Switched over to using irrxml for parsing XML, rather than expat.

7 files changed, 115 insertions, 69 deletions

diff --git a/src/gnucash.cc b/src/gnucash.cc
index 60c72ed4..ed3a8457 100644
--- a/src/gnucash.cc
+++ b/src/gnucash.cc
@@ -41,10 +41,12 @@ typedef std::pair<const string, account_t *> accounts_pair;
 typedef std::map<account_t *, commodity_t *>  account_comm_map;
 typedef std::pair<account_t *, commodity_t *> account_comm_pair;
 
+#if 0
+
 static journal_t *	curr_journal;
 static account_t *	master_account;
 static account_t *	curr_account;
-static string	curr_account_id;
+static string		curr_account_id;
 static entry_t *	curr_entry;
 static commodity_t *	entry_comm;
 static commodity_t *	curr_comm;
@@ -54,7 +56,7 @@ static XML_Parser	current_parser;
 static accounts_map	accounts_by_id;
 static account_comm_map	account_comms;
 static unsigned int	count;
-static string	have_error;
+static string		have_error;
 
 static std::istream *   instreamp;
 static unsigned int     offset;
@@ -350,14 +352,28 @@ static void dataHandler(void *, const char *s, int len)
   }
 }
 
+#endif
+
 bool gnucash_parser_t::test(std::istream& in) const
 {
-  char buf[5];
-  in.read(buf, 5);
+  char	 buf[80];
+  char * p;
+
+  in.read(buf, 11);
+  if (utf8::is_bom(buf))
+    p = &buf[3];
+  else
+    p = buf;
+
+  if (std::strncmp(p, "<?xml", 5) != 0) {
+    in.clear();
+    in.seekg(0, std::ios::beg);
+    return false;
+  }
+
   in.clear();
   in.seekg(0, std::ios::beg);
-
-  return std::strncmp(buf, "<?xml", 5) == 0;
+  return true;
 }
 
 unsigned int gnucash_parser_t::parse(std::istream& in,
@@ -366,6 +382,7 @@ unsigned int gnucash_parser_t::parse(std::istream& in,
 				     account_t *   master,
 				     const path *  original_file)
 {
+#if 0
   char buf[BUFSIZ];
 
 #if 0
@@ -428,6 +445,9 @@ unsigned int gnucash_parser_t::parse(std::istream& in,
   curr_account_id.clear();
 
   return count;
+#else
+  return 0;
+#endif
 }
 
 } // namespace ledger
diff --git a/src/main.cc b/src/main.cc
index d49cfbeb..2f51a5bf 100644
--- a/src/main.cc
+++ b/src/main.cc
@@ -39,10 +39,8 @@
 
 #include "textual.h"
 #include "qif.h"
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
 #include "xml.h"
 #include "gnucash.h"
-#endif
 #ifdef HAVE_LIBOFX
 #include "ofx.h"
 #endif
@@ -469,10 +467,8 @@ int main(int argc, char * argv[], char * envp[])
 #if 0
     session->register_parser(new ledger::journal_t::binary_parser_t);
 #endif
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
     session->register_parser(new ledger::xml_parser_t);
     session->register_parser(new ledger::gnucash_parser_t);
-#endif
 #ifdef HAVE_LIBOFX
     session->register_parser(new ledger::ofx_parser_t);
 #endif
diff --git a/src/session.h b/src/session.h
index a0637b2e..84e19531 100644
--- a/src/session.h
+++ b/src/session.h
@@ -188,9 +188,7 @@ public:
 This program is made available under the terms of the BSD Public License.\n\
 See LICENSE file included with the distribution for details and disclaimer.\n";
     std::cout << "\n(modules: gmp, pcre";
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
     std::cout << ", xml";
-#endif
 #ifdef HAVE_LIBOFX
     std::cout << ", ofx";
 #endif
diff --git a/src/system.hh b/src/system.hh
index 0dc4d33d..b44678d6 100644
--- a/src/system.hh
+++ b/src/system.hh
@@ -122,13 +122,8 @@ typedef std::ostream::pos_type ostream_pos_type;
 
 #include <gmp.h>
 
-extern "C" {
-#if defined(HAVE_EXPAT)
-#include <expat.h>		// expat XML parser
-#elif defined(HAVE_XMLPARSE)
-#include <xmlparse.h>		// expat XML parser
-#endif
-}
+#include "irrXML.h"		// XML parser
+#include "CXMLReaderImpl.h"
 
 #if defined(HAVE_LIBOFX)
 #include <libofx.h>
diff --git a/src/textual.cc b/src/textual.cc
index 4600f3f3..95c90708 100644
--- a/src/textual.cc
+++ b/src/textual.cc
@@ -605,13 +605,9 @@ bool textual_parser_t::test(std::istream& in) const
   else
     p = buf;
 
-  if (std::strncmp(p, "<?xml", 5) == 0) {
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
-    throw parse_error("Ledger file contains XML data, but format was not recognized");
-#else
-    throw parse_error("Ledger file contains XML data, but no XML support present");
-#endif
-  }
+  if (std::strncmp(p, "<?xml", 5) == 0)
+    throw_(parse_error,
+	   "Ledger file contains XML data, but format was not recognized");
 
   in.clear();
   in.seekg(0, std::ios::beg);
diff --git a/src/xml.cc b/src/xml.cc
index ccaa65df..0d04c158 100644
--- a/src/xml.cc
+++ b/src/xml.cc
@@ -35,10 +35,8 @@
 
 namespace ledger {
 
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
-
-static XML_Parser    current_parser;
-static unsigned int  count;
+static irr::io::IrrXMLReader * current_parser;
+static unsigned int	       count;
 
 static journal_t *   curr_journal;
 static entry_t *     curr_entry;
@@ -51,7 +49,7 @@ static string data;
 static bool   ignore;
 static string have_error;
 
-static void startElement(void *userData, const char *name, const char **attrs)
+static void startElement(const char *name)
 {
   if (ignore)
     return;
@@ -68,15 +66,15 @@ static void startElement(void *userData, const char *name, const char **attrs)
       curr_entry->xacts.back()->set_state(curr_state);
   }
   else if (std::strcmp(name, "commodity") == 0) {
-    if (string(attrs[0]) == "flags")
-      comm_flags = attrs[1];
+    if (const char * p = current_parser->getAttributeValue("flags"))
+      comm_flags = p;
   }
   else if (std::strcmp(name, "total") == 0) {
     ignore = true;
   }
 }
 
-static void endElement(void *userData, const char *name)
+static void endElement(const char *name)
 {
   if (ignore) {
     if (std::strcmp(name, "total") == 0)
@@ -181,27 +179,32 @@ static void endElement(void *userData, const char *name)
   }
 }
 
-static void dataHandler(void *userData, const char *s, int len)
-{
-  if (! ignore)
-    data = string(s, len);
-}
-
 bool xml_parser_t::test(std::istream& in) const
 {
-  char buf[80];
+  char	 buf[80];
+  char * p;
 
-  in.getline(buf, 79);
-  if (std::strncmp(buf, "<?xml", 5) != 0) {
+  DEBUG("xml.parse", "Testing whether the file is XML...");
+
+  in.read(buf, 10);
+  if (utf8::is_bom(buf))
+    p = &buf[3];
+  else
+    p = buf;
+
+  if (std::strncmp(p, "<?xml", 5) != 0) {
     in.clear();
     in.seekg(0, std::ios::beg);
+    DEBUG("xml.parse", "Does not begin with <?xml");
     return false;
   }
 
+  in.getline(buf, 79);		// skip rest of <?xml line
   in.getline(buf, 79);
   if (! std::strstr(buf, "<ledger")) {
     in.clear();
     in.seekg(0, std::ios::beg);
+    DEBUG("xml.parse", "Next line does not begin with <ledger");
     return false;
   }
 
@@ -211,11 +214,13 @@ bool xml_parser_t::test(std::istream& in) const
 }
 
 unsigned int xml_parser_t::parse(std::istream& in,
-				 session_t&     session,
-				 journal_t&   journal,
+				 session_t&    session,
+				 journal_t&    journal,
 				 account_t *   master,
 				 const path *  original_file)
 {
+  TRACE_START(xml_parsing_total, 1, "Total time spent parsing XML:");
+
   char buf[BUFSIZ];
 
   count        = 0;
@@ -224,47 +229,55 @@ unsigned int xml_parser_t::parse(std::istream& in,
   curr_comm    = NULL;
   ignore       = false;
 
-  XML_Parser   parser = XML_ParserCreate(NULL);
+  irr::io::IrrXMLReader * parser =
+    new irr::io::CXMLReaderImpl<char, irr::io::IXMLBase>(new CStreamReadCallBack(in)); 
   current_parser = parser;
 
-  XML_SetElementHandler(parser, startElement, endElement);
-  XML_SetCharacterDataHandler(parser, dataHandler);
+  while (parser->read()) {
+    switch (parser->getNodeType()) {
+    case irr::io::EXN_TEXT:
+      DEBUG("xml.parse", "Read text: " << parser->getNodeData());
+      if (! ignore) {
+	DEBUG("xml.parse", "  but ignoring it");
+	data = parser->getNodeData();
+      }
+      break;
+
+    case irr::io::EXN_ELEMENT:
+      DEBUG("xml.parse", "Read element: " << parser->getNodeName());
+      startElement(parser->getNodeName());
+      break;
+    case irr::io::EXN_ELEMENT_END:
+      DEBUG("xml.parse", "End element: " << parser->getNodeName());
+      endElement(parser->getNodeName());
+      break;
 
-  while (! in.eof()) {
-    in.getline(buf, BUFSIZ - 1);
-    std::strcat(buf, "\n");
-    bool result;
-    try {
-      result = XML_Parse(parser, buf, std::strlen(buf), in.eof());
-    }
-    catch (const std::exception& err) {
-      //unsigned long line = XML_GetCurrentLineNumber(parser) - offset++;
-      XML_ParserFree(parser);
-      throw parse_error(err.what());
+    default:			// ignore: COMMENT, CDATA, UNKNOWN
+      break;
     }
 
     if (! have_error.empty()) {
-      //unsigned long line = XML_GetCurrentLineNumber(parser) - offset++;
       parse_error err(have_error);
       std::cerr << "Error: " << err.what() << std::endl;
       have_error = "";
     }
 
+#if 0
     if (! result) {
-      //unsigned long line = XML_GetCurrentLineNumber(parser) - offset++;
       const char *  err  = XML_ErrorString(XML_GetErrorCode(parser));
       XML_ParserFree(parser);
       throw parse_error(err);
     }
+#endif
   }
 
-  XML_ParserFree(parser);
+  delete parser;
+
+  TRACE_FINISH(xml_parsing_total, 1);
 
   return count;
 }
 
-#endif // defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
-
 void xml_write_amount(std::ostream& out, const amount_t& amount,
 		      const int depth = 0)
 {
diff --git a/src/xml.h b/src/xml.h
index 6630146f..8e1c5af7 100644
--- a/src/xml.h
+++ b/src/xml.h
@@ -38,7 +38,37 @@
 
 namespace ledger {
 
-#if defined(HAVE_EXPAT) || defined(HAVE_XMLPARSE)
+class CStreamReadCallBack : public irr::io::IFileReadCallBack
+{
+  std::istream& in;
+  std::size_t	size;
+
+public:
+  //! construct from filename
+  CStreamReadCallBack(std::istream& _in) : in(_in), size(0) {
+    TRACE_CTOR(CStreamReadCallBack, "std::istream&");
+  }
+  virtual ~CStreamReadCallBack() {
+    TRACE_DTOR(CStreamReadCallBack);
+  }
+
+  virtual int read(void * buffer, int sizeToRead)
+  {
+    in.read(static_cast<char *>(buffer), sizeToRead);
+    return in.gcount();
+  }
+
+  virtual int getSize()
+  {
+    if (size == 0) {
+      std::ifstream::pos_type pos = in.tellg();
+      in.seekg(0, std::ios_base::end);
+      size = in.tellg() - pos;
+      in.seekg(pos, std::ios_base::beg);
+    }
+    return size;
+  }
+};
 
 class xml_parser_t : public journal_t::parser_t
 {
@@ -46,14 +76,12 @@ class xml_parser_t : public journal_t::parser_t
   virtual bool test(std::istream& in) const;
 
   virtual unsigned int parse(std::istream& in,
-			     session_t&     session,
-			     journal_t&   journal,
+			     session_t&    session,
+			     journal_t&	   journal,
 			     account_t *   master        = NULL,
 			     const path *  original_file = NULL);
 };
 
-#endif
-
 class format_xml_entries : public format_entries
 {
   bool show_totals;