Add parser support for XLSX format (untested), partial fix for issue #28

faisalman · Nov 3, 2014 · 9ecd4f1 · 9ecd4f1
1 parent a850802
commit 9ecd4f1
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 18 deletions.
diff --git a/src/SimpleExcel/Parser/XLSXParser.php b/src/SimpleExcel/Parser/XLSXParser.php
@@ -35,28 +35,53 @@ class XLSXParser extends BaseParser
     */
     public function loadFile ($file_path, $options = NULL) {
         if ($this->checkFile($file_path)) {
-            // TODO: extract zip & map string to worksheet
-            $zip = zip_open($file_path);    
+
+            // read uncompressed xlsx contents
+            $zip = zip_open($file_path);
+            $xml_worksheets = array();
+            $xml_sharedstrings = array(); 
             while($zip_entry = zip_read($zip))
             {
-                if(preg_match("/xl\/worksheets\/sheet\d/", zip_entry_name($zip_entry)))
+                if(preg_match("/xl\/worksheets\/sheet\d+\.xml/", zip_entry_name($zip_entry)))
                 {
-                    print_r(zip_entry_read($zip_entry));
+                    if(zip_entry_open($zip, $zip_entry))
+                    {
+                        $xml_length = zip_entry_filesize($zip_entry);
+                        $xml_read = zip_entry_read($zip_entry, $xml_length);
+                        $xml_simplexml = simplexml_load_string($xml_read);
+                        $xml_array = (array)$xml_simplexml->children();
+                        array_push($xml_worksheets, json_decode(json_encode((array)$xml_array['sheetData']), 1));
+                    }
                 }
-            }    
+                if(preg_match("/xl\/sharedStrings\.xml/", zip_entry_name($zip_entry)))
+                {
+                    if(zip_entry_open($zip, $zip_entry))
+                    {
+                        $xml_length = zip_entry_filesize($zip_entry);
+                        $xml_read = zip_entry_read($zip_entry, $xml_length);
+                        $xml_simplexml = simplexml_load_string($xml_read);
+                        $xml_array = (array)$xml_simplexml->children();
+                        $xml_sharedstrings = json_decode(json_encode((array)$xml_array['si']), 1);
+                    }
+                }
+            }
             zip_close($zip);
+
+            // map sheets <-> sharedstrings into simpleexcel workbook
+            $this->Workbook = new Workbook();
+            foreach ($xml_worksheets as $worksheet) {
+                $sheet = new Worksheet();
+                foreach ($worksheet['row'] as $row) {
+                    $record = array();
+                    foreach ($row['c'] as $cell) {
+                        if (array_key_exists('v', $cell) && array_key_exists($cell['v'], $xml_sharedstrings) && array_key_exists('t', $xml_sharedstrings[$cell['v']])) {
+                            array_push($record, new Cell($xml_sharedstrings[$cell['v']]['t']));
+                        }
+                    }
+                    $sheet->insertRecord($record);
+                }
+                $this->Workbook->insertWorksheet($sheet);
+            }
         }
     }
-
-    /**
-    * Load the string to be parsed
-    * 
-    * @param    string  $str        String with XLSX format
-    * @param    array   $options    Options
-    */
-    public function loadString($str, $options) {
-        $this->workbook = new Workbook();
-        $xml = simplexml_load_string($str);
-        // TODO: implement xlsx parser
-    }
 }
diff --git a/src/SimpleExcel/SimpleExcel.php b/src/SimpleExcel/SimpleExcel.php
@@ -96,7 +96,7 @@ class SimpleExcel
     */
     public function __construct ($filetype = NULL) {
         $this->workbook = new Workbook();
-        $this->validParserTypes = array('XML', 'CSV', 'TSV', 'HTML', 'JSON');
+        $this->validParserTypes = array('XML', 'CSV', 'TSV', 'HTML', 'JSON', 'XLSX');
         $this->validWriterTypes = array('XML', 'CSV', 'TSV', 'HTML', 'JSON');
         if (isset($filetype)) {
             $this->setParserType($filetype);