download eform resources directly from data gov and fix some validations

src/main/java/digital/slovensko/autogram/core/eforms/EFormResources.java

@@ -15,10 +15,12 @@
 import eu.europa.esig.dss.enumerations.ASiCContainerType;
 import eu.europa.esig.dss.enumerations.DigestAlgorithm;
 import eu.europa.esig.dss.model.DSSDocument;
+import eu.europa.esig.dss.model.InMemoryDocument;
+
 import org.w3c.dom.Node;
 
 public class EFormResources {
-    private static final String SOURCE_URL = "https://test-autogram-eforms-marek.s3.eu-central-1.amazonaws.com/v1/eforms/";
+    private static final String SOURCE_URL = "https://data.gov.sk/doc/egov/eform/";
     private static final Charset ENCODING = StandardCharsets.UTF_8;
 
     private final String url;
@@ -41,8 +43,11 @@ public static EFormAttributes tryToLoadEFormAttributes(DSSDocument document, Str
         if (eformResources == null)
             return null;
 
-        var transformation = eformResources.findTransformation();
-        var schema = eformResources.findSchema();
+        var transformation = eformResources.getTransformation();
+        var schema = eformResources.getSchema();
+        if (transformation == null || schema == null)
+            throw new XMLValidationException("Zlyhala príprava elektronického formulára", "Nepodarilo sa nájsť XSLT transformáciu alebo XSD schému");
+
         var identifier = eformResources.getIdentifier();
         var containerXmlns = "http://data.gov.sk/def/container/xmldatacontainer+xml/1.1";
         var container = ASiCContainerType.ASiC_E;
@@ -79,7 +84,7 @@ private static EFormResources buildEFormResourcesFromEformXml(Node xml, String c
         var parts = formUri.split("/");
         var formVersion = parts[parts.length - 1];
         var formIdentifier = parts[parts.length - 2];
-        var formDirectory = formIdentifier + "/" + formVersion;
+        var formDirectory = SOURCE_URL + formIdentifier + "/" + formVersion;
 
         return new EFormResources(formDirectory, xsdDigest, xsltDigest);
     }
@@ -90,12 +95,40 @@ private EFormResources(String url, String xsdDigest, String xsltDigest) {
         this.xsltDigest = xsltDigest;
     }
 
-    public String getIdentifier() {
+    private String getIdentifier() {
         return "http://data.gov.sk/doc/eform/" + url;
     }
 
-    public String findTransformation() throws XMLValidationException {
-        var xsltString = getTransformation(SOURCE_URL + url);
+    private String getTransformation() throws XMLValidationException {
+        var manifest_xml = getResource(url + "/META-INF/manifest.xml");
+        if (manifest_xml == null)
+            return null;
+
+        var fullPath = "";
+        try {
+            var parsed_manifest_xml = getXmlFromDocument(new InMemoryDocument(manifest_xml, "manifest.xml"));
+            var nodes = parsed_manifest_xml.getElementsByTagName("manifest:file-entry");
+            for (int i = 0; i < nodes.getLength(); i++) {
+                var node = nodes.item(i);
+
+                var mediaDestination = node.getAttributes().getNamedItem("media-destination");
+                if (mediaDestination != null && mediaDestination.getNodeValue().equals("sign")) {
+                    fullPath = node.getAttributes().getNamedItem("full-path").getNodeValue();
+
+                    fullPath = fullPath.replace("\\", "/");
+                    if (fullPath != null && fullPath.length() > 0)
+                        break;
+                }
+            }
+
+        } catch (XMLValidationException e) {
+            return null;
+        }
+
+        if (fullPath == null || fullPath.length() == 0)
+            return null;
+
+        var xsltString = getResource(url + "/" + fullPath);
         if (xsltString == null)
             return null;
 
@@ -106,8 +139,8 @@ public String findTransformation() throws XMLValidationException {
         return new String(xsltString, ENCODING);
     }
 
-    public String findSchema() throws XMLValidationException {
-        var xsdString = getSchema(SOURCE_URL + url);
+    private String getSchema() throws XMLValidationException {
+        var xsdString = getResource(url + "/schema.xsd");
         if (xsdString == null)
             return null;
 

src/main/java/digital/slovensko/autogram/core/eforms/EFormUtils.java

@@ -22,6 +22,7 @@
 import org.w3c.dom.Node;
 import org.xml.sax.InputSource;
 import org.xml.sax.SAXException;
+import org.xml.sax.SAXParseException;
 
 import digital.slovensko.autogram.core.errors.XMLValidationException;
 import digital.slovensko.autogram.core.errors.MultipleOriginalDocumentsFoundException;
@@ -124,17 +125,9 @@ public static String getNamespaceFromEformXml(Node xml) {
         return xmlns.getNodeValue();
     }
 
-    public static byte[] getTransformation(String url) {
-        return getResource(url + "/form.sb.xslt");
-    }
-
-    public static byte[] getSchema(String url) {
-        return getResource(url + "/form.xsd");
-    }
-
-    private static byte[] getResource(String url) {
+    public static byte[] getResource(String url) {
         var offlineFileLoader = new FileCacheDataLoader();
-        offlineFileLoader.setCacheExpirationTime(21600000);
+        offlineFileLoader.setCacheExpirationTime(21600000);  // 6 hours
         offlineFileLoader.setDataLoader(new CommonsDataLoader());
 
         DSSDocument xsltDoc;
@@ -194,6 +187,8 @@ public static Document getEformXmlFromXdcDocument(DSSDocument document) throws X
             throw new XMLValidationException("Zlyhala validácia XML Datacontainera", "Element XMLData je prázdny");
 
         var idk = xmlData.getFirstChild();
+        // In an indented XML, whitespaces between XMLData and its content are considered as text nodes.
+        // If there is a text node, validate it against all-whitespace regex and skip it.
         if (idk.getNodeType() == Node.TEXT_NODE) {
             if (!idk.getNodeValue().matches("\\s*"))
                 throw new XMLValidationException("Zlyhala validácia XML Datacontainera", "XMLData obsahuje neplatný text");

src/main/java/digital/slovensko/autogram/core/eforms/XDCValidator.java

@@ -31,7 +31,7 @@ public static boolean isXDCContent(DSSDocument document) {
 
             return validateXmlContentAgainstXsd(docString, new String(xdcSchema.readAllBytes(), ENCODING));
 
-        } catch (IOException | NullPointerException e) {
+        } catch (IOException | NullPointerException | XMLValidationException e) {
             return false;
         }
     }

src/test/java/digital/slovensko/autogram/SigningParametersTests.java

@@ -168,7 +168,24 @@ void testThrowsAutogramExceptionWithUnknownEformXml(DSSDocument document) {
     @ParameterizedTest
     @MethodSource("digital.slovensko.autogram.TestMethodSources#unknownEfomXmlProvider")
     void testThrowsAutogramExceptionWithUnknownEformXmlWithAutoLoadEform(DSSDocument document) {
-        Assertions.assertThrows(SigningParametersException.class,
+        Assertions.assertThrows(XMLValidationException.class,
+                () -> SigningParameters.buildFromRequest(SignatureLevel.XAdES_BASELINE_B, null, null, null, null,
+                        false, null, null, null, null, null, null, false, 800, true, document));
+    }
+
+    @ParameterizedTest
+    @MethodSource("digital.slovensko.autogram.TestMethodSources#mismatchedDigestsXmlProvider")
+    void testThrowsAutogramExceptionWithMismatchedDigestsXml(DSSDocument document) {
+        Assertions.assertThrows(XMLValidationException.class,
+                () -> SigningParameters.buildFromRequest(SignatureLevel.XAdES_BASELINE_B, asice, xdcXmlns, null, null,
+                        false, null, null, null, xsdSchema, xsltTransformation, identifier, false, 800, false,
+                        document));
+    }
+
+    @ParameterizedTest
+    @MethodSource("digital.slovensko.autogram.TestMethodSources#mismatchedDigestsXmlProvider")
+    void testThrowsAutogramExceptionWithMismatchedDigestsXmlWithAutoLoadEform(DSSDocument document) {
+        Assertions.assertThrows(XMLValidationException.class,
                 () -> SigningParameters.buildFromRequest(SignatureLevel.XAdES_BASELINE_B, null, null, null, null,
                         false, null, null, null, null, null, null, false, 800, true, document));
     }

src/test/java/digital/slovensko/autogram/TestMethodSources.java

@@ -153,6 +153,60 @@ public static Stream<InMemoryDocument> validXadesDocumentsProvider() throws IOEx
         );
     }
 
+    public static Stream<InMemoryDocument> xdcDocumentsProvider() throws IOException {
+        var generalAgendaXdcIndented = cls.getResourceAsStream("general_agenda_xdc_indented.xml").readAllBytes();
+        var generalAgendaXdc = cls.getResourceAsStream("general_agenda_xdc.xml").readAllBytes();
+        var mismatchedXsdGAXdc = cls.getResourceAsStream("mismatched_xsd_ga_xdc.xml").readAllBytes();
+        var mismatchedXsltGAXdc = cls.getResourceAsStream("mismatched_xslt_ga_xdc.xml").readAllBytes();
+        var unknownEfomXdc = cls.getResourceAsStream("unknown_eform_xdc.xml").readAllBytes();
+        var wrongSchemaGAXdc = cls.getResourceAsStream("wrong_schema_ga_xdc.xml").readAllBytes();
+
+        return Stream.of(
+            new InMemoryDocument(generalAgendaXdcIndented, "generalAgendaXdcIndented.xml", AutogramMimeType.XML_DATACONTAINER),
+            new InMemoryDocument(generalAgendaXdc, "generalAgendaXdc.xml", AutogramMimeType.XML_DATACONTAINER),
+            new InMemoryDocument(mismatchedXsdGAXdc, "mismatchedXsdGAXdc.xml", AutogramMimeType.XML_DATACONTAINER),
+            new InMemoryDocument(mismatchedXsltGAXdc, "mismatchedXsltGAXdc.xml", AutogramMimeType.XML_DATACONTAINER),
+            new InMemoryDocument(unknownEfomXdc, "unknownEfomXdc.xml", AutogramMimeType.XML_DATACONTAINER),
+            new InMemoryDocument(wrongSchemaGAXdc, "wrongSchemaGAXdc.xml", AutogramMimeType.XML_DATACONTAINER)
+        );
+    }
+
+    public static Stream<InMemoryDocument> xdcDocumentsWithXmlMimetypeProvider() throws IOException {
+        var generalAgendaXdcIndented = cls.getResourceAsStream("general_agenda_xdc_indented.xml").readAllBytes();
+        var generalAgendaXdc = cls.getResourceAsStream("general_agenda_xdc.xml").readAllBytes();
+        var mismatchedXsdGAXdc = cls.getResourceAsStream("mismatched_xsd_ga_xdc.xml").readAllBytes();
+        var mismatchedXsltGAXdc = cls.getResourceAsStream("mismatched_xslt_ga_xdc.xml").readAllBytes();
+        var unknownEfomXdc = cls.getResourceAsStream("unknown_eform_xdc.xml").readAllBytes();
+        var wrongSchemaGAXdc = cls.getResourceAsStream("wrong_schema_ga_xdc.xml").readAllBytes();
+
+        return Stream.of(
+            new InMemoryDocument(generalAgendaXdcIndented, "generalAgendaXdcIndented.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(generalAgendaXdc, "generalAgendaXdc.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(mismatchedXsdGAXdc, "mismatchedXsdGAXdc.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(mismatchedXsltGAXdc, "mismatchedXsltGAXdc.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(unknownEfomXdc, "unknownEfomXdc.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(wrongSchemaGAXdc, "wrongSchemaGAXdc.xml", MimeTypeEnum.XML)
+        );
+    }
+
+    public static Stream<InMemoryDocument> nonXdcXmlDocumentsProvider() throws IOException {
+        var documentContentNoUsedXSDReference = cls.getResourceAsStream("document-content-no-UsedXSDReference.xml").readAllBytes();
+        var documentContentUsedXSDReferenceNoAttributes = cls.getResourceAsStream("document-content-UsedXSDReference-no-attributes.xml").readAllBytes();
+        var documentContentUsedXSDReferenceNoDigestValue = cls.getResourceAsStream("document-content-UsedXSDReference-no-DigestValue.xml").readAllBytes();
+        var emptyXml = cls.getResourceAsStream("empty_xml.xml").readAllBytes();
+        var nonEformXml = cls.getResourceAsStream("non_eform.xml").readAllBytes();
+        var wrongSchemaXdcXml = cls.getResourceAsStream("wrong_schema_xdc.xml").readAllBytes();
+
+        return Stream.of(
+            new InMemoryDocument(documentContentNoUsedXSDReference, "documentContentNoUsedXSDReference.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(documentContentUsedXSDReferenceNoAttributes, "documentContentUsedXSDReferenceNoAttributes.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(documentContentUsedXSDReferenceNoDigestValue, "documentContentUsedXSDReferenceNoDigestValue.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(emptyXml, "emptyXml.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(nonEformXml, "nonEformXml.xml", MimeTypeEnum.XML),
+            new InMemoryDocument(wrongSchemaXdcXml, "wrongSchemaXdcXml.xml", MimeTypeEnum.XML)
+        );
+    }
+
     public static Stream<InMemoryDocument> validCadesDocumentsProvider() throws IOException {
         var samplePdfCadesAsice = cls.getResourceAsStream("sample_pdf_cades.asice").readAllBytes();
 

src/test/java/digital/slovensko/autogram/core/eforms/XDCValidatorTests.java

@@ -0,0 +1,22 @@
+package digital.slovensko.autogram.core.eforms;
+
+import org.junit.jupiter.api.Assertions;
+import org.junit.jupiter.params.ParameterizedTest;
+import org.junit.jupiter.params.provider.MethodSource;
+
+import eu.europa.esig.dss.model.DSSDocument;
+
+public class XDCValidatorTests {
+    @ParameterizedTest
+    @MethodSource({"digital.slovensko.autogram.TestMethodSources#xdcDocumentsProvider",
+            "digital.slovensko.autogram.TestMethodSources#xdcDocumentsWithXmlMimetypeProvider"})
+    void testReturnsTrueForAllXDCsRegerdlessOfMimeType(DSSDocument document) {
+        Assertions.assertTrue(XDCValidator.isXDCContent(document));
+    }
+
+    @ParameterizedTest
+    @MethodSource({"digital.slovensko.autogram.TestMethodSources#nonXdcXmlDocumentsProvider"})
+    void testReturnsFalseForAllNonXDCsRegerdlessOfMimeType(DSSDocument document) {
+        Assertions.assertFalse(XDCValidator.isXDCContent(document));
+    }
+}

src/test/resources/digital/slovensko/autogram/mismatched_xsd_ga_xdc.xml

@@ -8,7 +8,7 @@
         </GeneralAgenda>
     </xdc:XMLData>
     <xdc:UsedSchemasReferenced>
-        <xdc:UsedXSDReference DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="wrongValueMismatch" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xsd</xdc:UsedXSDReference>
+        <xdc:UsedXSDReference DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="Qo1jYX1JWydvM/OL/rnirphk1rM1z41fPRXBEgp/qbg=" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xsd</xdc:UsedXSDReference>
         <xdc:UsedPresentationSchemaReference ContentType="application/xslt+xml" DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="Qo1jYX1JWydvM/OL/rnirphk1rM1z41fPRXBEgp/qbg=" Language="sk" MediaDestinationTypeDescription="TXT" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xslt</xdc:UsedPresentationSchemaReference>
     </xdc:UsedSchemasReferenced>
 </xdc:XMLDataContainer>

src/test/resources/digital/slovensko/autogram/mismatched_xslt_ga_xdc.xml

@@ -9,6 +9,6 @@
     </xdc:XMLData>
     <xdc:UsedSchemasReferenced>
         <xdc:UsedXSDReference DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="/Ctn0B9D7HKn6URFR8iPUKfyGe4mBYpK+25dc1iYWuE=" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xsd</xdc:UsedXSDReference>
-        <xdc:UsedPresentationSchemaReference ContentType="application/xslt+xml" DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="wrongValueMismatch" Language="sk" MediaDestinationTypeDescription="TXT" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xslt</xdc:UsedPresentationSchemaReference>
+        <xdc:UsedPresentationSchemaReference ContentType="application/xslt+xml" DigestMethod="urn:oid:2.16.840.1.101.3.4.2.1" DigestValue="/Ctn0B9D7HKn6URFR8iPUKfyGe4mBYpK+25dc1iYWuE=" Language="sk" MediaDestinationTypeDescription="TXT" TransformAlgorithm="http://www.w3.org/TR/2001/REC-xml-c14n-20010315">http://schemas.gov.sk/form/App.GeneralAgenda/1.9/form.xslt</xdc:UsedPresentationSchemaReference>
     </xdc:UsedSchemasReferenced>
 </xdc:XMLDataContainer>