Definition Parser now consumes from the new Reader

Next steps are to move ParseUtils functionality to DefinitinoEx, sort
things into the namespaces in this library, and then prepare it to
handle overloading like nlohmann_json does. On the other side, it's
time for starshatter::data.

3 files changed, 353 insertions, 0 deletions

diff --git a/DefinitionEx/include/Parser_ss.h b/DefinitionEx/include/Parser_ss.h
new file mode 100644
index 0000000..50659b9
--- /dev/null
+++ b/DefinitionEx/include/Parser_ss.h
@@ -0,0 +1,43 @@
+/*  Starshatter: The Open Source Project
+    Copyright (c) 2021-2024, Starshatter: The Open Source Project Contributors
+    Copyright (c) 2011-2012, Starshatter OpenSource Distribution Contributors
+    Copyright (c) 1997-2006, Destroyer Studios LLC.
+
+    AUTHOR:       John DiCamillo
+
+
+    OVERVIEW
+    ========
+    Declaration of the generic Parser class
+*/
+
+#pragma once
+
+#include <starshatter/foundation/reader.h>
+#include <Text.h>
+
+#include "Term.h"
+
+// +-------------------------------------------------------------------+
+
+class Scanner;
+
+// +-------------------------------------------------------------------+
+
+class Parser
+{
+public:
+    Parser();
+    Parser(starshatter::foundation::Reader r);
+    ~Parser();
+
+    Term*          ParseTerm();
+    Term*          ParseTermBase();
+    Term*          ParseTermRest(Term* base);
+    TermList*      ParseTermList(int for_struct);
+    TermArray*     ParseArray();
+    TermStruct*    ParseStruct();
+
+private:
+    Scanner* lexer;
+};
diff --git a/DefinitionEx/include/Term.h b/DefinitionEx/include/Term.h
new file mode 100644
index 0000000..1bc30b6
--- /dev/null
+++ b/DefinitionEx/include/Term.h
@@ -0,0 +1,167 @@
+/*  Starshatter: The Open Source Project
+    Copyright (c) 2021-2024, Starshatter: The Open Source Project Contributors
+    Copyright (c) 2011-2012, Starshatter OpenSource Distribution Contributors
+    Copyright (c) 1997-2006, Destroyer Studios LLC.
+
+    AUTHOR:       John DiCamillo
+
+
+    OVERVIEW
+    ========
+    Declaration of the Abstract Syntax Tree classes
+*/
+
+#pragma once
+
+#include <Text.h>
+#include <List.h>
+
+// +-------------------------------------------------------------------+
+
+class Term;
+class TermBool;
+class TermNumber;
+class TermText;
+class TermArray;
+class TermDef;
+class TermStruct;
+
+// +-------------------------------------------------------------------+
+
+class Term
+{
+public:
+     static const char* TYPENAME() { return "Term"; }
+
+     Term()            { }
+     virtual ~Term()   { }
+
+     virtual int operator==(const Term& rhs) const { return 0; }
+
+     virtual void print(int level=10) { }
+
+     // conversion tests
+     virtual Term*              touch()     { return this; }
+     virtual TermBool*          isBool()    { return 0; }
+     virtual TermNumber*        isNumber()  { return 0; }
+     virtual TermText*          isText()    { return 0; }
+     virtual TermArray*         isArray()   { return 0; }
+     virtual TermDef*           isDef()     { return 0; }
+     virtual TermStruct*        isStruct()  { return 0; }
+};
+
+Term* error(const char*, const char* = 0);
+
+// +-------------------------------------------------------------------+
+
+typedef List<Term>      TermList;
+typedef ListIter<Term>  TermListIter;
+
+// +-------------------------------------------------------------------+
+
+class TermBool : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermBool"; }
+
+    TermBool(bool v) : val(v) { }
+
+    virtual void      print(int level=10);
+    virtual TermBool* isBool()      { return this; }
+    bool              value() const { return val;  }
+
+private:
+    bool val;
+};
+
+// +-------------------------------------------------------------------+
+
+class TermNumber : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermNumber"; }
+
+    TermNumber(double v) : val(v) { }
+
+    virtual void         print(int level=10);
+    virtual TermNumber*  isNumber()     { return this; }
+    double               value() const  { return val;  }
+
+private:
+    double val;
+};
+
+// +-------------------------------------------------------------------+
+
+class TermText : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermText"; }
+
+    TermText(const Text& v) : val(v)   { }
+
+    virtual void      print(int level=10);
+    virtual TermText* isText()       { return this; }
+    Text              value() const  { return val;  }
+
+private:
+    Text val;
+};
+
+// +-------------------------------------------------------------------+
+
+class TermArray : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermArray"; }
+
+    TermArray(TermList* elist);
+    virtual ~TermArray();
+
+    virtual void         print(int level=10);
+    virtual TermArray*   isArray()   { return this;  }
+    TermList*            elements()  { return elems; }
+
+private:
+    TermList*   elems;
+};
+
+// +-------------------------------------------------------------------+
+
+class TermStruct : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermStruct"; }
+
+    TermStruct(TermList* elist);
+    virtual ~TermStruct();
+
+    virtual void         print(int level=10);
+
+    virtual TermStruct*  isStruct()  { return this; }
+    TermList*            elements()  { return elems; }
+
+private:
+    TermList*   elems;
+};
+
+// +-------------------------------------------------------------------+
+
+class TermDef : public Term
+{
+public:
+    static const char* TYPENAME() { return "TermDef"; }
+
+    TermDef(TermText* n, Term* v) : mname(n), mval(v) { }
+    virtual ~TermDef();
+
+    virtual void         print(int level=10);
+    virtual TermDef*     isDef()     { return this; }
+
+    virtual TermText*    name()      { return mname; }
+    virtual Term*        term()      { return mval;  }
+
+private:
+    TermText*   mname;
+    Term*       mval;
+};
diff --git a/DefinitionEx/include/Token.h b/DefinitionEx/include/Token.h
new file mode 100644
index 0000000..3dd8eb9
--- /dev/null
+++ b/DefinitionEx/include/Token.h
@@ -0,0 +1,143 @@
+/*  Starshatter: The Open Source Project
+    Copyright (c) 2021-2024, Starshatter: The Open Source Project Contributors
+    Copyright (c) 2011-2012, Starshatter OpenSource Distribution Contributors
+    Copyright (c) 1997-2006, Destroyer Studios LLC.
+
+    AUTHOR:       John DiCamillo
+
+
+    OVERVIEW
+    ========
+    Scanner class definition
+*/
+
+#pragma once
+
+#include <Dictionary.h>
+#include <starshatter/foundation/reader.h>
+#include <Text.h>
+
+#pragma warning( disable : 4237)
+
+// +-------------------------------------------------------------------+
+
+class Token;
+class Scanner;
+
+// +-------------------------------------------------------------------+
+
+class Token
+{
+    friend class Scanner;
+
+public:
+    // keywords must be alphanumeric identifiers or symbolic identifiers
+    enum Types { Undefined, Keyword, AlphaIdent, SymbolicIdent, Comment,
+                     IntLiteral, FloatLiteral, StringLiteral, CharLiteral,
+                     Dot, Comma, Colon, Semicolon,
+                     LParen, RParen, LBracket, RBracket, LBrace, RBrace,
+                     EOT, LastTokenType };
+
+    enum Alias { CompoundSeparator   = Dot,
+                     ItemSeparator       = Comma,
+                     StatementTerminator = Semicolon,
+                     TypeIndicator       = Colon,
+                     Lambda              = LastTokenType + 1 };
+
+    Token();
+    Token(const Token& rhs);
+    Token(int t);
+    Token(const char* s, int t, int k=0, int l=0, int c=0);
+    Token(const Text& s, int t, int k=0, int l=0, int c=0);
+    ~Token();
+
+    Token& operator = (const Token& rhs);
+
+    bool     match(const Token& ref) const;
+
+    Text     symbol()   const;
+    int      type()     const { return mType;   }
+    int      key()      const { return mKey;    }
+    int      line()     const { return mLine;   }
+    int      column()   const { return mColumn; }
+
+    Text     typestr()  const;
+
+    static Text   describe(const Text& tok);
+    static void   addKey(const Text& k, int v);
+    static void   addKeys(Dictionary<int>& keys);
+    static bool   findKey(const Text& k, int& v);
+    static void   comments(const Text& begin, const Text& end);
+    static void   altComments(const Text& begin, const Text& end);
+    static void   hideComments(bool hide = true) { hidecom = hide; }
+
+    static char   comBeg(unsigned int i) { return combeg[i]; }
+    static char   comEnd(unsigned int i) { return comend[i]; }
+    static char   altBeg(unsigned int i) { return altbeg[i]; }
+    static char   altEnd(unsigned int i) { return altend[i]; }
+
+    static void   close();
+
+protected:
+    int      mLength;
+    union {
+        char  mSymbol[8];
+        char* mFullSymbol;
+    };
+    int      mType;
+    int      mKey;
+    int      mLine;
+    int      mColumn;
+
+    static bool             hidecom;
+    static char             combeg[3];
+    static char             comend[3];
+    static char             altbeg[3];
+    static char             altend[3];
+
+    static Dictionary<int> keymap;
+};
+
+// +-------------------------------------------------------------------+
+
+class Scanner
+{
+public:
+	Scanner();
+    Scanner(starshatter::foundation::Reader r);
+    Scanner(const Text&  s);
+    Scanner(const Scanner& rhs);
+    virtual ~Scanner();
+
+    Scanner& operator = (const Scanner& rhs);
+
+    void           Load(const Text& s);
+
+    enum Need { Demand, Request };
+    virtual Token  Get(Need n = Demand);
+
+    void           PutBack()     { index = old_index; line = old_line; }
+    int            GetCursor()   { return index;     }
+    int            GetLine()     { return line;      }
+    void           Reset(int c, int l) { index = old_index = c; line = old_line = l; }
+    Token          Best() const  { return best;      }
+
+protected:
+    virtual int    GetNumeric();
+    virtual bool   IsSymbolic(char c);
+    virtual bool   IsAlpha(char c);
+
+    starshatter::foundation::Reader reader;
+    char* str;
+
+    const char* p;
+    const char* eos;
+
+    size_t      index;
+    size_t      old_index;
+    Token       best;
+    size_t      length;
+    size_t      line;
+    size_t      old_line;
+    size_t      lineStart;
+};