LibOFX
lib/ofx_sgml.cpp
Go to the documentation of this file.
1 /***************************************************************************
2  ofx_sgml.cpp
3  -------------------
4  copyright : (C) 2002 by Benoit GrĂ©goire
5  email : benoitg@coeus.ca
6 ***************************************************************************/
12 /***************************************************************************
13  * *
14  * This program is free software; you can redistribute it and/or modify *
15  * it under the terms of the GNU General Public License as published by *
16  * the Free Software Foundation; either version 2 of the License, or *
17  * (at your option) any later version. *
18  * *
19  ***************************************************************************/
20 
21 #ifdef HAVE_CONFIG_H
22 #include <config.h>
23 #endif
24 
25 #include <iostream>
26 #include <stdlib.h>
27 #include <string>
28 #include <cassert>
29 #include "ParserEventGeneratorKit.h"
30 #include "libofx.h"
31 #include "ofx_utilities.hh"
32 #include "messages.hh"
33 #include "ofx_containers.hh"
34 #include "ofx_sgml.hh"
35 
36 using namespace std;
37 
38 OfxMainContainer * MainContainer = NULL;
39 extern SGMLApplication::OpenEntityPtr entity_ptr;
40 extern SGMLApplication::Position position;
41 
42 
45 class OFXApplication : public SGMLApplication
46 {
47 private:
48  OfxGenericContainer *curr_container_element;
49  OfxGenericContainer *tmp_container_element;
50  bool is_data_element;
51  string incoming_data;
52  LibofxContext * libofx_context;
53 
54 public:
55 
56  OFXApplication (LibofxContext * p_libofx_context)
57  {
58  MainContainer = NULL;
59  curr_container_element = NULL;
60  is_data_element = false;
61  libofx_context = p_libofx_context;
62  }
64  {
65  message_out(DEBUG, "Entering the OFXApplication's destructor");
66  }
67 
72  void startElement (const StartElementEvent & event)
73  {
74  string identifier;
75  CharStringtostring (event.gi, identifier);
76  message_out(PARSER, "startElement event received from OpenSP for element " + identifier);
77 
78  position = event.pos;
79 
80  switch (event.contentType)
81  {
82  case StartElementEvent::empty:
83  message_out(ERROR, "StartElementEvent::empty\n");
84  break;
85  case StartElementEvent::cdata:
86  message_out(ERROR, "StartElementEvent::cdata\n");
87  break;
88  case StartElementEvent::rcdata:
89  message_out(ERROR, "StartElementEvent::rcdata\n");
90  break;
91  case StartElementEvent::mixed:
92  message_out(PARSER, "StartElementEvent::mixed");
93  is_data_element = true;
94  break;
95  case StartElementEvent::element:
96  message_out(PARSER, "StartElementEvent::element");
97  is_data_element = false;
98  break;
99  default:
100  message_out(ERROR, "Unknown SGML content type?!?!?!? OpenSP interface changed?");
101  }
102 
103  if (is_data_element == false)
104  {
105  /*------- The following are OFX entities ---------------*/
106 
107  if (identifier == "OFX")
108  {
109  message_out (PARSER, "Element " + identifier + " found");
110  MainContainer = new OfxMainContainer (libofx_context, curr_container_element, identifier);
111  curr_container_element = MainContainer;
112  }
113  else if (identifier == "STATUS")
114  {
115  message_out (PARSER, "Element " + identifier + " found");
116  curr_container_element = new OfxStatusContainer (libofx_context, curr_container_element, identifier);
117  }
118  else if (identifier == "STMTRS" ||
119  identifier == "CCSTMTRS" ||
120  identifier == "INVSTMTRS")
121  {
122  message_out (PARSER, "Element " + identifier + " found");
123  curr_container_element = new OfxStatementContainer (libofx_context, curr_container_element, identifier);
124  }
125  else if (identifier == "BANKTRANLIST" || identifier == "INVTRANLIST")
126  {
127  message_out (PARSER, "Element " + identifier + " found");
128  //BANKTRANLIST ignored, we will process it's attributes directly inside the STATEMENT,
129  if (curr_container_element && curr_container_element->type != "STATEMENT")
130  {
131  message_out(ERROR, "Element " + identifier + " found while not inside a STATEMENT container");
132  }
133  else
134  {
135  curr_container_element = new OfxPushUpContainer (libofx_context, curr_container_element, identifier);
136  }
137  }
138  else if (identifier == "STMTTRN")
139  {
140  message_out (PARSER, "Element " + identifier + " found");
141  if (curr_container_element->type == "INVESTMENT")
142  {
143  //push up to the INVBANKTRAN OfxInvestmentTransactionContainer
144  curr_container_element = new OfxPushUpContainer (libofx_context, curr_container_element, identifier);
145  }
146  else
147  {
148  curr_container_element = new OfxBankTransactionContainer (libofx_context, curr_container_element, identifier);
149  }
150  }
151  else if (identifier == "BUYDEBT" ||
152  identifier == "BUYMF" ||
153  identifier == "BUYOPT" ||
154  identifier == "BUYOTHER" ||
155  identifier == "BUYSTOCK" ||
156  identifier == "CLOSUREOPT" ||
157  identifier == "INCOME" ||
158  identifier == "INVEXPENSE" ||
159  identifier == "JRNLFUND" ||
160  identifier == "JRNLSEC" ||
161  identifier == "MARGININTEREST" ||
162  identifier == "REINVEST" ||
163  identifier == "RETOFCAP" ||
164  identifier == "SELLDEBT" ||
165  identifier == "SELLMF" ||
166  identifier == "SELLOPT" ||
167  identifier == "SELLOTHER" ||
168  identifier == "SELLSTOCK" ||
169  identifier == "SPLIT" ||
170  identifier == "TRANSFER" ||
171  identifier == "INVBANKTRAN" )
172  {
173  message_out (PARSER, "Element " + identifier + " found");
174  curr_container_element = new OfxInvestmentTransactionContainer (libofx_context, curr_container_element, identifier);
175  }
176  /*The following is a list of OFX elements whose attributes will be processed by the parent container*/
177  else if (identifier == "INVBUY" ||
178  identifier == "INVSELL" ||
179  identifier == "INVTRAN" ||
180  identifier == "SECINFO" ||
181  identifier == "SECID" ||
182  identifier == "CURRENCY" ||
183  identifier == "ORIGCURRENCY")
184  {
185  message_out (PARSER, "Element " + identifier + " found");
186  curr_container_element = new OfxPushUpContainer (libofx_context, curr_container_element, identifier);
187  }
188 
189  /* provide a parent for the account list response so its ACCTFROM can be recognized */
190  else if (identifier == "BANKACCTINFO" || identifier == "CCACCTINFO" || identifier == "INVACCTINFO")
191  {
192  message_out (PARSER, "Element " + identifier + " found");
193  curr_container_element = new OfxPushUpContainer (libofx_context, curr_container_element, identifier);
194  }
195 
196  /* The different types of accounts */
197  else if (identifier == "BANKACCTFROM" || identifier == "CCACCTFROM" || identifier == "INVACCTFROM")
198  {
199  message_out (PARSER, "Element " + identifier + " found");
200  /* check the container to avoid creating multiple statements for TRANSFERs */
201  if (curr_container_element->type == "STATEMENT"
202  || curr_container_element->tag_identifier == "BANKACCTINFO"
203  || curr_container_element->tag_identifier == "CCACCTINFO"
204  || curr_container_element->tag_identifier == "INVACCTINFO")
205  curr_container_element = new OfxAccountContainer (libofx_context, curr_container_element, identifier);
206  else
207  // no new account or statement for a <TRANSFER>
208  curr_container_element = new OfxDummyContainer (libofx_context, curr_container_element, identifier);
209  }
210  else if (identifier == "STOCKINFO" || identifier == "OPTINFO" ||
211  identifier == "DEBTINFO" || identifier == "MFINFO" || identifier == "OTHERINFO")
212  {
213  message_out (PARSER, "Element " + identifier + " found");
214  curr_container_element = new OfxSecurityContainer (libofx_context, curr_container_element, identifier);
215  }
216  /* The different types of balances */
217  else if (identifier == "LEDGERBAL" ||
218  identifier == "AVAILBAL" ||
219  identifier == "INVBAL")
220  {
221  message_out (PARSER, "Element " + identifier + " found");
222  curr_container_element = new OfxBalanceContainer (libofx_context, curr_container_element, identifier);
223  }
224  else if (identifier == "INVPOS")
225  {
226  message_out (PARSER, "Element " + identifier + " found");
227  curr_container_element = new OfxPositionContainer (libofx_context, curr_container_element, identifier);
228  }
229  else
230  {
231  /* We dont know this OFX element, so we create a dummy container */
232  curr_container_element = new OfxDummyContainer(libofx_context, curr_container_element, identifier);
233  }
234  }
235  else
236  {
239  if (identifier == "INV401K")
240  {
241  /* Minimal handler for this section to discard <DTASOF>, <DTSTART> and <DTEND> that need to be ignored */
242  message_out (PARSER, "Element " + identifier + " found");
243  curr_container_element = new OfxInv401kContainer (libofx_context, curr_container_element, identifier);
244  }
245  if (identifier == "INV401KBAL")
246  {
247  message_out (PARSER, "Element " + identifier + " found");
248  curr_container_element = new OfxBalanceContainer (libofx_context, curr_container_element, identifier);
249  }
250  else
251  {
252  /* The element was a data element. OpenSP will call one or several data() callback with the data */
253  message_out (PARSER, "Data element " + identifier + " found");
254  /* There is a bug in OpenSP 1.3.4, which won't send endElement Event for some elements, and will instead send an error like "document type does not allow element "MESSAGE" here". Incoming_data should be empty in such a case, but it will not be if the endElement event was skiped. So we empty it, so at least the last element has a chance of having valid data */
255  if (incoming_data != "")
256  {
257  message_out (ERROR, "startElement: incoming_data should be empty! You are probably using OpenSP <= 1.3.4. The following data was lost: " + incoming_data );
258  incoming_data.assign ("");
259  }
260  }
261  }
262  }
263 
268  void endElement (const EndElementEvent & event)
269  {
270  string identifier;
271  bool end_element_for_data_element;
272 
273  CharStringtostring (event.gi, identifier);
274  end_element_for_data_element = is_data_element;
275  message_out(PARSER, "endElement event received from OpenSP for element " + identifier);
276 
277  position = event.pos;
278  if (curr_container_element == NULL)
279  {
280  message_out (ERROR, "Tried to close a " + identifier + " without a open element (NULL pointer)");
281  incoming_data.assign ("");
282  }
283  else //curr_container_element != NULL
284  {
285  if (end_element_for_data_element == true)
286  {
287  incoming_data = strip_whitespace(incoming_data);
288 
289  curr_container_element->add_attribute (identifier, incoming_data);
290  message_out (PARSER, "endElement: Added data '" + incoming_data + "' from " + identifier + " to " + curr_container_element->type + " container_element");
291  incoming_data.assign ("");
292  is_data_element = false;
293  }
294  else
295  {
296  if (identifier == curr_container_element->tag_identifier)
297  {
298  if (incoming_data != "")
299  {
300  message_out(ERROR, "End tag for non data element " + identifier + ", incoming data should be empty but contains: " + incoming_data + " DATA HAS BEEN LOST SOMEWHERE!");
301  }
302 
303  if (identifier == "OFX")
304  {
305  /* The main container is a special case */
306  tmp_container_element = curr_container_element;
307  curr_container_element = curr_container_element->getparent ();
308  if (curr_container_element == NULL)
309  {
310  //Defensive coding, this isn't supposed to happen
311  curr_container_element = tmp_container_element;
312  }
313  if (MainContainer != NULL)
314  {
315  MainContainer->gen_event();
316  delete MainContainer;
317  MainContainer = NULL;
318  curr_container_element = NULL;
319  message_out (DEBUG, "Element " + identifier + " closed, MainContainer destroyed");
320  }
321  else
322  {
323  message_out (DEBUG, "Element " + identifier + " closed, but there was no MainContainer to destroy (probably a malformed file)!");
324  }
325  }
326  else
327  {
328  tmp_container_element = curr_container_element;
329  curr_container_element = curr_container_element->getparent ();
330  if (MainContainer != NULL)
331  {
335  if (identifier == "CURRENCY" || identifier == "ORIGCURRENCY")
336  {
337  tmp_container_element->add_attribute (identifier, incoming_data);
338  message_out (DEBUG, "Element " + identifier + " closed, container " + tmp_container_element->type + " updated");
339  }
340  else
341  {
342  tmp_container_element->add_to_main_tree();
343  message_out (PARSER, "Element " + identifier + " closed, object added to MainContainer");
344  }
345  }
346  else
347  {
348  message_out (ERROR, "MainContainer is NULL trying to add element " + identifier);
349  }
350  }
351  }
352  else
353  {
354  message_out (ERROR, "Tried to close a " + identifier + " but a " + curr_container_element->type + " is currently open.");
355  }
356  }
357  }
358  }
359 
364  void data (const DataEvent & event)
365  {
366  string tmp;
367  position = event.pos;
368  AppendCharStringtostring (event.data, incoming_data);
369  message_out(PARSER, "data event received from OpenSP, incoming_data is now: " + incoming_data);
370  }
371 
376  void error (const ErrorEvent & event)
377  {
378  string message;
379  string string_buf;
380  OfxMsgType error_type = ERROR;
381 
382  position = event.pos;
383  message = message + "OpenSP parser: ";
384  switch (event.type)
385  {
386  case SGMLApplication::ErrorEvent::quantity:
387  message = message + "quantity (Exceeding a quantity limit):";
388  error_type = ERROR;
389  break;
390  case SGMLApplication::ErrorEvent::idref:
391  message = message + "idref (An IDREF to a non-existent ID):";
392  error_type = ERROR;
393  break;
394  case SGMLApplication::ErrorEvent::capacity:
395  message = message + "capacity (Exceeding a capacity limit):";
396  error_type = ERROR;
397  break;
398  case SGMLApplication::ErrorEvent::otherError:
399  message = message + "otherError (misc parse error):";
400  error_type = ERROR;
401  break;
402  case SGMLApplication::ErrorEvent::warning:
403  message = message + "warning (Not actually an error.):";
404  error_type = WARNING;
405  break;
406  case SGMLApplication::ErrorEvent::info:
407  message = message + "info (An informationnal message. Not actually an error):";
408  error_type = INFO;
409  break;
410  default:
411  message = message + "OpenSP sent an unknown error to LibOFX (You probably have a newer version of OpenSP):";
412  }
413  message = message + "\n" + CharStringtostring (event.message, string_buf);
414  message_out (error_type, message);
415  }
416 
421  void openEntityChange (const OpenEntityPtr & para_entity_ptr)
422  {
423  message_out(DEBUG, "openEntityChange()\n");
424  entity_ptr = para_entity_ptr;
425 
426  };
427 
428 private:
429 };
430 
434 int ofx_proc_sgml(LibofxContext * libofx_context, int argc, char * const* argv)
435 {
436  message_out(DEBUG, "Begin ofx_proc_sgml()");
437  assert(argc >= 3);
438  message_out(DEBUG, argv[0]);
439  message_out(DEBUG, argv[1]);
440  message_out(DEBUG, argv[2]);
441 
442  ParserEventGeneratorKit parserKit;
443  parserKit.setOption (ParserEventGeneratorKit::showOpenEntities);
444  EventGenerator *egp = parserKit.makeEventGenerator (argc, argv);
445  egp->inhibitMessages (true); /* Error output is handled by libofx not OpenSP */
446  OFXApplication *app = new OFXApplication(libofx_context);
447  unsigned nErrors = egp->run (*app); /* Begin parsing */
448  delete egp; //Note that this is where bug is triggered
449  return nErrors > 0;
450 }
SGMLApplication::Position position
int ofx_proc_sgml(LibofxContext *libofx_context, int argc, char *const *argv)
Parses a DTD and OFX file(s)
Represents a security, such as a stock or bond.
OfxGenericContainer * getparent()
Returns the parent container object (the one representing the containing OFX SGML element) ...
A generic container for an OFX SGML element. Every container inherits from OfxGenericContainer.
string CharStringtostring(const SGMLApplication::CharString source, string &dest)
Convert OpenSP CharString to a C++ STL string.
Various simple functions for type conversion & al.
void openEntityChange(const OpenEntityPtr &para_entity_ptr)
Callback: Receive internal OpenSP state.
virtual void add_attribute(const string identifier, const string value)
Add data to a container object.
string strip_whitespace(const string para_string)
Sanitize a string coming from OpenSP.
SGMLApplication::OpenEntityPtr entity_ptr
void error(const ErrorEvent &event)
Callback: SGML parse error.
int message_out(OfxMsgType error_type, const string message)
Message output function.
void data(const DataEvent &event)
Callback: Data from an OFX element.
string AppendCharStringtostring(const SGMLApplication::CharString source, string &dest)
Append an OpenSP CharString to an existing C++ STL string.
LibOFX internal object code.
Represents a statement for either a bank account or a credit card account.
Message IO functionality.
virtual int add_to_main_tree()
Add this container to the main tree.
Represents a bank or credid card transaction.
OFX/SGML parsing functionnality.
A container to hold a OFX SGML element for which you want the parent to process it&#39;s data elements...
void startElement(const StartElementEvent &event)
Callback: Start of an OFX element.
Represents an investment position, such as a stock or bond.
Represents a bank or credid card transaction.
Represents a bank account or a credit card account.
The root container. Created by the <OFX> OFX element or by the export functions.
int gen_event()
Generate libofx.h events.
This object is driven by OpenSP as it parses the SGML from the ofx file(s)
void endElement(const EndElementEvent &event)
Callback: End of an OFX element.
A container to hold OFX SGML elements that LibOFX knows nothing about.
Represents the <BALANCE>, <INVBAL> or <INV401KBAL> OFX SGML entity.
A container to hold OFX SGML elements for <INV401K>
Represents the <STATUS> OFX SGML entity.