use provenance logs to find head if no anchor is provided

bio-guoda · Aug 22, 2024 · a19951a · a19951a
1 parent 087fa29
commit a19951a
Show file tree

Hide file tree

Showing 6 changed files with 73 additions and 4 deletions.
diff --git a/preston-cmd/src/main/java/bio/guoda/preston/cmd/AnchorUtil.java b/preston-cmd/src/main/java/bio/guoda/preston/cmd/AnchorUtil.java
@@ -54,14 +54,14 @@ private static void findHead(Persisting persisting, AtomicReference<IRI> head, P
         }
     }
 
-    public static AtomicReference<IRI> findHeadOrThrow(Persisting persisting) {
+    public static IRI findHeadOrThrow(Persisting persisting) {
         AtomicReference<IRI> head = findHead(persisting);
 
         if (head.get() == null) {
             throw new RuntimeException("Cannot find most recent version: no provenance logs found.");
         }
 
-        return head;
+        return head.get();
     }
 
 }
diff --git a/preston-cmd/src/main/java/bio/guoda/preston/cmd/BlobStoreUtil.java b/preston-cmd/src/main/java/bio/guoda/preston/cmd/BlobStoreUtil.java
@@ -29,6 +29,7 @@
 import java.util.Map;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicLong;
+import java.util.concurrent.atomic.AtomicReference;
 
 public class BlobStoreUtil {
     private static final Logger LOG = LoggerFactory.getLogger(BlobStoreUtil.class);
@@ -71,8 +72,9 @@ private static Map<String, String> buildIndexedBlobStore(Persisting persisting)
         File tmpDir = persisting.getTmpDir();
         IRI provenanceAnchor = persisting.getProvenanceAnchor();
         if (CmdWithProvenance.PROVENANCE_ANCHOR_DEFAULT.equals(provenanceAnchor)) {
-            throw new IllegalArgumentException("--anchor provenance anchor not set; please set provenance anchor");
+            provenanceAnchor = AnchorUtil.findHeadOrThrow(persisting);
         }
+
         // indexing
         DBMaker maker = newTmpFileDB(tmpDir);
         Map<String, String> treeMap = maker

diff --git a/preston-cmd/src/test/java/bio/guoda/preston/cmd/BlobStoreUtilTest.java b/preston-cmd/src/test/java/bio/guoda/preston/cmd/BlobStoreUtilTest.java
@@ -37,6 +37,34 @@ public void indexedBlobStore() throws IOException, URISyntaxException {
         assertThat(IOUtils.toString(inputStream, StandardCharsets.UTF_8), Is.is("foo"));
     }
 
+    @Test
+    public void indexedBlobStoreWithoutProvenanceAnchor() throws IOException, URISyntaxException {
+        File dataDir = getDataDir();
+
+        Persisting persisting = getPersisting(dataDir);
+        persisting.setProvenanceArchor(CmdWithProvenance.PROVENANCE_ANCHOR_DEFAULT);
+        BlobStoreReadOnly blobStoreIndexed = BlobStoreUtil.createIndexedBlobStoreFor(getBlobStore(), persisting);
+
+        InputStream inputStream = blobStoreIndexed.get(RefNodeFactory.toIRI("https://example.org"));
+
+        assertThat(IOUtils.toString(inputStream, StandardCharsets.UTF_8), Is.is("foo"));
+    }
+
+    @Test(expected = RuntimeException.class)
+    public void indexedBlobStoreWithoutProvenanceAnchorNoProvenanceIndex() throws IOException, URISyntaxException {
+        File dataDir = getDataDir("index-data-no-provenance/d3/b0/d3b07384d113edec49eaa6238ad5ff00");
+
+        Persisting persisting = getPersisting(dataDir);
+        persisting.setProvenanceArchor(CmdWithProvenance.PROVENANCE_ANCHOR_DEFAULT);
+
+        try {
+            BlobStoreUtil.createIndexedBlobStoreFor(getBlobStore(), persisting);
+        } catch(RuntimeException ex) {
+            assertThat(ex.getMessage(), Is.is("Cannot find most recent version: no provenance logs found."));
+            throw ex;
+        }
+    }
+
     @Test
     public void resolvingBlobStoreWithProvenanceAnchor() throws IOException, URISyntaxException {
         File dataDir = getDataDir();
@@ -87,7 +115,12 @@ private Persisting getPersisting(File dataDir) throws IOException {
 
 
     private File getDataDir() throws URISyntaxException {
-        URL resource = getClass().getResource("index-data/27/f5/27f552c25bc733d05a5cc67e9ba63850");
+        String s = "index-data/27/f5/27f552c25bc733d05a5cc67e9ba63850";
+        return getDataDir(s);
+    }
+
+    private File getDataDir(String s) throws URISyntaxException {
+        URL resource = getClass().getResource(s);
         File root = new File(resource.toURI());
         return root.getParentFile().getParentFile().getParentFile();
     }

diff --git a/...ces/bio/guoda/preston/cmd/index-data-no-provenance/d3/b0/d3b07384d113edec49eaa6238ad5ff00 b/...ces/bio/guoda/preston/cmd/index-data-no-provenance/d3/b0/d3b07384d113edec49eaa6238ad5ff00
@@ -0,0 +1 @@
+foo
diff --git a/...ces/bio/guoda/preston/cmd/index-data-no-provenance/d8/ae/d8ae3c306edbac7a0d598ad0bc711a9a b/...ces/bio/guoda/preston/cmd/index-data-no-provenance/d8/ae/d8ae3c306edbac7a0d598ad0bc711a9a
@@ -0,0 +1,19 @@
+<https://preston.guoda.bio> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#SoftwareAgent> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<https://preston.guoda.bio> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Agent> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<https://preston.guoda.bio> <http://purl.org/dc/terms/description> "Preston is a software program that finds, archives and provides access to biodiversity datasets."@en <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Activity> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <http://purl.org/dc/terms/description> "A crawl event that discovers biodiversity archives."@en <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <http://www.w3.org/ns/prov#startedAtTime> "2024-08-22T13:19:17.894Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <http://www.w3.org/ns/prov#wasStartedBy> <https://preston.guoda.bio> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<https://doi.org/10.5281/zenodo.1410543> <http://www.w3.org/ns/prov#usedBy> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<https://doi.org/10.5281/zenodo.1410543> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://purl.org/dc/dcmitype/Software> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<https://doi.org/10.5281/zenodo.1410543> <http://purl.org/dc/terms/bibliographicCitation> "Jorrit Poelen, Icaro Alzuru, & Michael Elliott. 2018-2024. Preston: a biodiversity dataset tracker (Version 0.8.7-SNAPSHOT) [Software]. Zenodo. https://doi.org/10.5281/zenodo.1410543"@en <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:0659a54f-b713-4f86-a917-5be166a14110> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Entity> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<urn:uuid:0659a54f-b713-4f86-a917-5be166a14110> <http://purl.org/dc/terms/description> "A biodiversity dataset graph archive."@en <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> .
+<hash://md5/d3b07384d113edec49eaa6238ad5ff00> <http://www.w3.org/ns/prov#wasGeneratedBy> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<hash://md5/d3b07384d113edec49eaa6238ad5ff00> <http://www.w3.org/ns/prov#qualifiedGeneration> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <http://www.w3.org/ns/prov#generatedAtTime> "2024-08-22T13:19:17.988Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Generation> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <http://www.w3.org/ns/prov#wasInformedBy> <urn:uuid:4d4df7fa-e890-458c-ae52-c58926bd5b67> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> <http://www.w3.org/ns/prov#used> <urn:uuid:c5b68b8b-ad93-4de7-85e3-4f325b24712c> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
+<urn:uuid:c5b68b8b-ad93-4de7-85e3-4f325b24712c> <http://purl.org/pav/hasVersion> <hash://md5/d3b07384d113edec49eaa6238ad5ff00> <urn:uuid:8bcd4aeb-af38-4840-acf3-a829668c0757> .
diff --git a/...ces/bio/guoda/preston/cmd/index-data-no-provenance/ec/99/ec998a9c63a64ac7bfef04c91ee84f16 b/...ces/bio/guoda/preston/cmd/index-data-no-provenance/ec/99/ec998a9c63a64ac7bfef04c91ee84f16
@@ -0,0 +1,14 @@
+<https://preston.guoda.bio> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#SoftwareAgent> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://preston.guoda.bio> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Agent> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://preston.guoda.bio> <http://purl.org/dc/terms/description> "Preston is a software program that finds, archives and provides access to biodiversity datasets."@en <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Activity> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <http://purl.org/dc/terms/description> "An activity that assigns an alias to a content hash"@en <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <http://www.w3.org/ns/prov#startedAtTime> "2024-08-22T13:19:49.902Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <http://www.w3.org/ns/prov#wasStartedBy> <https://preston.guoda.bio> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://doi.org/10.5281/zenodo.1410543> <http://www.w3.org/ns/prov#usedBy> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://doi.org/10.5281/zenodo.1410543> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://purl.org/dc/dcmitype/Software> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://doi.org/10.5281/zenodo.1410543> <http://purl.org/dc/terms/bibliographicCitation> "Jorrit Poelen, Icaro Alzuru, & Michael Elliott. 2018-2024. Preston: a biodiversity dataset tracker (Version 0.8.7-SNAPSHOT) [Software]. Zenodo. https://doi.org/10.5281/zenodo.1410543"@en <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:0659a54f-b713-4f86-a917-5be166a14110> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Entity> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<urn:uuid:0659a54f-b713-4f86-a917-5be166a14110> <http://purl.org/dc/terms/description> "A biodiversity dataset graph archive."@en <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<hash://md5/d8ae3c306edbac7a0d598ad0bc711a9a> <http://www.w3.org/ns/prov#usedBy> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .
+<https://example.org> <http://purl.org/pav/hasVersion> <hash://md5/d3b07384d113edec49eaa6238ad5ff00> <urn:uuid:cfcbb9c0-5adf-48a1-84b6-18ed348d89f0> .