Merge pull request #112 from liberu-genealogy/sweep/_3b4fc

Sweep: (✓ Sandbox Passed)
liberu-genealogy · Mar 12, 2024 · 78a634f · 78a634f
2 parents b437c4a + e5ebcc9
commit 78a634f
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 13 deletions.
diff --git a/src/Snps/IO/CsvReader.php b/src/Snps/IO/CsvReader.php
@@ -35,6 +35,12 @@ public function setColumnNames($columnNames)
     }
 
     public function setColumnTypes($columnTypes)
+    private $enclosure;
+
+    public function setEnclosure($enclosure)
+    {
+        $this->enclosure = $enclosure;
+    }
     {
         $this->columnTypes = $columnTypes;
     }
@@ -45,10 +51,10 @@ public function read()
 
         if (($handle = fopen($this->filePath, "r")) !== false) {
             if ($this->header) {
-                fgetcsv($handle, 0, $this->separator); // Skip the header row
+                fgetcsv($handle, 0, $this->separator, $this->enclosure); // Skip the header row
             }
 
-            while (($row = fgetcsv($handle, 0, $this->separator)) !== false) {
+            while (($row = fgetcsv($handle, 0, $this->separator, $this->enclosure)) !== false) {
                 if (!empty($this->columnNames)) {
                     print_r($row);
                     $row = array_combine($this->columnNames, $row);

diff --git a/src/Snps/IO/ExtraTabsFilter.php b/src/Snps/IO/ExtraTabsFilter.php
@@ -17,3 +17,4 @@ public function filter($in, $out, int &$consumed, bool $closing): int
         return PSFS_PASS_ON;
     }
 }
+stream_filter_register("extra_tabs_filter", "Dna\\Snps\\IO\\ExtraTabsFilter");
diff --git a/src/Snps/IO/Reader.php b/src/Snps/IO/Reader.php
@@ -37,29 +37,27 @@ public function __construct(
         private ?SNPsResources $resources,
         private array $rsids
     ) {}
+    }
 
     /**
      * Read and parse a raw data / genotype file.
      *
-     * @return array
-     *   An array with the following items:
-     *   - snps (pandas.DataFrame): Dataframe of parsed SNPs.
-     *   - source (string): Detected source of SNPs.
-     *   - phased (bool): Flag indicating if SNPs are phased.
+     * @return array An array with the following items:
+     *               - 'snps': Array of parsed SNPs.
+     *               - 'source': Detected source of SNPs.
+     *               - 'phased': Flag indicating if SNPs are phased.
+     *               - 'build': Detected build of SNPs.
      */
     public function read(): array
     {
         $file = $this->file;
-        $compression = "infer";
         $read_data = [];
-        $d = array(
-            "snps" => IO::get_empty_snps_dataframe(),
+        $d = [
+            "snps" => [],
             "source" => "",
             "phased" => false,
             "build" => 0,
-        );
-        // Peek into files to determine the data format
-        // var_dump($file);
+        ];
         if (is_string($file) && file_exists($file)) {
             if (strpos($file, ".zip") !== false) {
                 $zip = new ZipArchive(ZipArchive::RDONLY);