
index a25c55a..3a23dac 100644
--- a/core/SqlLexer.cpp
+++ b/core/SqlLexer.cpp
@@ -110,7 +110,11 @@ bool SqlLexer::nextBasicToken(int &startpos, int &length, BasicTokenType &tokent
 			}
 			if (isSelf(c)) {
 				length = m_pos - startpos;
-				tokentype = BasicTokenType::Self;
+				if (c == ',')
+					tokentype = BasicTokenType::Comma;
+				else
+					tokentype = BasicTokenType::Self;
+
 				QStringRef sr(&m_block, startpos, length);
 				out = sr.toString();
 				return true;
diff --git a/core/SqlLexer.h b/core/SqlLexer.h
index dc23db8..3641714 100644
--- a/core/SqlLexer.h
+++ b/core/SqlLexer.h
@@ -15,7 +15,8 @@ enum class BasicTokenType {
 	QuotedIdentifier,
 	Parameter,
 	Operator,
-	Self, // single char representing it self
+	Self, // single char representing it self, maybe remove this and replace with token for each possibility
+	Comma,
 	Cast
 };
 
@@ -25,6 +26,14 @@ enum class LexerState {
 	InBlockComment
 };
 
+class SqlToken {
+public:
+	bool ok;
+	int startPos;
+	int length;
+	BasicTokenType tokenType;
+	QString out;
+};
 
 class SqlLexer {
 public:
@@ -40,6 +49,12 @@ public:
 	 * @return false when input seems invalid, it will return what it did recognize but something wasn't right, parser should try to recover
 	 */
 	bool nextBasicToken(int &startpos, int &length, BasicTokenType &tokentype, QString &out);
+	SqlToken nextBasicToken()
+	{
+		SqlToken token;
+		token.ok = !nextBasicToken(token.startPos, token.length, token.tokenType, token.out);
+		return token;
+	}
 
 	LexerState currentState() const { return m_state; }
 private:
diff --git a/core/SqlParser.cpp b/core/SqlParser.cpp
new file mode 100644
index 0000000..8ff1ddf
--- /dev/null
+++ b/core/SqlParser.cpp
@@ -0,0 +1,75 @@
+#include "SqlParser.h"
+#include "SqlAstSelect.h"
+#include <unordered_map>
+
+using namespace SqlAst;
+
+Keyword isKeyword(QString symbol)
+{
+	static std::unordered_map<std::string, Keyword> lookup_map = {
+		{ "as",     Keyword::As },
+		{ "by",     Keyword::By },
+		{ "delete", Keyword::Delete },
+		{ "from",   Keyword::From },
+		{ "group",  Keyword::Group },
+		{ "insert", Keyword::Insert },
+		{ "order",  Keyword::Order },
+		{ "select", Keyword::Select },
+		{ "update", Keyword::Update },
+		{ "where",  Keyword::Where }
+	};
+
+	auto res = lookup_map.find(symbol.toLower().toUtf8().data());
+	if (res != lookup_map.end())
+		return res->second;
+	else
+		return Keyword::NotAKeyword;
+}
+
+
+SqlParser::SqlParser(SqlLexer &lexer)
+	: lexer(lexer)
+{
+
+}
+
+void SqlParser::parse()
+{
+	// Basic algo:
+	// LOOP
+	//   GET token
+	//   IF NOT try_reduce(token)
+	//   THEN SHIFT
+	// END LOOP
+	while (1) {
+		SqlToken token = lexer.nextBasicToken();
+		if (token.ok) {
+			if (token.tokenType == BasicTokenType::Symbol) {
+				Keyword kw = isKeyword(token.out);
+				switch (kw) {
+				case Keyword::Select:
+					parseSelect(*this);
+					break;
+
+				case Keyword::NotAKeyword:
+				default:
+					// unexpected
+					break;
+				}
+			}
+		}
+		else {
+			// error during lexical analysis, need to recover
+
+		}
+	}
+}
+
+
+
+
+//bool try_reduce(SqkToken token)
+//{
+//	// what state are we in? what are we expecting
+
+//}
diff --git a/core/SqlParser.h b/core/SqlParser.h
new file mode 100644
index 0000000..621675b
--- /dev/null
+++ b/core/SqlParser.h
@@ -0,0 +1,72 @@
+#ifndef SQLPARSER_H
+#define SQLPARSER_H
+
+#include "SqlLexer.h"
+#include <memory>
+#include <optional>
+
+enum class Keyword {
+	NotAKeyword,
+	As,
+	By,
+	Delete,
+	From,
+	Group,
+	Insert,
+	Order,
+	Select,
+	Update,
+	Where,
+
+};
+
+namespace SqlAst {
+
+	class Select;
+	class SelectList;
+
+}
+// The parsing works by calling functions that know either a global part
+// or a smaller specific part and is thus recursive. At certain points a function parsing something specific
+// will reach a point where it is in a valid terminal state and it encounters something that cannot be a continuation
+// of what it is parsing and thus returns succesfully or it is in a non terminal state and encounters something unexpected
+// In both cases it will return hoping that one of the functions above can continue (and recover from the error if needed)
+
+class SqlParser
+{
+public:
+	explicit SqlParser(SqlLexer &lexer);
+
+	void parse();
+
+	/** Checks to see if the next token is the expected keyword.
+	 *
+	 * If it is the token is consumed and the function returns true.
+	 * Otherwise false is returned
+	 */
+	std::optional<SqlToken> expectKeyword(Keyword kw);
+	std::optional<SqlToken> expectSymbol();
+	std::optional<SqlToken> expectToken(BasicTokenType tt);
+
+	/** If the next token is Keyword kw consume it otherwise do nothing.
+	 * In some cases the return value is unimportant as the keyword is completely optional
+	 * in other cases the optional keywords presence might force the next token to be something specific
+	 *
+	 * \return true if the token was found
+	 */
+	bool consumeOptionalKeyword(Keyword kw);
+private:
+
+	//using TokenStack = std::stack<SqlToken>;
+
+	//TokenStack tokenStack;
+
+	SqlLexer &lexer;
+
+	//bool try_reduce(SqkToken token);
+
+
+};
+
+
+#endif // SQLPARSER_H
diff --git a/core/core.pro b/core/core.pro
index d716f57..7b165a1 100644
--- a/core/core.pro
+++ b/core/core.pro
@@ -39,7 +39,13 @@ SOURCES += my_boost_assert_handler.cpp \
 	QueuedBackgroundTask.cpp \
     ExplainTreeModelItem.cpp \
     jsoncpp.cpp \
-    WorkManager.cpp
+    WorkManager.cpp \
+    SqlParser.cpp \
+    SqlAstNode.cpp \
+    SqlAstSelectList.cpp \
+    SqlAstSelectListEntry.cpp \
+    SqlAstSelect.cpp \
+    SqlAstExpression.cpp
 
 HEADERS += PasswordManager.h \
     SqlLexer.h \
@@ -54,7 +60,13 @@ HEADERS += PasswordManager.h \
     TaskControl.h \
     ControllableTask.h \
     RunControllableTask.h \
-    TaskExecutor.h
+    TaskExecutor.h \
+    SqlParser.h \
+    SqlAstNode.h \
+    SqlAstSelectList.h \
+    SqlAstSelectListEntry.h \
+    SqlAstSelect.h \
+    SqlAstExpression.h
 	
 unix {
     target.path = /usr/lib