xref: /AOO41X/main/xml2cmp/source/xcd/parse.cxx (revision 2afa04a6106dcd4652fbf0ba99fc521070b33a24)
1 /**************************************************************
2  *
3  * Licensed to the Apache Software Foundation (ASF) under one
4  * or more contributor license agreements.  See the NOTICE file
5  * distributed with this work for additional information
6  * regarding copyright ownership.  The ASF licenses this file
7  * to you under the Apache License, Version 2.0 (the
8  * "License"); you may not use this file except in compliance
9  * with the License.  You may obtain a copy of the License at
10  *
11  *   http://www.apache.org/licenses/LICENSE-2.0
12  *
13  * Unless required by applicable law or agreed to in writing,
14  * software distributed under the License is distributed on an
15  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
16  * KIND, either express or implied.  See the License for the
17  * specific language governing permissions and limitations
18  * under the License.
19  *
20  *************************************************************/
21 
22 
23 
24 
25 #include <parse.hxx>
26 
27 #include <string.h>
28 #include <iostream>
29 #include <xmlelem.hxx>
30 
31 #if (_MSC_VER >=1400)
32 #pragma warning(disable:4365)
33 #endif
34 
35 #ifdef UNX
36 #define strnicmp strncasecmp
37 #endif
38 
39 
40 
41 // NOT FULLY DEFINED SERVICES
42 
43 
44 
45 #define AssertionOf(x)  \
46     {if (!(x)) {std::cerr << "Assertion failed: " << #x << __FILE__ << __LINE__ << std::endl; exit(3); }}
47 
48 
49 
X2CParser(XmlElement & o_rDocumentData)50 X2CParser::X2CParser( XmlElement & o_rDocumentData )
51     :   // sFileName,
52         nFileLine(0),
53         pDocumentData(&o_rDocumentData),
54         // sWord,
55         text(0)
56 {
57 }
58 
~X2CParser()59 X2CParser::~X2CParser()
60 {
61 }
62 
63 
64 bool
LoadFile(const char * i_sFilename)65 X2CParser::LoadFile( const char * i_sFilename )
66 {
67     sFileName = i_sFilename;
68     nFileLine = 1;
69 
70     // Load file:
71     if ( ! LoadXmlFile( aFile, i_sFilename ) )
72         return false;
73 
74     // Test correct end:
75     const char * pLastTag = strrchr(aFile.operator const char *(),'<');
76     if (pLastTag == 0)
77         return false;
78     if ( strnicmp(pLastTag+2, pDocumentData->Name().str(), pDocumentData->Name().l()) != 0
79          || strnicmp(pLastTag, "</", 2) != 0 )
80         return false;
81     if (strchr(pLastTag,'>') == 0)
82         return false;
83     return true;
84 }
85 
86 void
Parse()87 X2CParser::Parse()
88 {
89     // Parse:
90     text = aFile.operator const char *();
91 
92     Parse_XmlDeclaration();
93     Parse_Doctype();
94 
95     // skip XML comment
96     Goto('<');
97     if ( IsText("<!--") )
98         Goto_And_Pass('>');
99 
100     pDocumentData->Parse(*this);
101 }
102 
103 bool
Parse(const char * i_sFilename)104 X2CParser::Parse( const char * i_sFilename )
105 {
106     bool ret = LoadFile(i_sFilename);
107     if (ret)
108         Parse();
109     return ret;
110 }
111 
112 void
Parse_XmlDeclaration()113 X2CParser::Parse_XmlDeclaration()
114 {
115     Goto('<');
116     if ( IsText("<?xml") )
117     {
118         Goto_And_Pass('>');
119     }
120 }
121 
122 void
Parse_Doctype()123 X2CParser::Parse_Doctype()
124 {
125     Goto('<');
126     if ( IsText("<!DOCTYPE") )
127         Goto_And_Pass('>');
128 }
129 
130 void
Parse_Sequence(DynamicList<XmlElement> & o_rElements,const Simstr & i_sElementName)131 X2CParser::Parse_Sequence( DynamicList<XmlElement> & o_rElements,
132                            const Simstr &            i_sElementName  )
133 {
134     CheckAndPassBeginTag(i_sElementName.str());
135 
136     unsigned int i_max = o_rElements.size();
137     for (unsigned i = 0; i < i_max; ++i)
138     {
139         o_rElements[i]->Parse(*this);
140     }  // end for
141 
142     CheckAndPassEndTag(i_sElementName.str());
143 }
144 
145 void
Parse_FreeChoice(DynamicList<XmlElement> & o_rElements)146 X2CParser::Parse_FreeChoice( DynamicList<XmlElement> & o_rElements )
147 {
148     unsigned        nSize = o_rElements.size();
149 
150     for ( bool bBreak = false;  !bBreak; )
151     {
152         bBreak = true;
153         for ( unsigned i = 0; i < nSize; ++i )
154         {
155             Goto('<');
156             if ( IsBeginTag(o_rElements[i]->Name().str()) )
157             {
158                 o_rElements[i]->Parse(*this);
159                 bBreak = false;
160                 break;
161             }
162         }   // end for i
163     }   // end for !bBreak
164 }
165 
166 void
Parse_List(ListElement & o_rListElem)167 X2CParser::Parse_List( ListElement &  o_rListElem )
168 {
169 
170     for ( Goto('<'); IsBeginTag(o_rListElem.Name().str()); Goto('<') )
171     {
172         XmlElement * pNew = o_rListElem.Create_and_Add_NewElement();
173         pNew->Parse(*this);
174     }
175 }
176 
177 void
Parse_Text(Simstr & o_sText,const Simstr & i_sElementName,bool i_bReverseName)178 X2CParser::Parse_Text( Simstr &         o_sText,
179                        const Simstr &   i_sElementName,
180                        bool             i_bReverseName )
181 {
182 
183     if ( ! CheckAndPassBeginTag(i_sElementName.str()) )
184         return;
185 
186     // Add new Element
187     GetTextTill( o_sText, '<', i_bReverseName );
188     o_sText.remove_trailing_blanks();
189 
190     CheckAndPassEndTag(i_sElementName.str());
191 }
192 
193 void
Parse_MultipleText(List<Simstr> & o_rTexts,const Simstr & i_sElementName,bool i_bReverseName)194 X2CParser::Parse_MultipleText( List<Simstr> &   o_rTexts,
195                                const Simstr &   i_sElementName,
196                                bool             i_bReverseName )
197 {
198     for ( Goto('<'); IsBeginTag(i_sElementName.str()); Goto('<') )
199     {
200         Simstr sNew;
201         Parse_Text(sNew, i_sElementName, i_bReverseName);
202         if (sNew.l() > 0)
203             o_rTexts.push_back(sNew);
204     }
205 }
206 
207 void
Parse_SglAttr(Simstr & o_sAttrValue,const Simstr & i_sElementName,const Simstr & i_sAttrName)208 X2CParser::Parse_SglAttr( Simstr &          o_sAttrValue,
209                           const Simstr &    i_sElementName,
210                           const Simstr &    i_sAttrName )
211 {
212     Goto('<');
213     if ( !IsBeginTag(i_sElementName.str()) )
214         SyntaxError("unexpected element");
215     Move( i_sElementName.l() + 1 );
216 
217     Pass_White();
218     if (*text == '>')
219         SyntaxError("no attribute found, where one was expected");
220     Simstr sAttrName;
221     Get_Attribute(o_sAttrValue, sAttrName);
222     if (sAttrName != i_sAttrName)
223         SyntaxError("unknown attribute found");
224     Pass_White();
225     if (strncmp(text,"/>",2) != 0)
226         SyntaxError("missing \"/>\" at end of empty element");
227     Move(2);
228 }
229 
230 void
Parse_MultipleAttr(List<Simstr> & o_rAttrValues,const Simstr & i_sElementName,const List<Simstr> & i_rAttrNames)231 X2CParser::Parse_MultipleAttr( List<Simstr> &       o_rAttrValues,
232                                const Simstr &       i_sElementName,
233                                const List<Simstr> & i_rAttrNames )
234 {
235     Goto('<');
236     if ( !IsBeginTag(i_sElementName.str()) )
237         SyntaxError("unexpected element");
238     Move( i_sElementName.l() + 1 );
239     Simstr sAttrName;
240     Simstr sAttrValue;
241     unsigned nSize = i_rAttrNames.size();
242     unsigned i;
243 
244     for ( Pass_White(); *text != '/'; Pass_White() )
245     {
246 
247         Get_Attribute(sAttrValue, sAttrName);
248 
249         for ( i = 0; i < nSize; ++i )
250         {
251             if ( i_rAttrNames[i] == sAttrName )
252             {
253                 o_rAttrValues[i] = sAttrValue;
254                 break;
255             }
256         }
257         if (i == nSize)
258             SyntaxError("unknown attribute found");
259     }
260     Move(2);
261 }
262 
263 
264 void
Get_Attribute(Simstr & o_rAttrValue,Simstr & o_rAttrName)265 X2CParser::Get_Attribute( Simstr & o_rAttrValue,
266                           Simstr & o_rAttrName )
267 {
268     GetTextTill( o_rAttrName, '=');
269 
270     while (*(++text) != '"')
271     {
272         if (*text == '\0')
273             SyntaxError("unexpected end of file");
274     }
275 
276     ++text;
277     GetTextTill( o_rAttrValue, '"');
278     ++text;
279 }
280 
281 bool
IsText(const char * i_sComparedText)282 X2CParser::IsText( const char * i_sComparedText )
283 {
284     return strnicmp( text, i_sComparedText, strlen(i_sComparedText) ) == 0;
285 }
286 
287 bool
IsBeginTag(const char * i_sTagName)288 X2CParser::IsBeginTag( const char * i_sTagName )
289 {
290     return strnicmp( text+1, i_sTagName, strlen(i_sTagName) ) == 0
291            && *text == '<';
292 }
293 
294 bool
IsEndTag(const char * i_sTagName)295 X2CParser::IsEndTag( const char * i_sTagName )
296 {
297     return strnicmp( text+2, i_sTagName, strlen(i_sTagName) ) == 0
298            && strnicmp( text, "</", 2 ) == 0;
299 }
300 
301 void
Goto(char i_cNext)302 X2CParser::Goto( char i_cNext )
303 {
304     while (*text != i_cNext)
305     {
306         TestCurChar();
307         ++text;
308     }
309 }
310 
311 void
Goto_And_Pass(char i_cNext)312 X2CParser::Goto_And_Pass( char i_cNext )
313 {
314     Goto(i_cNext);
315     ++text;
316 }
317 
318 void
Move(int i_nForward)319 X2CParser::Move( int i_nForward )
320 {
321     text += i_nForward;
322 }
323 
324 void
Pass_White()325 X2CParser::Pass_White()
326 {
327     while (*text <= 32)
328     {
329         TestCurChar();
330         ++text;
331     }
332 }
333 
334 void
GetTextTill(Simstr & o_rText,char i_cEnd,bool i_bReverseName)335 X2CParser::GetTextTill( Simstr & o_rText,
336                         char     i_cEnd,
337                         bool     i_bReverseName )
338 {
339     char * pResult = &sWord[0];
340     char * pSet;
341 
342     for ( pSet = pResult;
343           *text != i_cEnd;
344           ++text )
345     {
346         TestCurChar();
347         *pSet++ = *text;
348     }
349 
350     while ( *pResult < 33 && *pResult > 0 )
351         ++pResult;
352     while ( pSet > pResult ? *(pSet-1) < 33 : false )
353         pSet--;
354     *pSet = '\0';
355 
356 
357     if (i_bReverseName)
358     {
359         const unsigned int nMaxLen = 1000;
360         if (strlen(pResult) < nMaxLen)
361         {
362             char * sBreak = strrchr(pResult,'.');
363             if (sBreak != 0)
364             {
365                 static char sScope[nMaxLen+10];
366                 static char sName[nMaxLen+10];
367 
368                 unsigned nScopeLen = sBreak - pResult;
369                 strncpy ( sScope, pResult, nScopeLen ); // STRNCPY SAFE HERE
370                 sScope[nScopeLen] = '\0';
371                 strcpy( sName, sBreak + 1 );            // STRCPY SAFE HERE
372                 strcat( sName, " in " );                // STRCAT SAFE HERE
373                 strcat( sName, sScope );                // STRCAT SAFE HERE
374 
375                 o_rText = sName;
376                 return;
377             }
378         }
379     }   // endif (i_bReverseName)
380 
381     o_rText = &sWord[0];
382 }
383 
384 bool
CheckAndPassBeginTag(const char * i_sElementName)385 X2CParser::CheckAndPassBeginTag( const char * i_sElementName )
386 {
387     bool ret = true;
388     Goto('<');
389     if ( ! IsBeginTag(i_sElementName) )
390         SyntaxError( "unexpected element");
391     if (IsAbsoluteEmpty())
392         ret = false;
393     Goto_And_Pass('>');
394     if (ret)
395         Pass_White();
396     return ret;
397 }
398 
399 void
CheckAndPassEndTag(const char * i_sElementName)400 X2CParser::CheckAndPassEndTag( const char * i_sElementName )
401 {
402     Pass_White();
403     if ( !IsEndTag(i_sElementName) )
404         SyntaxError("missing or not matching end tag");
405     Goto_And_Pass('>');
406 }
407 
408 bool
IsAbsoluteEmpty() const409 X2CParser::IsAbsoluteEmpty() const
410 {
411     const char * pEnd = strchr(text+1, '>');
412     if (pEnd != 0)
413     {
414         if ( *(pEnd-1) == '/' )
415         {
416             const char * pAttr = strchr(text+1, '"');
417             if (pAttr == 0)
418                 return true;
419             else if ( (pAttr-text) > (pEnd-text) )
420                 return true;
421         }
422     }
423     return false;
424 }
425 
426 void
SyntaxError(const char * i_sText)427 X2CParser::SyntaxError( const char * i_sText )
428 {
429     std::cerr
430          << "Syntax error "
431          << i_sText
432          << " in file: "
433          << sFileName.str()
434          << " in line "
435          << nFileLine
436          << "."
437          << std::endl;
438 
439      exit(3);
440 }
441 
442 void
TestCurChar()443 X2CParser::TestCurChar()
444 {
445 //  if (*text == '\0')
446 //      SyntaxError("unexpected end of file");
447 //  else
448 
449     if (*text == '\n')
450         nFileLine++;
451 }
452 
453 
454