contrib/src/xml/xmlexpat.cpp

   1 /////////////////////////////////////////////////////////////////////////////
   2 // Name:        xmlexpat.cpp
   3 // Purpose:     wxXmlDocument - XML reader via Expat
   4 // Author:      Vaclav Slavik
   5 // Created:     2001/04/30
   6 // RCS-ID:      $Id$
   7 // Copyright:   (c) 2001 Vaclav Slavik
   8 // Licence:     wxWindows licence
   9 /////////////////////////////////////////////////////////////////////////////
  10
  11 #ifdef __GNUG__
  12 // nothing - already in xml.cpp
  13 #endif
  14
  15 // For compilers that support precompilation, includes "wx.h".
  16 #include "wx/wxprec.h"
  17
  18 #ifdef __BORLANDC__
  19     #pragma hdrstop
  20 #endif
  21
  22 #include "wx/wfstream.h"
  23 #include "wx/intl.h"
  24 #include "wx/log.h"
  25 #include "wx/strconv.h"
  26 #include "wx/xml/xmlio.h"
  27
  28 #include "xmlparse.h"
  29
  30 /*
  31
  32     FIXME:
  33
  34      - handle unknown encodings
  35      - process all elements, including CDATA
  36      - XML resources should automatically select desired encoding besed on
  37        runtime environment (?) (would need BIN and BINZ formats modification,
  38        too)
  39
  40  */
  41
  42
  43 // converts Expat-produced string in UTF-8 into wxString.
  44 inline static wxString CharToString(const char *s, size_t len = wxSTRING_MAXLEN)
  45 {
  46 #if wxUSE_UNICODE
  47     return wxString(s, wxMBConvUTF8, len);
  48 #else
  49     return wxString(s, len);
  50 #endif
  51 }
  52
  53 bool wxXmlIOHandlerExpat::CanLoad(wxInputStream& stream)
  54 {
  55     char cheader[7];
  56     cheader[6] = 0;
  57     stream.Read(cheader, 6);
  58     stream.SeekI(-6, wxFromCurrent);
  59     return (strcmp(cheader, "<?xml ") == 0);
  60 }
  61
  62
  63 struct wxXmlParsingContext
  64 {
  65     wxXmlNode *root;
  66     wxXmlNode *node;
  67     wxXmlNode *lastAsText;
  68     wxString encoding;
  69     wxString version;
  70 };
  71
  72 static void StartElementHnd(void *userData, const char *name, const char **atts)
  73 {
  74     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
  75     wxXmlNode *node = new wxXmlNode(wxXML_ELEMENT_NODE, CharToString(name));
  76     const char **a = atts;
  77     while (*a)
  78     {
  79         node->AddProperty(CharToString(a[0]), CharToString(a[1]));
  80         a += 2;
  81     }
  82     if (ctx->root == NULL)
  83         ctx->root = node;
  84     else
  85         ctx->node->AddChild(node);
  86     ctx->node = node;
  87     ctx->lastAsText = NULL;
  88 }
  89
  90 static void EndElementHnd(void *userData, const char *name)
  91 {
  92     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
  93
  94     ctx->node = ctx->node->GetParent();
  95     ctx->lastAsText = NULL;
  96 }
  97
  98 static void TextHnd(void *userData, const char *s, int len)
  99 {
 100     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 101     char *buf = new char[len + 1];
 102
 103     buf[len] = '\0';
 104     memcpy(buf, s, (size_t)len);
 105
 106     if (ctx->lastAsText)
 107     {
 108         ctx->lastAsText->SetContent(ctx->lastAsText->GetContent() +
 109                                     CharToString(buf));
 110     }
 111     else
 112     {
 113         bool whiteOnly = TRUE;
 114         for (char *c = buf; *c != '\0'; c++)
 115             if (*c != ' ' && *c != '\t' && *c != '\n' && *c != '\r')
 116             {
 117                 whiteOnly = FALSE;
 118                 break;
 119             }
 120         if (!whiteOnly)
 121         {
 122             ctx->lastAsText = new wxXmlNode(wxXML_TEXT_NODE, wxT("text"),
 123                                             CharToString(buf));
 124             ctx->node->AddChild(ctx->lastAsText);
 125         }
 126     }
 127
 128     delete[] buf;
 129 }
 130
 131 static void CommentHnd(void *userData, const char *data)
 132 {
 133     wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 134
 135     ctx->node->AddChild(new wxXmlNode(wxXML_COMMENT_NODE,
 136                                       wxT("comment"), CharToString(data)));
 137     ctx->lastAsText = NULL;
 138 }
 139
 140 static void DefaultHnd(void *userData, const char *s, int len)
 141 {
 142     // XML header:
 143     if (len > 6 && memcmp(s, "<?xml ", 6) == 0)
 144     {
 145         wxXmlParsingContext *ctx = (wxXmlParsingContext*)userData;
 146
 147         wxString buf = CharToString(s, (size_t)len);
 148         int pos;
 149         pos = buf.Find(wxT("encoding="));
 150         if (pos != wxNOT_FOUND)
 151             ctx->encoding = buf.Mid(pos + 10).BeforeFirst(buf.c_str()[pos+9]);
 152         pos = buf.Find(wxT("version="));
 153         if (pos != wxNOT_FOUND)
 154             ctx->version = buf.Mid(pos + 9).BeforeFirst(buf.c_str()[pos+8]);
 155     }
 156 }
 157
 158 bool wxXmlIOHandlerExpat::Load(wxInputStream& stream, wxXmlDocument& doc)
 159 {
 160     const size_t BUFSIZE = 1024;
 161     char buf[BUFSIZE];
 162     wxXmlParsingContext ctx;
 163     bool done;
 164     XML_Parser parser = XML_ParserCreate(NULL);
 165
 166     ctx.root = ctx.node = NULL;
 167     XML_SetUserData(parser, (void*)&ctx);
 168     XML_SetElementHandler(parser, StartElementHnd, EndElementHnd);
 169     XML_SetCharacterDataHandler(parser, TextHnd);
 170     XML_SetCommentHandler(parser, CommentHnd);
 171     XML_SetDefaultHandler(parser, DefaultHnd);
 172
 173     do
 174     {
 175         size_t len = stream.Read(buf, BUFSIZE).LastRead();
 176         done = (len < BUFSIZE);
 177         if (!XML_Parse(parser, buf, len, done))
 178         {
 179             wxLogError(_("XML parsing error: '%s' at line %d"),
 180                        XML_ErrorString(XML_GetErrorCode(parser)),
 181                        XML_GetCurrentLineNumber(parser));
 182           return FALSE;
 183         }
 184     } while (!done);
 185
 186     doc.SetVersion(ctx.version);
 187     doc.SetEncoding(ctx.encoding);
 188     doc.SetRoot(ctx.root);
 189
 190     XML_ParserFree(parser);
 191     return TRUE;
 192 }