24522677438

Committed 16 Apr 2026 04:48PM UTC coverage: 64.192% (-3.9%) from 68.052%

Build # 24522677438

Build Type

push

github

Committed by

web-flow

Commit Message

Merge pull request #66 from knowledgepixels/feature/65-trust-state-repos

feat: mirror registry trust state into the trust repo (#65)

Coverage Stats

281 of 488 branches covered (57.58%)

Branch coverage included in aggregate %.

800 of 1196 relevant lines covered (66.89%)

9.54 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

30.48

src/main/java/com/knowledgepixels/query/JellyNanopubLoader.java

package com.knowledgepixels.query;

import java.io.IOException;
import java.util.concurrent.atomic.AtomicLong;

import org.apache.http.client.methods.CloseableHttpResponse;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.client.methods.HttpHead;
import org.apache.http.impl.client.CloseableHttpClient;
import org.apache.http.impl.client.HttpClientBuilder;
import org.apache.http.util.EntityUtils;
import org.nanopub.NanopubUtils;
import org.nanopub.jelly.NanopubStream;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

/**
 * Loads nanopubs from the attached Nanopub Registry via a restartable Jelly stream.
 */
public class JellyNanopubLoader {
    static final String registryUrl;
    private static long lastCommittedCounter = -1;
    private static Long lastKnownSetupId = null;
    // Latest registry metadata fields, updated on each metadata fetch and forwarded to clients
    static volatile String lastCoverageTypes = null;
    static volatile String lastCoverageAgents = null;
    static volatile String lastTestInstance = null;
    static volatile String lastNanopubCount = null;
    private static final CloseableHttpClient metadataClient;
    private static final CloseableHttpClient jellyStreamClient;

    private static final int MAX_RETRIES_METADATA = 10;
    private static final int RETRY_DELAY_METADATA = 3000;
    private static final int RETRY_DELAY_JELLY = 5000;

    private static final Logger log = LoggerFactory.getLogger(JellyNanopubLoader.class);

    /**
     * Registry metadata returned by a HEAD request.
     */
    record RegistryMetadata(long loadCounter, Long setupId, String coverageTypes,
                            String coverageAgents, String testInstance, String nanopubCount,
                            String trustStateHash) {}

    /**
     * The interval in milliseconds at which the updates loader should poll for new nanopubs.
     */
    public static final int UPDATES_POLL_INTERVAL = 2000;

    enum LoadingType {
        INITIAL,
        UPDATE,
    }

    static {
        // Initialize registryUrl
        var url = Utils.getEnvString(
                "REGISTRY_FIXED_URL", "https://registry.knowledgepixels.com/"
        );
        if (!url.endsWith("/")) url += "/";
        registryUrl = url;

        metadataClient = HttpClientBuilder.create().setDefaultRequestConfig(Utils.getHttpRequestConfig()).build();
        jellyStreamClient = NanopubUtils.getHttpClient();
    }

    /**
     * Start or continue (after restart) the initial loading procedure. This simply loads all
     * nanopubs from the attached Registry.
     *
     * @param afterCounter which counter to start from (-1 for the beginning)
     */
    public static void loadInitial(long afterCounter) {
        RegistryMetadata metadata = fetchRegistryMetadata();
        updateForwardingMetadata(metadata);
        TrustStateLoader.maybeUpdate(metadata.trustStateHash());
        long targetCounter = metadata.loadCounter();
        log.info("Fetched Registry load counter: {}", targetCounter);
        // Store setupId on initial load
        if (metadata.setupId() != null && lastKnownSetupId == null) {
            lastKnownSetupId = metadata.setupId();
            StatusController.get().setRegistrySetupId(metadata.setupId());
        }
        lastCommittedCounter = afterCounter;
        while (lastCommittedCounter < targetCounter) {
            try {
                loadBatch(lastCommittedCounter, LoadingType.INITIAL);
                log.info("Initial load: loaded batch up to counter {}", lastCommittedCounter);
            } catch (Exception e) {
                log.info("Failed to load batch starting from counter {}", lastCommittedCounter);
                log.info("Failure reason: ", e);
                try {
                    Thread.sleep(RETRY_DELAY_JELLY);
                } catch (InterruptedException e2) {
                    throw new RuntimeException("Interrupted while waiting to retry loading batch.");
                }
            }
        }
        log.info("Initial load complete.");
    }

    /**
     * Check if the Registry has any new nanopubs. If it does, load them.
     * This method should be called periodically, and you should wait for it to finish before
     * calling it again.
     */
    public static void loadUpdates() {
        try {
            final var status = StatusController.get().getState();
            lastCommittedCounter = status.loadCounter;
            RegistryMetadata metadata = fetchRegistryMetadata();
            updateForwardingMetadata(metadata);
            TrustStateLoader.maybeUpdate(metadata.trustStateHash());
            long targetCounter = metadata.loadCounter();
            Long currentSetupId = metadata.setupId();

            // Detect reset via setupId change
            if (lastKnownSetupId != null && currentSetupId != null
                    && !lastKnownSetupId.equals(currentSetupId)) {
                log.warn("Registry reset detected: setupId {} -> {}", lastKnownSetupId, currentSetupId);
                performResync(currentSetupId);
                return;
            }
            // Detect reset via counter decrease (also covers first run after upgrade
            // where no setupId was persisted yet but the registry has already been reset)
            if (lastCommittedCounter > 0 && targetCounter >= 0
                    && targetCounter < lastCommittedCounter) {
                log.warn("Registry counter decreased {} -> {}, triggering resync",
                        lastCommittedCounter, targetCounter);
                performResync(currentSetupId);
                return;
            }

            // Update lastKnownSetupId on first successful poll
            if (currentSetupId != null && lastKnownSetupId == null) {
                if (lastCommittedCounter > 0) {
                    // Upgrade from a version without setupId tracking. The DB has data but
                    // we can't verify it matches the current registry. Force a resync.
                    log.warn("No stored setupId but DB has data (counter: {}). "
                            + "Forcing resync to ensure data consistency.", lastCommittedCounter);
                    performResync(currentSetupId);
                    return;
                }
                lastKnownSetupId = currentSetupId;
                StatusController.get().setRegistrySetupId(currentSetupId);
            }

            StatusController.get().setLoadingUpdates(status.loadCounter);
            if (lastCommittedCounter >= targetCounter) {
                StatusController.get().setReady();
                return;
            }
            loadBatch(lastCommittedCounter, LoadingType.UPDATE);
            log.info("Loaded {} update(s). Counter: {}, target was: {}",
                    lastCommittedCounter - status.loadCounter, lastCommittedCounter, targetCounter);
            if (lastCommittedCounter < targetCounter) {
                log.info("Warning: expected to load nanopubs up to (inclusive) counter " +
                        targetCounter + " based on the counter reported in Registry's headers, " +
                        "but loaded only up to {}.", lastCommittedCounter);
            }
        } catch (Exception e) {
            log.info("Failed to load updates. Current counter: {}", lastCommittedCounter);
            log.info("Failure Reason: ", e);
        } finally {
            try {
                StatusController.get().setReady();
            } catch (Exception e) {
                log.info("Update loader: failed to set status to READY.");
                log.info("Failure Reason: ", e);
            }
        }
    }

    /**
     * Re-stream all nanopubs from the registry after a reset is detected.
     * Existing nanopubs are skipped by NanopubLoader's per-repo dedup.
     *
     * @param newSetupId the new setup ID from the registry, or null if unknown
     */
    private static void performResync(Long newSetupId) {
        log.warn("Starting resync with registry. New setupId: {}", newSetupId);
        StatusController.get().setResetting();
        lastKnownSetupId = newSetupId;
        if (newSetupId != null) {
            StatusController.get().setRegistrySetupId(newSetupId);
        }
        StatusController.get().setLoadingInitial(-1);
        loadInitial(-1);
        StatusController.get().setReady();
        log.warn("Resync complete. Counter: {}", lastCommittedCounter);
    }

    /**
     * Load a batch of nanopubs from the Jelly stream.
     * <p>
     * The method requests the list of all nanopubs from the Registry and reads it for as long
     * as it can. If the stream is interrupted, the method will throw an exception, and you
     * can resume loading from the last known counter.
     *
     * @param afterCounter the last known nanopub counter to have been committed in the DB
     * @param type         the type of loading operation (initial or update)
     */
    static void loadBatch(long afterCounter, LoadingType type) {
        CloseableHttpResponse response;
        try {
            var request = new HttpGet(makeStreamFetchUrl(afterCounter));
            response = jellyStreamClient.execute(request);
        } catch (IOException e) {
            throw new RuntimeException("Failed to fetch Jelly stream from the Registry (I/O error).", e);
        }

        int httpStatus = response.getStatusLine().getStatusCode();
        if (httpStatus < 200 || httpStatus >= 300) {
            EntityUtils.consumeQuietly(response.getEntity());
            throw new RuntimeException("Jelly stream HTTP status is not 2xx: " + httpStatus + ".");
        }

        try (
                var is = response.getEntity().getContent();
                var npStream = NanopubStream.fromByteStream(is).getAsNanopubs()
        ) {
            AtomicLong checkpointTime = new AtomicLong(System.currentTimeMillis());
            AtomicLong checkpointCounter = new AtomicLong(lastCommittedCounter);
            AtomicLong lastSavedCounter = new AtomicLong(lastCommittedCounter);
            AtomicLong loaded = new AtomicLong(0L);

            npStream.forEach(m -> {
                if (!m.isSuccess()) throw new RuntimeException("Failed to load " +
                        "nanopub from Jelly stream. Last known counter: " + lastCommittedCounter,
                        m.getException()
                );
                if (m.getCounter() < lastCommittedCounter) {
                    throw new RuntimeException("Received a nanopub with a counter lower than " +
                            "the last known counter. Last known counter: " + lastCommittedCounter +
                            ", received counter: " + m.getCounter());
                }
                NanopubLoader.load(m.getNanopub(), m.getCounter());
                if (m.getCounter() % 10 == 0) {
                    // Save the committed counter only every 10 nanopubs to reduce DB load
                    saveCommittedCounter(type);
                    lastSavedCounter.set(m.getCounter());
                }
                lastCommittedCounter = m.getCounter();
                loaded.getAndIncrement();

                if (loaded.get() % 50 == 0) {
                    long currTime = System.currentTimeMillis();
                    double speed = 50 / ((currTime - checkpointTime.get()) / 1000.0);
                    log.info("Loading speed: " + String.format("%.2f", speed) +
                            " np/s. Counter: " + lastCommittedCounter);
                    checkpointTime.set(currTime);
                    checkpointCounter.set(lastCommittedCounter);
                }
            });
            // Make sure to save the last committed counter at the end of the batch
            if (lastCommittedCounter >= lastSavedCounter.get()) {
                saveCommittedCounter(type);
            }
        } catch (IOException e) {
            throw new RuntimeException("I/O error while reading the response Jelly stream.", e);
        } finally {
            try {
                response.close();
            } catch (IOException e) {
                log.info("Failed to close the Jelly stream response.");
            }
        }
    }

    /**
     * Save the last committed counter to the DB. Do this every N nanopubs to reduce DB load.
     * Remember to call this method at the end of the batch as well.
     *
     * @param type the type of loading operation (initial or update)
     */
    private static void saveCommittedCounter(LoadingType type) {
        try {
            if (type == LoadingType.INITIAL) {
                StatusController.get().setLoadingInitial(lastCommittedCounter);
            } else {
                StatusController.get().setLoadingUpdates(lastCommittedCounter);
            }
        } catch (Exception e) {
            throw new RuntimeException("Could not update the nanopub counter in DB", e);
        }
    }

    /**
     * Set the last known setup ID. Called from MainVerticle on startup to restore persisted state.
     *
     * @param setupId the setup ID to set, or null if not known
     */
    static void setLastKnownSetupId(Long setupId) {
        lastKnownSetupId = setupId;
    }

    /**
     * Update the cached metadata fields used for forwarding to clients.
     */
    private static void updateForwardingMetadata(RegistryMetadata metadata) {
        lastCoverageTypes = metadata.coverageTypes();
        lastCoverageAgents = metadata.coverageAgents();
        lastTestInstance = metadata.testInstance();
        lastNanopubCount = metadata.nanopubCount();
    }

    /**
     * Run a HEAD request to the Registry to fetch its current metadata (load counter and setup ID).
     *
     * @return the registry metadata
     */
    static RegistryMetadata fetchRegistryMetadata() {
        int tries = 0;
        RegistryMetadata metadata = null;
        while (metadata == null && tries < MAX_RETRIES_METADATA) {
            try {
                metadata = fetchRegistryMetadataInner();
            } catch (Exception e) {
                tries++;
                log.info("Failed to fetch registry metadata, try " + tries +
                        ". Retrying in {}ms...", RETRY_DELAY_METADATA);
                log.info("Failure Reason: ", e);
                try {
                    Thread.sleep(RETRY_DELAY_METADATA);
                } catch (InterruptedException e2) {
                    throw new RuntimeException(
                            "Interrupted while waiting to retry fetching registry metadata.");
                }
            }
        }
        if (metadata == null) {
            throw new RuntimeException("Failed to fetch registry metadata after " +
                    MAX_RETRIES_METADATA + " retries.");
        }
        return metadata;
    }

    /**
     * Inner logic for fetching the registry metadata via HEAD request.
     *
     * @return the registry metadata (load counter and setup ID)
     * @throws IOException if the HTTP request fails
     */
    private static RegistryMetadata fetchRegistryMetadataInner() throws IOException {
        var request = new HttpHead(registryUrl);
        try (var response = metadataClient.execute(request)) {
            int status = response.getStatusLine().getStatusCode();
            EntityUtils.consumeQuietly(response.getEntity());
            if (status < 200 || status >= 300) {
                throw new RuntimeException("Registry metadata HTTP status is not 2xx: " +
                        status + ".");
            }

            // Check if the registry is ready
            var hStatus = response.getHeaders("Nanopub-Registry-Status");
            if (hStatus.length == 0) {
                throw new RuntimeException("Registry did not return a Nanopub-Registry-Status header.");
            }
            if (!"ready".equals(hStatus[0].getValue()) && !"updating".equals(hStatus[0].getValue())) {
                throw new RuntimeException("Registry is not in ready state.");
            }

            // Get the load counter
            var hCounter = response.getHeaders("Nanopub-Registry-Load-Counter");
            if (hCounter.length == 0) {
                throw new RuntimeException("Registry did not return a Nanopub-Registry-Load-Counter header.");
            }
            long loadCounter = Long.parseLong(hCounter[0].getValue());

            // Get the setup ID (optional — older registries may not have it)
            Long setupId = null;
            var hSetupId = response.getHeaders("Nanopub-Registry-Setup-Id");
            if (hSetupId.length > 0) {
                try {
                    setupId = Long.parseLong(hSetupId[0].getValue());
                } catch (NumberFormatException e) {
                    log.info("Could not parse Nanopub-Registry-Setup-Id header: {}", hSetupId[0].getValue());
                }
            }

            // Read metadata headers for forwarding to clients
            String coverageTypes = getHeaderValue(response, "Nanopub-Registry-Coverage-Types");
            String coverageAgents = getHeaderValue(response, "Nanopub-Registry-Coverage-Agents");
            String testInstance = getHeaderValue(response, "Nanopub-Registry-Test-Instance");
            String nanopubCount = getHeaderValue(response, "Nanopub-Registry-Nanopub-Count");
            // Optional — older registries (without trust calculation) won't set this header.
            String trustStateHash = getHeaderValue(response, "Nanopub-Registry-Trust-State-Hash");

            return new RegistryMetadata(loadCounter, setupId, coverageTypes, coverageAgents,
                    testInstance, nanopubCount, trustStateHash);
        }
    }

    private static String getHeaderValue(CloseableHttpResponse response, String name) {
        var headers = response.getHeaders(name);
        return headers.length > 0 ? headers[0].getValue() : null;
    }

    /**
     * Construct the URL for fetching the Jelly stream.
     *
     * @param afterCounter the last known counter to have been committed in the DB
     * @return the URL for fetching the Jelly stream
     */
    private static String makeStreamFetchUrl(long afterCounter) {
        return registryUrl + "nanopubs.jelly?afterCounter=" + afterCounter;
    }
}

1	package com.knowledgepixels.query;
2
3	import java.io.IOException;
4	import java.util.concurrent.atomic.AtomicLong;
5
6	import org.apache.http.client.methods.CloseableHttpResponse;
7	import org.apache.http.client.methods.HttpGet;
8	import org.apache.http.client.methods.HttpHead;
9	import org.apache.http.impl.client.CloseableHttpClient;
10	import org.apache.http.impl.client.HttpClientBuilder;
11	import org.apache.http.util.EntityUtils;
12	import org.nanopub.NanopubUtils;
13	import org.nanopub.jelly.NanopubStream;
14	import org.slf4j.Logger;
15	import org.slf4j.LoggerFactory;
16
17	/**
18	* Loads nanopubs from the attached Nanopub Registry via a restartable Jelly stream.
19	*/
20	public class JellyNanopubLoader {	×
21	static final String registryUrl;
22	private static long lastCommittedCounter = -1;	6✔
23	private static Long lastKnownSetupId = null;	6✔
24	// Latest registry metadata fields, updated on each metadata fetch and forwarded to clients
25	static volatile String lastCoverageTypes = null;	6✔
26	static volatile String lastCoverageAgents = null;	6✔
27	static volatile String lastTestInstance = null;	6✔
28	static volatile String lastNanopubCount = null;	6✔
29	private static final CloseableHttpClient metadataClient;
30	private static final CloseableHttpClient jellyStreamClient;
31
32	private static final int MAX_RETRIES_METADATA = 10;
33	private static final int RETRY_DELAY_METADATA = 3000;
34	private static final int RETRY_DELAY_JELLY = 5000;
35
36	private static final Logger log = LoggerFactory.getLogger(JellyNanopubLoader.class);	9✔
37
38	/**
39	* Registry metadata returned by a HEAD request.
40	*/
41	record RegistryMetadata(long loadCounter, Long setupId, String coverageTypes,	72✔
42	String coverageAgents, String testInstance, String nanopubCount,
43	String trustStateHash) {}
44
45	/**
46	* The interval in milliseconds at which the updates loader should poll for new nanopubs.
47	*/
48	public static final int UPDATES_POLL_INTERVAL = 2000;
49
50	enum LoadingType {	×
51	INITIAL,	×
52	UPDATE,	×
53	}
54
55	static {
56	// Initialize registryUrl
57	var url = Utils.getEnvString(	12✔
58	"REGISTRY_FIXED_URL", "https://registry.knowledgepixels.com/"
59	);
60	if (!url.endsWith("/")) url += "/";	12!
61	registryUrl = url;	6✔
62
63	metadataClient = HttpClientBuilder.create().setDefaultRequestConfig(Utils.getHttpRequestConfig()).build();	15✔
64	jellyStreamClient = NanopubUtils.getHttpClient();	6✔
65	}	3✔
66
67	/**
68	* Start or continue (after restart) the initial loading procedure. This simply loads all
69	* nanopubs from the attached Registry.
70	*
71	* @param afterCounter which counter to start from (-1 for the beginning)
72	*/
73	public static void loadInitial(long afterCounter) {
74	RegistryMetadata metadata = fetchRegistryMetadata();	6✔
75	updateForwardingMetadata(metadata);	6✔
76	TrustStateLoader.maybeUpdate(metadata.trustStateHash());	9✔
77	long targetCounter = metadata.loadCounter();	9✔
78	log.info("Fetched Registry load counter: {}", targetCounter);	15✔
79	// Store setupId on initial load
80	if (metadata.setupId() != null && lastKnownSetupId == null) {	9!
81	lastKnownSetupId = metadata.setupId();	×
82	StatusController.get().setRegistrySetupId(metadata.setupId());	×
83	}
84	lastCommittedCounter = afterCounter;	6✔
85	while (lastCommittedCounter < targetCounter) {	12!
86	try {
87	loadBatch(lastCommittedCounter, LoadingType.INITIAL);	×
88	log.info("Initial load: loaded batch up to counter {}", lastCommittedCounter);	×
89	} catch (Exception e) {	×
90	log.info("Failed to load batch starting from counter {}", lastCommittedCounter);	×
91	log.info("Failure reason: ", e);	×
92	try {
93	Thread.sleep(RETRY_DELAY_JELLY);	×
94	} catch (InterruptedException e2) {	×
95	throw new RuntimeException("Interrupted while waiting to retry loading batch.");	×
96	}	×
97	}	×
98	}
99	log.info("Initial load complete.");	9✔
100	}	3✔
101
102	/**
103	* Check if the Registry has any new nanopubs. If it does, load them.
104	* This method should be called periodically, and you should wait for it to finish before
105	* calling it again.
106	*/
107	public static void loadUpdates() {
108	try {
109	final var status = StatusController.get().getState();	×
110	lastCommittedCounter = status.loadCounter;	×
111	RegistryMetadata metadata = fetchRegistryMetadata();	×
112	updateForwardingMetadata(metadata);	×
113	TrustStateLoader.maybeUpdate(metadata.trustStateHash());	×
114	long targetCounter = metadata.loadCounter();	×
115	Long currentSetupId = metadata.setupId();	×
116
117	// Detect reset via setupId change
118	if (lastKnownSetupId != null && currentSetupId != null	×
119	&& !lastKnownSetupId.equals(currentSetupId)) {	×
120	log.warn("Registry reset detected: setupId {} -> {}", lastKnownSetupId, currentSetupId);	×
121	performResync(currentSetupId);	×
122	return;	×
123	}
124	// Detect reset via counter decrease (also covers first run after upgrade
125	// where no setupId was persisted yet but the registry has already been reset)
126	if (lastCommittedCounter > 0 && targetCounter >= 0	×
127	&& targetCounter < lastCommittedCounter) {
128	log.warn("Registry counter decreased {} -> {}, triggering resync",	×
129	lastCommittedCounter, targetCounter);	×
130	performResync(currentSetupId);	×
131	return;	×
132	}
133
134	// Update lastKnownSetupId on first successful poll
135	if (currentSetupId != null && lastKnownSetupId == null) {	×
136	if (lastCommittedCounter > 0) {	×
137	// Upgrade from a version without setupId tracking. The DB has data but
138	// we can't verify it matches the current registry. Force a resync.
139	log.warn("No stored setupId but DB has data (counter: {}). "	×
140	+ "Forcing resync to ensure data consistency.", lastCommittedCounter);	×
141	performResync(currentSetupId);	×
142	return;	×
143	}
144	lastKnownSetupId = currentSetupId;	×
145	StatusController.get().setRegistrySetupId(currentSetupId);	×
146	}
147
148	StatusController.get().setLoadingUpdates(status.loadCounter);	×
149	if (lastCommittedCounter >= targetCounter) {	×
150	StatusController.get().setReady();	×
151	return;	×
152	}
153	loadBatch(lastCommittedCounter, LoadingType.UPDATE);	×
154	log.info("Loaded {} update(s). Counter: {}, target was: {}",	×
155	lastCommittedCounter - status.loadCounter, lastCommittedCounter, targetCounter);	×
156	if (lastCommittedCounter < targetCounter) {	×
157	log.info("Warning: expected to load nanopubs up to (inclusive) counter " +	×
158	targetCounter + " based on the counter reported in Registry's headers, " +
159	"but loaded only up to {}.", lastCommittedCounter);	×
160	}
161	} catch (Exception e) {	×
162	log.info("Failed to load updates. Current counter: {}", lastCommittedCounter);	×
163	log.info("Failure Reason: ", e);	×
164	} finally {
165	try {
166	StatusController.get().setReady();	×
167	} catch (Exception e) {	×
168	log.info("Update loader: failed to set status to READY.");	×
169	log.info("Failure Reason: ", e);	×
170	}	×
171	}
172	}	×
173
174	/**
175	* Re-stream all nanopubs from the registry after a reset is detected.
176	* Existing nanopubs are skipped by NanopubLoader's per-repo dedup.
177	*
178	* @param newSetupId the new setup ID from the registry, or null if unknown
179	*/
180	private static void performResync(Long newSetupId) {
181	log.warn("Starting resync with registry. New setupId: {}", newSetupId);	×
182	StatusController.get().setResetting();	×
183	lastKnownSetupId = newSetupId;	×
184	if (newSetupId != null) {	×
185	StatusController.get().setRegistrySetupId(newSetupId);	×
186	}
187	StatusController.get().setLoadingInitial(-1);	×
188	loadInitial(-1);	×
189	StatusController.get().setReady();	×
190	log.warn("Resync complete. Counter: {}", lastCommittedCounter);	×
191	}	×
192
193	/**
194	* Load a batch of nanopubs from the Jelly stream.
195	* <p>
196	* The method requests the list of all nanopubs from the Registry and reads it for as long
197	* as it can. If the stream is interrupted, the method will throw an exception, and you
198	* can resume loading from the last known counter.
199	*
200	* @param afterCounter the last known nanopub counter to have been committed in the DB
201	* @param type the type of loading operation (initial or update)
202	*/
203	static void loadBatch(long afterCounter, LoadingType type) {
204	CloseableHttpResponse response;
205	try {
206	var request = new HttpGet(makeStreamFetchUrl(afterCounter));	×
207	response = jellyStreamClient.execute(request);	×
208	} catch (IOException e) {	×
209	throw new RuntimeException("Failed to fetch Jelly stream from the Registry (I/O error).", e);	×
210	}	×
211
212	int httpStatus = response.getStatusLine().getStatusCode();	×
213	if (httpStatus < 200 \|\| httpStatus >= 300) {	×
214	EntityUtils.consumeQuietly(response.getEntity());	×
215	throw new RuntimeException("Jelly stream HTTP status is not 2xx: " + httpStatus + ".");	×
216	}
217
218	try (
219	var is = response.getEntity().getContent();	×
220	var npStream = NanopubStream.fromByteStream(is).getAsNanopubs()	×
221	) {
222	AtomicLong checkpointTime = new AtomicLong(System.currentTimeMillis());	×
223	AtomicLong checkpointCounter = new AtomicLong(lastCommittedCounter);	×
224	AtomicLong lastSavedCounter = new AtomicLong(lastCommittedCounter);	×
225	AtomicLong loaded = new AtomicLong(0L);	×
226
227	npStream.forEach(m -> {	×
228	if (!m.isSuccess()) throw new RuntimeException("Failed to load " +	×
229	"nanopub from Jelly stream. Last known counter: " + lastCommittedCounter,
230	m.getException()	×
231	);
232	if (m.getCounter() < lastCommittedCounter) {	×
233	throw new RuntimeException("Received a nanopub with a counter lower than " +	×
234	"the last known counter. Last known counter: " + lastCommittedCounter +
235	", received counter: " + m.getCounter());	×
236	}
237	NanopubLoader.load(m.getNanopub(), m.getCounter());	×
238	if (m.getCounter() % 10 == 0) {	×
239	// Save the committed counter only every 10 nanopubs to reduce DB load
240	saveCommittedCounter(type);	×
241	lastSavedCounter.set(m.getCounter());	×
242	}
243	lastCommittedCounter = m.getCounter();	×
244	loaded.getAndIncrement();	×
245
246	if (loaded.get() % 50 == 0) {	×
247	long currTime = System.currentTimeMillis();	×
248	double speed = 50 / ((currTime - checkpointTime.get()) / 1000.0);	×
249	log.info("Loading speed: " + String.format("%.2f", speed) +	×
250	" np/s. Counter: " + lastCommittedCounter);
251	checkpointTime.set(currTime);	×
252	checkpointCounter.set(lastCommittedCounter);	×
253	}
254	});	×
255	// Make sure to save the last committed counter at the end of the batch
256	if (lastCommittedCounter >= lastSavedCounter.get()) {	×
257	saveCommittedCounter(type);	×
258	}
259	} catch (IOException e) {	×
260	throw new RuntimeException("I/O error while reading the response Jelly stream.", e);	×
261	} finally {
262	try {
263	response.close();	×
264	} catch (IOException e) {	×
265	log.info("Failed to close the Jelly stream response.");	×
266	}	×
267	}
268	}	×
269
270	/**
271	* Save the last committed counter to the DB. Do this every N nanopubs to reduce DB load.
272	* Remember to call this method at the end of the batch as well.
273	*
274	* @param type the type of loading operation (initial or update)
275	*/
276	private static void saveCommittedCounter(LoadingType type) {
277	try {
278	if (type == LoadingType.INITIAL) {	×
279	StatusController.get().setLoadingInitial(lastCommittedCounter);	×
280	} else {
281	StatusController.get().setLoadingUpdates(lastCommittedCounter);	×
282	}
283	} catch (Exception e) {	×
284	throw new RuntimeException("Could not update the nanopub counter in DB", e);	×
285	}	×
286	}	×
287
288	/**
289	* Set the last known setup ID. Called from MainVerticle on startup to restore persisted state.
290	*
291	* @param setupId the setup ID to set, or null if not known
292	*/
293	static void setLastKnownSetupId(Long setupId) {
294	lastKnownSetupId = setupId;	×
295	}	×
296
297	/**
298	* Update the cached metadata fields used for forwarding to clients.
299	*/
300	private static void updateForwardingMetadata(RegistryMetadata metadata) {
301	lastCoverageTypes = metadata.coverageTypes();	9✔
302	lastCoverageAgents = metadata.coverageAgents();	9✔
303	lastTestInstance = metadata.testInstance();	9✔
304	lastNanopubCount = metadata.nanopubCount();	9✔
305	}	3✔
306
307	/**
308	* Run a HEAD request to the Registry to fetch its current metadata (load counter and setup ID).
309	*
310	* @return the registry metadata
311	*/
312	static RegistryMetadata fetchRegistryMetadata() {
313	int tries = 0;	6✔
314	RegistryMetadata metadata = null;	6✔
315	while (metadata == null && tries < MAX_RETRIES_METADATA) {	15!
316	try {
317	metadata = fetchRegistryMetadataInner();	4✔
318	} catch (Exception e) {	1✔
319	tries++;	1✔
320	log.info("Failed to fetch registry metadata, try " + tries +	5✔
321	". Retrying in {}ms...", RETRY_DELAY_METADATA);	1✔
322	log.info("Failure Reason: ", e);	4✔
323	try {
324	Thread.sleep(RETRY_DELAY_METADATA);	2✔
325	} catch (InterruptedException e2) {	×
326	throw new RuntimeException(	×
327	"Interrupted while waiting to retry fetching registry metadata.");
328	}	1✔
329	}	3✔
330	}
331	if (metadata == null) {	6!
332	throw new RuntimeException("Failed to fetch registry metadata after " +	5✔
333	MAX_RETRIES_METADATA + " retries.");
334	}
335	return metadata;	4✔
336	}
337
338	/**
339	* Inner logic for fetching the registry metadata via HEAD request.
340	*
341	* @return the registry metadata (load counter and setup ID)
342	* @throws IOException if the HTTP request fails
343	*/
344	private static RegistryMetadata fetchRegistryMetadataInner() throws IOException {
345	var request = new HttpHead(registryUrl);	15✔
346	try (var response = metadataClient.execute(request)) {	8✔
347	int status = response.getStatusLine().getStatusCode();	8✔
348	EntityUtils.consumeQuietly(response.getEntity());	6✔
349	if (status < 200 \|\| status >= 300) {	12!
350	throw new RuntimeException("Registry metadata HTTP status is not 2xx: " +	×
351	status + ".");
352	}
353
354	// Check if the registry is ready
355	var hStatus = response.getHeaders("Nanopub-Registry-Status");	8✔
356	if (hStatus.length == 0) {	6!
357	throw new RuntimeException("Registry did not return a Nanopub-Registry-Status header.");	×
358	}
359	if (!"ready".equals(hStatus[0].getValue()) && !"updating".equals(hStatus[0].getValue())) {	14!
360	throw new RuntimeException("Registry is not in ready state.");	×
361	}
362
363	// Get the load counter
364	var hCounter = response.getHeaders("Nanopub-Registry-Load-Counter");	8✔
365	if (hCounter.length == 0) {	6!
366	throw new RuntimeException("Registry did not return a Nanopub-Registry-Load-Counter header.");	×
367	}
368	long loadCounter = Long.parseLong(hCounter[0].getValue());	12✔
369
370	// Get the setup ID (optional — older registries may not have it)
371	Long setupId = null;	4✔
372	var hSetupId = response.getHeaders("Nanopub-Registry-Setup-Id");	8✔
373	if (hSetupId.length > 0) {	6!
374	try {
375	setupId = Long.parseLong(hSetupId[0].getValue());	14✔
376	} catch (NumberFormatException e) {	×
377	log.info("Could not parse Nanopub-Registry-Setup-Id header: {}", hSetupId[0].getValue());	×
378	}	2✔
379	}
380
381	// Read metadata headers for forwarding to clients
382	String coverageTypes = getHeaderValue(response, "Nanopub-Registry-Coverage-Types");	8✔
383	String coverageAgents = getHeaderValue(response, "Nanopub-Registry-Coverage-Agents");	8✔
384	String testInstance = getHeaderValue(response, "Nanopub-Registry-Test-Instance");	8✔
385	String nanopubCount = getHeaderValue(response, "Nanopub-Registry-Nanopub-Count");	8✔
386	// Optional — older registries (without trust calculation) won't set this header.
387	String trustStateHash = getHeaderValue(response, "Nanopub-Registry-Trust-State-Hash");	8✔
388
389	return new RegistryMetadata(loadCounter, setupId, coverageTypes, coverageAgents,	26✔
390	testInstance, nanopubCount, trustStateHash);
391	}
392	}
393
394	private static String getHeaderValue(CloseableHttpResponse response, String name) {
395	var headers = response.getHeaders(name);	8✔
396	return headers.length > 0 ? headers[0].getValue() : null;	18!
397	}
398
399	/**
400	* Construct the URL for fetching the Jelly stream.
401	*
402	* @param afterCounter the last known counter to have been committed in the DB
403	* @return the URL for fetching the Jelly stream
404	*/
405	private static String makeStreamFetchUrl(long afterCounter) {
406	return registryUrl + "nanopubs.jelly?afterCounter=" + afterCounter;	×
407	}
408	}

knowledgepixels / nanopub-query / 24522677438

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous