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