From 54485fa2520c9188f30a38413e7a7ef5cbf17752 Mon Sep 17 00:00:00 2001
From: Jan van Mansum <janvanmansum@gmail.com>
Date: Wed, 28 Aug 2024 15:49:35 +0200
Subject: [PATCH] Added examples with invalid terms and codes

---
 .../invalid/invalid-codes/README.md           |   5 +
 .../invalid/invalid-codes/bag-info.txt        |   4 +
 .../invalid/invalid-codes/bagit.txt           |   2 +
 .../invalid/invalid-codes/data/file1.txt      |   1 +
 .../invalid-codes/data/subdir/file2.txt       |   1 +
 .../invalid/invalid-codes/manifest-sha1.txt   |   2 +
 .../invalid-codes/metadata/dataset.xml        | 149 +++++++++++++
 .../invalid/invalid-codes/metadata/files.xml  |  27 +++
 .../invalid-codes/tagmanifest-sha1.txt        |   5 +
 .../invalid/invalid-terms/README.md           |   5 +
 .../invalid/invalid-terms/bag-info.txt        |   4 +
 .../invalid/invalid-terms/bagit.txt           |   2 +
 .../invalid/invalid-terms/data/file1.txt      |   1 +
 .../invalid-terms/data/subdir/file2.txt       |   1 +
 .../invalid/invalid-terms/manifest-sha1.txt   |   2 +
 .../invalid-terms/metadata/dataset.xml        | 197 ++++++++++++++++++
 .../invalid/invalid-terms/metadata/files.xml  |  27 +++
 .../invalid-terms/tagmanifest-sha1.txt        |   5 +
 18 files changed, 440 insertions(+)
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/README.md
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/bag-info.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/bagit.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/data/file1.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/data/subdir/file2.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/manifest-sha1.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/metadata/dataset.xml
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/metadata/files.xml
 create mode 100644 src/main/resources/example-bags/invalid/invalid-codes/tagmanifest-sha1.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/README.md
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/bag-info.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/bagit.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/data/file1.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/data/subdir/file2.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/manifest-sha1.txt
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/metadata/dataset.xml
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/metadata/files.xml
 create mode 100644 src/main/resources/example-bags/invalid/invalid-terms/tagmanifest-sha1.txt

diff --git a/src/main/resources/example-bags/invalid/invalid-codes/README.md b/src/main/resources/example-bags/invalid/invalid-codes/README.md
new file mode 100644
index 0000000..8ad3801
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/README.md
@@ -0,0 +1,5 @@
+all-mappings
+============
+
+This bags contains one or more examples of each [mapping rule](https://docs.google.com/spreadsheets/d/1G5YHSDg3a91nI9NgRjbz11iRFU9qgnNkde6K84j1NWI/).
+The codes in the comments reference the rule identifiers in the spreadsheet.
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/bag-info.txt b/src/main/resources/example-bags/invalid/invalid-codes/bag-info.txt
new file mode 100644
index 0000000..27fb5e7
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/bag-info.txt
@@ -0,0 +1,4 @@
+Payload-Oxum: 240.3
+Bagging-Date: 2022-10-23
+Bag-Size: 3.1 MB
+Has-Organizational-Identifier: TESTPREFIX:1234
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/bagit.txt b/src/main/resources/example-bags/invalid/invalid-codes/bagit.txt
new file mode 100644
index 0000000..c4aebb4
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/bagit.txt
@@ -0,0 +1,2 @@
+BagIt-Version: 0.97
+Tag-File-Character-Encoding: UTF-8
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/data/file1.txt b/src/main/resources/example-bags/invalid/invalid-codes/data/file1.txt
new file mode 100644
index 0000000..31beb17
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/data/file1.txt
@@ -0,0 +1 @@
+Test file in root of dataset
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/data/subdir/file2.txt b/src/main/resources/example-bags/invalid/invalid-codes/data/subdir/file2.txt
new file mode 100644
index 0000000..15528f4
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/data/subdir/file2.txt
@@ -0,0 +1 @@
+Test file in subdirectory.
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/manifest-sha1.txt b/src/main/resources/example-bags/invalid/invalid-codes/manifest-sha1.txt
new file mode 100644
index 0000000..6de22e0
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/manifest-sha1.txt
@@ -0,0 +1,2 @@
+a5c5c4051724b655863c517a15c56e45753c3e5a  data/file1.txt
+0d57a5bc9f5af7e8edcc90d64fd3c24dfc23e727  data/subdir/file2.txt
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/metadata/dataset.xml b/src/main/resources/example-bags/invalid/invalid-codes/metadata/dataset.xml
new file mode 100644
index 0000000..e612a0c
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/metadata/dataset.xml
@@ -0,0 +1,149 @@
+<ddm:DDM
+        xmlns:dc="http://purl.org/dc/elements/1.1/"
+        xmlns:dcx-dai="http://easy.dans.knaw.nl/schemas/dcx/dai/"
+        xmlns:dcx-gml="http://easy.dans.knaw.nl/schemas/dcx/gml/"
+        xmlns:ddm="http://schemas.dans.knaw.nl/dataset/ddm-v2/"
+        xmlns:dcterms="http://purl.org/dc/terms/"
+        xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+        xmlns:id-type="http://easy.dans.knaw.nl/schemas/vocab/identifier-type/">
+
+    <ddm:profile>
+        <dc:title>A bag containing invalid vocab codes</dc:title>
+        <dcterms:description>
+            Some of the codes specified for ABR are invalid
+        </dcterms:description>
+        <dcx-dai:creatorDetails>
+            <dcx-dai:author>
+                <dcx-dai:titles>Dr</dcx-dai:titles>
+                <dcx-dai:initials>I</dcx-dai:initials>
+                <dcx-dai:surname>Lastname</dcx-dai:surname>
+                <dcx-dai:ISNI>0000 0001 2281 955X</dcx-dai:ISNI>
+                <dcx-dai:ORCID>https://orcid.org/0000-0001-9183-9538</dcx-dai:ORCID>
+                <dcx-dai:organization>
+                    <dcx-dai:name xml:lang="en">Example Org</dcx-dai:name>
+                </dcx-dai:organization>
+            </dcx-dai:author>
+        </dcx-dai:creatorDetails>
+        <ddm:created>2024-08-28</ddm:created>
+        <ddm:available>2024-08-28</ddm:available>
+        <ddm:audience>D13400</ddm:audience>
+        <ddm:audience>D16300</ddm:audience>
+        <ddm:accessRights>OPEN_ACCESS</ddm:accessRights>
+        <ddm:personalData present="No"/>
+    </ddm:profile>
+    <ddm:dcmiMetadata>
+        <dcterms:rightsHolder>I Lastname</dcterms:rightsHolder>
+        <dcterms:abstract xml:lang="ka">Georgian</dcterms:abstract>
+
+
+        <!-- REL001 see profile -->
+
+        <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:reportNumber
+                subjectScheme="ABR Rapporten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/7a99aaba-c1e7-49a4-9dd8-d295dbcc870e"
+                valueCode="CECI.N.EST.PAS.UN.RAPPORT"
+                reportNo="123-A">
+            Not a report
+        </ddm:reportNumber>
+
+        <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:acquisitionMethod
+                subjectScheme="ABR verwervingswijzen"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/554ca1ec-3ed8-42d3-ae4b-47bcb848b238"
+                valueCode="NOT.A.VALID.AQC.METHOD">
+            Not a valid acquisition method
+        </ddm:acquisitionMethod>
+
+    <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:subject
+                subjectScheme="ABR Complextypen"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/e9546020-4b28-4819-b0c2-29e7c864c5c0"
+                valueCode="INVAL.COMPLEX.TYPE">
+            Not a valid complex type
+        </ddm:subject>
+
+        <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:subject
+                subjectScheme="ABR Artefacten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/22cbb070-6542-48f0-8afe-7d98d398cc0b"
+                valueCode="OBJECT.INVALID">
+            Not a valid artifact
+        </ddm:subject>
+
+        <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:temporal
+                subjectScheme="ABR Periodes"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/9b688754-1315-484b-9c89-8817e87c1e84"
+                valueCode="MIDDLE.EARTH.3RD.AGE">
+            Not a valid period
+        </ddm:temporal>
+
+        <!-- TS001 -->
+        <dcterms:temporal>Het Romeinse Rijk</dcterms:temporal>
+        <dcterms:temporal>De Oudheid</dcterms:temporal>
+
+        <!-- TS002 -->
+        <dcx-gml:spatial srsName="http://www.opengis.net/def/crs/EPSG/0/28992">
+            <Point xmlns="http://www.opengis.net/gml">
+                <pos>126466 529006</pos>
+            </Point>
+        </dcx-gml:spatial>
+
+        <!-- TS003, defaults to latitude/longitude-->
+        <dcx-gml:spatial>
+            <Point xmlns="http://www.opengis.net/gml">
+                <pos>52.078663 4.288788</pos>
+            </Point>
+        </dcx-gml:spatial>
+
+        <!-- TS004 -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="http://www.opengis.net/def/crs/EPSG/0/28992">
+                    <lowerCorner>102000 335000</lowerCorner>
+                    <upperCorner>140000 628000</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- TS005 -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="http://www.opengis.net/def/crs/EPSG/0/4326">
+                    <lowerCorner>51.46343658020442 3.5621054065986075</lowerCorner>
+                    <upperCorner>53.23074335194507 6.563118076315912</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- Not RD or lat/lon, so should be ignored -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="XXX">
+                    <lowerCorner>1 2</lowerCorner>
+                    <upperCorner>3 4</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- TS006, note that the text must exactly match the vocabulary term, so South-Africa will be handled by TS007 -->
+        <dcterms:spatial>South Africa</dcterms:spatial>
+        <dcterms:spatial>Japan</dcterms:spatial>
+
+        <!-- TS007, not in the controlled list, so mapped to the free text field -->
+        <dcterms:spatial>Roman Empire</dcterms:spatial>
+
+        <!--TRM001 -->
+        <dcterms:license xsi:type="dcterms:URI">http://opensource.org/licenses/MIT</dcterms:license>
+
+        <!-- TRM002 see example restricted-files-no-access-request -->
+
+        <!-- TRM005 and TRM006 -->
+        <dcterms:accessRights>Restricted files accessible under the following conditions: ...</dcterms:accessRights>
+
+        <!-- VLT005A -->
+        <dcterms:identifier xsi:type="id-type:DOI">10.17026/dans-z6y-5y2e</dcterms:identifier>
+
+    </ddm:dcmiMetadata>
+</ddm:DDM>
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/metadata/files.xml b/src/main/resources/example-bags/invalid/invalid-codes/metadata/files.xml
new file mode 100644
index 0000000..1455279
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/metadata/files.xml
@@ -0,0 +1,27 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<files xmlns="http://easy.dans.knaw.nl/schemas/bag/metadata/files/" xmlns:dcterms="http://purl.org/dc/terms/">
+    <!-- TRM004 applies because no accessibleToRights NONE is present and ddm:accessRights in dataset.xml is set to OPEN_ACCESS -->
+    <file filepath="data/file1.txt">
+        <!-- FIL004 -->
+        <dcterms:description>A file with a simple description</dcterms:description>
+    </file>
+    <file filepath="data/subdir/file2.txt">
+        <!-- TRM006 (see also TRM006 in dataset.xml) -->
+        <accessibleToRights>RESTRICTED_REQUEST</accessibleToRights>
+    </file>
+    <!-- FIL001, FIL002, FIL003 -->
+    <!-- This is tested in the example 'forbidden-chars-in-filepath', because it requires a file system that supports non-ascii characters in filenames.
+    <file filepath="data/subdir_υποφάκελο/c:a*q?d&quot;l&lt;g&gt;p|s;h#.txt">
+        <dcterms:description>A file with a problematic name</dcterms:description>
+    </file>
+    -->
+    <!-- FIL005, FIL006 see examples 'default-open' and 'default-restricted' -->
+
+    <!-- FIL008 see example 'embargoed' -->
+
+    <!--
+    An extra, unrestricted (FIL007), uncompressed zip file called original-metadata.zip with no embargo (FIL009) is created,
+    containing dataset.xml and files.xml.
+    -->
+</files>
+
diff --git a/src/main/resources/example-bags/invalid/invalid-codes/tagmanifest-sha1.txt b/src/main/resources/example-bags/invalid/invalid-codes/tagmanifest-sha1.txt
new file mode 100644
index 0000000..1e565af
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-codes/tagmanifest-sha1.txt
@@ -0,0 +1,5 @@
+7e9cb7c9bcb9402ca1b68fb9b80114a520427d65  metadata/dataset.xml
+33d507e3ced80f1b83f8e73ae780e870e159de9a  metadata/files.xml
+e2924b081506bac23f5fffe650ad1848a1c8ac1d  bagit.txt
+fcebcebafb062a34ecee8f475cfa81cb745fbed6  manifest-sha1.txt
+f1532a6307e82f76b7911304a6c16a4b15414e4b  README.md
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/README.md b/src/main/resources/example-bags/invalid/invalid-terms/README.md
new file mode 100644
index 0000000..8ad3801
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/README.md
@@ -0,0 +1,5 @@
+all-mappings
+============
+
+This bags contains one or more examples of each [mapping rule](https://docs.google.com/spreadsheets/d/1G5YHSDg3a91nI9NgRjbz11iRFU9qgnNkde6K84j1NWI/).
+The codes in the comments reference the rule identifiers in the spreadsheet.
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/bag-info.txt b/src/main/resources/example-bags/invalid/invalid-terms/bag-info.txt
new file mode 100644
index 0000000..27fb5e7
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/bag-info.txt
@@ -0,0 +1,4 @@
+Payload-Oxum: 240.3
+Bagging-Date: 2022-10-23
+Bag-Size: 3.1 MB
+Has-Organizational-Identifier: TESTPREFIX:1234
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/bagit.txt b/src/main/resources/example-bags/invalid/invalid-terms/bagit.txt
new file mode 100644
index 0000000..c4aebb4
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/bagit.txt
@@ -0,0 +1,2 @@
+BagIt-Version: 0.97
+Tag-File-Character-Encoding: UTF-8
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/data/file1.txt b/src/main/resources/example-bags/invalid/invalid-terms/data/file1.txt
new file mode 100644
index 0000000..31beb17
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/data/file1.txt
@@ -0,0 +1 @@
+Test file in root of dataset
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/data/subdir/file2.txt b/src/main/resources/example-bags/invalid/invalid-terms/data/subdir/file2.txt
new file mode 100644
index 0000000..15528f4
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/data/subdir/file2.txt
@@ -0,0 +1 @@
+Test file in subdirectory.
\ No newline at end of file
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/manifest-sha1.txt b/src/main/resources/example-bags/invalid/invalid-terms/manifest-sha1.txt
new file mode 100644
index 0000000..6de22e0
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/manifest-sha1.txt
@@ -0,0 +1,2 @@
+a5c5c4051724b655863c517a15c56e45753c3e5a  data/file1.txt
+0d57a5bc9f5af7e8edcc90d64fd3c24dfc23e727  data/subdir/file2.txt
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/metadata/dataset.xml b/src/main/resources/example-bags/invalid/invalid-terms/metadata/dataset.xml
new file mode 100644
index 0000000..e51a821
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/metadata/dataset.xml
@@ -0,0 +1,197 @@
+<ddm:DDM
+        xmlns:dc="http://purl.org/dc/elements/1.1/"
+        xmlns:dcx-dai="http://easy.dans.knaw.nl/schemas/dcx/dai/"
+        xmlns:dcx-gml="http://easy.dans.knaw.nl/schemas/dcx/gml/"
+        xmlns:ddm="http://schemas.dans.knaw.nl/dataset/ddm-v2/"
+        xmlns:dcterms="http://purl.org/dc/terms/"
+        xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+        xmlns:id-type="http://easy.dans.knaw.nl/schemas/vocab/identifier-type/">
+
+    <ddm:profile>
+        <dc:title>A bag containing invalid vocab terms</dc:title>
+        <dcterms:description>
+            Some of the terms specified for ABR and DANS Collections are invalid.
+        </dcterms:description>
+        <dcx-dai:creatorDetails>
+            <dcx-dai:author>
+                <dcx-dai:titles>Dr</dcx-dai:titles>
+                <dcx-dai:initials>I</dcx-dai:initials>
+                <dcx-dai:surname>Lastname</dcx-dai:surname>
+                <dcx-dai:ISNI>0000 0001 2281 955X</dcx-dai:ISNI>
+                <dcx-dai:ORCID>https://orcid.org/0000-0001-9183-9538</dcx-dai:ORCID>
+                <dcx-dai:organization>
+                    <dcx-dai:name xml:lang="en">Example Org</dcx-dai:name>
+                </dcx-dai:organization>
+            </dcx-dai:author>
+        </dcx-dai:creatorDetails>
+        <ddm:created>2024-08-28</ddm:created>
+        <ddm:available>2024-08-28</ddm:available>
+        <ddm:audience>D13400</ddm:audience>
+        <ddm:audience>D16300</ddm:audience>
+        <ddm:accessRights>OPEN_ACCESS</ddm:accessRights>
+        <ddm:personalData present="No"/>
+    </ddm:profile>
+    <ddm:dcmiMetadata>
+        <dcterms:rightsHolder>I Lastname</dcterms:rightsHolder>
+        <dcterms:abstract xml:lang="ka">Georgian</dcterms:abstract>
+
+
+        <!-- REL001 see profile -->
+
+        <!-- REL002 -->
+        <ddm:inCollection
+                subjectScheme="DANS Collection"
+                schemeURI="https://vocabularies.dans.knaw.nl/collections"
+                valueURI="https://vocabularies.dans.knaw.nl/collections/ssh/MADE-UP-COLLECTION"/>
+
+        <ddm:reportNumber
+                subjectScheme="ABR Rapporten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/7a99aaba-c1e7-49a4-9dd8-d295dbcc870e"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/abr/MADE-UP-REPORT"
+                reportNo="123-A">
+            Fantastic report
+        </ddm:reportNumber>
+
+        <!-- Legacy support: selection by "hasCode" attribute instead of the term URI -->
+        <ddm:reportNumber
+                subjectScheme="ABR Rapporten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/7a99aaba-c1e7-49a4-9dd8-d295dbcc870e"
+                valueCode="CECI.N.EST.PAS.UN.RAPPORT"
+                reportNo="123-A">
+            Not a report
+        </ddm:reportNumber>
+
+        <!-- AR005 -->
+        <!-- Preferred: selection by term URI -->
+        <ddm:acquisitionMethod
+                subjectScheme="ABR verwervingswijzen"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/554ca1ec-3ed8-42d3-ae4b-47bcb848b238"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/abr/MADE-UP-ACQUISITION-METHOD">
+            Not a valid acquisition method
+        </ddm:acquisitionMethod>
+
+
+
+        <!-- AR006 -->
+        <!-- Preferred: selection by term URI -->
+        <ddm:subject
+                subjectScheme="ABR Complextypen"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/e9546020-4b28-4819-b0c2-29e7c864c5c0"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/abr/MADE-UP-COMPLEXTYPE">
+            Not a valid complex type
+        </ddm:subject>
+
+        <!-- Preferred: selection by term URI -->
+        <ddm:subject
+                subjectScheme="ABR Complextypen"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/e9546020-4b28-4819-b0c2-29e7c864c5c0"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/rn/60bc20cd-0010-48ea-8d6f-42d333bfb39d">
+            Not a valid complex type
+        </ddm:subject>
+
+
+        <!-- AR007 -->
+        <!-- Preferred: selection by term URI -->
+        <ddm:subject
+                subjectScheme="ABR Artefacten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/22cbb070-6542-48f0-8afe-7d98d398cc0b"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/abr/MADE-UP-ARTIFACT">
+            Not a valid artifact
+        </ddm:subject>
+
+
+        <!-- Legacy term URI (with "rn") will be converted to new term URI-->
+        <ddm:subject
+                subjectScheme="ABR Artefacten"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/22cbb070-6542-48f0-8afe-7d98d398cc0b"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/rn/MADE-UP-ARTIFACT">
+            Not a valid artifact
+        </ddm:subject>
+
+        <!-- AR007A -->
+        <!-- Legacy subjectScheme/schemeURI for overall ABR will be taken as artifact -->
+        <ddm:subject
+                subjectScheme="Archeologisch Basis Register"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/rn/a4a7933c-e096-4bcf-a921-4f70a78749fe"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/rn/MADE-UP-ARTIFACT">
+            Not a valid artifact
+        </ddm:subject>
+
+        <!-- AR008 -->
+        <!-- Preferred: selection by term URI -->
+        <ddm:temporal
+                subjectScheme="ABR Periodes"
+                schemeURI="https://data.cultureelerfgoed.nl/term/id/abr/9b688754-1315-484b-9c89-8817e87c1e84"
+                valueURI="https://data.cultureelerfgoed.nl/term/id/abr/MADE-UP-PERIOD">
+            Not a valid period
+        </ddm:temporal>
+
+
+        <!-- TS001 -->
+        <dcterms:temporal>Het Romeinse Rijk</dcterms:temporal>
+        <dcterms:temporal>De Oudheid</dcterms:temporal>
+
+        <!-- TS002 -->
+        <dcx-gml:spatial srsName="http://www.opengis.net/def/crs/EPSG/0/28992">
+            <Point xmlns="http://www.opengis.net/gml">
+                <pos>126466 529006</pos>
+            </Point>
+        </dcx-gml:spatial>
+
+        <!-- TS003, defaults to latitude/longitude-->
+        <dcx-gml:spatial>
+            <Point xmlns="http://www.opengis.net/gml">
+                <pos>52.078663 4.288788</pos>
+            </Point>
+        </dcx-gml:spatial>
+
+        <!-- TS004 -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="http://www.opengis.net/def/crs/EPSG/0/28992">
+                    <lowerCorner>102000 335000</lowerCorner>
+                    <upperCorner>140000 628000</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- TS005 -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="http://www.opengis.net/def/crs/EPSG/0/4326">
+                    <lowerCorner>51.46343658020442 3.5621054065986075</lowerCorner>
+                    <upperCorner>53.23074335194507 6.563118076315912</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- Not RD or lat/lon, so should be ignored -->
+        <dcx-gml:spatial>
+            <boundedBy xmlns="http://www.opengis.net/gml">
+                <Envelope srsName="XXX">
+                    <lowerCorner>1 2</lowerCorner>
+                    <upperCorner>3 4</upperCorner>
+                </Envelope>
+            </boundedBy>
+        </dcx-gml:spatial>
+
+        <!-- TS006, note that the text must exactly match the vocabulary term, so South-Africa will be handled by TS007 -->
+        <dcterms:spatial>South Africa</dcterms:spatial>
+        <dcterms:spatial>Japan</dcterms:spatial>
+
+        <!-- TS007, not in the controlled list, so mapped to the free text field -->
+        <dcterms:spatial>Roman Empire</dcterms:spatial>
+
+        <!--TRM001 -->
+        <dcterms:license xsi:type="dcterms:URI">http://opensource.org/licenses/MIT</dcterms:license>
+
+        <!-- TRM002 see example restricted-files-no-access-request -->
+
+        <!-- TRM005 and TRM006 -->
+        <dcterms:accessRights>Restricted files accessible under the following conditions: ...</dcterms:accessRights>
+
+        <!-- VLT005A -->
+        <dcterms:identifier xsi:type="id-type:DOI">10.17026/dans-z6y-5y2e</dcterms:identifier>
+
+    </ddm:dcmiMetadata>
+</ddm:DDM>
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/metadata/files.xml b/src/main/resources/example-bags/invalid/invalid-terms/metadata/files.xml
new file mode 100644
index 0000000..1455279
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/metadata/files.xml
@@ -0,0 +1,27 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<files xmlns="http://easy.dans.knaw.nl/schemas/bag/metadata/files/" xmlns:dcterms="http://purl.org/dc/terms/">
+    <!-- TRM004 applies because no accessibleToRights NONE is present and ddm:accessRights in dataset.xml is set to OPEN_ACCESS -->
+    <file filepath="data/file1.txt">
+        <!-- FIL004 -->
+        <dcterms:description>A file with a simple description</dcterms:description>
+    </file>
+    <file filepath="data/subdir/file2.txt">
+        <!-- TRM006 (see also TRM006 in dataset.xml) -->
+        <accessibleToRights>RESTRICTED_REQUEST</accessibleToRights>
+    </file>
+    <!-- FIL001, FIL002, FIL003 -->
+    <!-- This is tested in the example 'forbidden-chars-in-filepath', because it requires a file system that supports non-ascii characters in filenames.
+    <file filepath="data/subdir_υποφάκελο/c:a*q?d&quot;l&lt;g&gt;p|s;h#.txt">
+        <dcterms:description>A file with a problematic name</dcterms:description>
+    </file>
+    -->
+    <!-- FIL005, FIL006 see examples 'default-open' and 'default-restricted' -->
+
+    <!-- FIL008 see example 'embargoed' -->
+
+    <!--
+    An extra, unrestricted (FIL007), uncompressed zip file called original-metadata.zip with no embargo (FIL009) is created,
+    containing dataset.xml and files.xml.
+    -->
+</files>
+
diff --git a/src/main/resources/example-bags/invalid/invalid-terms/tagmanifest-sha1.txt b/src/main/resources/example-bags/invalid/invalid-terms/tagmanifest-sha1.txt
new file mode 100644
index 0000000..77c4a81
--- /dev/null
+++ b/src/main/resources/example-bags/invalid/invalid-terms/tagmanifest-sha1.txt
@@ -0,0 +1,5 @@
+3f8d298632083b92c86a7442b91244afcafa1add  metadata/dataset.xml
+33d507e3ced80f1b83f8e73ae780e870e159de9a  metadata/files.xml
+e2924b081506bac23f5fffe650ad1848a1c8ac1d  bagit.txt
+fcebcebafb062a34ecee8f475cfa81cb745fbed6  manifest-sha1.txt
+f1532a6307e82f76b7911304a6c16a4b15414e4b  README.md