Merge pull request #131 from TripalCultivate/g4.115-delimitedCaseStrings

Add cases to valid file delimiter validator
TripalCultivate · Jan 8, 2025 · 3d934ca · 3d934ca
2 parents b762537 + 970e3a0
commit 3d934ca
Show file tree

Hide file tree

Showing 3 changed files with 59 additions and 34 deletions.
diff --git a/...ultivate_phenotypes/src/Plugin/TripalImporter/TripalCultivatePhenotypesTraitsImporter.php b/...ultivate_phenotypes/src/Plugin/TripalImporter/TripalCultivatePhenotypesTraitsImporter.php
@@ -1086,7 +1086,8 @@ public function processValidDelimitedFileFailures(array $failures) {
           ($validation_result['case'] == 'None of the delimiters supported by the file type was used')) {
         $table_case = 'unsupported';
       }
-      elseif ($validation_result['case'] == 'Raw row is not delimited') {
+      elseif (($validation_result['case'] == 'Raw row exceeds number of strict columns') ||
+            ($validation_result['case'] == 'Raw row has insufficient number of columns')) {
         $table_case = 'delimited';
       }
 

diff --git a/trpcultivate_phenotypes/src/Plugin/Validators/ValidDelimitedFile.php b/trpcultivate_phenotypes/src/Plugin/Validators/ValidDelimitedFile.php
@@ -97,36 +97,33 @@ public function validateRawRow(string $raw_row) {
       ];
     }
 
-    // With the list of delimiters identified in the raw row, try each delimiter
-    // separately to see if number of values is the expected number of columns.
-    // Store every delimiter that failed into the failed delimiters array.
-    $delimiters_failed = [];
-
-    foreach ($delimiters_used as $delimiter) {
-      $columns = TripalCultivatePhenotypesValidatorBase::splitRowIntoColumns($raw_row, $input_file_mime_type);
+    $columns = TripalCultivatePhenotypesValidatorBase::splitRowIntoColumns($raw_row, $input_file_mime_type);
+    $no_cols = count($columns);
 
+    if ($no_cols > $expected_columns['number_of_columns']) {
+      // The line has more columns than expected.
       if ($expected_columns['strict']) {
-        // A strict comparison - exact match only.
-        if (count($columns) != $expected_columns['number_of_columns']) {
-          array_push($delimiters_failed, $delimiter);
-        }
-      }
-      else {
-        // Not a strict comparison - at least x number of columns.
-        if (count($columns) < $expected_columns['number_of_columns']) {
-          array_push($delimiters_failed, $delimiter);
-        }
+        return [
+          'case' => 'Raw row exceeds number of strict columns',
+          'valid' => FALSE,
+          'failedItems' => [
+            'raw_row' => $raw_row,
+            'expected_columns' => $expected_columns['number_of_columns'],
+            'strict' => $expected_columns['strict'],
+          ],
+        ];
       }
     }
 
-    // If the failed delimiters array contains the same number of delimiters
-    // attempted, then every delimiter failed to split the line as required.
-    if ($delimiters_used == $delimiters_failed) {
+    if ($no_cols < $expected_columns['number_of_columns']) {
+      // The line has less column than expected.
       return [
-        'case' => 'Raw row is not delimited',
+        'case' => 'Raw row has insufficient number of columns',
         'valid' => FALSE,
         'failedItems' => [
           'raw_row' => $raw_row,
+          'expected_columns' => $expected_columns['number_of_columns'],
+          'strict' => $expected_columns['strict'],
         ],
       ];
     }

diff --git a/trpcultivate_phenotypes/tests/src/Kernel/Validators/ValidatorValidDelimitedFileTest.php b/trpcultivate_phenotypes/tests/src/Kernel/Validators/ValidatorValidDelimitedFileTest.php
@@ -107,22 +107,45 @@ public function provideRawRowToDelimitedFileValidator() {
         ],
       ],
 
-      // #2: Not the expected number of columns (strict comparison).
+      // #2: Not the expected number of columns (more and strict comparison).
       [
-        'column number mismatch',
+        'column number mismatch - more',
+        "Data Value One\tData Value Two\tData Value Three\tData Value Four\tData Value Five",
+        [
+          'number_of_columns' => 4,
+          'strict' => TRUE,
+        ],
+        [
+          'case' => 'Raw row exceeds number of strict columns',
+          'valid' => FALSE,
+          'failedItems' => [
+            'raw_row' => "Data Value One\tData Value Two\tData Value Three\tData Value Four\tData Value Five",
+            'expected_columns' => 4,
+            'strict' => TRUE,
+          ],
+        ],
+      ],
+
+      // #3: Not the expected number of columns (less and strict comparison).
+      [
+        'column number mismatch - less',
         "Data Value One\tData Value Two\tData Value Three",
         [
           'number_of_columns' => 4,
           'strict' => TRUE,
         ],
         [
-          'case' => 'Raw row is not delimited',
+          'case' => 'Raw row has insufficient number of columns',
           'valid' => FALSE,
-          'failedItems' => ['raw_row' => "Data Value One\tData Value Two\tData Value Three"],
+          'failedItems' => [
+            'raw_row' => "Data Value One\tData Value Two\tData Value Three",
+            'expected_columns' => 4,
+            'strict' => TRUE,
+          ],
         ],
       ],
 
-      // #3: Not the expected number of columns (not strict comparison).
+      // #4: Not the expected number of columns (not strict comparison).
       [
         'column number failed minimum',
         "Data Value One\tData Value Two\tData Value Three",
@@ -131,13 +154,17 @@ public function provideRawRowToDelimitedFileValidator() {
           'strict' => FALSE,
         ],
         [
-          'case' => 'Raw row is not delimited',
+          'case' => 'Raw row has insufficient number of columns',
           'valid' => FALSE,
-          'failedItems' => ['raw_row' => "Data Value One\tData Value Two\tData Value Three"],
+          'failedItems' => [
+            'raw_row' => "Data Value One\tData Value Two\tData Value Three",
+            'expected_columns' => 4,
+            'strict' => FALSE,
+          ],
         ],
       ],
 
-      // #4: Line has 2 different delimiters (tab + comma) where one is used to
+      // #5: Line has 2 different delimiters (tab + comma) where one is used to
       // delimit values and the other exists within the values.
       [
         'two delimiters used',
@@ -153,7 +180,7 @@ public function provideRawRowToDelimitedFileValidator() {
         ],
       ],
 
-      // #5: Valid raw row and expecting exactly 4 columns.
+      // #6: Valid raw row and expecting exactly 4 columns.
       [
         'valid raw row with exact columns',
         "Data Value One\tData Value Two\tData Value Three\tData Value Four",
@@ -168,7 +195,7 @@ public function provideRawRowToDelimitedFileValidator() {
         ],
       ],
 
-      // #6: Valid raw row and expecting at least 3 columns.
+      // #7: Valid raw row and expecting at least 3 columns.
       [
         'valid raw row with minimum columns',
         "Data Value One\tData Value Two\tData Value Three\tData Value Four",
@@ -183,7 +210,7 @@ public function provideRawRowToDelimitedFileValidator() {
         ],
       ],
 
-      // #7: Raw row has one column with strict flag set to FALSE (mininum).
+      // #8: Raw row has one column with strict flag set to FALSE (mininum).
       [
         'one column with strict set to false',
         "Data Value One",
@@ -198,7 +225,7 @@ public function provideRawRowToDelimitedFileValidator() {
         ],
       ],
 
-      // #8: Raw row has one column with strict flag set to TRUE (exact match).
+      // #9: Raw row has one column with strict flag set to TRUE (exact match).
       [
         'one column with strict flag set to true',
         "Data Value One",