#120 ZugferdDocumentPdfReader now contains additional methods for ext…

…racting the XML content
horstoeko · Sep 24, 2024 · 10fc7d3 · 10fc7d3
1 parent 1422fc1
commit 10fc7d3
Show file tree

Hide file tree

Showing 4 changed files with 112 additions and 26 deletions.
diff --git a/build/phpunit.xml b/build/phpunit.xml
@@ -29,13 +29,13 @@
             <file>../tests/testcases/BuilderExtendedTest.php</file>
         </testsuite>
         <testsuite name="PDFReader">
+            <file>../tests/testcases/PdfReaderGeneralTest.php</file>
             <file>../tests/testcases/PdfReaderMinimumSimpleTest.php</file>
             <file>../tests/testcases/PdfReaderEn16931AllowanceChargeTest.php</file>
             <file>../tests/testcases/PdfReaderEn16931SimpleTest.php</file>
             <file>../tests/testcases/PdfReaderExtendedTest.php</file>
             <file>../tests/testcases/PdfReaderExtended2Test.php</file>
             <file>../tests/testcases/PdfReaderXRechnungSimpleTest.php</file>
-            <file>../tests/testcases/PdfReaderInvalidTest.php</file>
             <file>../tests/testcases/PdfReaderMultipleAttachmentsTest.php</file>
         </testsuite>
         <testsuite name="PDFBuilder">
@@ -45,11 +45,9 @@
             <file>../tests/testcases/ValidatorValidTest.php</file>
             <file>../tests/testcases/ValidatorInvalidTest.php</file>
         </testsuite>
-        <!--
         <testsuite name="Exporter">
             <file>../tests/testcases/JsonExporterTest.php</file>
         </testsuite>
-        -->
         <testsuite name="issues">
             <file>../tests/testcases/issues/Issue10Test.php</file>
             <file>../tests/testcases/issues/Issue18Test.php</file>

diff --git a/src/ZugferdDocumentPdfReader.php b/src/ZugferdDocumentPdfReader.php
@@ -53,13 +53,70 @@ public static function readAndGuessFromFile(string $pdfFilename): ?ZugferdDocume
     }
 
     /**
-     * Load a PDF content (ZUGFeRD/Factur-X)
+     * Tries to load an attachment content from PDF and return a ZugferdDocumentReader
+     * If any erros occured or no attachments were found null is returned
      *
      * @param  string $pdfContent
      * String Containing the binary pdf data
      * @return ZugferdDocumentReader|null
      */
     public static function readAndGuessFromContent(string $pdfContent): ?ZugferdDocumentReader
+    {
+        $xmlContent = static::extractXMLFromContent($pdfContent);
+
+        if (is_null($xmlContent)) {
+            return null;
+        }
+
+        try {
+            return ZugferdDocumentReader::readAndGuessFromContent($xmlContent);
+        } catch (\Exception $e) {
+            return null;
+        }
+    }
+
+    /**
+     * Returns a XML content from a PDF file
+     *
+     * @param string $pdfFilename
+     * Contains a full-qualified filename which must exist and must be readable
+     * @return string|null
+     */
+    public static function getXmlFromFile(string $pdfFilename) : ?string
+    {
+        if (!file_exists($pdfFilename)) {
+            throw new ZugferdFileNotFoundException($pdfFilename);
+        }
+
+        $pdfContent = file_get_contents($pdfFilename);
+
+        if ($pdfContent === false) {
+            throw new ZugferdFileNotReadableException($pdfFilename);
+        }
+
+        return static::getXmlFromContent($pdfContent);
+    }
+
+    /**
+     * Returns a XML content from a PDF binary stream (string)
+     *
+     * @param string $pdfContent
+     * String Containing the binary pdf data
+     * @return string|null
+     */
+    public static function getXmlFromContent(string $pdfContent) : ?string
+    {
+        return static::extractXMLFromContent($pdfContent);
+    }
+
+    /**
+     * Get the attachment content from XML.
+     * See the allowed filenames which are supported
+     *
+     * @param string $pdfContent
+     * @return string|null
+     */
+    private static function extractXMLFromContent(string $pdfContent): ?string
     {
         $pdfParser = new PdfParser();
         $pdfParsed = $pdfParser->parseContent($pdfContent);
@@ -87,7 +144,7 @@ public static function readAndGuessFromContent(string $pdfContent): ?ZugferdDocu
                 $embeddedFiles = $pdfParsed->getObjectsByType('EmbeddedFile');
                 foreach ($embeddedFiles as $embeddedFile) {
                     if ($attachmentIndex == $embeddedFileIndex) {
-                        $returnValue = ZugferdDocumentReader::readAndGuessFromContent($embeddedFile->getContent());
+                        $returnValue = $embeddedFile->getContent();
                         break;
                     }
                     $embeddedFileIndex++;

diff --git a/tests/testcases/PdfReaderGeneralTest.php b/tests/testcases/PdfReaderGeneralTest.php
@@ -0,0 +1,52 @@
+<?php
+
+namespace horstoeko\zugferd\tests\testcases;
+
+use horstoeko\zugferd\exception\ZugferdFileNotFoundException;
+use \horstoeko\zugferd\tests\TestCase;
+use horstoeko\zugferd\ZugferdDocument;
+use \horstoeko\zugferd\ZugferdDocumentPdfReader;
+
+class PdfReaderGeneralTest extends TestCase
+{
+    public function testCanReadPdf(): void
+    {
+        $document = ZugferdDocumentPdfReader::readAndGuessFromFile(dirname(__FILE__) . "/../assets/InvalidPDF.pdf");
+        $this->assertNull($document);
+    }
+
+    public function testFileNotFound(): void
+    {
+        $this->expectException(ZugferdFileNotFoundException::class);
+        $document = ZugferdDocumentPdfReader::readAndGuessFromFile(dirname(__FILE__) . "/../assets/unknown.pdf");
+    }
+
+    public function testCanReadPdf2(): void
+    {
+        $document = ZugferdDocumentPdfReader::getXmlFromFile(dirname(__FILE__) . "/../assets/InvalidPDF.pdf");
+        $this->assertNull($document);
+    }
+
+    public function testFileNotFound2(): void
+    {
+        $this->expectException(ZugferdFileNotFoundException::class);
+        $document = ZugferdDocumentPdfReader::getXmlFromFile(dirname(__FILE__) . "/../assets/unknown.pdf");
+    }
+
+    public function testCanReadPdf3(): void
+    {
+        $document = ZugferdDocumentPdfReader::readAndGuessFromFile(dirname(__FILE__) . "/../assets/zugferd_2p1_EN16931_Einfach.pdf");
+        $this->assertNotNull($document);
+        $this->assertInstanceOf(ZugferdDocument::class, $document);
+    }
+
+    public function testCanReadPdf4(): void
+    {
+        $xmlString = ZugferdDocumentPdfReader::getXmlFromFile(dirname(__FILE__) . "/../assets/zugferd_2p1_EN16931_Einfach.pdf");
+        $this->assertNotNull($xmlString);
+        $this->assertIsString($xmlString);
+        $this->assertStringContainsString("<?xml version='1.0'", $xmlString);
+        $this->assertStringContainsString("<rsm:CrossIndustryInvoice", $xmlString);
+        $this->assertStringContainsString("</rsm:CrossIndustryInvoice>", $xmlString);
+    }
+}
diff --git a/tests/testcases/PdfReaderInvalidTest.php b/tests/testcases/PdfReaderInvalidTest.php