libraries/import/ods.php

   1 <?php
   2 /* vim: set expandtab sw=4 ts=4 sts=4: */
   3 /**
   4  * OpenDocument Spreadsheet import plugin for phpMyAdmin
   5  *
   6  * @todo    Pretty much everything
   7  * @todo    Importing of accented characters seems to fail
   8  * @version 0.5-beta
   9  * @package phpMyAdmin-Import
  10  */
  11
  12 if (! defined('PHPMYADMIN')) {
  13     exit;
  14 }
  15
  16 /**
  17  * The possible scopes for $plugin_param are: 'table', 'database', and 'server'
  18  */
  19
  20 if (isset($plugin_list)) {
  21     $plugin_list['ods'] = array(
  22         'text' => __('Open Document Spreadsheet'),
  23         'extension' => 'ods',
  24         'options' => array(
  25                 array('type' => 'begin_group', 'name' => 'general_opts'),
  26                 array('type' => 'bool', 'name' => 'col_names', 'text' => __('The first line of the file contains the table column names <i>(if this is unchecked, the first line will become part of the data)</i>')),
  27                 array('type' => 'bool', 'name' => 'empty_rows', 'text' => __('Do not import empty rows')),
  28                 array('type' => 'bool', 'name' => 'recognize_percentages', 'text' => __('Import percentages as proper decimals <i>(ex. 12.00% to .12)</i>')),
  29                 array('type' => 'bool', 'name' => 'recognize_currency', 'text' => __('Import currencies <i>(ex. $5.00 to 5.00)</i>')),
  30                 array('type' => 'end_group')
  31             ),
  32         'options_text' => __('Options'),
  33         );
  34     /* We do not define function when plugin is just queried for information above */
  35     return;
  36 }
  37
  38 ini_set('memory_limit', '128M');
  39 set_time_limit(120);
  40
  41 $i = 0;
  42 $len = 0;
  43 $buffer = "";
  44
  45 /**
  46  * Read in the file via PMA_importGetNextChunk so that
  47  * it can process compressed files
  48  */
  49 while (! ($finished && $i >= $len) && ! $error && ! $timeout_passed) {
  50     $data = PMA_importGetNextChunk();
  51     if ($data === FALSE) {
  52         /* subtract data we didn't handle yet and stop processing */
  53         $offset -= strlen($buffer);
  54         break;
  55     } elseif ($data === TRUE) {
  56         /* Handle rest of buffer */
  57     } else {
  58         /* Append new data to buffer */
  59         $buffer .= $data;
  60         unset($data);
  61     }
  62 }
  63
  64 unset($data);
  65
  66 /**
  67  * Load the XML string
  68  *
  69  * The option LIBXML_COMPACT is specified because it can
  70  * result in increased performance without the need to
  71  * alter the code in any way. It's basically a freebee.
  72  */
  73 $xml = simplexml_load_string($buffer, "SimpleXMLElement", LIBXML_COMPACT);
  74
  75 unset($buffer);
  76
  77 $sheets = $xml->children('office', true)->{'body'}->{'spreadsheet'}->children('table', true);
  78
  79 $tables = array();
  80
  81 $max_cols = 0;
  82
  83 $row_count = 0;
  84 $col_count = 0;
  85 $col_names = array();
  86
  87 $tempRow = array();
  88 $tempRows = array();
  89 $rows = array();
  90
  91 /* Iterate over tables */
  92 foreach ($sheets as $sheet) {
  93     $col_names_in_first_row = isset($_REQUEST['ods_col_names']);
  94
  95     /* Iterate over rows */
  96     foreach ($sheet as $row) {
  97         $type = $row->getName();
  98         if (! strcmp('table-row', $type)) {
  99             /* Iterate over columns */
 100             foreach ($row as $cell) {
 101                 $text = $cell->children('text', true);
 102                 $cell_attrs = $cell->attributes('office', true);
 103
 104                 if (count($text) != 0) {
 105                     if (! $col_names_in_first_row) {
 106                         if ($_REQUEST['ods_recognize_percentages'] && !strcmp('percentage', $cell_attrs['value-type'])) {
 107                             $tempRow[] = (double)$cell_attrs['value'];
 108                         } elseif ($_REQUEST['ods_recognize_currency'] && !strcmp('currency', $cell_attrs['value-type'])) {
 109                             $tempRow[] = (double)$cell_attrs['value'];
 110                         } else {
 111                             $tempRow[] = (string)$text;
 112                         }
 113                     } else {
 114                         if ($_REQUEST['ods_recognize_percentages'] && !strcmp('percentage', $cell_attrs['value-type'])) {
 115                             $col_names[] = (double)$cell_attrs['value'];
 116                         } else if ($_REQUEST['ods_recognize_currency'] && !strcmp('currency', $cell_attrs['value-type'])) {
 117                             $col_names[] = (double)$cell_attrs['value'];
 118                         } else {
 119                             $col_names[] = (string)$text;
 120                         }
 121                     }
 122
 123                     ++$col_count;
 124                 } else {
 125                     /* Number of blank columns repeated */
 126                     if ($col_count < count($row->children('table', true)) - 1) {
 127                         $attr = $cell->attributes('table', true);
 128                         $num_null = (int)$attr['number-columns-repeated'];
 129
 130                         if ($num_null) {
 131                             if (! $col_names_in_first_row) {
 132                                 for ($i = 0; $i < $num_null; ++$i) {
 133                                     $tempRow[] = 'NULL';
 134                                     ++$col_count;
 135                                 }
 136                             } else {
 137                                 for ($i = 0; $i < $num_null; ++$i) {
 138                                     $col_names[] = PMA_getColumnAlphaName($col_count + 1);
 139                                     ++$col_count;
 140                                 }
 141                             }
 142                         } else {
 143                             if (! $col_names_in_first_row) {
 144                                 $tempRow[] = 'NULL';
 145                             } else {
 146                                 $col_names[] = PMA_getColumnAlphaName($col_count + 1);
 147                             }
 148
 149                             ++$col_count;
 150                         }
 151                     }
 152                 }
 153             }
 154
 155             /* Find the widest row */
 156             if ($col_count > $max_cols) {
 157                 $max_cols = $col_count;
 158             }
 159
 160             /* Don't include a row that is full of NULL values */
 161             if (! $col_names_in_first_row) {
 162                 if ($_REQUEST['ods_empty_rows']) {
 163                     foreach ($tempRow as $cell) {
 164                         if (strcmp('NULL', $cell)) {
 165                             $tempRows[] = $tempRow;
 166                             break;
 167                         }
 168                     }
 169                 } else {
 170                     $tempRows[] = $tempRow;
 171                 }
 172             }
 173
 174             $col_count = 0;
 175             $col_names_in_first_row = false;
 176             $tempRow = array();
 177         }
 178     }
 179
 180     /* Skip over empty sheets */
 181     if (count($tempRows) == 0 || count($tempRows[0]) == 0) {
 182         $col_names = array();
 183         $tempRow = array();
 184         $tempRows = array();
 185         continue;
 186     }
 187
 188     /**
 189      * Fill out each row as necessary to make
 190      * every one exactly as wide as the widest
 191      * row. This included column names.
 192      */
 193
 194     /* Fill out column names */
 195     for ($i = count($col_names); $i < $max_cols; ++$i) {
 196         $col_names[] = PMA_getColumnAlphaName($i + 1);
 197     }
 198
 199     /* Fill out all rows */
 200     $num_rows = count($tempRows);
 201     for ($i = 0; $i < $num_rows; ++$i) {
 202         for ($j = count($tempRows[$i]); $j < $max_cols; ++$j) {
 203             $tempRows[$i][] = 'NULL';
 204         }
 205     }
 206
 207     /* Store the table name so we know where to place the row set */
 208     $tbl_attr = $sheet->attributes('table', true);
 209     $tables[] = array((string)$tbl_attr['name']);
 210
 211     /* Store the current sheet in the accumulator */
 212     $rows[] = array((string)$tbl_attr['name'], $col_names, $tempRows);
 213     $tempRows = array();
 214     $col_names = array();
 215     $max_cols = 0;
 216 }
 217
 218 unset($tempRow);
 219 unset($tempRows);
 220 unset($col_names);
 221 unset($sheets);
 222 unset($xml);
 223
 224 /**
 225  * Bring accumulated rows into the corresponding table
 226  */
 227 $num_tbls = count($tables);
 228 for ($i = 0; $i < $num_tbls; ++$i) {
 229     for ($j = 0; $j < count($rows); ++$j) {
 230         if (! strcmp($tables[$i][TBL_NAME], $rows[$j][TBL_NAME])) {
 231             if (! isset($tables[$i][COL_NAMES])) {
 232                 $tables[$i][] = $rows[$j][COL_NAMES];
 233             }
 234
 235             $tables[$i][ROWS] = $rows[$j][ROWS];
 236         }
 237     }
 238 }
 239
 240 /* No longer needed */
 241 unset($rows);
 242
 243 /* Obtain the best-fit MySQL types for each column */
 244 $analyses = array();
 245
 246 $len = count($tables);
 247 for ($i = 0; $i < $len; ++$i) {
 248     $analyses[] = PMA_analyzeTable($tables[$i]);
 249 }
 250
 251 /**
 252  * string $db_name (no backquotes)
 253  *
 254  * array $table = array(table_name, array() column_names, array()() rows)
 255  * array $tables = array of "$table"s
 256  *
 257  * array $analysis = array(array() column_types, array() column_sizes)
 258  * array $analyses = array of "$analysis"s
 259  *
 260  * array $create = array of SQL strings
 261  *
 262  * array $options = an associative array of options
 263  */
 264
 265 /* Set database name to the currently selected one, if applicable */
 266 if (strlen($db)) {
 267     $db_name = $db;
 268     $options = array('create_db' => false);
 269 } else {
 270     $db_name = 'ODS_DB';
 271     $options = NULL;
 272 }
 273
 274 /* Non-applicable parameters */
 275 $create = NULL;
 276
 277 /* Created and execute necessary SQL statements from data */
 278 PMA_buildSQL($db_name, $tables, $analyses, $create, $options);
 279
 280 unset($tables);
 281 unset($analyses);
 282
 283 /* Commit any possible data in buffers */
 284 PMA_importRunQuery();
 285 ?>