Add basic validation support (#5)

* coding style fixes, minor refactoring * added DatapackageValidator (depends on SchemaValidator from frictionlessdata/tableschema) * added basic datapackage validation functionality
frictionlessdata · Apr 18, 2017 · 34593be · 34593be
1 parent 9929166
commit 34593be
Show file tree

Hide file tree

Showing 18 changed files with 949 additions and 205 deletions.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -4,6 +4,10 @@ The project follows the [Open Knowledge International coding standards](https://
 
 All PHP Code should conform to [PHP-FIG](http://www.php-fig.org/psr/) accepted PSRs.
 
+Flow Framework has a nice guide regarding coding standards:
+* [Printable summary of most important coding guidelines on one page **(.pdf)**](http://flowframework.readthedocs.io/en/stable/_downloads/Flow_Coding_Guidelines_on_one_page.pdf)
+* [The full guide **(.html)**](http://flowframework.readthedocs.io/en/stable/TheDefinitiveGuide/PartV/CodingGuideLines/PHP.html)
+
 
 ## Getting Started
 

diff --git a/README.md b/README.md
@@ -23,7 +23,8 @@ $ composer require frictionlessdata/datapackage
 ```php
 use frictionlessdata\datapackage;
 
-$datapackage = new Datapackage("tests/fixtures/multi_data_datapackage.json");
+// iterate over the data
+$datapackage = new datapackage\Datapackage("tests/fixtures/multi_data_datapackage.json");
 foreach ($datapackage as $resource) {
     print("-- ".$resource->name()." --");
     $i = 0;
@@ -34,6 +35,14 @@ foreach ($datapackage as $resource) {
         }
     }
 }
+
+// validate the descriptor
+$validationErrors = datapackage\Datapackage::validate("tests/fixtures/simple_invalid_datapackage.json");
+if (count($validationErrors) == 0) {
+    print("descriptor is valid");
+} else {
+    print(datapackage\DatapackageValidationError::getErrorMessages($validationErrors));
+}
 ```
 
 

diff --git a/composer.json b/composer.json
@@ -3,7 +3,9 @@
     "description": "A utility library for working with Data Packages",
     "license": "MIT",
     "require": {
-        "php": ">=5.4"
+        "php": ">=5.4",
+        "justinrainbow/json-schema": "^5.2",
+        "frictionlessdata/tableschema": "^0.1.2"
     },
     "require-dev": {
         "phpunit/phpunit": "^4.8.35",
@@ -15,6 +17,6 @@
         }
     },
     "scripts": {
-        "test": "phpunit --debug tests/ --coverage-clover coverage-clover.xml"
+        "test": "phpunit --debug --coverage-clover coverage-clover.xml --bootstrap tests/autoload.php tests/"
     }
 }
diff --git a/src/DataStream.php b/src/DataStream.php
@@ -1,34 +1,30 @@
-<?php namespace frictionlessdata\datapackage;
-
+<?php
+namespace frictionlessdata\datapackage;
 
 class DataStream implements \Iterator
 {
-    protected $_currentLineNumber = 0;
-    protected $_fopenResource;
-    protected $_dataSource;
-
     public function __construct($dataSource)
     {
         try {
-            $this->_fopenResource = fopen($dataSource, "r");
+            $this->fopenResource = fopen($dataSource, "r");
         } catch (\Exception $e) {
-            throw new DataStreamOpenException("Failed to open source ".json_encode($dataSource));
+            throw new Exceptions\DataStreamOpenException("Failed to open source ".json_encode($dataSource).": ".json_encode($e->getMessage()));
         }
     }
 
     public function __destruct()
     {
-        fclose($this->_fopenResource);
+        fclose($this->fopenResource);
     }
 
     public function rewind() {
-        if ($this->_currentLineNumber != 0) {
+        if ($this->currentLineNumber != 0) {
             throw new \Exception("DataStream does not support rewind, sorry");
         }
     }
 
     public function current() {
-        $line = fgets($this->_fopenResource);
+        $line = fgets($this->fopenResource);
         if ($line === false) {
             return "";
         } else {
@@ -37,17 +33,18 @@ public function current() {
     }
 
     public function key() {
-        return $this->_currentLineNumber;
+        return $this->currentLineNumber;
     }
 
     public function next() {
-        $this->_currentLineNumber++;
+        $this->currentLineNumber++;
     }
 
     public function valid() {
-        return (!feof($this->_fopenResource));
+        return (!feof($this->fopenResource));
     }
-}
 
-
-class DataStreamOpenException extends \Exception {};
+    protected $currentLineNumber = 0;
+    protected $fopenResource;
+    protected $dataSource;
+}
diff --git a/src/Datapackage.php b/src/Datapackage.php
@@ -1,5 +1,5 @@
-<?php namespace frictionlessdata\datapackage;
-
+<?php
+namespace frictionlessdata\datapackage;
 
 /**
  * Datapackage representation, supports loading from the following sources:
@@ -10,80 +10,121 @@
  */
 class Datapackage implements \Iterator
 {
-    protected $_descriptor;
-    protected $_currentResourcePosition = 0;
-    protected $_basePath;
-
     public function __construct($source, $basePath=null)
     {
         if (is_object($source)) {
-            $this->_descriptor = $source;
-            $this->_basePath = $basePath;
+            $this->descriptor = $source;
+            $this->basePath = $basePath;
         } elseif (is_string($source)) {
-            if (Utils::is_json_string($source)) {
+            if (Utils::isJsonString($source)) {
                 try {
-                    $this->_descriptor = json_decode($source);
+                    $this->descriptor = json_decode($source);
                 } catch (\Exception $e) {
-                    throw new DatapackageInvalidSourceException("Failed to load source: ".json_encode($source).": ".$e->getMessage());
+                    throw new Exceptions\DatapackageInvalidSourceException(
+                        "Failed to load source: ".json_encode($source).": ".$e->getMessage()
+                    );
                 }
-                $this->_basePath = $basePath;
-            } elseif ($this->_isHttpSource($source)) {
+                $this->basePath = $basePath;
+            } elseif ($this->isHttpSource($source)) {
                 try {
-                    $this->_descriptor = json_decode(file_get_contents($this->_normalizeHttpSource($source)));
+                    $this->descriptor = json_decode(file_get_contents($this->normalizeHttpSource($source)));
                 } catch (\Exception $e) {
-                    throw new DatapackageInvalidSourceException("Failed to load source: ".json_encode($source).": ".$e->getMessage());
+                    throw new Exceptions\DatapackageInvalidSourceException(
+                        "Failed to load source: ".json_encode($source).": ".$e->getMessage()
+                    );
                 }
                 // http sources don't allow relative paths, hence basePath should remain null
-                $this->_basePath = null;
+                $this->basePath = null;
             } else {
                 if (empty($basePath)) {
-                    $this->_basePath = dirname($source);
+                    $this->basePath = dirname($source);
                 } else {
-                    $this->_basePath = $basePath;
-                    $absPath = $this->_basePath.DIRECTORY_SEPARATOR.$source;
+                    $this->basePath = $basePath;
+                    $absPath = $this->basePath.DIRECTORY_SEPARATOR.$source;
                     if (file_exists($absPath)) {
                         $source = $absPath;
                     }
                 }
                 try {
-                    $this->_descriptor = json_decode(file_get_contents($source));
+                    $this->descriptor = json_decode(file_get_contents($source));
                 } catch (\Exception $e) {
-                    throw new DatapackageInvalidSourceException("Failed to load source: ".json_encode($source).": ".$e->getMessage());
+                    throw new Exceptions\DatapackageInvalidSourceException(
+                        "Failed to load source: ".json_encode($source).": ".$e->getMessage()
+                    );
                 }
 
             }
         } else {
-            throw new DatapackageInvalidSourceException("Invalid source: ".json_encode($source));
+            throw new Exceptions\DatapackageInvalidSourceException(
+                "Invalid source: ".json_encode($source)
+            );
         }
     }
 
-    protected function _normalizeHttpSource($source)
+    public static function validate($source, $basePath=null)
     {
-        return $source;
+        try {
+            $datapackage = new self($source, $basePath);
+            return DatapackageValidator::validate($datapackage->descriptor());
+        } catch (\Exception $e) {
+            return [new DatapackageValidationError(DatapackageValidationError::LOAD_FAILED, $e->getMessage())];
+        }
+
     }
 
-    protected function _isHttpSource($source)
+    /**
+     * get the descriptor as a native PHP object
+     *
+     * @return object
+     */
+    public function descriptor()
     {
-        return Utils::is_http_source($source);
+        return $this->descriptor;
     }
 
-    protected function _initResource($resourceDescriptor)
+    // standard iterator functions - to iterate over the resources
+    public function rewind() {$this->currentResourcePosition = 0;}
+    public function current() { return $this->initResource($this->descriptor()->resources[$this->currentResourcePosition]); }
+    public function key() { return $this->currentResourcePosition; }
+    public function next() { $this->currentResourcePosition++; }
+    public function valid() { return isset($this->descriptor()->resources[$this->currentResourcePosition]); }
+
+    protected $descriptor;
+    protected $currentResourcePosition = 0;
+    protected $basePath;
+
+    /**
+     * allows extending classes to add custom sources
+     * used by unit tests to add a mock http source
+     *
+     * @param string $source
+     * @return string
+     */
+    protected function normalizeHttpSource($source)
     {
-        return new Resource($resourceDescriptor, $this->_basePath);
+        return $source;
     }
 
-    public function descriptor()
+    /**
+     * allows extending classes to add custom sources
+     * used by unit tests to add a mock http source
+     *
+     * @param string $source
+     * @return bool
+     */
+    protected function isHttpSource($source)
     {
-        return $this->_descriptor;
+        return Utils::isHttpSource($source);
     }
 
-    // standard iterator functions - to iterate over the resources
-    public function rewind() { $this->_currentResourcePosition = 0; }
-    public function current() { return $this->_initResource($this->descriptor()->resources[$this->_currentResourcePosition]); }
-    public function key() { return $this->_currentResourcePosition; }
-    public function next() { $this->_currentResourcePosition++; }
-    public function valid() { return isset($this->descriptor()->resources[$this->_currentResourcePosition]); }
+    /**
+     * called by the resources iterator for each iteration
+     *
+     * @param object $resourceDescriptor
+     * @return \frictionlessdata\datapackage\Resource
+     */
+    protected function initResource($resourceDescriptor)
+    {
+        return new Resource($resourceDescriptor, $this->basePath);
+    }
 }
-
-
-class DatapackageInvalidSourceException extends \Exception {};
diff --git a/src/DatapackageValidationError.php b/src/DatapackageValidationError.php
@@ -0,0 +1,8 @@
+<?php namespace frictionlessdata\datapackage;
+
+// TODO: move the SchemaValidation logic to it's own independent package
+use frictionlessdata\tableschema;
+
+class DatapackageValidationError extends tableschema\SchemaValidationError
+{
+}
diff --git a/src/DatapackageValidator.php b/src/DatapackageValidator.php
@@ -0,0 +1,32 @@
+<?php namespace frictionlessdata\datapackage;
+
+// TODO: move the SchemaValidation logic to it's own independent package
+use frictionlessdata\tableschema;
+
+class DatapackageValidator extends tableschema\SchemaValidator
+{
+    public static function validate($descriptor)
+    {
+        $validator = new self($descriptor);
+        return $validator->get_validation_errors();
+    }
+
+    protected function _validateSchema()
+    {
+        // Validate
+        $validator = new \JsonSchema\Validator();
+        $validator->validate(
+            $this->descriptor, (object)[
+                '$ref' => 'file://' . realpath(dirname(__FILE__)).'/schemas/data-package.json'
+            ]
+        );
+        if (!$validator->isValid()) {
+            foreach ($validator->getErrors() as $error) {
+                $this->_addError(
+                    DatapackageValidationError::SCHEMA_VIOLATION,
+                    sprintf("[%s] %s", $error['property'], $error['message'])
+                );
+            }
+        }
+    }
+}
diff --git a/src/Exceptions/DataStreamOpenException.php b/src/Exceptions/DataStreamOpenException.php
@@ -0,0 +1,6 @@
+<?php
+namespace frictionlessdata\datapackage\Exceptions;
+
+class DataStreamOpenException extends \Exception {
+
+}
diff --git a/src/Exceptions/DatapackageInvalidSourceException.php b/src/Exceptions/DatapackageInvalidSourceException.php
@@ -0,0 +1,6 @@
+<?php
+namespace frictionlessdata\datapackage\Exceptions;
+
+class DatapackageInvalidSourceException extends \Exception {
+
+}