parser.h

一个开源的网络开发库ACE

// -*- C++ -*-

//=============================================================================
/**
 *  @file    Parser.h
 *
 *  Parser.h,v 1.11 2002/10/13 03:58:57 kitty Exp
 *
 *  @author Nanbor Wang <[e-mail]>
 */
//=============================================================================

#ifndef _ACEXML_BASIC_PARSER_H_
#define _ACEXML_BASIC_PARSER_H_

#include "ace/pre.h"
#include "ACEXML/parser/parser/Parser_export.h"

#if !defined (ACE_LACKS_PRAGMA_ONCE)
#pragma once
#endif /* ACE_LACKS_PRAGMA_ONCE */

#include "ACEXML/common/XMLReader.h"
#include "ACEXML/common/LocatorImpl.h"
#include "ACEXML/common/NamespaceSupport.h"
#include "ACEXML/common/CharStream.h"
#include "ace/Obstack.h"
#include "ace/Functor.h"
#include "ace/SString.h"
#include "ace/Hash_Map_Manager.h"
#include "ace/Containers_T.h"
#include "ace/Auto_Ptr.h"
#include "ACEXML/parser/parser/Entity_Manager.h"

/**
 * @class ACEXML_Parser Parser.h "ACEXML/parser/parser/Parser.h"
 *
 * @brief A SAX based parser.
 *
 */
class ACEXML_PARSER_Export ACEXML_Parser : public ACEXML_XMLReader
{
public:
  /// Default constructor.
  ACEXML_Parser (void);

  /// Destructor.
  virtual ~ACEXML_Parser (void);

  /*
   * Return the current content handler.
   */
  virtual ACEXML_ContentHandler *getContentHandler (void) const;

  /*
   * Return the current DTD handler.
   */
  virtual ACEXML_DTDHandler *getDTDHandler (void) const;

  /*
   * Return the current entity resolver.
   */
  virtual ACEXML_EntityResolver *getEntityResolver (void) const;

  /*
   * Return the current error handler.
   */
  virtual ACEXML_ErrorHandler *getErrorHandler (void) const;

  /**
   * Look up the value of a feature.  This method allows
   * programmers to check whether a specific feature has been
   * activated in the parser.
   */
  virtual int getFeature (const ACEXML_Char *name ACEXML_ENV_ARG_DECL)
          ACE_THROW_SPEC ((ACEXML_SAXNotRecognizedException,
                           ACEXML_SAXNotSupportedException)) ;

  /**
   * Activating or deactivating a feature.
   */
  virtual void setFeature (const ACEXML_Char *name,
                           int boolean_value ACEXML_ENV_ARG_DECL)
    ACE_THROW_SPEC ((ACEXML_SAXNotRecognizedException,
                     ACEXML_SAXNotSupportedException)) ;

  /*
   * Look up the value of a property.
   */
  virtual void * getProperty (const ACEXML_Char *name ACEXML_ENV_ARG_DECL)
    ACE_THROW_SPEC ((ACEXML_SAXNotRecognizedException,
                     ACEXML_SAXNotSupportedException)) ;

  /*
   * Set the value of a property.
   */
  virtual void setProperty (const ACEXML_Char *name,
                            void *value ACEXML_ENV_ARG_DECL)
    ACE_THROW_SPEC ((ACEXML_SAXNotRecognizedException,
                     ACEXML_SAXNotSupportedException)) ;

  /*
   * Parse an XML document.
   */
  virtual void parse (ACEXML_InputSource *input ACEXML_ENV_ARG_DECL)
    ACE_THROW_SPEC ((ACEXML_SAXException)) ;

  /*
   * Parse an XML document from a system identifier (URI).
   */
  virtual void parse (const ACEXML_Char *systemId ACEXML_ENV_ARG_DECL)
        ACE_THROW_SPEC ((ACEXML_SAXException))
    ;

  /*
   * Allow an application to register a content event handler.
   */
  virtual void setContentHandler (ACEXML_ContentHandler *handler);

  /*
   * Allow an application to register a DTD event handler.
   */
  virtual void setDTDHandler (ACEXML_DTDHandler *handler);

  /*
   * Allow an application to register an entity resolver.
   */
  virtual void setEntityResolver (ACEXML_EntityResolver *resolver);

  /*
   * Allow an application to register an error event handler.
   */
  virtual void setErrorHandler (ACEXML_ErrorHandler *handler);

  // *** Helper functions for parsing XML

  /**
   * Skip any whitespaces encountered until the first non-whitespace
   * character is encountered and consumed from the current input
   * CharStream.
   *
   * @param whitespace Return a pointer to the string of skipped
   * whitespace after proper conversion.  Null if there's no
   * whitespace found.
   *
   * @return The first none-white space characters (which will be
   * consumed from the CharStream.)  If no whitespace is found, it
   * returns 0.
   *
   * @sa skip_whitespace_count
   */
  ACEXML_Char skip_whitespace (ACEXML_Char **whitespace);

  /**
   * Skip any whitespaces encountered until the first non-whitespace
   * character.  The first non-whitespace character is not consumed.
   * This method does peek into the input CharStream and therefore
   * is more expensive than @ref skip_whitespace.
   *
   * @param peek If non-null, @a peek points to a ACEXML_Char where
   *        skip_whitespace_count stores the first non-whitespace
   *        character it sees (character is not removed from the stream.)
   *
   * @return The number of whitespace characters consumed.
   *
   * @sa skip_whitespace
   */
  int skip_whitespace_count (ACEXML_Char *peek = 0);

  /**
   * Check if a character @a c is a whitespace.
   *
   * @retval 1 if @a c is a valid white space character. 0 otherwise.
   */
  int is_whitespace (ACEXML_Char c);

  /**
   * Check if a character @a c is a whitespace or '='.
   *
   * @retval 1 if true, 0 otherwise.
   */
  int is_whitespace_or_equal (ACEXML_Char c);

  /**
   * Check if a character @a c is a valid character for nonterminal NAME.
   *
   * @retval 1 if true, 0 otherwise.
   */
  int is_nonname (ACEXML_Char c);

  /**
   * Skip an equal sign.
   *
   * @retval 0 when succeeds, -1 if no equal sign is found.
   */
  int skip_equal (void);

  /**
   * Get a quoted string.  Quoted strings are used to specify
   * attribute values and this routine will replace character and
   * entity references on-the-fly.  Parameter entities are not allowed
   * (or replaced) in this function.  (But regular entities are.)
   *
   * @param str returns the un-quoted string.
   *
   * @retval 0 on success, -1 otherwise.
   */
  int get_quoted_string (ACEXML_Char *&str);

  /**
   * Parse a PI statement.  The first character encountered
   * should always be '?' in the PI prefix "@<?".
   *
   * @retval 0 on success, -1 otherwise.
   */
  int parse_processing_instruction (ACEXML_ENV_SINGLE_ARG_DECL);

  /**
   * Skip over a comment. The first character encountered
   * should always be the first '-' in the comment prefix
   * "@<@!--".
   */
  int grok_comment ();

  /**
   * Read a name from the input CharStream (until white space).
   * If @a ch @!= 0, then we have already consumed the first name
   * character from the input CharStream, otherwise, read_name
   * will use this->get() to acquire the initial character.
   *
   * @return A pointer to the string in the obstack, 0 if it's not
   * a valid name.
   */
  ACEXML_Char *read_name (ACEXML_Char ch = 0);

  /**
   * Parse the DOCTYPE declaration.  The first character encountered
   * should always be  'D' in doctype prefix: "@<@!DOCTYPE".
   */
  int parse_doctypedecl (ACEXML_ENV_SINGLE_ARG_DECL)
        ACE_THROW_SPEC ((ACEXML_SAXException))
    ;

  /**
   * Parse an XML element.  The first character encountered should
   * be the first character of the element "Name".
   *
   * @param is_root If not 0, then we are expecting to see the "root"
   * element now, and the next element's name need to match the name
   * defined in DOCTYPE definition, i.e., @a this->doctype_.
   *
   * @todo Instead of simply checking for the root element based on the
   * argument @a is_root, we should instead either pass in some sort
   * of validator or allow the function to return the element name so it
   * can be used in a validator.
   */
  void parse_element (int is_root ACEXML_ENV_ARG_DECL)
        ACE_THROW_SPEC ((ACEXML_SAXException))
    ;