pintos-os.org Git - pspp/blob - tests/language/data-io/get-data-spreadsheet.at

   1 dnl PSPP - a program for statistical analysis.
   2 dnl Copyright (C) 2017 Free Software Foundation, Inc.
   3 dnl
   4 dnl This program is free software: you can redistribute it and/or modify
   5 dnl it under the terms of the GNU General Public License as published by
   6 dnl the Free Software Foundation, either version 3 of the License, or
   7 dnl (at your option) any later version.
   8 dnl
   9 dnl This program is distributed in the hope that it will be useful,
  10 dnl but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 dnl GNU General Public License for more details.
  13 dnl
  14 dnl You should have received a copy of the GNU General Public License
  15 dnl along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16 dnl
  17 m4_define([SPREADSHEET_TEST_PREP],[dnl
  18  m4_if($1,[GNM],[dnl
  19     AT_CHECK([gzip -c $top_srcdir/tests/language/data-io/Book1.gnm.unzipped > Book1.gnumeric])dnl
  20     m4_define([testsheet],[Book1.gnumeric])dnl
  21     ]) dnl
  22  m4_if($1,[ODS],[dnl
  23     AT_CHECK([cp $top_srcdir/tests/language/data-io/test.ods test.ods])dnl
  24     m4_define([testsheet],[test.ods])dnl
  25     ])dnl
  26 ])
  27
  28 m4_define([CHECK_SPREADSHEET_READER],
  29  [dnl
  30 AT_SETUP([GET DATA /TYPE=$1 with CELLRANGE])
  31 SPREADSHEET_TEST_PREP($1)
  32 AT_DATA([get-data.sps], [dnl
  33 GET DATA /TYPE=$1 /FILE='testsheet'  /READNAMES=off /SHEET=name 'This' /CELLRANGE=range 'g9:i13' .
  34 DISPLAY VARIABLES.
  35 LIST.
  36 ])
  37 AT_CHECK([pspp -o pspp.csv get-data.sps])
  38 AT_CHECK([cat pspp.csv], [0], [dnl
  39 Variable,Description,Position
  40 VAR001,Format: F8.2,1
  41 VAR002,Format: A8,2
  42 VAR003,Format: F8.2,3
  43
  44 Table: Data List
  45 VAR001,VAR002,VAR003
  46 .00,fred    ,20.00
  47 1.00,11      ,21.00
  48 2.00,twelve  ,22.00
  49 3.00,13      ,23.00
  50 4.00,14      ,24.00
  51 ])
  52 AT_CLEANUP
  53
  54 AT_SETUP([GET DATA /TYPE=$1 with CELLRANGE and READNAMES])
  55 SPREADSHEET_TEST_PREP($1)
  56 AT_DATA([get-data.sps], [dnl
  57 GET DATA /TYPE=$1 /FILE='testsheet'  /READNAMES=on /SHEET=name 'This' /CELLRANGE=range 'g8:i13' .
  58 DISPLAY VARIABLES.
  59 LIST.
  60 ])
  61 AT_CHECK([pspp -o pspp.csv get-data.sps])
  62 AT_CHECK([cat pspp.csv], [0], [dnl
  63 Variable,Description,Position
  64 V1,Format: F8.2,1
  65 V2,Format: A8,2
  66 VAR001,Format: F8.2,3
  67
  68 Table: Data List
  69 V1,V2,VAR001
  70 .00,fred    ,20.00
  71 1.00,11      ,21.00
  72 2.00,twelve  ,22.00
  73 3.00,13      ,23.00
  74 4.00,14      ,24.00
  75 ])
  76 AT_CLEANUP
  77
  78 AT_SETUP([GET DATA /TYPE=$1 without CELLRANGE])
  79 SPREADSHEET_TEST_PREP($1)
  80 AT_DATA([get-data.sps], [dnl
  81 GET DATA /TYPE=$1 /FILE='testsheet' /SHEET=index 3.
  82 DISPLAY VARIABLES.
  83 LIST.
  84 ])
  85 AT_CHECK([pspp -O format=csv get-data.sps], [0], [dnl
  86 Variable,Description,Position
  87 name,Format: A8,1
  88 id,Format: F8.2,2
  89 height,Format: F8.2,3
  90
  91 warning: Cannot convert the value in the spreadsheet cell C4 to format (F8.2): Field contents are not numeric.
  92
  93 Table: Data List
  94 name,id,height
  95 fred    ,.00,23.40
  96 bert    ,1.00,.56
  97 charlie ,2.00,.  @&t@
  98 dick    ,3.00,-34.09
  99 ])
 100 AT_CLEANUP
 101
 102 AT_SETUP([GET DATA /TYPE=$1 with missing data])
 103 SPREADSHEET_TEST_PREP($1)
 104 AT_DATA([get-data.sps], [dnl
 105 * This sheet has no data in one of its variables
 106 GET DATA /TYPE=$1 /FILE='testsheet' /READNAMES=on /SHEET=index 5.
 107 DISPLAY VARIABLES.
 108 LIST.
 109 ])
 110 AT_CHECK([pspp -o pspp.csv get-data.sps])
 111 AT_CHECK([cat pspp.csv], [0], [dnl
 112 Variable,Description,Position
 113 vone,Format: F8.2,1
 114 vtwo,Format: F8.2,2
 115 vthree,Format: A8,3
 116 v4,Format: F8.2,4
 117
 118 Table: Data List
 119 vone,vtwo,vthree,v4
 120 1.00,3.00,,5.00
 121 2.00,4.00,,6.00
 122 ])
 123 AT_CLEANUP
 124
 125 dnl This syntax doesn't do anything particularly useful.
 126 dnl It has been seen to cause a few crashes, so we check here that it
 127 dnl doesn't do anthing bad.
 128 AT_SETUP([GET DATA /TYPE=$1 with no options])
 129 SPREADSHEET_TEST_PREP($1)
 130 AT_DATA([get-data.sps], [dnl
 131 * This sheet is empty
 132 GET DATA /TYPE=$1 /FILE='testsheet'.
 133 DISPLAY DICTIONARY.
 134 LIST.
 135 ])
 136 AT_CHECK([pspp -o pspp.csv get-data.sps], [0], [ignore])
 137 AT_CLEANUP
 138
 139
 140
 141 AT_SETUP([GET DATA /TYPE=$1 with empty sheet])
 142 SPREADSHEET_TEST_PREP($1)
 143 AT_DATA([get-data.sps], [dnl
 144 * This sheet is empty
 145 GET DATA /TYPE=$1 /FILE='testsheet' /SHEET=name 'Empty'.
 146 ])
 147 AT_CHECK([pspp -o pspp.csv get-data.sps], [0], [dnl
 148 warning: Selected sheet or range of spreadsheet `testsheet' is empty.
 149 ])
 150 AT_CLEANUP
 151
 152 AT_SETUP([GET DATA /TYPE=$1 with nonexistent sheet])
 153 SPREADSHEET_TEST_PREP($1)
 154 AT_DATA([get-data.sps], [dnl
 155 * This sheet doesnt exist.
 156 GET DATA /TYPE=$1 /FILE='testsheet' /SHEET=name 'foobarxx'.
 157 ])
 158 AT_CHECK([pspp -o pspp.csv get-data.sps], [0], [dnl
 159 warning: Selected sheet or range of spreadsheet `testsheet' is empty.
 160 ])
 161 AT_CLEANUP
 162 ])
 163
 164
 165 AT_BANNER([GET DATA Spreadsheet /TYPE=GNM])
 166
 167 CHECK_SPREADSHEET_READER([GNM])
 168
 169 dnl Check for a bug where gnumeric files were interpreted incorrectly
 170 AT_SETUP([GET DATA /TYPE=GNM sheet index bug])
 171 AT_DATA([minimal3.gnumeric],[dnl
 172 <?xml version="1.0" encoding="UTF-8"?>
 173 <gnm:Workbook xmlns:gnm="http://www.gnumeric.org/v10.dtd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.gnumeric.org/v9.xsd">
 174   <gnm:Version Epoch="1" Major="10" Minor="8" Full="1.10.8"/>
 175   <gnm:SheetNameIndex>
 176     <gnm:SheetName gnm:Cols="256" gnm:Rows="65536">Sheet1</gnm:SheetName>
 177     <gnm:SheetName gnm:Cols="256" gnm:Rows="65536">Sheet2</gnm:SheetName>
 178     <gnm:SheetName gnm:Cols="256" gnm:Rows="65536">Sheet3</gnm:SheetName>
 179   </gnm:SheetNameIndex>
 180   <gnm:Sheets>
 181     <gnm:Sheet DisplayFormulas="0" HideZero="0" HideGrid="0" HideColHeader="0" HideRowHeader="0" DisplayOutlines="1" OutlineSymbolsBelow="1" OutlineSymbolsRight="1" Visibility="GNM_SHEET_VISIBILITY_VISIBLE" GridColor="0:0:0">
 182       <gnm:Name>Sheet1</gnm:Name>
 183       <gnm:MaxCol>2</gnm:MaxCol>
 184       <gnm:MaxRow>3</gnm:MaxRow>
 185       <gnm:Names>
 186         <gnm:Name>
 187           <gnm:name>Print_Area</gnm:name>
 188           <gnm:value>#REF!</gnm:value>
 189           <gnm:position>A1</gnm:position>
 190         </gnm:Name>
 191         <gnm:Name>
 192           <gnm:name>Sheet_Title</gnm:name>
 193           <gnm:value>&quot;Sheet1&quot;</gnm:value>
 194           <gnm:position>A1</gnm:position>
 195         </gnm:Name>
 196       </gnm:Names>
 197       <gnm:Cols DefaultSizePts="48">
 198         <gnm:ColInfo No="0" Unit="94.5" HardSize="1"/>
 199         <gnm:ColInfo No="1" Unit="48" Count="2"/>
 200       </gnm:Cols>
 201       <gnm:Rows DefaultSizePts="12.75">
 202         <gnm:RowInfo No="0" Unit="13.5" Count="4"/>
 203       </gnm:Rows>
 204       <gnm:Cells>
 205         <gnm:Cell Row="0" Col="0" ValueType="60">Name</gnm:Cell>
 206         <gnm:Cell Row="0" Col="1" ValueType="60">x</gnm:Cell>
 207         <gnm:Cell Row="0" Col="2" ValueType="60">y</gnm:Cell>
 208         <gnm:Cell Row="1" Col="0" ValueType="60">Sheet One</gnm:Cell>
 209         <gnm:Cell Row="1" Col="1" ValueType="40">1</gnm:Cell>
 210         <gnm:Cell Row="1" Col="2" ValueType="40">2</gnm:Cell>
 211         <gnm:Cell Row="2" Col="0" ValueType="60">foo</gnm:Cell>
 212         <gnm:Cell Row="2" Col="1" ValueType="40">3</gnm:Cell>
 213         <gnm:Cell Row="2" Col="2" ValueType="40">4</gnm:Cell>
 214         <gnm:Cell Row="3" Col="0" ValueType="60">bar</gnm:Cell>
 215         <gnm:Cell Row="3" Col="1" ValueType="40">5</gnm:Cell>
 216         <gnm:Cell Row="3" Col="2" ValueType="40">6</gnm:Cell>
 217       </gnm:Cells>
 218     </gnm:Sheet>
 219     <gnm:Sheet DisplayFormulas="0" HideZero="0" HideGrid="0" HideColHeader="0" HideRowHeader="0" DisplayOutlines="1" OutlineSymbolsBelow="1" OutlineSymbolsRight="1" Visibility="GNM_SHEET_VISIBILITY_VISIBLE" GridColor="0:0:0">
 220       <gnm:Name>Sheet2</gnm:Name>
 221       <gnm:MaxCol>2</gnm:MaxCol>
 222       <gnm:MaxRow>2</gnm:MaxRow>
 223       <gnm:Names>
 224         <gnm:Name>
 225           <gnm:name>Print_Area</gnm:name>
 226           <gnm:value>#REF!</gnm:value>
 227           <gnm:position>A1</gnm:position>
 228         </gnm:Name>
 229         <gnm:Name>
 230           <gnm:name>Sheet_Title</gnm:name>
 231           <gnm:value>&quot;Sheet2&quot;</gnm:value>
 232           <gnm:position>A1</gnm:position>
 233         </gnm:Name>
 234       </gnm:Names>
 235       <gnm:Cols DefaultSizePts="48">
 236         <gnm:ColInfo No="0" Unit="48"/>
 237         <gnm:ColInfo No="1" Unit="57.75"/>
 238         <gnm:ColInfo No="2" Unit="54.75"/>
 239       </gnm:Cols>
 240       <gnm:Rows DefaultSizePts="12.75">
 241         <gnm:RowInfo No="0" Unit="13.5" Count="3"/>
 242       </gnm:Rows>
 243       <gnm:Cells>
 244         <gnm:Cell Row="0" Col="0" ValueType="60">Comment</gnm:Cell>
 245         <gnm:Cell Row="0" Col="1" ValueType="60">DOB</gnm:Cell>
 246         <gnm:Cell Row="0" Col="2" ValueType="60">wealth</gnm:Cell>
 247         <gnm:Cell Row="1" Col="0" ValueType="60">Sheet Two</gnm:Cell>
 248         <gnm:Cell Row="1" Col="1" ValueType="60">24/5/1966</gnm:Cell>
 249         <gnm:Cell Row="1" Col="2" ValueType="40" ValueFormat="_($* 0.00_);_($* (0.00);_($* &quot;-&quot;??_);_(@_)">0.02</gnm:Cell>
 250         <gnm:Cell Row="2" Col="0" ValueType="60">wee</gnm:Cell>
 251         <gnm:Cell Row="2" Col="1" ValueType="40" ValueFormat="dd/mm/yyyy">37145</gnm:Cell>
 252         <gnm:Cell Row="2" Col="2" ValueType="40" ValueFormat="_($* 0.00_);_($* (0.00);_($* &quot;-&quot;??_);_(@_)">3000</gnm:Cell>
 253       </gnm:Cells>
 254     </gnm:Sheet>
 255     <gnm:Sheet DisplayFormulas="0" HideZero="0" HideGrid="0" HideColHeader="0" HideRowHeader="0" DisplayOutlines="1" OutlineSymbolsBelow="1" OutlineSymbolsRight="1" Visibility="GNM_SHEET_VISIBILITY_VISIBLE" GridColor="0:0:0">
 256       <gnm:Name>Sheet3</gnm:Name>
 257       <gnm:MaxCol>2</gnm:MaxCol>
 258       <gnm:MaxRow>2</gnm:MaxRow>
 259       <gnm:Names>
 260         <gnm:Name>
 261           <gnm:name>Print_Area</gnm:name>
 262           <gnm:value>#REF!</gnm:value>
 263           <gnm:position>A1</gnm:position>
 264         </gnm:Name>
 265         <gnm:Name>
 266           <gnm:name>Sheet_Title</gnm:name>
 267           <gnm:value>&quot;Sheet3&quot;</gnm:value>
 268           <gnm:position>A1</gnm:position>
 269         </gnm:Name>
 270       </gnm:Names>
 271       <gnm:Cols DefaultSizePts="48">
 272         <gnm:ColInfo No="0" Unit="48" Count="3"/>
 273       </gnm:Cols>
 274       <gnm:Rows DefaultSizePts="12.75">
 275         <gnm:RowInfo No="0" Unit="13.5"/>
 276         <gnm:RowInfo No="1" Unit="12.75" Count="2"/>
 277       </gnm:Rows>
 278       <gnm:Cells>
 279         <gnm:Cell Row="0" Col="0" ValueType="40">3</gnm:Cell>
 280         <gnm:Cell Row="0" Col="1" ValueType="40">4</gnm:Cell>
 281         <gnm:Cell Row="0" Col="2" ValueType="40">5</gnm:Cell>
 282         <gnm:Cell Row="1" Col="0" ValueType="40">6</gnm:Cell>
 283         <gnm:Cell Row="1" Col="1" ValueType="40">7</gnm:Cell>
 284         <gnm:Cell Row="1" Col="2" ValueType="40">8</gnm:Cell>
 285         <gnm:Cell Row="2" Col="0" ValueType="40">9</gnm:Cell>
 286         <gnm:Cell Row="2" Col="1" ValueType="40">10</gnm:Cell>
 287         <gnm:Cell Row="2" Col="2" ValueType="40">11</gnm:Cell>
 288       </gnm:Cells>
 289     </gnm:Sheet>
 290   </gnm:Sheets>
 291 </gnm:Workbook>
 292 ])
 293
 294 AT_DATA([gnum.sps], [dnl
 295 GET DATA
 296         /TYPE=GNM
 297         /FILE='minimal3.gnumeric'
 298         /SHEET=index 3
 299         /READNAMES=off
 300         .
 301
 302 LIST.
 303 ])
 304
 305 AT_CHECK([pspp -O format=csv gnum.sps], [0], [dnl
 306 Table: Data List
 307 VAR001,VAR002,VAR003
 308 3       ,4.00,5.00
 309 6       ,7.00,8.00
 310 9       ,10.00,11.00
 311 ])
 312
 313
 314 AT_CLEANUP
 315
 316
 317 dnl Check for a bug where certain gnumeric files failed an assertion
 318 AT_SETUP([GET DATA /TYPE=GNM assert-fail])
 319 AT_DATA([read.sps],[dnl
 320 GET DATA
 321         /TYPE=GNM
 322         /FILE='crash.gnumeric'
 323         .
 324 list.
 325 ])
 326
 327
 328 AT_DATA([crash.gnumeric],[dnl
 329 <?xml version="1.0" encoding="UTF-8"?>
 330 <gnm:Workbook xmlns:gnm="http://www.gnumeric.org/v10.dtd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.gnumeric.org/v9.xsd">
 331   <office:document-meta xmlns:office="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:meta="urn:oasis:names:tc:opendocument:xmlns:meta:1.0" xmlns:ooo="http://openoffice.org/2004/office" office:version="1.1">
 332   </office:document-meta>
 333   <gnm:SheetNameIndex>
 334     <gnm:SheetName gnm:Cols="256" gnm:Rows="65536">Sheet1</gnm:SheetName>
 335   </gnm:SheetNameIndex>
 336   <gnm:Sheets>
 337     <gnm:Sheet DisplayFormulas="0" HideZero="0" HideGrid="0" HideColHeader="0" HideRowHeader="0" DisplayOutlines="1" OutlineSymbolsBelow="1" OutlineSymbolsRight="1" Visibility="GNM_SHEET_VISIBILITY_VISIBLE" GridColor="0:0:0">
 338       <gnm:Name>Sheet1</gnm:Name>
 339       <gnm:MaxCol>2</gnm:MaxCol>
 340       <gnm:MaxRow>4</gnm:MaxRow>
 341       <gnm:Styles>
 342         <gnm:StyleRegion startCol="0" startRow="0" endCol="255" endRow="65535">
 343           <gnm:Style HAlign="1" VAlign="2" WrapText="0" ShrinkToFit="0" Rotation="0" Shade="0" Indent="0" Locked="1" Hidden="0" Fore="0:0:0" Back="FFFF:FFFF:FFFF" PatternColor="0:0:0" Format="General">
 344           </gnm:Style>
 345         </gnm:StyleRegion>
 346       </gnm:Styles>
 347       <gnm:Cells>
 348         <gnm:Cell Row="1" Col="1" ValueType="60">one</gnm:Cell>
 349         <gnm:Cell Row="1" Col="2" ValueType="60">two</gnm:Cell>
 350         <gnm:Cell Row="2" Col="1" ValueType="40">1</gnm:Cell>
 351         <gnm:Cell Row="2" Col="2" ValueType="40">2</gnm:Cell>
 352         <gnm:Cell Row="3" Col="1" ValueType="40">1</gnm:Cell>
 353         <gnm:Cell Row="3" Col="2" ValueType="40">2</gnm:Cell>
 354         <gnm:Cell Row="4" Col="1" ValueType="40">1</gnm:Cell>
 355         <gnm:Cell Row="4" Col="2" ValueType="40">2</gnm:Cell>
 356       </gnm:Cells>
 357     </gnm:Sheet>
 358   </gnm:Sheets>
 359 </gnm:Workbook>
 360 ])
 361
 362 AT_CHECK([pspp -O format=csv read.sps], [0], [ignore])
 363
 364
 365 AT_CLEANUP
 366
 367
 368
 369 AT_BANNER([GET DATA Spreadsheet /TYPE=ODS])
 370
 371 CHECK_SPREADSHEET_READER([ODS])
 372
 373
 374 AT_SETUP([GET DATA /TYPE=ODS crash])
 375
 376
 377 AT_CHECK([cp $top_srcdir/tests/language/data-io/newone.ods this.ods])dnl
 378
 379 AT_DATA([crash.sps],[dnl
 380 GET DATA /TYPE=ODS /FILE='this.ods' /CELLRANGE=RANGE 'A1:C8'  /READNAMES=ON
 381 LIST.
 382 ])
 383
 384 AT_CHECK([pspp -O format=csv crash.sps], [0], [ignore])
 385
 386 AT_CLEANUP
 387
 388
 389 AT_SETUP([GET DATA /TYPE=ODS readnames])
 390
 391 dnl Check for a bug where in the ODS reader /READNAMES incorrectly
 392 dnl dealt with repeated names.
 393 AT_CHECK([cp $top_srcdir/tests/language/data-io/readnames.ods this.ods])dnl
 394
 395 AT_DATA([readnames.sps],[dnl
 396 GET DATA /TYPE=ODS /FILE='this.ods' /CELLRANGE=RANGE 'A1:H8' /READNAMES=ON
 397 DISPLAY DICTIONARY.
 398 LIST.
 399 ])
 400
 401
 402 AT_CHECK([pspp -O format=csv readnames.sps], [0],
 403 [Variable,Description,Position
 404 freda,Format: F8.2,1
 405 fred,Format: F8.2,2
 406 fred_A,Format: F8.2,3
 407 fred_B,Format: F8.2,4
 408 fred_C,Format: F8.2,5
 409 fred_D,Format: F8.2,6
 410 fred_E,Format: F8.2,7
 411
 412 Table: Data List
 413 freda,fred,fred_A,fred_B,fred_C,fred_D,fred_E
 414 1.00,2.00,3.00,4.00,5.00,6.00,7.00
 415 8.00,9.00,10.00,11.00,12.00,13.00,14.00
 416 ])
 417
 418 AT_CLEANUP
 419