utils/configtool/src/htmlparser.h

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        htmlparser.cpp
   3 // Purpose:     Simple HTML parser
   4 // Author:      Julian Smart
   5 // Modified by:
   6 // Created:     2002-09-25
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) Julian Smart
   9 // Licence:     wxWindows license
  10 /////////////////////////////////////////////////////////////////////////////
  11
  12 #ifndef _HTMLPARSER_H_
  13 #define _HTMLPARSER_H_
  14
  15 #if defined(__GNUG__) && !defined(NO_GCC_PRAGMA)
  16     #pragma interface "htmlparser.cpp"
  17 #endif
  18
  19 //#include "wx/module.h"
  20 #include "wx/stream.h"
  21
  22 /*
  23  * wxSimpleHtmlAttribute
  24  * Representation of an attribute
  25  */
  26
  27 class wxSimpleHtmlAttribute
  28 {
  29     friend class wxSimpleHtmlTag;
  30 public:
  31     wxSimpleHtmlAttribute(const wxString& name, const wxString& value)
  32     {
  33         m_name = name; m_value = value; m_next = NULL;
  34     }
  35 //// Operations
  36
  37     // Write this attribute
  38     void Write(wxOutputStream& stream);
  39
  40 //// Accessors
  41     const wxString& GetName() const { return m_name; }
  42     const wxString& GetValue() const { return m_value; }
  43     void SetName(const wxString& name)  { m_name = name; }
  44     void SetValue(const wxString& value)  { m_value = value; }
  45
  46     wxSimpleHtmlAttribute* GetNextAttribute() { return m_next; }
  47     void SetNextAttribute(wxSimpleHtmlAttribute* attr) { m_next = attr; }
  48
  49     bool HasName(const wxString& name) const { return (0 == m_name.CmpNoCase(name)); }
  50     bool HasValue(const wxString& val) const { return (0 == m_value.CmpNoCase(val)); }
  51
  52 private:
  53     wxString                m_name;
  54     wxString                m_value;
  55     wxSimpleHtmlAttribute*  m_next;
  56 };
  57
  58
  59 /*
  60  * wxSimpleHtmlTag
  61  * Representation of a tag or chunk of text
  62  */
  63
  64 enum { wxSimpleHtmlTag_Text, wxSimpleHtmlTag_TopLevel, wxSimpleHtmlTag_Open, wxSimpleHtmlTag_Close, wxSimpleHtmlTag_Directive, wxSimpleHtmlTag_XMLDeclaration  };
  65
  66 class wxSimpleHtmlTag
  67 {
  68 public:
  69     wxSimpleHtmlTag(const wxString& tagName, int tagType);
  70     ~wxSimpleHtmlTag();
  71
  72 //// Operations
  73     void ClearAttributes();
  74     wxSimpleHtmlAttribute* FindAttribute(const wxString& name) const ;
  75     void AppendAttribute(const wxString& name, const wxString& value);
  76     void ClearChildren();
  77     // Remove 1 tag from the child list.
  78     void RemoveChild(wxSimpleHtmlTag *remove);
  79     // Appaned tag to the end of the child list.
  80     void AppendTag(wxSimpleHtmlTag* tag);
  81     // Insert tag after ourself in the parents child list.
  82     void AppendTagAfterUs(wxSimpleHtmlTag* tag);
  83     // Write this tag
  84     void Write(wxOutputStream& stream);
  85
  86     // Gets the text from this tag and its descendants
  87     wxString GetTagText();
  88
  89 //// Accessors
  90     const wxString& GetName() const { return m_name; }
  91     void SetName(const wxString& name) { m_name = name; }
  92
  93     int GetType() const { return m_type; }
  94     void SetType(int t) { m_type = t; }
  95
  96     // If type is wxSimpleHtmlTag_Text, m_text will contain some text.
  97     const wxString& GetText() const { return m_text; }
  98     void SetText(const wxString& text) { m_text = text; }
  99
 100     wxSimpleHtmlAttribute* GetFirstAttribute() { return m_attributes; }
 101     void SetFirstAttribute(wxSimpleHtmlAttribute* attr) { m_attributes = attr; }
 102
 103     int GetAttributeCount() const ;
 104     wxSimpleHtmlAttribute* GetAttribute(int i) const ;
 105
 106     wxSimpleHtmlTag* GetChildren() const { return m_children; }
 107     void SetChildren(wxSimpleHtmlTag* children) { m_children = children; }
 108
 109     wxSimpleHtmlTag* GetParent() const { return m_parent; }
 110     void SetParent(wxSimpleHtmlTag* parent) { m_parent = parent; }
 111     int GetChildCount() const;
 112     wxSimpleHtmlTag*    GetChild(int i) const;
 113     wxSimpleHtmlTag*    GetNext() const { return m_next; }
 114
 115 //// Convenience accessors & search functions
 116     bool NameIs(const wxString& name) { return (m_name.CmpNoCase(name) == 0); }
 117     bool HasAttribute(const wxString& name, const wxString& value) const;
 118     bool HasAttribute(const wxString& name) const;
 119     bool GetAttributeValue(wxString& value, const wxString& attrName);
 120
 121     // Search forward from this tag until we find a tag with this name & optionally attribute
 122     wxSimpleHtmlTag* FindTag(const wxString& tagName, const wxString& attrName = wxEmptyString);
 123
 124     // Gather the text until we hit the given close tag
 125     bool FindTextUntilTagClose(wxString& text, const wxString& tagName);
 126
 127 private:
 128     wxString                m_name;
 129     int                     m_type;
 130     wxString                m_text;
 131     wxSimpleHtmlAttribute*  m_attributes;
 132
 133     // List of children
 134     wxSimpleHtmlTag*        m_children;
 135     wxSimpleHtmlTag*        m_next; // Next sibling
 136     wxSimpleHtmlTag*        m_parent;
 137 };
 138
 139 /*
 140  * wxSimpleHtmlParser
 141  * Simple HTML parser, for such tasks as scanning HTML for keywords, contents, etc.
 142  */
 143
 144 class wxSimpleHtmlParser : public wxObject
 145 {
 146
 147 public:
 148     wxSimpleHtmlParser();
 149     ~wxSimpleHtmlParser();
 150
 151 //// Operations
 152     bool ParseFile(const wxString& filename);
 153     bool ParseString(const wxString& str);
 154     void Clear();
 155     // Write this file
 156     void Write(wxOutputStream& stream);
 157     bool WriteFile(wxString& filename);
 158
 159 //// Helpers
 160
 161     // Main recursive parsing function
 162     bool ParseHtml(wxSimpleHtmlTag* parent);
 163
 164     wxSimpleHtmlTag* ParseTagHeader();
 165     wxSimpleHtmlTag* ParseTagClose();
 166     bool ParseAttributes(wxSimpleHtmlTag* tag);
 167     wxSimpleHtmlTag* ParseDirective(); // e.g. <!DOCTYPE ....>
 168     wxSimpleHtmlTag* ParseXMLDeclaration(); // e.g. <?xml .... ?>
 169     bool ParseComment(); // Throw away comments
 170     // Plain text, up until an angled bracket
 171     bool ParseText(wxString& text);
 172
 173     bool EatWhitespace(); // Throw away whitespace
 174     bool EatWhitespace(int& pos); // Throw away whitespace: using 'pos'
 175     bool ReadString(wxString& str, bool eatIt = false);
 176     bool ReadWord(wxString& str, bool eatIt = false);
 177     bool ReadNumber(wxString& str, bool eatIt = false);
 178     // Could be number, string, whatever, but read up until whitespace.
 179     bool ReadLiteral(wxString& str, bool eatIt = false);
 180
 181     bool IsComment();
 182     bool IsDirective();
 183     bool IsXMLDeclaration();
 184     bool IsString();
 185     bool IsWord();
 186     bool IsTagClose();
 187     bool IsTagStartBracket(int ch);
 188     bool IsTagEndBracket(int ch);
 189     bool IsWhitespace(int ch);
 190     bool IsAlpha(int ch);
 191     bool IsWordChar(int ch);
 192     bool IsNumeric(int ch);
 193     // Check if a specific tag needs a close tag. If not this function should return false.
 194     // If no close tag is needed the result will be that the tag will be insert in a none
 195     // hierarchical way. i.e. if the function would return false all the time we would get
 196     // a flat list of all tags (like it used to be previously).
 197     virtual bool IsCloseTagNeeded(const wxString &name);
 198
 199     // Encode/Decode Special Characters like:
 200     // >    Begins a tag.      &gt;
 201     // <    Ends a tag.        &lt;
 202     // "    Quotation mark.    &quot;
 203     // '    Apostrophe.        &apos;
 204     // &    Ampersand.         &amp;
 205     static void DecodeSpecialChars(wxString &value);
 206     static wxString EncodeSpecialChars(const wxString &value);
 207
 208     // Matches this string (case insensitive)
 209     bool Matches(const wxString& tok, bool eatIt = false) ;
 210     bool Eof() const { return (m_pos >= m_length); }
 211     bool Eof(int pos) const { return (pos >= m_length); }
 212
 213     void SetPosition(int pos) { m_pos = pos; }
 214
 215
 216 //// Accessors
 217     wxSimpleHtmlTag* GetTopLevelTag() const { return m_topLevel; }
 218
 219     // Safe way of getting a character
 220     int GetChar(size_t i) const;
 221
 222 private:
 223
 224     wxSimpleHtmlTag*    m_topLevel;
 225     int                 m_pos;    // Position in string
 226     int                 m_length; // Length of string
 227     wxString            m_text;   // The actual text
 228
 229 };
 230
 231 /*
 232  * wxSimpleHtmlTagSpec
 233  * Describes a tag, and what type it is.
 234  * wxSimpleHtmlModule will initialise/cleanup a list of these, one per tag type
 235  */
 236
 237 #if 0
 238 class wxSimpleHtmlTagSpec : public wxObject
 239 {
 240
 241 public:
 242     wxSimpleHtmlTagSpec(const wxString& name, int type);
 243
 244 //// Operations
 245     static void AddTagSpec(wxSimpleHtmlTagSpec* spec);
 246     static void Clear();
 247
 248 //// Accessors
 249     const wxString& GetName() const { return m_name; }
 250     int GetType() const { return m_type; }
 251
 252 private:
 253
 254     wxString    m_name;
 255     int         m_type;
 256
 257     static wxList* sm_tagSpecs;
 258 };
 259
 260 /*
 261  * wxSimpleHtmlModule
 262  * Responsible for init/cleanup of appropriate data structures
 263  */
 264
 265 class wxSimpleHtmlModule : public wxModule
 266 {
 267 DECLARE_DYNAMIC_CLASS(wxSimpleHtmlModule)
 268
 269 public:
 270     wxSimpleHtmlModule() {};
 271
 272     bool OnInit() ;
 273     void OnExit() ;
 274 };
 275 #endif
 276
 277 #endif