Add CfdiUtils\Cleaner\Cleaner::removeIncompleteSchemaLocations

docs/CHANGELOG.md

@@ -13,7 +13,7 @@
 - Remove `trigger_error` on `\CfdiUtils\Elements\Cfdi33\Comprobante::getCfdiRelacionados` when called with arguments.
 
 
-## Version 2.7.1 2018-12-03
+## Version 2.7.1 2018-12-04
 
 - Fix wrong use of `escapeshellcmd` replacing with `escapeshellarg`
 - Add argument `-c|--clean` to script `tests/validate.php` to perform clean before validate
@@ -27,6 +27,8 @@
     - Add `Config::getServiceUrl`
     - Add `Config::wsdlLocation` property
 - Add `--local-wsdl` parameter to `tests/estadosat.php` script
+- Add a new step on `CfdiUtils\Cleaner\Cleaner` that removes from `xsi:schemaLocations` the namespaces that are
+  not followed by a string that ends on `.xsd`
 
 
 ## Version 2.7.0 2018-10-19

docs/leer/limpieza-cfdi.md

@@ -25,6 +25,8 @@ Para evitar estos errores se puede usar el objeto `CfdiUtils\Cleaner\Cleaner`.
 Este objeto requiere una cadena de texto con XML válido. Y limpia el XML siguiendo estos pasos:
 
 1. Remueve el nodo `cfdi:Addenda`.
+1. Remueve dentro de las locaciones de espacios de nombre `xsi:schemaLocation` los namespaces que no tengan
+    a continuación una uri que termine en `.xsd`.
 1. Remueve todos los nodos que no tengan relación con el SAT (los que no contengan `http://www.sat.gob.mx/`).
 1. Remueve todos los pares de espacio de nombre y archivo xsd de los `xsi:schemaLocation` que no tengan relación con el SAT.
 1. Remueve todos los espacios de nombres listados que no están en uso.
@@ -38,3 +40,6 @@ También se puede instanciar un objeto de la clase `CfdiUtils\Cleaner\Cleaner` y
 - `load(string $content)`: Carga un contenido XML "sucio"
 - `clean()`: Realiza la limpieza
 - `retrieveXml()`: Obtiene el contenido XML "limpio"
+
+Si deseas implementar tu propio orden, hacer o agregar nuevos limpiadores puedes extender la clase o sobrescribir
+el método `clean` o bien llamar a cada uno de los pasos de limpieza por tu propia cuenta.

src/CfdiUtils/Cleaner/Cleaner.php

@@ -87,6 +87,7 @@ public static function isNameSpaceAllowed(string $namespace): bool
     public function clean()
     {
         $this->removeAddenda();
+        $this->removeIncompleteSchemaLocations();
         $this->removeNonSatNSNodes();
         $this->removeNonSatNSschemaLocations();
         $this->removeUnusedNamespaces();
@@ -157,6 +158,37 @@ public function removeAddenda()
         }
     }
 
+    /**
+     * Procedure to drop schemaLocations where second part does not ends with '.xsd'
+     *
+     * @return void
+     */
+    public function removeIncompleteSchemaLocations()
+    {
+        $schemaLocations = $this->obtainXsiSchemaLocations();
+        for ($s = 0; $s < $schemaLocations->length; $s++) {
+            $element = $schemaLocations->item($s);
+            if (null !== $element) {
+                $element->nodeValue = $this->removeIncompleteSchemaLocation($element->nodeValue);
+            }
+        }
+    }
+
+    public function removeIncompleteSchemaLocation(string $source): string
+    {
+        $components = array_values(array_filter(array_map('trim', explode(' ', $source))));
+        $length = count($components);
+        for ($c = 0; $c < $length; $c = $c + 1) {
+            $xsd = $components[$c + 1] ?? '';
+            if ((0 === strcasecmp('.xsd', substr($xsd, -4, 4)))) {
+                $c = $c + 1;
+                continue;
+            }
+            $components[$c] = '';
+        }
+        return strval(implode(' ', array_filter($components)));
+    }
+
     /**
      * Procedure to drop schemaLocations that are not allowed
      * If the schemaLocation is empty then remove the attribute
@@ -165,12 +197,7 @@ public function removeAddenda()
      */
     public function removeNonSatNSschemaLocations()
     {
-        // Do not assume that prefix for http://www.w3.org/2001/XMLSchema-instance is "xsi"
-        $xsi = $this->dom()->lookupPrefix('http://www.w3.org/2001/XMLSchema-instance');
-        if (! $xsi) {
-            return;
-        }
-        $schemaLocations = $this->xpathQuery("//@$xsi:schemaLocation");
+        $schemaLocations = $this->obtainXsiSchemaLocations();
         for ($s = 0; $s < $schemaLocations->length; $s++) {
             $element = $schemaLocations->item($s);
             if (null !== $element) {
@@ -265,6 +292,16 @@ public function removeUnusedNamespaces()
         }
     }
 
+    private function obtainXsiSchemaLocations(): DOMNodeList
+    {
+        // Do not assume that prefix for http://www.w3.org/2001/XMLSchema-instance is "xsi"
+        $xsi = $this->dom()->lookupPrefix('http://www.w3.org/2001/XMLSchema-instance');
+        if (! $xsi) {
+            return new DOMNodeList();
+        }
+        return $this->xpathQuery("//@$xsi:schemaLocation");
+    }
+
     /**
      * Helper function to perform a XPath query using an element (or root element)
      * @param string $query

tests/CfdiUtilsTests/Cleaner/CleanerTest.php

@@ -68,10 +68,11 @@ public function testCleanOnDetail()
     {
         $basefile = $this->utilAsset('cleaner/v32-dirty.xml');
         $step1 = $this->utilAsset('cleaner/v32-no-addenda.xml');
-        $step2 = $this->utilAsset('cleaner/v32-no-nonsat-nodes.xml');
-        $step3 = $this->utilAsset('cleaner/v32-no-nonsat-schemalocations.xml');
-        $step4 = $this->utilAsset('cleaner/v32-no-nonsat-xmlns.xml');
-        foreach ([$basefile, $step1, $step2, $step3, $step4] as $filename) {
+        $step2 = $this->utilAsset('cleaner/v32-no-incomplete-schemalocations.xml');
+        $step3 = $this->utilAsset('cleaner/v32-no-nonsat-nodes.xml');
+        $step4 = $this->utilAsset('cleaner/v32-no-nonsat-schemalocations.xml');
+        $step5 = $this->utilAsset('cleaner/v32-no-nonsat-xmlns.xml');
+        foreach ([$basefile, $step1, $step3, $step2, $step4, $step5] as $filename) {
             $this->assertFileExists($basefile, "The file $filename for testing does not exists");
         }
         $cleaner = new Cleaner(file_get_contents($basefile));
@@ -88,29 +89,36 @@ public function testCleanOnDetail()
             'Compare that addenda was removed'
         );
 
-        $cleaner->removeNonSatNSNodes();
+        $cleaner->removeIncompleteSchemaLocations();
         $this->assertXmlStringEqualsXmlFile(
             $step2,
             $cleaner->retrieveXml(),
+            'Compare that incomplete schemaLocations were removed'
+        );
+
+        $cleaner->removeNonSatNSNodes();
+        $this->assertXmlStringEqualsXmlFile(
+            $step3,
+            $cleaner->retrieveXml(),
             'Compare that non SAT nodes were removed'
         );
 
         $cleaner->removeNonSatNSschemaLocations();
         $this->assertXmlStringEqualsXmlFile(
-            $step3,
+            $step4,
             $cleaner->retrieveXml(),
             'Compare that non SAT schemaLocations were removed'
         );
 
         $cleaner->removeUnusedNamespaces();
         $this->assertXmlStringEqualsXmlFile(
-            $step4,
+            $step5,
             $cleaner->retrieveXml(),
             'Compare that xmlns definitions were removed'
         );
 
         $this->assertXmlStringEqualsXmlFile(
-            $step4,
+            $step5,
             Cleaner::staticClean(file_get_contents($basefile)),
             'Check static method for cleaning is giving the same results as detailed execution'
         );
@@ -161,4 +169,13 @@ public function testRemoveNonSatNSschemaLocationsRemoveEmptySchemaLocation()
         $cleaner->removeNonSatNSschemaLocations();
         $this->assertXmlStringEqualsXmlString($xmlExpectedContent, $cleaner->retrieveXml());
     }
+
+    public function testRemoveIncompleteSchemaLocation()
+    {
+        // source include spaces to ensure that is working properly
+        $source = '  bleh  foo   foo.xsd bar baz      zoo zoo.xsd baa   xee  xee.xsd bah       ';
+        $expected = 'foo foo.xsd zoo zoo.xsd xee xee.xsd';
+        $cleaner = new Cleaner('');
+        $this->assertSame($expected, $cleaner->removeIncompleteSchemaLocation($source));
+    }
 }

tests/assets/cleaner/v32-dirty.xml

@@ -1,7 +1,7 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <cfdi:Comprobante xmlns:cfdi="http://www.sat.gob.mx/cfd/3" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
   xmlns:add="http://example.org/add" xmlns:xyz="http://example.org/xyz"
-  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add   http://example.org/xsd/add.xsd http://example.org/xyz http://example.org/xsd/xyz.xsd"
+  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add   http://example.org/xsd/add.xsd  http://www.sat.gob.mx/orphan1 http://www.sat.gob.mx/orphan2 http://example.org/xyz http://example.org/xsd/xyz.xsd http://www.sat.gob.mx/orphan3"
   version="3.2">
 <cfdi:Addenda>
     <ext:ElementosExtra xmlns:ext="http://example.com/xsd/ext">

tests/assets/cleaner/v32-no-addenda.xml

@@ -1,7 +1,7 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <cfdi:Comprobante xmlns:cfdi="http://www.sat.gob.mx/cfd/3" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
   xmlns:add="http://example.org/add" xmlns:xyz="http://example.org/xyz"
-  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add   http://example.org/xsd/add.xsd http://example.org/xyz http://example.org/xsd/xyz.xsd"
+  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add   http://example.org/xsd/add.xsd  http://www.sat.gob.mx/orphan1 http://www.sat.gob.mx/orphan2 http://example.org/xyz http://example.org/xsd/xyz.xsd http://www.sat.gob.mx/orphan3"
   version="3.2">
 
 <cfdi:Complemento>

tests/assets/cleaner/v32-no-incomplete-schemalocations.xml

@@ -0,0 +1,11 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<cfdi:Comprobante xmlns:cfdi="http://www.sat.gob.mx/cfd/3" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+  xmlns:add="http://example.org/add" xmlns:xyz="http://example.org/xyz"
+  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add http://example.org/xsd/add.xsd http://example.org/xyz http://example.org/xsd/xyz.xsd"
+  version="3.2">
+
+<cfdi:Complemento>
+    <tfd:TimbreFiscalDigital xmlns:tfd="http://www.sat.gob.mx/TimbreFiscalDigital" xsi:schemaLocation="http://www.sat.gob.mx/TimbreFiscalDigital http://www.sat.gob.mx/sitio_internet/cfd/TimbreFiscalDigital/TimbreFiscalDigital.xsd" version="1.0"/>
+    <add:point x="1" y="2" add:removed="Check if this attributte is removed"/>
+</cfdi:Complemento>
+</cfdi:Comprobante>

tests/assets/cleaner/v32-no-nonsat-nodes.xml

@@ -1,11 +1,11 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <cfdi:Comprobante xmlns:cfdi="http://www.sat.gob.mx/cfd/3" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
   xmlns:add="http://example.org/add" xmlns:xyz="http://example.org/xyz"
-  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add   http://example.org/xsd/add.xsd http://example.org/xyz http://example.org/xsd/xyz.xsd"
+  xsi:schemaLocation="http://www.sat.gob.mx/cfd/3 http://www.sat.gob.mx/sitio_internet/cfd/3/cfdv32.xsd http://example.org/add http://example.org/xsd/add.xsd http://example.org/xyz http://example.org/xsd/xyz.xsd"
   version="3.2">
 
 <cfdi:Complemento>
     <tfd:TimbreFiscalDigital xmlns:tfd="http://www.sat.gob.mx/TimbreFiscalDigital" xsi:schemaLocation="http://www.sat.gob.mx/TimbreFiscalDigital http://www.sat.gob.mx/sitio_internet/cfd/TimbreFiscalDigital/TimbreFiscalDigital.xsd" version="1.0"/>
-    
+
 </cfdi:Complemento>
 </cfdi:Comprobante>

tests/assets/cleaner/v32-no-nonsat-schemalocations.xml

@@ -6,6 +6,6 @@
 
 <cfdi:Complemento>
     <tfd:TimbreFiscalDigital xmlns:tfd="http://www.sat.gob.mx/TimbreFiscalDigital" xsi:schemaLocation="http://www.sat.gob.mx/TimbreFiscalDigital http://www.sat.gob.mx/sitio_internet/cfd/TimbreFiscalDigital/TimbreFiscalDigital.xsd" version="1.0"/>
-    
+
 </cfdi:Complemento>
 </cfdi:Comprobante>

tests/assets/cleaner/v32-no-nonsat-xmlns.xml

@@ -5,6 +5,6 @@
 
 <cfdi:Complemento>
     <tfd:TimbreFiscalDigital xmlns:tfd="http://www.sat.gob.mx/TimbreFiscalDigital" xsi:schemaLocation="http://www.sat.gob.mx/TimbreFiscalDigital http://www.sat.gob.mx/sitio_internet/cfd/TimbreFiscalDigital/TimbreFiscalDigital.xsd" version="1.0"/>
-    
+
 </cfdi:Complemento>
 </cfdi:Comprobante>