Fixed validation for csv.header: false (#80)

JBZoo · Mar 24, 2024 · b11dbfe · b11dbfe
1 parent c06a6d0
commit b11dbfe
Show file tree

Hide file tree

Showing 8 changed files with 121 additions and 27 deletions.
diff --git a/src/Csv/CsvFile.php b/src/Csv/CsvFile.php
@@ -96,6 +96,11 @@ public function validate(bool $quickStop = false): ErrorSuite
         return (new ValidatorCsv($this, $this->schema))->validate($quickStop);
     }
 
+    public function getRealColumNumber(): int
+    {
+        return \count($this->getRecordsChunk(0, 1)->first());
+    }
+
     private function prepareReader(): LeagueReader
     {
         $reader = LeagueReader::createFromPath($this->csvFilename)

diff --git a/src/Rules/Cell/AllowValues.php b/src/Rules/Cell/AllowValues.php
@@ -24,7 +24,7 @@ class AllowValues extends AbstractCellRule
 
     public function validateRule(string $cellValue): ?string
     {
-        $allowedValues = $this->getOptionAsArray();
+        $allowedValues = \array_map('\strval', $this->getOptionAsArray());
 
         if (\count($allowedValues) === 0) {
             return 'Allowed values are not defined';

diff --git a/src/Rules/Cell/NotAllowValues.php b/src/Rules/Cell/NotAllowValues.php
@@ -24,7 +24,7 @@ class NotAllowValues extends AbstractCellRule
 
     public function validateRule(string $cellValue): ?string
     {
-        $notAllowedValues = $this->getOptionAsArray();
+        $notAllowedValues = \array_map('\strval', $this->getOptionAsArray());
 
         if (\count($notAllowedValues) === 0) {
             return 'Not allowed values are not defined';

diff --git a/src/Schema.php b/src/Schema.php
@@ -94,8 +94,12 @@ public function getColumnsMappedByHeader(array $header): array
     {
         $map = [];
 
-        foreach ($header as $headerName) {
-            $map[$headerName] = $this->columns[$headerName] ?? null;
+        if ($this->getCsvStructure()->isHeader()) {
+            foreach ($header as $headerName) {
+                $map[$headerName] = $this->columns[$headerName] ?? null;
+            }
+        } else {
+            return $this->getColumns();
         }
 
         return $map;

diff --git a/src/Validators/ValidatorCsv.php b/src/Validators/ValidatorCsv.php
@@ -100,6 +100,9 @@ private function validateHeader(bool $quickStop = false): ErrorSuite
         return $errors;
     }
 
+    /**
+     * @SuppressWarnings(PHPMD.NPathComplexity)
+     */
     private function validateLines(bool $quickStop = false): ErrorSuite
     {
         $errors      = new ErrorSuite();
@@ -126,16 +129,30 @@ private function validateLines(bool $quickStop = false): ErrorSuite
             }
 
             foreach ($this->csv->getRecords() as $line => $record) {
-                if ($isAggRules) {  // Time & memory optimization
-                    $columValues[] = ValidatorColumn::prepareValue($record[$column->getKey()], $aggInputType);
-                }
+                $lineNum = (int)$line + 1;
 
                 if ($isRules) { // Time optimization
-                    $errors->addErrorSuit($colValidator->validateCell($record[$column->getKey()], (int)$line + 1));
+                    if (!isset($record[$column->getKey()])) {
+                        $errors->addError(
+                            new Error(
+                                'csv.column',
+                                "Column index:{$column->getKey()} not found",
+                                $column->getHumanName(),
+                                $lineNum,
+                            ),
+                        );
+                    } else {
+                        $errors->addErrorSuit($colValidator->validateCell($record[$column->getKey()], $lineNum));
+                    }
+
                     if ($quickStop && $errors->count() > 0) {
                         return $errors;
                     }
                 }
+
+                if ($isAggRules && isset($record[$column->getKey()])) {  // Time & memory optimization
+                    $columValues[] = ValidatorColumn::prepareValue($record[$column->getKey()], $aggInputType);
+                }
             }
 
             Utils::debug("<i>Col</i> aggregate: {$column->getKey()}");
@@ -182,27 +199,40 @@ private function validateColumn(bool $quickStop): ErrorSuite
     {
         $errors = new ErrorSuite();
 
-        if (!$this->schema->getCsvStructure()->isHeader()) {
-            return $errors;
-        }
+        if ($this->schema->getCsvStructure()->isHeader()) {
+            $realColumns   = $this->schema->getColumnsMappedByHeader($this->csv->getHeader());
+            $schemaColumns = $this->schema->getColumns();
 
-        $realColumns   = $this->schema->getColumnsMappedByHeader($this->csv->getHeader());
-        $schemaColumns = $this->schema->getColumns();
+            $notFoundColums = \array_diff(\array_keys($schemaColumns), \array_keys($realColumns));
 
-        $notFoundColums = \array_diff(\array_keys($schemaColumns), \array_keys($realColumns));
-
-        if (\count($notFoundColums) > 0) {
-            $error = new Error(
-                'csv.header',
-                'Columns not found in CSV: "<c>' . \implode(', ', $notFoundColums) . '</c>"',
-                '',
-                ValidatorColumn::FALLBACK_LINE,
-            );
+            if (\count($notFoundColums) > 0) {
+                $error = new Error(
+                    'csv.header',
+                    'Columns not found in CSV: "<c>' . \implode(', ', $notFoundColums) . '</c>"',
+                    '',
+                    ValidatorColumn::FALLBACK_LINE,
+                );
 
-            $errors->addError($error);
+                $errors->addError($error);
+                if ($quickStop) {
+                    return $errors;
+                }
+            }
+        } else {
+            $schemaColumns = \count($this->schema->getColumns());
+            $realColumns   = $this->csv->getRealColumNumber();
+            if ($realColumns < $schemaColumns) {
+                $error = new Error(
+                    'csv.header',
+                    'Real number of columns is less than schema: ' . $realColumns . ' < ' . $schemaColumns,
+                    '',
+                    ValidatorColumn::FALLBACK_LINE,
+                );
 
-            if ($quickStop) {
-                return $errors;
+                $errors->addError($error);
+                if ($quickStop) {
+                    return $errors;
+                }
             }
         }
 

diff --git a/tests/Commands/ValidateCsvBasicTest.php b/tests/Commands/ValidateCsvBasicTest.php
@@ -252,6 +252,47 @@ public function testInvalidSchemaAndNotFoundCSV(): void
         isSame($expected, $actual);
     }
 
+    public function testValidateOneCsvNoHeaderNegative(): void
+    {
+        [$actual, $exitCode] = Tools::virtualExecution('validate:csv', [
+            'csv'    => Tools::DEMO_CSV,
+            'schema' => './tests/schemas/simple_no_header.yml',
+        ]);
+
+        $expected = $expected = <<<'TXT'
+            Found Schemas   : 1
+            Found CSV files : 1
+            Pairs by pattern: 1
+            
+            Check schema syntax: 1
+            (1/1) OK: ./tests/schemas/simple_no_header.yml
+            
+            CSV file validation: 1
+            (1/1) Schema: ./tests/schemas/simple_no_header.yml
+            (1/1) CSV   : ./tests/fixtures/demo.csv
+            (1/1) Issues: 2
+            +------+-----------+---------- demo.csv -----------------------------+
+            | Line | id:Column | Rule             | Message                      |
+            +------+-----------+------------------+------------------------------+
+            | 2    | 0:        | not_allow_values | Value "Clyde" is not allowed |
+            | 5    | 2:        | not_allow_values | Value "74" is not allowed    |
+            +------+-----------+---------- demo.csv -----------------------------+
+            
+            
+            Summary:
+              1 pairs (schema to csv) were found based on `filename_pattern`.
+              No issues in 1 schemas.
+              Found 2 issues in 1 out of 1 CSV files.
+              Schemas have no filename_pattern and are applied to all CSV files found:
+                - ./tests/schemas/simple_no_header.yml
+            
+            
+            TXT;
+
+        isSame(1, $exitCode, $actual);
+        isSame($expected, $actual);
+    }
+
     public function testSchemaNotFound(): void
     {
         $this->expectExceptionMessage('Schema file(s) not found: invalid_schema_path.yml');

diff --git a/tests/Validators/CsvValidatorTest.php b/tests/Validators/CsvValidatorTest.php
@@ -36,10 +36,19 @@ public function testValidWithHeader(): void
         isSame('', \strip_tags((string)$csv->validate()));
     }
 
-    public function testValidWithoutHeader(): void
+    public function testInvalidWithoutHeader(): void
     {
         $csv = new CsvFile(Tools::CSV_SIMPLE_NO_HEADER, Tools::SCHEMA_SIMPLE_NO_HEADER);
-        isSame('', \strip_tags((string)$csv->validate()));
+        isSame(
+            <<<'TEXT'
+                "csv.header" at line 1. Real number of columns is less than schema: 2 < 3.
+                "csv.column" at line 1, column "2:". Column index:2 not found.
+                "csv.column" at line 2, column "2:". Column index:2 not found.
+                "csv.column" at line 3, column "2:". Column index:2 not found.
+                
+                TEXT,
+            \strip_tags((string)$csv->validate()),
+        );
     }
 
     public function testInvalidSchemaFile(): void

diff --git a/tests/schemas/simple_no_header.yml b/tests/schemas/simple_no_header.yml
@@ -14,7 +14,12 @@ csv:
   header: false
 
 columns:
+  - rules:
+      not_allow_values: [ Clyde ]
   - rules:
       not_empty: true
   - rules:
+      not_allow_values: [ 74 ]
       not_empty: true
+    aggregate_rules:
+      is_unique: true