X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fods-reader.c;h=8f307200b052bb5aa2a8b8d83a6aa93590662d3d;hb=1b62b2fae1214cee4bacf0cf8b9aea3a9723574e;hp=640e08d8f871f2755052133caf916d20185498b2;hpb=bb709687104979db2a1a27e7e943a16ab71eb417;p=pspp

diff --git a/src/data/ods-reader.c b/src/data/ods-reader.c
index 640e08d8f8..8f307200b0 100644
--- a/src/data/ods-reader.c
+++ b/src/data/ods-reader.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2011, 2012, 2013 Free Software Foundation, Inc.
+   Copyright (C) 2011, 2012, 2013, 2016 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
@@ -16,59 +16,42 @@
 
 #include <config.h>
 
-#include "libpspp/message.h"
-#include "libpspp/misc.h"
-#include "libpspp/assertion.h"
-
-#include "data/data-in.h"
-
-#include "gl/minmax.h"
-
-#include "gettext.h"
-#define _(msgid) gettext (msgid)
-#define N_(msgid) (msgid)
-
 #include "ods-reader.h"
 #include "spreadsheet-reader.h"
 
-#if !ODF_READ_SUPPORT
-
-struct casereader *
-ods_open_reader (const struct spreadsheet_read_options *opts, 
-		 struct dictionary **dict)
-{
-  msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "OpenDocument");
-
-  return NULL;
-}
-
-#else
-
-#include "libpspp/zip-reader.h"
-
-
 #include <assert.h>
 #include <stdbool.h>
 #include <errno.h>
 #include <libxml/xmlreader.h>
 #include <zlib.h>
 
-#include "data/format.h"
 #include "data/case.h"
 #include "data/casereader-provider.h"
+#include "data/data-in.h"
 #include "data/dictionary.h"
+#include "data/format.h"
 #include "data/identifier.h"
 #include "data/value.h"
 #include "data/variable.h"
+#include "libpspp/assertion.h"
 #include "libpspp/i18n.h"
+#include "libpspp/message.h"
+#include "libpspp/misc.h"
 #include "libpspp/str.h"
+#include "libpspp/zip-reader.h"
 
+#include "gl/c-strtod.h"
+#include "gl/minmax.h"
 #include "gl/xalloc.h"
 
-static void ods_file_casereader_destroy (struct casereader *, void *);
+#include "gettext.h"
+#define _(msgid) gettext (msgid)
+#define N_(msgid) (msgid)
 
+static void ods_file_casereader_destroy (struct casereader *, void *);
 static struct ccase *ods_file_casereader_read (struct casereader *, void *);
 
+
 static const struct casereader_class ods_file_casereader_class =
   {
     ods_file_casereader_read,
@@ -99,34 +82,51 @@ enum reader_state
     STATE_CELL_CONTENT     /* Found a the text within a cell */
   };
 
-struct ods_reader
+struct state_data
 {
-  struct spreadsheet spreadsheet;
-  struct zip_reader *zreader;
   xmlTextReaderPtr xtr;
-
+  struct zip_member *zm;
+  int node_type;
   enum reader_state state;
-  bool sheet_found;
   int row;
   int col;
-  int node_type;
   int current_sheet;
+  xmlChar *current_sheet_name;
+
+  int col_span;
+};
+
+static void
+state_data_destroy (struct state_data *sd)
+{
+  xmlFree (sd->current_sheet_name);
+  sd->current_sheet_name = NULL;
+
+  xmlFreeTextReader (sd->xtr);
+  sd->xtr = NULL;
+
+  zip_member_finish (sd->zm);
+  sd->zm = NULL;
+}
+
+struct ods_reader
+{
+  struct spreadsheet spreadsheet;
+  struct zip_reader *zreader;
 
-  const xmlChar *target_sheet;
   int target_sheet_index;
+  xmlChar *target_sheet_name;
 
+  /* State data for the meta data */
+  struct state_data msd;
 
-  int wanted_row_start;
-  int wanted_col_start;
+  /* State data for the reader */
+  struct state_data rsd;
 
-#if 0
   int start_row;
   int start_col;
   int stop_row;
   int stop_col;
-#endif
-
-  int col_span;
 
   struct sheet_detail *sheets;
   int n_allocated_sheets;
@@ -138,57 +138,104 @@ struct ods_reader
   bool read_names;
 
   struct string ods_errs;
+
+  struct string zip_errs;
 };
 
+void
+ods_unref (struct spreadsheet *s)
+{
+  struct ods_reader *r = (struct ods_reader *) s;
+
+  if (--s->ref_cnt == 0)
+    {
+      int i;
 
-static void process_node (struct ods_reader *r);
+      state_data_destroy (&r->msd);
+      for (i = 0; i < r->n_allocated_sheets; ++i)
+	{
+	  xmlFree (r->sheets[i].name);
+	}
+
+      dict_unref (r->dict);
+
+      zip_reader_destroy (r->zreader);
+      free (r->sheets);
+      free (s->file_name);
+      free (r);
+    }
+}
+
+
+
+static bool
+reading_target_sheet (const struct ods_reader *r, const struct state_data *msd)
+{
+  if (r->target_sheet_name != NULL)
+    {
+      if (0 == xmlStrcmp (r->target_sheet_name, msd->current_sheet_name))
+	return true;
+    }
+
+  if (r->target_sheet_index == msd->current_sheet + 1)
+    return true;
+
+  return false;
+}
+
+
+static void process_node (struct ods_reader *or, struct state_data *r);
 
 
 const char *
 ods_get_sheet_name (struct spreadsheet *s, int n)
 {
-  int ret;
-  struct ods_reader *or = (struct ods_reader *) s;
-  
+  struct ods_reader *r = (struct ods_reader *) s;
+  struct state_data *or = &r->msd;
+
   assert (n < s->n_sheets);
 
-  while ( 
-	 (or->n_allocated_sheets <= n)
-	 && 
-	 (1 == (ret = xmlTextReaderRead (or->xtr)))
-	  )
+  while (
+	  (r->n_allocated_sheets <= n)
+	  || or->state != STATE_SPREADSHEET
+	)
     {
-      process_node (or);
+      int ret = xmlTextReaderRead (or->xtr);
+      if (ret != 1)
+	break;
+
+      process_node (r, or);
     }
 
-  return or->sheets[n].name;
+  return r->sheets[n].name;
 }
 
 char *
 ods_get_sheet_range (struct spreadsheet *s, int n)
 {
-  int ret = -1;
-  struct ods_reader *or = (struct ods_reader *) s;
-  
+  struct ods_reader *r = (struct ods_reader *) s;
+  struct state_data *or = &r->msd;
+
   assert (n < s->n_sheets);
 
-  while ( 
-	 (
-	  (or->n_allocated_sheets <= n)
-	  || (or->sheets[n].stop_row == -1) )
-	 && 
-	 (1 == (ret = xmlTextReaderRead (or->xtr)))
-	  )
+  while (
+	  (r->n_allocated_sheets <= n)
+	  || (r->sheets[n].stop_row == -1)
+	  || or->state != STATE_SPREADSHEET
+	)
     {
-      process_node (or);
-    }
+      int ret = xmlTextReaderRead (or->xtr);
+      if (ret != 1)
+	break;
 
+      process_node (r, or);
+    }
 
-  return create_cell_ref (
-			  or->sheets[n].start_col,
-			  or->sheets[n].start_row,
-			  or->sheets[n].stop_col,
-			  or->sheets[n].stop_row);
+  return create_cell_range (
+			  r->sheets[n].start_col,
+			  r->sheets[n].start_row,
+			  r->sheets[n].stop_col,
+			  r->sheets[n].stop_row);
 }
 
 
@@ -196,28 +243,36 @@ static void
 ods_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
 {
   struct ods_reader *r = r_;
-  if ( r == NULL)
+  if (r == NULL)
     return ;
 
-  if (r->xtr)
-    xmlFreeTextReader (r->xtr);
-  r->xtr = NULL;
+  state_data_destroy (&r->rsd);
 
-  if ( ! ds_is_empty (&r->ods_errs))
+  if (! ds_is_empty (&r->ods_errs))
     msg (ME, "%s", ds_cstr (&r->ods_errs));
 
   ds_destroy (&r->ods_errs);
 
-  if ( ! r->used_first_case )
+  if (r->first_case && ! r->used_first_case)
     case_unref (r->first_case);
 
+
   caseproto_unref (r->proto);
+  r->proto = NULL;
+
+  xmlFree (r->target_sheet_name);
+  r->target_sheet_name = NULL;
+
 
-  //  free (r);
+  ods_unref (&r->spreadsheet);
 }
 
+
+
+
+
 static void
-process_node (struct ods_reader *r)
+process_node (struct ods_reader *or, struct state_data *r)
 {
   xmlChar *name = xmlTextReaderName (r->xtr);
   if (name == NULL)
@@ -234,89 +289,83 @@ process_node (struct ods_reader *r)
 	{
 	  r->state = STATE_SPREADSHEET;
 	  r->current_sheet = -1;
+	  r->current_sheet_name = NULL;
 	}
       break;
     case STATE_SPREADSHEET:
       if (0 == xmlStrcasecmp (name, _xml("table:table"))
-	  && 
+	  &&
 	  (XML_READER_TYPE_ELEMENT == r->node_type))
 	{
-	  xmlChar *value = xmlTextReaderGetAttribute (r->xtr, _xml ("table:name"));
+	  xmlFree (r->current_sheet_name);
+	  r->current_sheet_name = xmlTextReaderGetAttribute (r->xtr, _xml ("table:name"));
 
 	  ++r->current_sheet;
 
-	  if (r->current_sheet >= r->n_allocated_sheets)
+	  if (r->current_sheet >= or->n_allocated_sheets)
 	    {
-	      r->sheets = xrealloc (r->sheets, sizeof (*r->sheets) * ++r->n_allocated_sheets);
-	      r->sheets[r->n_allocated_sheets - 1].start_col = -1;
-	      r->sheets[r->n_allocated_sheets - 1].stop_col = -1;
-	      r->sheets[r->n_allocated_sheets - 1].start_row = -1;
-	      r->sheets[r->n_allocated_sheets - 1].stop_row = -1;
-	      r->sheets[r->n_allocated_sheets - 1].name = value;
+	      assert (r->current_sheet == or->n_allocated_sheets);
+	      or->sheets = xrealloc (or->sheets, sizeof (*or->sheets) * ++or->n_allocated_sheets);
+	      or->sheets[or->n_allocated_sheets - 1].start_col = -1;
+	      or->sheets[or->n_allocated_sheets - 1].stop_col = -1;
+	      or->sheets[or->n_allocated_sheets - 1].start_row = -1;
+	      or->sheets[or->n_allocated_sheets - 1].stop_row = -1;
+	      or->sheets[or->n_allocated_sheets - 1].name = CHAR_CAST (char *, xmlStrdup (r->current_sheet_name));
 	    }
 
 	  r->col = 0;
 	  r->row = 0;
 
-	  if ( r->target_sheet != NULL)
-	    {
-	      if ( 0 == xmlStrcmp (value, r->target_sheet))
-		{
-		  r->sheet_found = true;
-		}
-	    }
-	  else if (r->target_sheet_index == r->current_sheet)
-	    {
-	      r->sheet_found = true;
-	    }
 	  r->state = STATE_TABLE;
 	}
       else if (0 == xmlStrcasecmp (name, _xml("office:spreadsheet")) &&
 	       XML_READER_TYPE_ELEMENT  == r->node_type)
 	{
 	  r->state = STATE_INIT;
-	  printf ("%s:%d End of Workbook %d: Rows %d Cols %d\n", __FILE__, __LINE__,
-		  r->current_sheet,  r->row, r->col);
 	}
       break;
     case STATE_TABLE:
-      if (0 == xmlStrcasecmp (name, _xml("table:table-row")) && 
+      if (0 == xmlStrcasecmp (name, _xml("table:table-row")) &&
 	  (XML_READER_TYPE_ELEMENT  == r->node_type))
 	{
 	  xmlChar *value =
 	    xmlTextReaderGetAttribute (r->xtr,
 				       _xml ("table:number-rows-repeated"));
-	  
+
 	  int row_span = value ? _xmlchar_to_int (value) : 1;
 
 	  r->row += row_span;
 	  r->col = 0;
-	  
+
 	  if (! xmlTextReaderIsEmptyElement (r->xtr))
 	    r->state = STATE_ROW;
+
+	  xmlFree (value);
 	}
-      else if (0 == xmlStrcasecmp (name, _xml("table:table")) && 
+      else if (0 == xmlStrcasecmp (name, _xml("table:table")) &&
 	       (XML_READER_TYPE_END_ELEMENT  == r->node_type))
 	{
 	  r->state = STATE_SPREADSHEET;
 	}
       break;
     case STATE_ROW:
-      if ( (0 == xmlStrcasecmp (name, _xml ("table:table-cell")))
-	   && 
+      if ((0 == xmlStrcasecmp (name, _xml ("table:table-cell")))
+	   &&
 	   (XML_READER_TYPE_ELEMENT  == r->node_type))
 	{
 	  xmlChar *value =
 	    xmlTextReaderGetAttribute (r->xtr,
 				       _xml ("table:number-columns-repeated"));
-	  
+
 	  r->col_span = value ? _xmlchar_to_int (value) : 1;
 	  r->col += r->col_span;
 
 	  if (! xmlTextReaderIsEmptyElement (r->xtr))
 	    r->state = STATE_CELL;
+
+	  xmlFree (value);
 	}
-      else if ( (0 == xmlStrcasecmp (name, _xml ("table:table-row")))
+      else if ((0 == xmlStrcasecmp (name, _xml ("table:table-row")))
 		&&
 		(XML_READER_TYPE_END_ELEMENT  == r->node_type))
 	{
@@ -324,40 +373,40 @@ process_node (struct ods_reader *r)
 	}
       break;
     case STATE_CELL:
-      if ( (0 == xmlStrcasecmp (name, _xml("text:p")))
+      if ((0 == xmlStrcasecmp (name, _xml("text:p")))
 	    &&
-	   ( XML_READER_TYPE_ELEMENT  == r->node_type))
+	   (XML_READER_TYPE_ELEMENT  == r->node_type))
 	{
 	  if (! xmlTextReaderIsEmptyElement (r->xtr))
 	    r->state = STATE_CELL_CONTENT;
 	}
       else if
-	( (0 == xmlStrcasecmp (name, _xml("table:table-cell")))
+	((0 == xmlStrcasecmp (name, _xml("table:table-cell")))
 	  &&
 	  (XML_READER_TYPE_END_ELEMENT  == r->node_type)
-	  )
+	)
 	{
 	  r->state = STATE_ROW;
 	}
       break;
     case STATE_CELL_CONTENT:
       assert (r->current_sheet >= 0);
-      assert (r->current_sheet < r->n_allocated_sheets);
+      assert (r->current_sheet < or->n_allocated_sheets);
 
-      if (r->sheets[r->current_sheet].start_row == -1)
-	r->sheets[r->current_sheet].start_row = r->row - 1;
+      if (or->sheets[r->current_sheet].start_row == -1)
+	or->sheets[r->current_sheet].start_row = r->row - 1;
 
-      if ( 
-	  (r->sheets[r->current_sheet].start_col == -1)
+      if (
+	  (or->sheets[r->current_sheet].start_col == -1)
 	  ||
-	  (r->sheets[r->current_sheet].start_col >= r->col - 1)
-	   )
-	r->sheets[r->current_sheet].start_col = r->col - 1;
+	  (or->sheets[r->current_sheet].start_col >= r->col - 1)
+	)
+	or->sheets[r->current_sheet].start_col = r->col - 1;
 
-      r->sheets[r->current_sheet].stop_row = r->row - 1;
+      or->sheets[r->current_sheet].stop_row = r->row - 1;
 
-      if ( r->sheets[r->current_sheet].stop_col <  r->col - 1)
-	r->sheets[r->current_sheet].stop_col = r->col - 1;
+      if (or->sheets[r->current_sheet].stop_col < r->col - 1)
+	or->sheets[r->current_sheet].stop_col = r->col - 1;
 
       if (XML_READER_TYPE_END_ELEMENT  == r->node_type)
 	r->state = STATE_CELL;
@@ -370,8 +419,8 @@ process_node (struct ods_reader *r)
   xmlFree (name);
 }
 
-/* 
-   A struct containing the parameters of a cell's value 
+/*
+   A struct containing the parameters of a cell's value
    parsed from the xml
 */
 struct xml_value
@@ -398,13 +447,13 @@ xmv_to_width (const struct xml_value *xmv, int fallback)
   if (xmv->type != NULL && 0 != xmlStrcmp (xmv->type, _xml("string")))
     return 0;
 
-  if ( fallback != -1)
+  if (fallback != -1)
     return fallback;
 
-  if ( xmv->value )
+  if (xmv->value)
     width = ROUND_UP (xmlStrlen (xmv->value),
 		      SPREADSHEET_DEFAULT_WIDTH);
-  else if ( xmv->text)
+  else if (xmv->text)
     width = ROUND_UP (xmlStrlen (xmv->text),
 		      SPREADSHEET_DEFAULT_WIDTH);
 
@@ -416,35 +465,58 @@ xmv_to_width (const struct xml_value *xmv, int fallback)
  */
 static void
 convert_xml_to_value (struct ccase *c, const struct variable *var,
-		      const struct xml_value *xmv)
+		      const struct xml_value *xmv, int col, int row)
 {
   union value *v = case_data_rw (c, var);
 
   if (xmv->value == NULL && xmv->text == NULL)
     value_set_missing (v, var_get_width (var));
-  else if ( var_is_alpha (var))
+  else if (var_is_alpha (var))
     /* Use the text field, because it seems that there is no
        value field for strings */
     value_copy_str_rpad (v, var_get_width (var), xmv->text, ' ');
   else
     {
-      const char *text ;
       const struct fmt_spec *fmt = var_get_write_format (var);
       enum fmt_category fc  = fmt_get_category (fmt->type);
 
-      assert ( fc != FMT_CAT_STRING);
+      assert (fc != FMT_CAT_STRING);
+
+      if (0 == xmlStrcmp (xmv->type, _xml("float")))
+	{
+	  v->f = c_strtod (CHAR_CAST (const char *, xmv->value), NULL);
+	}
+      else
+	{
+	  const char *text = xmv->value ?
+	    CHAR_CAST (const char *, xmv->value) : CHAR_CAST (const char *, xmv->text);
 
-      text =
-        xmv->value ? CHAR_CAST (const char *, xmv->value) : CHAR_CAST (const char *, xmv->text);
+	  char *m = data_in (ss_cstr (text), "UTF-8",
+			 fmt->type,
+			 v,
+			 var_get_width (var),
+			 "UTF-8");
 
-      free (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);
+	}
     }
 }
 
+static int
+xml_reader_for_zip_member (void *zm_, char *buffer, int len)
+{
+  struct zip_member *zm = zm_;
+  return zip_member_read (zm, buffer, len);
+}
 
 /* Try to find out how many sheets there are in the "workbook" */
 static int
@@ -454,36 +526,44 @@ get_sheet_count (struct zip_reader *zreader)
   struct zip_member *meta = NULL;
   meta = zip_member_open (zreader, "meta.xml");
 
-  if ( meta == NULL)
+  if (meta == NULL)
     return -1;
 
-  mxtr = xmlReaderForIO ((xmlInputReadCallback) zip_member_read,
-			 (xmlInputCloseCallback) zip_member_finish,
-			 meta,   NULL, NULL, 0);
+  mxtr = xmlReaderForIO (xml_reader_for_zip_member, NULL, meta, NULL, NULL, 0);
 
   while (1 == xmlTextReaderRead (mxtr))
     {
       xmlChar *name = xmlTextReaderName (mxtr);
-      if ( 0 == xmlStrcmp (name, _xml("meta:document-statistic")))
+      if (0 == xmlStrcmp (name, _xml("meta:document-statistic")))
 	{
 	  xmlChar *attr = xmlTextReaderGetAttribute (mxtr, _xml ("meta:table-count"));
-	    
-	  if ( attr != NULL)
+
+	  if (attr != NULL)
 	    {
 	      int s = _xmlchar_to_int (attr);
+	      xmlFreeTextReader (mxtr);
+              zip_member_finish (meta);
+	      xmlFree (name);
+	      xmlFree (attr);
 	      return s;
 	    }
+	  xmlFree (attr);
 	}
+      xmlFree (name);
     }
+
+  xmlFreeTextReader (mxtr);
+  zip_member_finish (meta);
   return -1;
 }
 
 static void
 ods_error_handler (void *ctx, const char *mesg,
-			UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
+		   xmlParserSeverities sev UNUSED,
+		   xmlTextReaderLocatorPtr loc)
 {
   struct ods_reader *r = ctx;
-       
+
   msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
        "ODF",
        r->spreadsheet.file_name,
@@ -493,74 +573,82 @@ ods_error_handler (void *ctx, const char *mesg,
 
 
 static bool
-init_reader (struct ods_reader *r, bool report_errors)
+init_reader (struct ods_reader *r, bool report_errors,
+             struct state_data *state)
 {
   struct zip_member *content = zip_member_open (r->zreader, "content.xml");
   xmlTextReaderPtr xtr;
 
-  if ( content == NULL)
-    return false;
-
-  zip_member_ref (content);
+  if (content == NULL)
+    return NULL;
 
+  xtr = xmlReaderForIO (xml_reader_for_zip_member, NULL, content, NULL, NULL,
+			report_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING));
 
-  xtr = xmlReaderForIO ((xmlInputReadCallback) zip_member_read,
-			(xmlInputCloseCallback) zip_member_finish,
-			content,   NULL, NULL,
-			report_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
+  if (xtr == NULL)
+      return false;
 
-  if ( xtr == NULL)
-    return false;
+  *state = (struct state_data) { .xtr = xtr,
+                                 .zm = content,
+                                 .state = STATE_INIT };
 
-  r->xtr = xtr;
   r->spreadsheet.type = SPREADSHEET_ODS;
 
-  if (report_errors) 
+  if (report_errors)
     xmlTextReaderSetErrorHandler (xtr, ods_error_handler, r);
 
   return true;
 }
 
 
+
 struct spreadsheet *
 ods_probe (const char *filename, bool report_errors)
 {
-  struct ods_reader *r;
-  struct string errs = DS_EMPTY_INITIALIZER;
   int sheet_count;
-  struct zip_reader *zr = zip_reader_create (filename, &errs);
+  struct ods_reader *r = xzalloc (sizeof *r);
+  struct zip_reader *zr;
+
+  ds_init_empty (&r->zip_errs);
+
+  zr = zip_reader_create (filename, &r->zip_errs);
 
   if (zr == NULL)
-    return NULL;
+    {
+      if (report_errors)
+	{
+	  msg (ME, _("Cannot open %s as a OpenDocument file: %s"),
+	       filename, ds_cstr (&r->zip_errs));
+	}
+      ds_destroy (&r->zip_errs);
+      free (r);
+      return NULL;
+    }
 
   sheet_count = get_sheet_count (zr);
 
-  r = xzalloc (sizeof *r);
   r->zreader = zr;
+  r->spreadsheet.ref_cnt = 1;
 
-  if (! init_reader (r, report_errors))
-    {
-      goto error;
-    }
+  if (!init_reader (r, report_errors, &r->msd))
+    goto error;
 
   r->spreadsheet.n_sheets = sheet_count;
   r->n_allocated_sheets = 0;
   r->sheets = NULL;
 
-  ds_destroy (&errs);
-
-  r->spreadsheet.file_name = filename;
+  r->spreadsheet.file_name = strdup (filename);
   return &r->spreadsheet;
 
  error:
+  ds_destroy (&r->zip_errs);
   zip_reader_destroy (r->zreader);
-  ds_destroy (&errs);
   free (r);
   return NULL;
 }
 
 struct casereader *
-ods_make_reader (struct spreadsheet *spreadsheet, 
+ods_make_reader (struct spreadsheet *spreadsheet,
 		 const struct spreadsheet_read_options *opts)
 {
   intf ret = 0;
@@ -577,15 +665,17 @@ ods_make_reader (struct spreadsheet *spreadsheet,
   assert (r);
   r->read_names = opts->read_names;
   ds_init_empty (&r->ods_errs);
+  ++r->spreadsheet.ref_cnt;
 
-
-  if ( !init_reader (r, true))
+  if (!init_reader (r, true, &r->rsd))
     goto error;
 
-#if 0
-  if ( opts->cell_range )
+  r->used_first_case = false;
+  r->first_case = NULL;
+
+  if (opts->cell_range)
     {
-      if ( ! convert_cell_ref (opts->cell_range,
+      if (! convert_cell_ref (opts->cell_range,
 			       &r->start_col, &r->start_row,
 			       &r->stop_col, &r->stop_row))
 	{
@@ -601,36 +691,20 @@ ods_make_reader (struct spreadsheet *spreadsheet,
       r->stop_col = -1;
       r->stop_row = -1;
     }
-#endif
 
-  r->state = STATE_INIT;
-  r->target_sheet = BAD_CAST opts->sheet_name;
+  r->target_sheet_name = xmlStrdup (BAD_CAST opts->sheet_name);
   r->target_sheet_index = opts->sheet_index;
-  r->row = r->col = 0;
-
-#if 0
-  /* If CELLRANGE was given, then we know how many variables should be read */
-  if ( r->stop_col != -1 )
-    {
-      assert (var_spec == NULL);
-      n_var_specs =  r->stop_col - r->start_col + 1;
-      var_spec = xrealloc (var_spec, sizeof (*var_spec) * n_var_specs);
-      memset (var_spec, '\0', sizeof (*var_spec) * n_var_specs);
-    }
-#endif
 
   /* Advance to the start of the cells for the target sheet */
-  while ( r->current_sheet < r->target_sheet_index - 1 ||
-	  r->state != STATE_TABLE
-	  )
+  while (! reading_target_sheet (r, &r->rsd)
+	  || r->rsd.state != STATE_ROW || r->rsd.row <= r->start_row)
     {
-      if (1 != (ret = xmlTextReaderRead (r->xtr)))
+      if (1 != (ret = xmlTextReaderRead (r->rsd.xtr)))
 	   break;
 
-      process_node (r);
+      process_node (r, &r->rsd);
     }
 
-
   if (ret < 1)
     {
       msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
@@ -638,77 +712,97 @@ ods_make_reader (struct spreadsheet *spreadsheet,
       goto error;
     }
 
-  if ( opts->read_names)
+  if (opts->read_names)
     {
-      while (1 == (ret = xmlTextReaderRead (r->xtr)))
+      while (1 == xmlTextReaderRead (r->rsd.xtr))
 	{
-	  int idx;
-
-	  process_node (r);
+	  process_node (r, &r->rsd);
 
 	  /* If the row is finished then stop for now */
-	  if (r->state == STATE_TABLE && r->row > r->wanted_row_start)
+	  if (r->rsd.state == STATE_TABLE && r->rsd.row > r->start_row)
 	    break;
 
-	  idx = r->col - r->wanted_col_start - 1;
+	  int idx = r->rsd.col - r->start_col - 1;
 
-	  if (r->state == STATE_CELL_CONTENT 
+	  if (idx < 0)
+	    continue;
+
+	  if (r->stop_col != -1 && idx > r->stop_col - r->start_col)
+	    continue;
+
+	  if (r->rsd.state == STATE_CELL_CONTENT
 	      &&
-	      XML_READER_TYPE_TEXT  == r->node_type)
+	      XML_READER_TYPE_TEXT  == r->rsd.node_type)
 	    {
-	      xmlChar *value = xmlTextReaderValue (r->xtr);
-
-	      if ( idx >= n_var_specs)
+	      xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
+	      if (idx >= n_var_specs)
 		{
 		  var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
 
 		  /* xrealloc (unlike realloc) doesn't initialise its memory to 0 */
 		  memset (var_spec + n_var_specs,
-			  0, 
-			  (n_var_specs - idx + 1) * sizeof (*var_spec));
+			  0,
+			  (idx - n_var_specs + 1) * sizeof (*var_spec));
 		  n_var_specs = idx + 1;
 		}
-	      var_spec[idx].firstval.text = 0;
-	      var_spec[idx].firstval.value = 0;
-	      var_spec[idx].firstval.type = 0;
+	      for (int i = 0; i < r->rsd.col_span; ++i)
+		{
+		  var_spec[idx - i].firstval.text = 0;
+		  var_spec[idx - i].firstval.value = 0;
+		  var_spec[idx - i].firstval.type = 0;
+		  var_spec[idx - i].name =
+		    strdup (CHAR_CAST (const char *, value));
+		}
 
-	      var_spec [idx].name = strdup (CHAR_CAST (const char *, value));
 	      xmlFree (value);
 	    }
 	}
     }
 
   /* Read in the first row of data */
-  while (1 == xmlTextReaderRead (r->xtr))
+  while (1 == xmlTextReaderRead (r->rsd.xtr))
     {
       int idx;
-      process_node (r);
+      process_node (r, &r->rsd);
+
+      if (! reading_target_sheet (r, &r->rsd))
+	break;
 
       /* If the row is finished then stop for now */
-      if (r->state == STATE_TABLE && r->row > r->wanted_row_start + (opts->read_names ? 1 : 0))
+      if (r->rsd.state == STATE_TABLE &&
+	  r->rsd.row > r->start_row + (opts->read_names ? 1 : 0))
 	break;
 
-      idx = r->col - r->wanted_col_start - 1;
+      idx = r->rsd.col - r->start_col - 1;
+      if (idx < 0)
+	continue;
+
+      if (r->stop_col != -1 && idx > r->stop_col - r->start_col)
+	continue;
 
-      if ( r->state == STATE_CELL &&
-	   XML_READER_TYPE_ELEMENT  == r->node_type)
+      if (r->rsd.state == STATE_CELL &&
+	   XML_READER_TYPE_ELEMENT  == r->rsd.node_type)
 	{
-	  type = xmlTextReaderGetAttribute (r->xtr, _xml ("office:value-type"));
-	  val_string = xmlTextReaderGetAttribute (r->xtr, _xml ("office:value"));
+	  type = xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("office:value-type"));
+	  val_string = xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("office:value"));
 	}
 
-      if ( r->state == STATE_CELL_CONTENT &&
-	   XML_READER_TYPE_TEXT  == r->node_type)
+      if (r->rsd.state == STATE_CELL_CONTENT &&
+	   XML_READER_TYPE_TEXT  == r->rsd.node_type)
 	{
-	  if ( idx >= n_var_specs) 
+	  if (idx >= n_var_specs)
 	    {
 	      var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
+	      memset (var_spec + n_var_specs,
+		      0,
+		      (idx - n_var_specs + 1) * sizeof (*var_spec));
+
 	      var_spec [idx].name = NULL;
 	      n_var_specs = idx + 1;
 	    }
-	    
+
 	  var_spec [idx].firstval.type = type;
-	  var_spec [idx].firstval.text = xmlTextReaderValue (r->xtr);
+	  var_spec [idx].firstval.text = xmlTextReaderValue (r->rsd.xtr);
 	  var_spec [idx].firstval.value = val_string;
 
 	  val_string = NULL;
@@ -716,11 +810,12 @@ ods_make_reader (struct spreadsheet *spreadsheet,
 	}
     }
 
+
   /* Create the dictionary and populate it */
   r->spreadsheet.dict = r->dict = dict_create (
-    CHAR_CAST (const char *, xmlTextReaderConstEncoding (r->xtr)));
+    CHAR_CAST (const char *, xmlTextReaderConstEncoding (r->rsd.xtr)));
 
-  for (i = 0; i < n_var_specs ; ++i )
+  for (i = 0; i < n_var_specs ; ++i)
     {
       struct fmt_spec fmt;
       struct variable *var = NULL;
@@ -731,7 +826,7 @@ ods_make_reader (struct spreadsheet *spreadsheet,
 
       var = dict_get_var (r->dict, i);
 
-      if ( 0 == xmlStrcmp (var_spec[i].firstval.type, _xml("date")))
+      if (0 == xmlStrcmp (var_spec[i].firstval.type, _xml("date")))
 	{
 	  fmt.type = FMT_DATE;
 	  fmt.d = 0;
@@ -743,16 +838,14 @@ ods_make_reader (struct spreadsheet *spreadsheet,
       var_set_both_formats (var, &fmt);
     }
 
-  /* Create the first case, and cache it */
-  r->used_first_case = false;
-
-  if ( n_var_specs ==  0 )
+  if (n_var_specs ==  0)
     {
       msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
            spreadsheet->file_name);
       goto error;
     }
 
+  /* Create the first case, and cache it */
   r->proto = caseproto_ref (dict_get_proto (r->dict));
   r->first_case = case_create (r->proto);
   case_set_missing (r->first_case);
@@ -761,22 +854,22 @@ ods_make_reader (struct spreadsheet *spreadsheet,
     {
       const struct variable *var = dict_get_var (r->dict, i);
 
-      convert_xml_to_value (r->first_case, var,  &var_spec[i].firstval);
+      convert_xml_to_value (r->first_case, var,  &var_spec[i].firstval,
+			    r->rsd.col - n_var_specs + i,
+			    r->rsd.row - 1);
     }
 
   /* Read in the first row of data */
-  while (1 == xmlTextReaderRead (r->xtr))
+  while (1 == xmlTextReaderRead (r->rsd.xtr))
     {
-      process_node (r);
+      process_node (r, &r->rsd);
 
-      if (r->state == STATE_ROW)
+      if (r->rsd.state == STATE_ROW)
 	break;
     }
 
-  //  zip_reader_destroy (zreader);
 
-#if 0
-  for ( i = 0 ; i < n_var_specs ; ++i )
+  for (i = 0 ; i < n_var_specs ; ++i)
     {
       free (var_spec[i].firstval.type);
       free (var_spec[i].firstval.value);
@@ -785,7 +878,7 @@ ods_make_reader (struct spreadsheet *spreadsheet,
     }
 
   free (var_spec);
-#endif
+
 
   return casereader_create_sequential
     (NULL,
@@ -794,10 +887,8 @@ ods_make_reader (struct spreadsheet *spreadsheet,
      &ods_file_casereader_class, r);
 
  error:
-  
-  // zip_reader_destroy (zreader);
 
-  for ( i = 0 ; i < n_var_specs ; ++i )
+  for (i = 0 ; i < n_var_specs ; ++i)
     {
       free (var_spec[i].firstval.type);
       free (var_spec[i].firstval.value);
@@ -807,11 +898,8 @@ ods_make_reader (struct spreadsheet *spreadsheet,
 
   free (var_spec);
 
-  dict_destroy (r->spreadsheet.dict);
-  r->spreadsheet.dict = NULL;
   ods_file_casereader_destroy (NULL, r);
 
-
   return NULL;
 }
 
@@ -822,76 +910,89 @@ static struct ccase *
 ods_file_casereader_read (struct casereader *reader UNUSED, void *r_)
 {
   struct ccase *c = NULL;
-  xmlChar *val_string = NULL;
   struct ods_reader *r = r_;
-  int current_row = r->row;
 
-  if ( !r->used_first_case )
+  xmlChar *val_string = NULL;
+  xmlChar *type = NULL;
+
+  if (!r->used_first_case)
     {
       r->used_first_case = true;
       return r->first_case;
     }
 
-  if ( r->state > STATE_INIT)
+
+  /* Advance to the start of a row. (If there is one) */
+  while (r->rsd.state != STATE_ROW
+	 && 1 == xmlTextReaderRead (r->rsd.xtr)
+	)
     {
-      c = case_create (r->proto);
-      case_set_missing (c);
+      process_node (r, &r->rsd);
     }
 
-  if (r->state == STATE_SPREADSHEET 
-      && 
-      r->current_sheet == r->target_sheet_index - 1
-      )
+
+  if (! reading_target_sheet (r, &r->rsd)
+       ||  r->rsd.state < STATE_TABLE
+       ||  (r->stop_row != -1 && r->rsd.row > r->stop_row + 1)
+)
     {
       return NULL;
     }
 
-  while (1 == xmlTextReaderRead (r->xtr))
+  c = case_create (r->proto);
+  case_set_missing (c);
+
+  while (1 == xmlTextReaderRead (r->rsd.xtr))
     {
-      process_node (r);
+      process_node (r, &r->rsd);
 
-      if (r->row > current_row && r->state == STATE_ROW)
+      if (r->stop_row != -1 && r->rsd.row > r->stop_row + 1)
 	break;
 
-      if ( r->state == STATE_CELL &&
-	   r->node_type == XML_READER_TYPE_ELEMENT )
+      if (r->rsd.state == STATE_CELL &&
+	   r->rsd.node_type == XML_READER_TYPE_ELEMENT)
 	{
-	  val_string = xmlTextReaderGetAttribute (r->xtr, _xml ("office:value"));
+	  type = xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("office:value-type"));
+	  val_string = xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("office:value"));
 	}
 
-      if ( r->state == STATE_CELL_CONTENT && 
-	   r->node_type == XML_READER_TYPE_TEXT )
+      if (r->rsd.state == STATE_CELL_CONTENT &&
+	   r->rsd.node_type == XML_READER_TYPE_TEXT)
 	{
 	  int col;
 	  struct xml_value *xmv = xzalloc (sizeof *xmv);
-	  xmv->text = xmlTextReaderValue (r->xtr);
+	  xmv->text = xmlTextReaderValue (r->rsd.xtr);
 	  xmv->value = val_string;
 	  val_string = NULL;
+	  xmv->type = type;
+	  type = NULL;
 
-	  for (col = 0; col < r->col_span; ++col)
+	  for (col = 0; col < r->rsd.col_span; ++col)
 	    {
-	      const int idx = r->col + col - r->wanted_col_start - 1;
-	      const struct variable *var = dict_get_var (r->dict, idx);
-	      convert_xml_to_value (c, var, xmv);
+	      const struct variable *var;
+	      const int idx = r->rsd.col - col - r->start_col - 1;
+	      if (idx < 0)
+		continue;
+	      if (r->stop_col != -1 && idx > r->stop_col - r->start_col)
+		break;
+	      if (idx >= dict_get_var_cnt (r->dict))
+		break;
+
+              var = dict_get_var (r->dict, idx);
+	      convert_xml_to_value (c, var, xmv, idx + r->start_col, r->rsd.row - 1);
 	    }
 
-	  free (xmv->text);
-	  free (xmv->value);
+	  xmlFree (xmv->text);
+	  xmlFree (xmv->value);
+	  xmlFree (xmv->type);
 	  free (xmv);
 	}
-
-      if ( r->state < STATE_TABLE)
+      if (r->rsd.state <= STATE_TABLE)
 	break;
     }
 
-  if (NULL == c)
-    {
-      case_unref (c);
-      return NULL;
-    }
-  else
-    {
-      return c;
-    }
+  xmlFree (type);
+  xmlFree (val_string);
+
+  return c;
 }
-#endif