X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fgnumeric-reader.c;h=0783e10f7c1b33452222b82238d95d51ae582004;hb=df2866a475d36298a14cc1e97c6118e088ac85da;hp=f166ee6aba86f1a00acefb409c466e253635a574;hpb=b5c82cc9aabe7e641011130240ae1b2e84348e23;p=pspp

diff --git a/src/data/gnumeric-reader.c b/src/data/gnumeric-reader.c
index f166ee6aba..0783e10f7c 100644
--- a/src/data/gnumeric-reader.c
+++ b/src/data/gnumeric-reader.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2007, 2009 Free Software Foundation, Inc.
+   Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013  Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -14,59 +14,77 @@
    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>. */
 
-
-
 #include <config.h>
 
-#include <libpspp/message.h>
-#include <libpspp/misc.h>
+#include "libpspp/message.h"
+#include "libpspp/misc.h"
 
-#include "minmax.h"
+#include "gl/minmax.h"
+#include "gl/c-strtod.h"
 
 #include "gettext.h"
 #define _(msgid) gettext (msgid)
 #define N_(msgid) (msgid)
 
+#include "spreadsheet-reader.h"
 
-#if !GNM_SUPPORT
+#if !GNM_READ_SUPPORT
 
 struct casereader *
-gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
+gnumeric_open_reader (const struct spreadsheet_read_options *opts, struct dictionary **dict)
 {
-  msg (ME, _("Support for Gnumeric files was not compiled into this installation of PSPP"));
+  msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "Gnumeric");
 
   return NULL;
 }
 
 #else
 
-#include <data/casereader-provider.h>
-#include <errno.h>
-#include <libpspp/str.h>
-#include <libpspp/i18n.h>
-#include <data/dictionary.h>
-#include <data/variable.h>
-#include <xalloc.h>
+#include "data/gnumeric-reader.h"
 
+#include <assert.h>
+#include <stdbool.h>
 #include <errno.h>
 #include <libxml/xmlreader.h>
 #include <zlib.h>
-#include <stdbool.h>
 
-#include <data/case.h>
-#include <data/value.h>
+#include "data/format.h"
+#include "data/data-in.h"
+#include "data/case.h"
+#include "data/casereader-provider.h"
+#include "data/dictionary.h"
+#include "data/identifier.h"
+#include "data/value.h"
+#include "data/variable.h"
+#include "libpspp/i18n.h"
+#include "libpspp/str.h"
+
+#include "gl/xalloc.h"
+
+
+/* Shamelessly lifted from the Gnumeric sources:
+   https://git.gnome.org/browse/gnumeric/tree/src/value.h
+ */
+enum gnm_value_type
+{
+  VALUE_EMPTY   = 10,
+  VALUE_BOOLEAN = 20,
+  VALUE_INTEGER = 30, /* Note, this was removed from gnumeric in 2006 - old versions may of
+			 course still be around. New ones are supposed to use float.*/
+  VALUE_FLOAT   = 40,
+  VALUE_ERROR   = 50,
+  VALUE_STRING  = 60,
+  VALUE_CELLRANGE  = 70,
+  VALUE_ARRAY   = 80
+};
 
-#include "gnumeric-reader.h"
-#include <data/identifier.h>
-#include <assert.h>
 
-/* Default width of string variables. */
-#define GNUMERIC_DEFAULT_WIDTH 8
 
 static void gnm_file_casereader_destroy (struct casereader *, void *);
 
 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
 
+
 static const struct casereader_class gnm_file_casereader_class =
   {
     gnm_file_casereader_read,
@@ -75,235 +93,352 @@ static const struct casereader_class gnm_file_casereader_class =
     NULL,
   };
 
-/* Convert a string, which is an integer encoded in base26
-   IE, A=0, B=1, ... Z=25 to the integer it represents.
-   ... except that in this scheme, digits with an exponent
-   greater than 1 are implicitly incremented by 1, so
-   AA  = 0 + 1*26, AB = 1 + 1*26,
-   ABC = 2 + 2*26 + 1*26^2 ....
-*/
-static int
-pseudo_base26 (const char *str)
-{
-  int i;
-  int multiplier = 1;
-  int result = 0;
-  int len = strlen (str);
-
-  for ( i = len - 1 ; i >= 0; --i)
-    {
-      int mantissa = (str[i] - 'A');
-
-      if ( mantissa < 0 || mantissa > 25 )
-	return -1;
-
-      if ( i != len - 1)
-	mantissa++;
-
-      result += mantissa * multiplier;
-
-      multiplier *= 26;
-    }
-
-  return result;
-}
-
-
-
-/* Convert a cell reference in the form "A1:B2", to
-   integers.  A1 means column zero, row zero.
-   B1 means column 1 row 0. AA1 means column 26, row 0.
-*/
-static bool
-convert_cell_ref (const char *ref,
-		  int *col0, int *row0,
-		  int *coli, int *rowi)
-{
-  char startcol[5];
-  char stopcol [5];
-
-  int startrow;
-  int stoprow;
-
-  int n = sscanf (ref, "%4[a-zA-Z]%d:%4[a-zA-Z]%d",
-	      startcol, &startrow,
-	      stopcol, &stoprow);
-  if ( n != 4)
-    return false;
-
-  str_uppercase (startcol);
-  *col0 = pseudo_base26 (startcol);
-  str_uppercase (stopcol);
-  *coli = pseudo_base26 (stopcol);
-  *row0 = startrow - 1;
-  *rowi = stoprow - 1 ;
-
-  return true;
-}
-
-
 enum reader_state
   {
-    STATE_INIT = 0,        /* Initial state */
+    STATE_PRE_INIT = 0,        /* Initial state */
+    STATE_SHEET_COUNT,      /* Found the sheet index */
+    STATE_INIT ,           /* Other Initial state */
     STATE_SHEET_START,     /* Found the start of a sheet */
     STATE_SHEET_NAME,      /* Found the sheet name */
     STATE_MAXROW,
+    STATE_MAXCOL,
     STATE_SHEET_FOUND,     /* Found the sheet that we actually want */
     STATE_CELLS_START,     /* Found the start of the cell array */
     STATE_CELL             /* Found a cell */
   };
 
+struct sheet_detail
+{
+  /* The name of the sheet (utf8 encoding) */
+  char *name;
 
-struct gnumeric_reader
+  int start_col;
+  int stop_col;
+  int start_row;
+  int stop_row;
+
+  int maxcol;
+  int maxrow;
+};
+
+struct state_data 
 {
+  /* The libxml reader for this instance */
   xmlTextReaderPtr xtr;
 
+  /* An internal state variable */
   enum reader_state state;
+
+  int node_type;
+  int current_sheet;
+
   int row;
   int col;
-  int node_type;
-  int sheet_index;
 
+  int min_col;
+};
 
-  const xmlChar *target_sheet;
-  int target_sheet_index;
 
-  int start_row;
+static void
+state_data_destroy (struct state_data *sd)
+{
+  xmlFreeTextReader (sd->xtr);
+}
+
+
+struct gnumeric_reader
+{
+  struct spreadsheet spreadsheet;
+
+  struct state_data rsd;
+  struct state_data msd;
+
   int start_col;
-  int stop_row;
   int stop_col;
+  int start_row;
+  int stop_row;
+  
+  struct sheet_detail *sheets;
+
+  const xmlChar *target_sheet;
+  int target_sheet_index;
 
   struct caseproto *proto;
   struct dictionary *dict;
   struct ccase *first_case;
   bool used_first_case;
+
+  enum gnm_value_type vtype;
 };
 
-static void process_node (struct gnumeric_reader *r);
 
-#define _xml(X) (const xmlChar *)(X)
+void
+gnumeric_unref (struct spreadsheet *s)
+{
+  struct gnumeric_reader *r = (struct gnumeric_reader *) s;
+
+  if (0 == --s->ref_cnt)
+    {
+      int i;
+
+      for (i = 0; i < s->n_sheets; ++i)
+	{
+	  xmlFree (r->sheets[i].name);
+	}
+    
+      free (r->sheets);
+      state_data_destroy (&r->msd);
+
+      free (s->file_name);
+
+      free (r);
+    }
+}
+
+
+const char *
+gnumeric_get_sheet_name (struct spreadsheet *s, int n)
+{
+  struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
+  assert (n < s->n_sheets);
+
+  return gr->sheets[n].name; 
+}
+
+
+static void process_node (struct gnumeric_reader *r, struct state_data *sd);
+
+
+
+char *
+gnumeric_get_sheet_range (struct spreadsheet *s, int n)
+{
+  int ret;
+  struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
+  
+  assert (n < s->n_sheets);
+
+  while ( 
+	 (gr->sheets[n].stop_col == -1)
+	 && 
+	 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
+	  )
+    {
+      process_node (gr, &gr->msd);
+    }
+
+  return create_cell_range (
+			  gr->sheets[n].start_col,
+			  gr->sheets[n].start_row,
+			  gr->sheets[n].stop_col,
+			  gr->sheets[n].stop_row);
+}
 
-#define _xmlchar_to_int(X) atoi((const char *)X)
 
 static void
 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
 {
   struct gnumeric_reader *r = r_;
+
   if ( r == NULL)
 	return ;
 
-  if ( r->xtr)
-    xmlFreeTextReader (r->xtr);
+  state_data_destroy (&r->rsd);
 
-  if ( ! r->used_first_case )
+  if (r->first_case &&  ! r->used_first_case )
     case_unref (r->first_case);
 
-  caseproto_unref (r->proto);
+  if (r->proto) 
+    caseproto_unref (r->proto);
 
-  free (r);
+  gnumeric_unref (&r->spreadsheet);
 }
 
+
 static void
-process_node (struct gnumeric_reader *r)
+process_node (struct gnumeric_reader *r, struct state_data *sd)
 {
-  xmlChar *name = xmlTextReaderName (r->xtr);
+  xmlChar *name = xmlTextReaderName (sd->xtr);
   if (name == NULL)
     name = xmlStrdup (_xml ("--"));
 
+  sd->node_type = xmlTextReaderNodeType (sd->xtr);
 
-  r->node_type = xmlTextReaderNodeType (r->xtr);
-
-  switch ( r->state)
+  switch (sd->state)
     {
+    case STATE_PRE_INIT:
+      sd->current_sheet = -1;
+      if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_SHEET_COUNT;
+	}
+      break;
+
+    case STATE_SHEET_COUNT:
+      if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
+	{
+	  ++sd->current_sheet;
+	  if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
+	    {
+	      struct sheet_detail *detail ;
+	      r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
+	      detail = &r->sheets[sd->current_sheet];
+	      detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
+	      detail->name = NULL;
+	      r->spreadsheet.n_sheets = sd->current_sheet + 1;
+	    }
+	}
+      else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_INIT;
+	  sd->current_sheet = -1;
+	}
+      else if (XML_READER_TYPE_TEXT == sd->node_type)
+	{
+	  if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
+	    r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
+	}
+      break;
+
     case STATE_INIT:
       if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
-	  XML_READER_TYPE_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_SHEET_START;
+	  ++sd->current_sheet;
+	  sd->state = STATE_SHEET_START;
 	}
       break;
     case STATE_SHEET_START:
       if (0 == xmlStrcasecmp (name, _xml("gnm:Name"))  &&
-	  XML_READER_TYPE_ELEMENT  == r->node_type)
-	{
-	  r->state = STATE_SHEET_NAME;
-	}
-      else if (0 == xmlStrcasecmp (name, _xml("gnm:Name"))  &&
-	       XML_READER_TYPE_END_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_INIT;
+	  sd->state = STATE_SHEET_NAME;
 	}
       break;
     case STATE_SHEET_NAME:
       if (0 == xmlStrcasecmp (name, _xml("gnm:Name"))  &&
-	  XML_READER_TYPE_END_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_SHEET_START;
+	  sd->state = STATE_INIT;
 	}
-      else if (XML_READER_TYPE_TEXT == r->node_type)
+      else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet"))  &&
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_INIT;
+	}
+      else if (XML_READER_TYPE_TEXT == sd->node_type)
 	{
-	  ++r->sheet_index;
-	  if ( r->target_sheet != NULL)
+       	  if ( r->target_sheet != NULL)
 	    {
-	      xmlChar *value = xmlTextReaderValue (r->xtr);
+	      xmlChar *value = xmlTextReaderValue (sd->xtr);
 	      if ( 0 == xmlStrcmp (value, r->target_sheet))
-		r->state = STATE_SHEET_FOUND;
+		sd->state = STATE_SHEET_FOUND;
 	      free (value);
 	    }
-	  else if (r->target_sheet_index == r->sheet_index)
+	  else if (r->target_sheet_index == sd->current_sheet + 1)
+	    {
+	      sd->state = STATE_SHEET_FOUND;
+	    }
+	  else if (r->target_sheet_index == -1)
 	    {
-	      r->state = STATE_SHEET_FOUND;
+	      sd->state = STATE_SHEET_FOUND;
 	    }
 	}
       break;
     case STATE_SHEET_FOUND:
       if (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))  &&
-	  XML_READER_TYPE_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
 	{
-	  if (! xmlTextReaderIsEmptyElement (r->xtr))
-	    r->state = STATE_CELLS_START;
+	  sd->min_col = INT_MAX;
+	  if (! xmlTextReaderIsEmptyElement (sd->xtr))
+	    sd->state = STATE_CELLS_START;
 	}
       else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow"))  &&
-	  XML_READER_TYPE_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_MAXROW;
+	}
+      else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol"))  &&
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_MAXROW;
+	  sd->state = STATE_MAXCOL;
 	}
       else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet"))  &&
-	  XML_READER_TYPE_END_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_INIT;
+      	  sd->state = STATE_INIT;
 	}
       break;
     case STATE_MAXROW:
       if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow"))  &&
-	  XML_READER_TYPE_END_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
 	{
-	  r->state = STATE_SHEET_FOUND;
+	  sd->state = STATE_SHEET_FOUND;
 	}
+      else if (sd->node_type == XML_READER_TYPE_TEXT)
+	{
+	  xmlChar *value = xmlTextReaderValue (sd->xtr);
+	  r->sheets[sd->current_sheet].maxrow = _xmlchar_to_int (value);
+	  xmlFree (value);
+	}
+      break;
+    case STATE_MAXCOL:
+      if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol"))  &&
+	  XML_READER_TYPE_END_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_SHEET_FOUND;
+	}
+      else if (sd->node_type == XML_READER_TYPE_TEXT)
+	{
+	  xmlChar *value = xmlTextReaderValue (sd->xtr);
+	  r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
+	  xmlFree (value);
+	}
+      break;
     case STATE_CELLS_START:
       if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell"))  &&
-	  XML_READER_TYPE_ELEMENT  == r->node_type)
+	  XML_READER_TYPE_ELEMENT  == sd->node_type)
 	{
 	  xmlChar *attr = NULL;
-	  r->state = STATE_CELL;
 
-	  attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
-	  r->col =  _xmlchar_to_int (attr);
+	  attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
+	  sd->col =  _xmlchar_to_int (attr);
 	  free (attr);
 
-	  attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
-	  r->row = _xmlchar_to_int (attr);
+	  if (sd->col < sd->min_col)
+	    sd->min_col = sd->col;
+
+	  attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
+	  sd->row = _xmlchar_to_int (attr);
 	  free (attr);
-	}
-      else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))  &&
-	       XML_READER_TYPE_END_ELEMENT  == r->node_type)
-	r->state = STATE_SHEET_NAME;
 
+	  if (r->sheets[sd->current_sheet].start_row == -1)
+	    {
+	      r->sheets[sd->current_sheet].start_row = sd->row;
+	    }
+
+	  if (r->sheets[sd->current_sheet].start_col == -1)
+	    {
+	      r->sheets[sd->current_sheet].start_col = sd->col;
+	    }
+	  if (! xmlTextReaderIsEmptyElement (sd->xtr))
+	    sd->state = STATE_CELL;
+	}
+      else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells")))  &&  (XML_READER_TYPE_END_ELEMENT  == sd->node_type) )
+	{
+	  r->sheets[sd->current_sheet].stop_col = sd->col;
+	  r->sheets[sd->current_sheet].stop_row = sd->row;
+	  sd->state = STATE_SHEET_NAME;
+	}
       break;
     case STATE_CELL:
-      if (0 == xmlStrcasecmp (name, _xml("gnm:Cell"))  &&
-			      XML_READER_TYPE_END_ELEMENT  == r->node_type)
-	r->state = STATE_CELLS_START;
+      if (0 == xmlStrcasecmp (name, _xml("gnm:Cell"))  && XML_READER_TYPE_END_ELEMENT  == sd->node_type)
+	{
+	  sd->state = STATE_CELLS_START;
+	}
       break;
     default:
       break;
@@ -318,28 +453,47 @@ process_node (struct gnumeric_reader *r)
  */
 static void
 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
-			     const xmlChar *xv)
+			     const xmlChar *xv, enum gnm_value_type type, int col, int row)
 {
-  int n_bytes = 0;
   union value *v = case_data_rw (c, var);
 
-  const char *text = (const char *) xv;
-
-  if ( text)
-    n_bytes = MIN (var_get_width (var), strlen (text));
-
-  if ( var_is_alpha (var))
-    {
-      memcpy (value_str_rw (v, var_get_width (var)), text, n_bytes);
-    }
-  else
+  if (xv == NULL)
+    value_set_missing (v, var_get_width (var));
+  else if ( var_is_alpha (var))
+    value_copy_str_rpad (v, var_get_width (var), xv, ' ');
+  else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
     {
+      const char *text = CHAR_CAST (const char *, xv);
       char *endptr;
+
       errno = 0;
-      v->f = strtod (text, &endptr);
+      v->f = c_strtod (text, &endptr);
       if ( errno != 0 || endptr == text)
 	v->f = SYSMIS;
     }
+  else
+    {
+      const char *text = CHAR_CAST (const char *, xv);
+
+      const struct fmt_spec *fmt = var_get_write_format (var);
+
+      char *m = data_in (ss_cstr (text), "UTF-8",
+			 fmt->type,
+			 v,
+			 var_get_width (var),
+			 "UTF-8");
+      
+      if (m)
+	{
+	  char buf [FMT_STRING_LEN_MAX + 1];
+	  char *cell = create_cell_ref (col, row);
+	  
+	  msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"), 
+	       cell, fmt_to_string (fmt, buf), m);
+	  free (cell);
+	}
+      free (m);
+    }
 }
 
 struct var_spec
@@ -347,127 +501,266 @@ struct var_spec
   char *name;
   int width;
   xmlChar *first_value;
+  int first_type;
 };
 
-struct casereader *
-gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
+
+static void
+gnumeric_error_handler (void *ctx, const char *mesg,
+			UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
 {
-  unsigned long int vstart = 0;
-  int ret;
-  casenumber n_cases = CASENUMBER_MAX;
-  int i;
-  struct var_spec *var_spec = NULL;
-  int n_var_specs = 0;
+  struct gnumeric_reader *r = ctx;
+       
+  msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
+       "Gnumeric",
+       r->spreadsheet.file_name,
+       xmlTextReaderLocatorLineNumber (loc),
+       mesg);
+}
 
-  struct gnumeric_reader *r = NULL;
+static struct gnumeric_reader *
+gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
+{  
+  int ret = -1;
+  struct state_data *sd;
 
-  gzFile gz = gzopen (gri->file_name, "r");
+  xmlTextReaderPtr xtr;
+  gzFile gz;
+
+  assert (r == NULL || filename == NULL);
 
-  if ( NULL == gz)
+  if (filename)
+    {
+      gz = gzopen (filename, "r");
+    }
+  else
     {
-      msg (ME, _("Error opening \"%s\" for reading as a Gnumeric file: %s."),
-           gri->file_name, strerror (errno));
+      gz = gzopen (r->spreadsheet.file_name, "r");
+    }
 
-      goto error;
+  if (NULL == gz)
+    return NULL;
+
+
+  xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
+			(xmlInputCloseCallback) gzclose, gz,
+			NULL, NULL,
+			show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
+
+  if (xtr == NULL)
+    {
+      gzclose (gz);
+      return NULL;
     }
 
-  r = xzalloc (sizeof *r);
+  if (r == NULL)
+    {
+      r = xzalloc (sizeof *r);
+      r->spreadsheet.n_sheets = -1;
+      r->spreadsheet.file_name = strdup (filename);
+      sd = &r->msd;
+    }
+  else
+    {
+      sd = &r->rsd;
+    }
+  
+  if (show_errors) 
+    xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
+
+  r->target_sheet = NULL;
+  r->target_sheet_index = -1;
+
+  sd->row = sd->col = -1;
+  sd->state = STATE_PRE_INIT;
+  sd->xtr = xtr;
+  r->spreadsheet.ref_cnt++;
+
 
-  r->xtr = xmlReaderForIO ((xmlInputReadCallback) gzread, gzclose, gz,
-			   NULL, NULL, 0);
+  /* Advance to the start of the workbook.
+     This gives us some confidence that we are actually dealing with a gnumeric
+     spreadsheet.
+   */
+  while ( (sd->state != STATE_INIT )
+	  && 1 == (ret = xmlTextReaderRead (sd->xtr)))
+    {
+      process_node (r, sd);
+    }
+
+
+  if ( ret != 1)
+    {
+      /* Does not seem to be a gnumeric file */
+      gnumeric_unref (&r->spreadsheet);
+      return NULL;
+    }
 
-  if ( r->xtr == NULL)
-    goto error;
+  r->spreadsheet.type = SPREADSHEET_GNUMERIC;
 
-  if ( gri->cell_range )
+  if (show_errors)
     {
-      if ( ! convert_cell_ref (gri->cell_range,
+      const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
+      xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
+
+      if ( XML_CHAR_ENCODING_UTF8 != xce)
+	{
+	  /* I have been told that ALL gnumeric files are UTF8 encoded.  If that is correct, this 
+	     can never happen. */
+	  msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
+		     "Any non-ascii characters will be incorrectly imported."),
+	       r->spreadsheet.file_name,
+	       enc);
+	}
+    }
+
+  return r;
+}
+
+
+struct spreadsheet *
+gnumeric_probe (const char *filename, bool report_errors)
+{
+  struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
+
+  return &r->spreadsheet;
+}
+
+
+struct casereader *
+gnumeric_make_reader (struct spreadsheet *spreadsheet,
+		      const struct spreadsheet_read_options *opts)
+{
+  int type = 0;
+  int x = 0;
+  struct gnumeric_reader *r = NULL;
+  unsigned long int vstart = 0;
+  int ret;
+  casenumber n_cases = CASENUMBER_MAX;
+  int i;
+  struct var_spec *var_spec = NULL;
+  int n_var_specs = 0;
+
+  r = (struct gnumeric_reader *) (spreadsheet);
+
+  r = gnumeric_reopen (r, NULL, true);
+
+  if ( opts->cell_range )
+    {
+      if ( ! convert_cell_ref (opts->cell_range,
 			       &r->start_col, &r->start_row,
 			       &r->stop_col, &r->stop_row))
 	{
-	  msg (SE, _("Invalid cell range \"%s\""),
-	       gri->cell_range);
+	  msg (SE, _("Invalid cell range `%s'"),
+	       opts->cell_range);
 	  goto error;
 	}
     }
   else
     {
-      r->start_col = 0;
+      r->start_col = -1;
       r->start_row = 0;
       r->stop_col = -1;
       r->stop_row = -1;
     }
 
-  r->state = STATE_INIT;
-  r->target_sheet = BAD_CAST gri->sheet_name;
-  r->target_sheet_index = gri->sheet_index;
-  r->row = r->col = -1;
-  r->sheet_index = 0;
+  r->target_sheet = BAD_CAST opts->sheet_name;
+  r->target_sheet_index = opts->sheet_index;
+  r->rsd.row = r->rsd.col = -1;
+  r->rsd.current_sheet = -1;
+  r->first_case = NULL;
+  r->proto = NULL;
 
   /* Advance to the start of the cells for the target sheet */
-  while ( (r->state != STATE_CELL || r->row < r->start_row )
-	  && 1 == (ret = xmlTextReaderRead (r->xtr)))
+  while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
+	  && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
     {
       xmlChar *value ;
-      process_node (r);
-      value = xmlTextReaderValue (r->xtr);
+      process_node (r, &r->rsd);
+      value = xmlTextReaderValue (r->rsd.xtr);
 
-      if ( r->state == STATE_MAXROW  && r->node_type == XML_READER_TYPE_TEXT)
+      if ( r->rsd.state == STATE_MAXROW  && r->rsd.node_type == XML_READER_TYPE_TEXT)
 	{
 	  n_cases = 1 + _xmlchar_to_int (value) ;
 	}
       free (value);
     }
 
-
   /* If a range has been given, then  use that to calculate the number
      of cases */
-  if ( gri->cell_range)
+  if ( opts->cell_range)
     {
       n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
     }
 
-  if ( gri->read_names )
+  if ( opts->read_names )
     {
       r->start_row++;
       n_cases --;
     }
 
+
   /* Read in the first row of cells,
      including the headers if read_names was set */
   while (
-	 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
-	 && (ret = xmlTextReaderRead (r->xtr))
+	 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
+	 && (ret = xmlTextReaderRead (r->rsd.xtr))
 	 )
     {
       int idx;
-      process_node (r);
 
-      if ( r->row > r->start_row ) break;
+      if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
+	{
+	  xmlChar *attr =
+	    xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+
+	  type  =  _xmlchar_to_int (attr);
+
+	  xmlFree (attr);
+	}
+
+      process_node (r, &r->rsd);
+
+      if ( r->rsd.row > r->start_row ) 
+	{
+	  xmlChar *attr =
+	    xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+	  
+	  r->vtype  =  _xmlchar_to_int (attr);
+	  
+	  xmlFree (attr);
+	  break;
+	}
 
-      if ( r->col < r->start_col ||
-	   (r->stop_col != -1 && r->col > r->stop_col))
+      if ( r->rsd.col < r->start_col ||
+	   (r->stop_col != -1 && r->rsd.col > r->stop_col))
 	continue;
 
-      idx = r->col - r->start_col;
+      idx = r->rsd.col - r->start_col;
 
       if ( idx  >= n_var_specs )
 	{
+	  int i;
+	  var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
+	  for (i = n_var_specs; i <= idx; ++i)
+	  {
+	    var_spec [i].name = NULL;
+	    var_spec [i].width = -1;
+	    var_spec [i].first_value = NULL;
+	    var_spec [i].first_type = -1;
+	  }
 	  n_var_specs =  idx + 1 ;
-	  var_spec = xrealloc (var_spec, sizeof (*var_spec) * n_var_specs);
-	  var_spec [idx].name = NULL;
-	  var_spec [idx].width = -1;
-	  var_spec [idx].first_value = NULL;
 	}
 
-      if ( r->node_type == XML_READER_TYPE_TEXT )
+      var_spec [idx].first_type = type;
+
+      if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
 	{
-	  xmlChar *value = xmlTextReaderValue (r->xtr);
-	  const char *text  = (const char *) value;
+	  xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
+	  const char *text  = CHAR_CAST (const char *, value);
 
-	  if ( r->row < r->start_row)
+	  if ( r->rsd.row < r->start_row)
 	    {
-	      if ( gri->read_names )
+	      if ( opts->read_names )
 		{
 		  var_spec [idx].name = xstrdup (text);
 		}
@@ -477,21 +770,21 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
 	      var_spec [idx].first_value = xmlStrdup (value);
 
 	      if (-1 ==  var_spec [idx].width )
-		var_spec [idx].width = (gri->asw == -1) ?
-		  ROUND_UP (strlen(text), GNUMERIC_DEFAULT_WIDTH) : gri->asw;
+		var_spec [idx].width = (opts->asw == -1) ?
+		  ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
 	    }
 
 	  free (value);
 	}
-      else if ( r->node_type == XML_READER_TYPE_ELEMENT
-		&& r->state == STATE_CELL)
+      else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
+		&& r->rsd.state == STATE_CELL)
 	{
-	  if ( r->row == r->start_row )
+	  if ( r->rsd.row == r->start_row )
 	    {
 	      xmlChar *attr =
-		xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
+		xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
 
-	      if ( NULL == attr || 60 !=  _xmlchar_to_int (attr))
+	      if ( NULL == attr || VALUE_STRING !=  _xmlchar_to_int (attr))
 		var_spec [idx].width = 0;
 
 	      free (attr);
@@ -499,29 +792,29 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
 	}
     }
 
-
-  /* Create the dictionary and populate it */
-  *dict = r->dict = dict_create ();
-
-  dict_set_encoding (r->dict, (const char *) xmlTextReaderConstEncoding (r->xtr));
+  {
+    const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
+    if ( enc == NULL)
+      goto error;
+    /* Create the dictionary and populate it */
+    spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
+  }
 
   for (i = 0 ; i < n_var_specs ; ++i )
     {
-      char name[VAR_NAME_LEN + 1];
+      char *name;
+
+      if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
+	continue;
 
       /* Probably no data exists for this variable, so allocate a
 	 default width */
       if ( var_spec[i].width == -1 )
-	var_spec[i].width = GNUMERIC_DEFAULT_WIDTH;
-
-      if  ( ! dict_make_unique_var_name (r->dict, var_spec[i].name,
-					 &vstart, name))
-	{
-	  msg (ME, _("Cannot create variable name from %s"), var_spec[i].name);
-	  goto error;
-	}
+	var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
 
+      name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
       dict_create_var (r->dict, name, var_spec[i].width);
+      free (name);
     }
 
   /* Create the first case, and cache it */
@@ -529,8 +822,8 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
 
   if ( n_var_specs ==  0 )
     {
-      msg (MW, _("Selected sheet or range of spreadsheet \"%s\" is empty."),
-           gri->file_name);
+      msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
+           spreadsheet->file_name);
       goto error;
     }
 
@@ -538,12 +831,21 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
   r->first_case = case_create (r->proto);
   case_set_missing (r->first_case);
 
+
   for ( i = 0 ; i < n_var_specs ; ++i )
     {
-      const struct variable *var = dict_get_var (r->dict, i);
+      const struct variable *var;
+
+      if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
+	continue;
 
+      var = dict_get_var (r->dict, x++);
+      
       convert_xml_string_to_value (r->first_case, var,
-				   var_spec[i].first_value);
+				   var_spec[i].first_value, 
+				   var_spec[i].first_type,
+				   r->rsd.col + i - 1,
+				   r->rsd.row - 1);
     }
 
   for ( i = 0 ; i < n_var_specs ; ++i )
@@ -553,6 +855,7 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
     }
 
   free (var_spec);
+  
 
   return casereader_create_sequential
     (NULL,
@@ -569,7 +872,8 @@ gnumeric_open_reader (struct gnumeric_read_info *gri, struct dictionary **dict)
     }
 
   free (var_spec);
-  dict_destroy (*dict);
+  dict_destroy (spreadsheet->dict);
+  spreadsheet->dict = NULL;
 
   gnm_file_casereader_destroy (NULL, r);
 
@@ -586,7 +890,7 @@ gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
   int ret = 0;
 
   struct gnumeric_reader *r = r_;
-  int current_row = r->row;
+  int current_row = r->rsd.row;
 
   if ( !r->used_first_case )
     {
@@ -597,34 +901,47 @@ gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
   c = case_create (r->proto);
   case_set_missing (c);
 
-  while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
-	 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
+  if (r->start_col == -1)
+    r->start_col = r->rsd.min_col;
+
+
+  while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
+	 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
     {
-      process_node (r);
+      process_node (r, &r->rsd);
+
+      if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
+	{
+	  xmlChar *attr =
+	    xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+
+	  r->vtype  = _xmlchar_to_int (attr);
 
-      if ( r->col < r->start_col || (r->stop_col != -1 &&
-				     r->col > r->stop_col))
+	  xmlFree (attr);
+	}
+
+      if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
+				     r->rsd.col > r->stop_col))
 	continue;
 
-      if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
+      if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
 	continue;
 
-      if ( r->stop_row != -1 && r->row > r->stop_row)
+      if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
 	break;
 
-      if ( r->node_type == XML_READER_TYPE_TEXT )
-	{
-	  xmlChar *value = xmlTextReaderValue (r->xtr);
-
-	  const int idx = r->col - r->start_col;
 
+      if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
+	{
+	  xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
+	  const int idx = r->rsd.col - r->start_col;
 	  const struct variable *var = dict_get_var (r->dict, idx);
 
-	  convert_xml_string_to_value (c, var, value);
+	  convert_xml_string_to_value (c, var, value, r->vtype, 
+				       r->rsd.col, r->rsd.row);
 
-	  free (value);
+	  xmlFree (value);
 	}
-
     }
 
   if (ret == 1)
@@ -637,4 +954,4 @@ gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
 }
 
 
-#endif /* GNM_SUPPORT */
+#endif /* GNM_READ_SUPPORT */