BaseType.h

// -*- mode: c++; c-basic-offset:4 -*-

// This file is part of libdap, A C++ implementation of the OPeNDAP Data
// Access Protocol.

// Copyright (c) 2002,2003 OPeNDAP, Inc.
// Author: James Gallagher <jgallagher@opendap.org>
//         Dan Holloway <dan@hollywood.gso.uri.edu>
//         Reza Nekovei <reza@intcomm.net>
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public
// License as published by the Free Software Foundation; either
// version 2.1 of the License, or (at your option) any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
// Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
//
// You can contact OPeNDAP, Inc. at PO Box 112, Saunderstown, RI. 02874-0112.

// (c) COPYRIGHT URI/MIT 1994-1999
// Please read the full copyright statement in the file COPYRIGHT_URI.
//
// Authors:
//      jhrg,jimg       James Gallagher <jgallagher@gso.uri.edu>
//      dan             Dan Holloway <dan@hollywood.gso.uri.edu>
//      reza            Reza Nekovei <reza@intcomm.net>

// Abstract base class for the variables in a dataset. This is used to store
// the type-invariant information that describes a variable as given in the
// DODS API.
//
// jhrg 9/6/94

#ifndef _basetype_h
#define _basetype_h 1

#include <vector>
#include <stack>
#include <iostream>
#include <string>

#include "AttrTable.h"

#include "InternalErr.h"

#include "dods-datatypes.h"
#include "Type.h"

#include "DapObj.h"

using namespace std;

class Crc32;

namespace libdap
{

class ConstraintEvaluator;

class DDS;
class Marshaller;
class UnMarshaller;

class Constructor;
class XMLWrter;

class DMR;
class D4Group;
class XMLWriter;
class D4StreamMarshaller;
class D4StreamUnMarshaller;

class D4Attributes;

/** This defines the basic data type features for the DODS data access
    protocol (DAP) data types. All the DAP type classes (Float64, Array,
    etc.) subclass it. This class is an abstract one; no variables will ever
    be stored as BaseType instances, only as instances of its child classes.

    These classes and their methods give a user the capacity to set up
    sophisticated data types. They do <i>not</i> provide sophisticated ways to
    access and use this data. On the server side, in many cases, the class
    instances will have no data in them at all until the
    <tt>serialize</tt> function
    is called to send data to the client. On the client side, most DAP
    application programs will unpack the data promptly into whatever local
    data structure the programmer deems the most useful.

    In order to use these classes on the server side of a DAP
    client/server connection, you must write a <tt>read</tt> method
    for each of the data types you expect to encounter in the
    application. This function, whose purpose is to read data from a
    local source into the class instance data buffer, is called in
    <tt>serialize</tt>, when the data is about to be sent to the
    client.  The <tt>read</tt> function may be called earlier, in the
    case of data subset requests (constraint expressions) whose
    evaluation requires it. (For example, the constraint expression
    ``<tt>a,b&b>c</tt>'' requires that <tt>c</tt> be read even though
    it will not be sent.)

    For some data types, the <tt>read</tt> function must be aware of
    the constraints
    to be returned. These cautions are outlined where they occur.

	@note This class is ued by both DAP2 and DAP4.

    @brief The basic data type for the DODS DAP types.  */

class BaseType : public DapObj
{
private:
    string d_name;  // name of the instance
    Type d_type;   // instance's type
    string d_dataset; // name of the dataset used to create this BaseType

    bool d_is_read;  // true if the value has been read
    bool d_is_send;  // Is the variable in the projection?

    // d_parent points to the Constructor or Vector which holds a particular
    // variable. It is null for simple variables. The Vector and Constructor
    // classes must maintain this variable.
    BaseType *d_parent;

    // Attributes for this variable. Added 05/20/03 jhrg
    AttrTable d_attr;

    D4Attributes *d_attributes;

    bool d_is_dap4;         // True if this is a DAP4 variable, false ... DAP2

    // These are non-empty only for DAP4 variables. Added 9/27/12 jhrg

protected:
    // These were/are used for DAP2 CEs, but not for DAP4 ones
    bool d_in_selection; // Is the variable in the selection?
    bool d_is_synthesized; // true if the variable is synthesized

    void m_duplicate(const BaseType &bt);

public:
    typedef stack<BaseType *> btp_stack;

    // These ctors assume is_dap4 is false
    BaseType(const string &n, const Type &t, bool is_dap4 = false);
    BaseType(const string &n, const string &d, const Type &t, bool is_dap4 = false);

    BaseType(const BaseType &copy_from);
    virtual ~BaseType();

    virtual string toString();

    virtual void transform_to_dap4(D4Group *root, Constructor *container);
    virtual std::vector<BaseType *> *transform_to_dap2(AttrTable *parent_attr_table);

    virtual void dump(ostream &strm) const ;

    BaseType &operator=(const BaseType &rhs);

    /**
     * Remove any read or set data in the private data of the variable,
     * setting read_p() to false. Used to clear any dynamically allocated
     * storage that holds (potentially large) data. For the simple types,
     * this no-op version is all that's needed. Vector and some other classes
     * define a special version and have serialize() implementations that
     * call it to free data as soon as possible after sending it.
     *
     * @note Added 7/5/15 jhrg
     * @note Any specialization of this should make sure to reset the read_p
     * property.
     */
    virtual void clear_local_data() { set_read_p(false); }

    virtual bool is_dap4() const { return d_is_dap4; }
    virtual void set_is_dap4(const bool v) { d_is_dap4 = v;}

    /** Clone this instance. Allocate a new instance and copy \c *this into
	it. This method must perform a deep copy.

        @note This method should \e not copy data values, but must copy all
        other fields in the object.
	@return A newly allocated copy of \c this. */
    virtual BaseType *ptr_duplicate() = 0;

    virtual string name() const;
    virtual void set_name(const string &n);
    virtual std::string FQN() const;

    virtual Type type() const;
    virtual void set_type(const Type &t);
    virtual string type_name() const;

    virtual string dataset() const ;

    /**
     * @brief How many elements are in this variable.
     * @todo change the return type to int64_t
     * @return The number of elements; 1 for scalars
     */
    virtual int length() const { return 1; }

    /**
     * @brief Set the number of elements for this variable
     * @todo change param type to int64_t
     * @param l The number of elements
     */
    virtual void set_length(int) { }

    virtual bool is_simple_type() const;
    virtual bool is_vector_type() const;
    virtual bool is_constructor_type() const;

    virtual bool synthesized_p();
    virtual void set_synthesized_p(bool state);

    virtual int element_count(bool leaves = false);

    virtual bool read_p();
    virtual void set_read_p(bool state);

    virtual bool send_p();
    virtual void set_send_p(bool state);

    virtual AttrTable &get_attr_table();
    virtual void set_attr_table(const AttrTable &at);

    // DAP4 attributes
    virtual D4Attributes *attributes();
    virtual void set_attributes(D4Attributes *);
    virtual void set_attributes_nocopy(D4Attributes *);

    virtual bool is_in_selection();
    virtual void set_in_selection(bool state);

    virtual void set_parent(BaseType *parent);
    virtual BaseType *get_parent() const;

    virtual void transfer_attributes(AttrTable *at);

    // I put this comment here because the version in BaseType.cc does not
    // include the exact_match or s variables since they are not used. Doxygen
    // was gaging on the comment.

    /** Returns a pointer to the contained variable in a composite class. The
        composite classes are those made up of aggregated simple data types.
        Array, Grid, and Structure are composite types, while Int and Float are
        simple types. This function is only used by composite classes. The
        BaseType implementation always returns null.

        Several of the subclasses provide alternate access methods
        that make sense for that particular data type. For example,
        the Array class defines a <tt>*var(int i)</tt> method that
        returns the ith entry in the Array data, and the Structure
        provides a <tt>*var(Vars_iter)</tt> function using a
        pseudo-index to access the different members of the structure.

        @brief Returns a pointer to a member of a constructor class.
        @param name The name of the class member.  Defaults to ""
        @param exact_match True if only interested in variables whose
        full names match \e n exactly. If false, returns the first
        variable whose name matches \e name. For example, if \e name
        is \c x and \c point.x is a variable, then var("x", false)
        would return a BaseType pointer to \c point.x. If \e
        exact_match was <tt>true</tt> then \e name would need to be \c
        "point.x" for var to return that pointer. This feature
        simplifies constraint expressions for datasets which have
        complex, nested, constructor variables. Defaults to true.
        @param s Record the path to \e name. Defaults to null, in
        which case it is not used.
        @return A pointer to the member named in the \e n argument. If
        no name is given, the function returns the first (only)
        variable. For example, an Array has only one variable, while a
        Structure can have many. */
    virtual BaseType *var(const string &name = "", bool exact_match = true, btp_stack *s = 0);
    virtual BaseType *var(const string &name, btp_stack &s);

    virtual void add_var(BaseType *bt, Part part = nil);
    virtual void add_var_nocopy(BaseType *bt, Part part = nil);

    virtual bool read();

    virtual bool check_semantics(string &msg, bool all = false);

    virtual bool ops(BaseType *b, int op);
    virtual bool d4_ops(BaseType *b, int op);

    virtual unsigned int width(bool constrained = false) const;

    virtual void print_decl(FILE *out, string space = "    ",
                            bool print_semi = true,
                            bool constraint_info = false,
                            bool constrained = false);

    virtual void print_xml(FILE *out, string space = "    ",
                           bool constrained = false);

    virtual void print_decl(ostream &out, string space = "    ",
                            bool print_semi = true,
                            bool constraint_info = false,
                            bool constrained = false);

    virtual void print_xml(ostream &out, string space = "    ",
                           bool constrained = false);

    virtual void print_xml_writer(XMLWriter &xml, bool constrained = false);

    virtual void print_dap4(XMLWriter &xml, bool constrained = false);

    /** @name Abstract Methods */
    //@{
#if 0
    /** Return the number of bytes that are required to hold the
	instance's value. In the case of simple types such as Int32,
	this is the size of one Int32 (four bytes). For a String or
	Url type, <tt>width(bool constrained = false)</tt> returns the number of bytes needed
	for a <tt>String *</tt> variable, not the bytes needed for all
	the characters, since that value cannot be determined from
	type information alone. For Structure, and other constructor
	types size() returns the number of bytes needed to store
	pointers to the C++ objects.

	@brief Returns the size of the class instance data. */
    virtual unsigned int width(bool constrained = false) = 0;
#endif
    /** Reads the class data into the memory referenced by <i>val</i>.
	The caller should either allocate enough storage to <i>val</i>
	to hold the class data or set \c *val to null. If <i>*val</i>
	is NULL, memory will be allocated by this function with
	<tt>new()</tt>. If the memory is allocated this way, the
	caller is responsible for deallocating that memory. Array and
	values for simple types are stored as C would store an array.

    @deprecated Use value() in the leaf classes.

	@brief Reads the class data.

	@param val A pointer to a pointer to the memory into which the
	class data will be copied. If the value pointed to is NULL,
	memory will be allocated to hold the data, and the pointer
	value modified accordingly. The calling program is responsible
	for deallocating the memory references by this pointer.

	@return The size (in bytes) of the information copied to <i>val</i>.
    */
    virtual unsigned int buf2val(void **val) = 0;

    /** Store the value pointed to by <i>val</i> in the object's
	internal buffer. This function does not perform any checks, so
	users must be sure that the thing pointed to can actually be
	stored in the object's buffer.

	Only simple objects (Int, Float, Byte, and so on) and arrays
	of these simple objects may be stored using this function. To
	put data into more complex constructor types, use the
	functions provided by that class.

    @deprecated Use set_value() in the leaf classes.

	@brief Loads class data.

	@param val A pointer to the data to be inserted into the class
	data buffer.

	@param reuse A boolean value, indicating whether the class
	internal data storage can be reused or not. If this argument
	is TRUE, the class buffer is assumed to be large enough to
	hold the incoming data, and it is <i>not</i> reallocated. If
	FALSE, new storage is allocated. If the internal buffer has
	not been allocated at all, this argument has no effect. This
	is currently used only in the Vector class.

	@return The size (in bytes) of the information copied from
	<i>val</i>.
	@see Grid
	@see Vector::val2buf */
    virtual unsigned int val2buf(void *val, bool reuse = false) = 0;

    /** Similar to using serialize() and deserialize() together in one object.
        Data are read as for serialize and those values are stored in the
        objects as deserialize() does but does not write and then read data
        to/from a stream.

        This method is defined by the various data type classes. It calls the
        read() abstract method. Unlike serialize(), this method does not
        clear the memory use to hold the data values, so the caller should
        make sure to delete the DDS or the variable as soon as possible.

        @param eval Use this as the constraint expression evaluator.
        @param dds The Data Descriptor Structure object corresponding
        to this dataset. See <i>The DODS User Manual</i> for
        information about this structure. */
    virtual void intern_data(ConstraintEvaluator &eval, DDS &dds);

    /** Sends the data from the indicated (local) dataset through the
	connection identified by the Marshaller parameter. If the
	data is not already incorporated into the DDS object, read the
	data from the dataset. Once the data are sent (written to the
	Marshaller), they are deleted from the object and the object
	state is reset so that they will be read again if the read()
	method is called.

	This function is only used on the server side of the
	client/server connection, and is generally only called from
	the ResponseBuilder functions. It has no BaseType
	implementation; each datatype child class supplies its own
	implementation.

	@brief Move data to the net, then remove them from the object.

        @param eval Use this as the constraint expression evaluator.
	@param dds The Data Descriptor Structure object corresponding
	to this dataset. See <i>The DODS User Manual</i> for
	information about this structure.
	@param m A marshaller used to serialize data types
	@param ce_eval A boolean value indicating whether to evaluate
	the DODS constraint expression that may accompany this
	dataset. The constraint expression is stored in the <i>dds</i>.
	@return This method always returns true. Older versions used
	the return value to signal success or failure.

	@note We changed the default behavior of this method so that it
	calls BaseType::clear_local_data() once the values are sent. This,
	combined with the behavior that read() is called by this method
	just before data are sent, means that data for any given variable
	remain in memory for the shortest time possible. Furthermore, since
	variables are serialized one at a time, no more than one variable's
	data will be in memory at any given time when using the default
	behavior. Some code - code that uses intern_data() or server functions -
	might alter this default behavior. Only Array (i.e. Vector), Sequence,
	D4Sequence and D4Opaque types actually hold data in dynamically allocated
	memory, so sonly those types have the new/changed behavior.
	This change was made on 7/5/15.

	@exception InternalErr.
	@exception Error.
	@see DDS */
    virtual bool serialize(ConstraintEvaluator &eval, DDS &dds, Marshaller &m, bool ce_eval = true);

#if 0
    /**
     * Provide a way to get the old behavior of serialize() - calling this
     * method will serialize the BaseType object's data but _not_ delete its
     * data storage.
     *
     * @note This method's behavior differs only for Array (i.e. Vector), Sequence,
     * D4Sequence and D4Opaque types; the other types do not use dynamic memory to
     * hold data values.
     *
     * @param eval Use this as the constraint expression evaluator.
     * @param dds The Data Descriptor Structure object corresponding
     * to this dataset. See <i>The DODS User Manual</i> for
     * information about this structure.
     * @param m A marshaller used to serialize data types
     * @param ce_eval A boolean value indicating whether to evaluate
     * the DODS constraint expression that may accompany this
     * @return This method always returns true. Older versions used
     * the return value to signal success or failure.
     * @param
     */
    virtual bool serialize_no_release(ConstraintEvaluator &eval, DDS &dds, Marshaller &m, bool ce_eval = true) {
        return serialize(eval, dds, m, ce_eval);
    }
#endif

    /**
     * @brief include the data for this variable in the checksum
     * DAP4 includes a checksum with every data response. This method adds the
     * variable's data to that checksum.
     * @param checksum A Crc32 instance that holds the current checksum.
     */
    virtual void compute_checksum(Crc32 &checksum) = 0;

    virtual void intern_data(/*Crc32 &checksum, DMR &dmr, ConstraintEvaluator &eval*/);

    /**
     * @brief The DAP4 serialization method.
     * Serialize a variable's values for DAP4. This does not write the DMR
     * persistent representation but does write that part of the binary
     * data blob that holds a variable's data. Once a variable's data are
     * serialized, that memory is reclaimed (by calling BaseType::clear_local_data())
     *
     * @param m
     * @param dmr
     * @param eval
     * @param filter True if there is one variable that should be 'filtered'
     * @exception Error or InternalErr
     */
    virtual void serialize(D4StreamMarshaller &m, DMR &dmr, bool filter = false);

#if 0
    /**
     * @brief Variation on the DAP4 serialization method - retain data after serialization
     * Serialize a variable's values for DAP4. This does not write the DMR
     * persistent representation but does write that part of the binary
     * data blob that holds a variable's data. Once a variable's data are
     * serialized, that memory is reclaimed (by calling BaseType::clear_local_data())
     *
     * @note This version does not delete the storage of Array, D4Sequence or
     * D4Opaque variables, as it the case with serialize(). For other types,
     * this method and serialize have the same beavior (since those types do
     * not us dynamic memory to hold data values).
     *
     * @param m
     * @param dmr
     * @param eval
     * @param filter True if there is one variable that should be 'filtered'
     * @exception Error or InternalErr
     */
    virtual void serialize_no_release(D4StreamMarshaller &m, DMR &dmr, bool filter = false) {
        serialize(m, dmr, filter);
    }
#endif

    /** Receives data from the network connection identified by the
	<tt>source</tt> parameter. The data is put into the class data
	buffer according to the input <tt>dds</tt>.

	This function is only used on the client side of the DODS
	client/server connection.

	@brief Receive data from the net.

	@param um An UnMarshaller that knows how to deserialize data types
	@param dds The Data Descriptor Structure object corresponding
	to this dataset. See <i>The DODS User Manual</i> for
	information about this structure. This would have been
	received from the server in an earlier transmission.
	@param reuse A boolean value, indicating whether the class
	internal data storage can be reused or not. If this argument
	is TRUE, the class buffer is assumed to be large enough to
	hold the incoming data, and it is <i>not</i> reallocated. If
	FALSE, new storage is allocated. If the internal buffer has
	not been allocated at all, this argument has no effect.
	@return Always returns TRUE.
	@exception Error when a problem reading from the UnMarshaller is
	found.
	@see DDS */
    virtual bool deserialize(UnMarshaller &um, DDS *dds, bool reuse = false);

    /**
     * The DAP4 deserialization method.
     * @param um
     * @param dmr
     * @exception Error or InternalErr
     */
    virtual void deserialize(D4StreamUnMarshaller &um, DMR &dmr);

    /** Prints the value of the variable, with its declaration. This
	function is primarily intended for debugging DODS
	applications. However, it can be overloaded and used to do
	some useful things. Take a look at the asciival and writeval
	clients, both of which overload this to output the values of
	variables in different ways.

	@brief Prints the value of the variable.

	@param out The output stream on which to print the value.
	@param space This value is passed to the print_decl()
	function, and controls the leading spaces of the output.
	@param print_decl_p A boolean value controlling whether the
	variable declaration is printed as well as the value. */

    virtual void print_val(FILE *out, string space = "",
                           bool print_decl_p = true);

    /** Prints the value of the variable, with its declaration. This
	function is primarily intended for debugging DODS
	applications. However, it can be overloaded and used to do
	some useful things. Take a look at the asciival and writeval
	clients, both of which overload this to output the values of
	variables in different ways.

	@brief Prints the value of the variable.

	@param out The output ostream on which to print the value.
	@param space This value is passed to the print_decl()
	function, and controls the leading spaces of the output.
	@param print_decl_p A boolean value controlling whether the
	variable declaration is printed as well as the value. */
    virtual void print_val(ostream &out, string space = "",
                           bool print_decl_p = true) = 0;
    //@}
};

} // namespace libdap

#endif // _basetype_h