From 026ef435e12b3220ee658147bf5e7741b0c70c1a Mon Sep 17 00:00:00 2001 From: Claire Lovisa Date: Wed, 22 Jan 2025 16:41:12 +0100 Subject: [PATCH] Revert "Revert "Chore/update leidinggevenden consumer"" --- CHANGELOG.md | 39 +++- README.md | 39 ++++ .../leidinggevenden-dispatching/config.js | 32 ---- .../delta-context-config.js | 72 -------- .../delta-sync-dispatching.js | 85 --------- .../initial-sync-dispatching.js | 85 --------- .../leidinggevenden-dispatching/util.js | 168 ------------------ .../queries/dl2op/data-cleanup/phone.rq | 21 +++ .../queries/dl2op/inheritance/Agent.rq | 15 ++ .../dl2op/inheritance/AgentInPositie.rq | 11 ++ .../queries/dl2op/inverse/holds-to-held-by.rq | 8 + .../mapping/queries/dl2op/pass/address.sparql | 10 ++ .../queries/dl2op/pass/bestuursfunctie.sparql | 7 + .../queries/dl2op/pass/contactpoint.sparql | 11 ++ .../queries/dl2op/pass/functionaris.sparql | 10 ++ .../mapping/queries/dl2op/pass/person.sparql | 16 ++ .../dl2op/predicate/person-foaf-name.sparql | 21 +++ .../person-gebruikte-voornaam.sparql | 21 +++ ...flush-existing-leidinggevenden-data.sparql | 25 +++ ...flush-existing-leidinggevenden-data.sparql | 25 +++ ...flush-existing-leidinggevenden-data.sparql | 37 ++++ ...flush-existing-leidinggevenden-data.sparql | 28 +++ ...flush-existing-leidinggevenden-data.sparql | 13 ++ .../20250113151947-flush-landing-graph.sparql | 1 + ...leidinggevenden-consumer-containers.sparql | 37 ++++ ...lush-leidinggevenden-consumer-tasks.sparql | 34 ++++ ...flush-leidinggevenden-consumer-jobs.sparql | 32 ++++ ...leidinggevenden-consumer-containers.sparql | 37 ++++ ...lush-leidinggevenden-consumer-tasks.sparql | 34 ++++ ...flush-leidinggevenden-consumer-jobs.sparql | 32 ++++ ...leidinggevenden-consumer-containers.sparql | 34 ++++ ...lush-leidinggevenden-consumer-tasks.sparql | 31 ++++ ...flush-leidinggevenden-consumer-jobs.sparql | 29 +++ docker-compose.yml | 16 +- 34 files changed, 665 insertions(+), 451 deletions(-) delete mode 100644 config/loket-consumer/leidinggevenden-dispatching/config.js delete mode 100644 config/loket-consumer/leidinggevenden-dispatching/delta-context-config.js delete mode 100644 config/loket-consumer/leidinggevenden-dispatching/delta-sync-dispatching.js delete mode 100644 config/loket-consumer/leidinggevenden-dispatching/initial-sync-dispatching.js delete mode 100644 config/loket-consumer/leidinggevenden-dispatching/util.js create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/data-cleanup/phone.rq create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/Agent.rq create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/AgentInPositie.rq create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inverse/holds-to-held-by.rq create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/address.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/bestuursfunctie.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/contactpoint.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/functionaris.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/person.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-foaf-name.sparql create mode 100644 config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-gebruikte-voornaam.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151927-flush-existing-leidinggevenden-data.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151928-flush-existing-leidinggevenden-data.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151929-flush-existing-leidinggevenden-data.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151930-flush-existing-leidinggevenden-data.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151931-flush-existing-leidinggevenden-data.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151947-flush-landing-graph.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152654-flush-leidinggevenden-consumer-containers.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152655-flush-leidinggevenden-consumer-tasks.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152656-flush-leidinggevenden-consumer-jobs.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152657-flush-leidinggevenden-consumer-containers.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152658-flush-leidinggevenden-consumer-tasks.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152659-flush-leidinggevenden-consumer-jobs.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152660-flush-leidinggevenden-consumer-containers.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152661-flush-leidinggevenden-consumer-tasks.sparql create mode 100644 config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152662-flush-leidinggevenden-consumer-jobs.sparql diff --git a/CHANGELOG.md b/CHANGELOG.md index f3ead38..16a6c2f 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,12 +1,47 @@ # Changelog + ## 1.28.5 (TODO) ### Frontend - Bump to [v1.28.2](https://github.com/lblod/frontend-organization-portal/releases/tag/v1.28.2) ### Backend - Datafix: correct merger change events for worship organisations (OP-3534) +#### Consumer +- Upgraded `leidinggevenden-consumer` [OP-3533], [OP-3511] ### Deploy notes -- `drc restart migrations-triggering-indexing; drc logs -ft --tail=200 migrations-triggering-indexing` -- `drc pull frontend; drc up -d frontend` +``` +drc down; +``` +Update `docker-compose.override.yml` to: +``` + leidinggevenden-consumer: + environment: + DCR_SYNC_BASE_URL: "https://loket.lokaalbestuur.vlaanderen.be" # or another endpoint + DCR_LANDING_ZONE_DATABASE: "triplestore" # for the initial sync, we go directly to virtuoso + DCR_REMAPPING_DATABASE: "triplestore" # for the initial sync, we go directly to virtuoso + DCR_DISABLE_DELTA_INGEST: "false" + DCR_DISABLE_INITIAL_SYNC: "false" +``` +Then: +``` +drc up -d migrations migrations-triggering-indexing; drc logs -ft --tail=200 migrations migrations-triggering-indexing +drc up -d db leidinggevenden-consumer +# Wait until success of the previous step +``` +Then, update `docker-compose.override.yml` to: +``` + leidinggevenden-consumer: + environment: + DCR_SYNC_BASE_URL: "https://loket.lokaalbestuur.vlaanderen.be" # choose the correct endpoint + DCR_LANDING_ZONE_DATABASE: "db" + DCR_REMAPPING_DATABASE: "db" + DCR_DISABLE_DELTA_INGEST: "false" + DCR_DISABLE_INITIAL_SYNC: "false" +``` +``` +drc pull frontend +drc up -d +sh scripts/reset-elastic.sh +``` ## 1.28.4 (2025-01-16) ### Backend diff --git a/README.md b/README.md index a59e0e3..0b9011f 100644 --- a/README.md +++ b/README.md @@ -11,6 +11,45 @@ Backend for the organization portal application, based on the mu.semte.ch micros You can shut down using `docker-compose stop` and remove everything using `docker-compose rm`. +#### Setup consumers +The application depends on external data. You'll have to perform extra steps if you want this data to be available. +These are multiple data sets, we'll try to be exhaustive on how to ingest these. Even though the procedures are similar. + +#### Setup http://data.lblod.info/vocabularies/leidinggevenden/Functionaris +Note: this is assuming you have never ran this before. +``` +drc down; +``` +Update `docker-compose.override.yml` to: +``` + leidinggevenden-consumer: + environment: + DCR_SYNC_BASE_URL: "https://loket.lokaalbestuur.vlaanderen.be" # or another endpoint + DCR_LANDING_ZONE_DATABASE: "triplestore" # for the initial sync, we go directly to virtuoso + DCR_REMAPPING_DATABASE: "triplestore" # for the initial sync, we go directly to virtuoso + DCR_DISABLE_DELTA_INGEST: "false" + DCR_DISABLE_INITIAL_SYNC: "false" +``` +Then: +``` +drc up -d migrations +drc up -d db leidinggevenden-consumer +# Wait until success of the previous step +``` +Then, update `docker-compose.override.yml` to: +``` + leidinggevenden-consumer: + environment: + DCR_SYNC_BASE_URL: "https://loket.lokaalbestuur.vlaanderen.be" # choose the correct endpoint + DCR_LANDING_ZONE_DATABASE: "db" + DCR_REMAPPING_DATABASE: "db" + DCR_DISABLE_DELTA_INGEST: "false" + DCR_DISABLE_INITIAL_SYNC: "false" +``` +``` +drc up -d +sh scripts/reset-elastic.sh +``` ### Setting up the delta-producers related services To make sure the app can share data, producers need to be set up. There is an intial sync, that is potentially very expensive, and must be started manually diff --git a/config/loket-consumer/leidinggevenden-dispatching/config.js b/config/loket-consumer/leidinggevenden-dispatching/config.js deleted file mode 100644 index b3ebc2a..0000000 --- a/config/loket-consumer/leidinggevenden-dispatching/config.js +++ /dev/null @@ -1,32 +0,0 @@ -const BATCH_SIZE = parseInt(process.env.BATCH_SIZE) || 100; -const MU_CALL_SCOPE_ID_INITIAL_SYNC = process.env.MU_CALL_SCOPE_ID_INITIAL_SYNC || 'http://redpencil.data.gift/id/concept/muScope/deltas/consumer/initialSync'; -const BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES = process.env.BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES == 'true' ? true : false; -const DIRECT_DATABASE_ENDPOINT = process.env.DIRECT_DATABASE_ENDPOINT || 'http://triplestore:8890/sparql'; -const MAX_DB_RETRY_ATTEMPTS = parseInt(process.env.MAX_DB_RETRY_ATTEMPTS || 5); -const MAX_REASONING_RETRY_ATTEMPTS = parseInt(process.env.MAX_REASONING_RETRY_ATTEMPTS || 5); -const SLEEP_BETWEEN_BATCHES = parseInt(process.env.SLEEP_BETWEEN_BATCHES || 1000); -const SLEEP_TIME_AFTER_FAILED_REASONING_OPERATION = parseInt(process.env.SLEEP_TIME_AFTER_FAILED_REASONING_OPERATION || 10000); -const SLEEP_TIME_AFTER_FAILED_DB_OPERATION = parseInt(process.env.SLEEP_TIME_AFTER_FAILED_DB_OPERATION || 60000); - -// Graphs -const TARGET_GRAPH = process.env.TARGET_GRAPH || `http://mu.semte.ch/graphs/administrative-unit`; -const LANDING_ZONE_GRAPH = process.env.DCR_LANDING_ZONE_GRAPH || `http://mu.semte.ch/graphs/landing-zone/leidinggevenden`; - -// ENDPOINTS -const LANDING_ZONE_DATABASE = process.env.DCR_LANDING_ZONE_DATABASE || 'db'; -const LANDING_ZONE_DATABASE_ENDPOINT = process.env.DCR_LANDING_ZONE_DATABASE_ENDPOINT || `http://${LANDING_ZONE_DATABASE}:8890/sparql`; - -module.exports = { - BATCH_SIZE, - MU_CALL_SCOPE_ID_INITIAL_SYNC, - BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES, - DIRECT_DATABASE_ENDPOINT, - MAX_DB_RETRY_ATTEMPTS, - MAX_REASONING_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION, - SLEEP_TIME_AFTER_FAILED_REASONING_OPERATION, - LANDING_ZONE_GRAPH, - TARGET_GRAPH, - LANDING_ZONE_DATABASE_ENDPOINT -}; diff --git a/config/loket-consumer/leidinggevenden-dispatching/delta-context-config.js b/config/loket-consumer/leidinggevenden-dispatching/delta-context-config.js deleted file mode 100644 index 973462f..0000000 --- a/config/loket-consumer/leidinggevenden-dispatching/delta-context-config.js +++ /dev/null @@ -1,72 +0,0 @@ - -const PREFIXES = ` -PREFIX org: -PREFIX mandaat: -PREFIX lblodlg: -PREFIX besluit: ` - -const contextConfig = { - addTypes: { - scope: 'all', // 'inserts, 'deletes', 'all' or none. To add rdf:type to subjects of inserts, deletes or both - exhausitive: false, // true or false: find all types for a subject, even if one is already present in delta - }, - contextQueries: [ - // Case Leidnggevenden - // They can be triggered when being inserted or when their link to a bestuursfunctie is added - { - trigger: { // subjectType or predicateValue - subjectType: "lblodlg:Functionaris" - }, - queryTemplate: (subject) => ` - ${PREFIXES} - CONSTRUCT { - ?functionaris org:holds ?bestuursfunctie ; - ?p ?o . - ?orgaanInTime lblodlg:heeftBestuursfunctie ?bestuursfunctie ; - mandaat:isTijdspecialisatieVan ?orgaan . - ?orgaan besluit:bestuurt ?eenheid . - ?eenheid besluit:classificatie ?classification . - } WHERE { - GRAPH { - VALUES ?functionaris { ${subject} } - ?functionaris org:holds ?bestuursfunctie ; - ?p ?o . - ?orgaanInTime lblodlg:heeftBestuursfunctie ?bestuursfunctie ; - mandaat:isTijdspecialisatieVan ?orgaan . - ?orgaan besluit:bestuurt ?eenheid . - ?eenheid besluit:classificatie ?classification . - } - }` - }, - { - trigger: { - predicateValue: "org:holds" - }, - queryTemplate: (subject) => ` - ${PREFIXES} - CONSTRUCT { - ?functionaris org:holds ?bestuursfunctie ; - ?p ?o . - ?orgaanInTime lblodlg:heeftBestuursfunctie ?bestuursfunctie ; - mandaat:isTijdspecialisatieVan ?orgaan . - ?orgaan besluit:bestuurt ?eenheid . - ?eenheid besluit:classificatie ?classification . - } WHERE { - GRAPH { - VALUES ?functionaris { ${subject} } - ?functionaris org:holds ?bestuursfunctie ; - ?p ?o . - ?orgaanInTime lblodlg:heeftBestuursfunctie ?bestuursfunctie ; - mandaat:isTijdspecialisatieVan ?orgaan . - ?orgaan besluit:bestuurt ?eenheid . - ?eenheid besluit:classificatie ?classification . - } - }` - } - ] -} - -module.exports = { - contextConfig, - PREFIXES -}; diff --git a/config/loket-consumer/leidinggevenden-dispatching/delta-sync-dispatching.js b/config/loket-consumer/leidinggevenden-dispatching/delta-sync-dispatching.js deleted file mode 100644 index b309f35..0000000 --- a/config/loket-consumer/leidinggevenden-dispatching/delta-sync-dispatching.js +++ /dev/null @@ -1,85 +0,0 @@ -const { - transformStatements, - deleteFromTargetGraph, - insertIntoTargetGraph, -} = require('./util'); - -/** -* Dispatch the fetched information to a target graph. -* Note: will be ADDED to it's own graph. -* We take only care of adding them, not updating triples, this is a TODO -* @param { mu, muAuthSudo } lib - The provided libraries from the host service. -* @param { termObjectChangeSets: { deletes, inserts } } data - The fetched changes sets, which objects of serialized Terms -* [ { -* graph: "", -* subject: "", -* predicate: "", -* object: "^^" -* } -* ] -* @return {void} Nothing -*/ -async function dispatch(lib, data) { - const { mu, fetch } = lib; - let { termObjectChangeSets, termObjectChangeSetsWithContext } = data; - - // Both arrays are the same length, so we can zip them together for easier processing - zippedChangeSets = termObjectChangeSets.map((o, i) => ({ - original: o, - withContext: termObjectChangeSetsWithContext[i] - })); - - console.log(`Received ${zippedChangeSets.length} change sets`) - - for (let { original, withContext } of zippedChangeSets) { - let originalDeletes, originalInserts, insertsWithContext, deletesWithContext; - try { - originalInserts = original.inserts.map(o => `${o.subject} ${o.predicate} ${o.object}.`); - originalDeletes = original.deletes.map(o => `${o.subject} ${o.predicate} ${o.object}.`); - } catch (e) { - console.log(`Error while processing change set: ${e}`); - console.log(`Original change set: ${JSON.stringify(original)}`); - throw e; - } - - // Extra context needed for mapping from DL to OP model and filtering based on type. - try { - insertsWithContext = withContext.inserts.map(o => `${o.subject} ${o.predicate} ${o.object}.`); - deletesWithContext = withContext.deletes.map(o => `${o.subject} ${o.predicate} ${o.object}.`); - } catch (e) { - console.log(`Error while processing change set with context: ${e}`); - console.log(`Change set with context: ${JSON.stringify(withContext)}`); - throw e; - } - - if (originalDeletes.length) { - // Map deletes from DL to OP model - const transformedDeletes = await transformStatements(fetch, deletesWithContext); - - if (!transformedDeletes.length) { - console.log(`Warn: Delete statements mapped to empty result.`); - console.log(`Input: ${deletesWithContext}`); - console.log(`Output: ${transformedDeletes}`); - } else { - await deleteFromTargetGraph(lib, transformedDeletes); - } - } - - if (originalInserts.length) { - // Map inserts from DL to OP model - const transformedInserts = await transformStatements(fetch, insertsWithContext); - - if (!transformedInserts.length) { - console.log(`Warn: Insert statements mapped to empty result.`); - console.log(`Input: ${insertsWithContext}`); - console.log(`Output: ${transformedInserts}`); - } else { - await insertIntoTargetGraph(lib, transformedInserts); - } - } - } -} - -module.exports = { - dispatch -}; diff --git a/config/loket-consumer/leidinggevenden-dispatching/initial-sync-dispatching.js b/config/loket-consumer/leidinggevenden-dispatching/initial-sync-dispatching.js deleted file mode 100644 index d5d672f..0000000 --- a/config/loket-consumer/leidinggevenden-dispatching/initial-sync-dispatching.js +++ /dev/null @@ -1,85 +0,0 @@ -const { batchedDbUpdate, transformLandingZoneGraph } = require('./util'); -const { - BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES, - DIRECT_DATABASE_ENDPOINT, - MU_CALL_SCOPE_ID_INITIAL_SYNC, - BATCH_SIZE, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION, - TARGET_GRAPH, - LANDING_ZONE_GRAPH, - LANDING_ZONE_DATABASE_ENDPOINT -} = require('./config'); - -const endpoint = BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES ? DIRECT_DATABASE_ENDPOINT : process.env.MU_SPARQL_ENDPOINT; - -/** -* Dispatch the fetched information to a target graph. -* @param { mu, muAuthSudo } lib - The provided libraries from the host service. -* @param { termObjects } data - The fetched quad information, which objects of serialized Terms -* [ { -* graph: "", -* subject: "", -* predicate: "", -* object: "^^" -* } -* ] -* @return {void} Nothing -*/ -async function dispatch(lib, data) { - const { mu, muAuthSudo } = lib; - const { termObjects } = data; - - // Steps: - // 1. Write the triples to the landing zone graph - withouth any mapping or filtering - // 2. One-time reasoning run with the full landing zone graph and write the results to the target graph (on-finish) - - if (BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES) { - console.warn(`Service configured to skip MU_AUTH!`); - } - console.log(`Using ${endpoint} to insert triples`); - - if (termObjects.length) { - const originalInsertTriples = termObjects.map(o => `${o.subject} ${o.predicate} ${o.object}.`); - - await batchedDbUpdate( - muAuthSudo.updateSudo, - LANDING_ZONE_GRAPH, - originalInsertTriples, - { 'mu-call-scope-id': MU_CALL_SCOPE_ID_INITIAL_SYNC }, - LANDING_ZONE_DATABASE_ENDPOINT, - BATCH_SIZE, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION - ); - } -} - -async function onFinishInitialIngest(lib) { - const { muAuthSudo, fetch } = lib; - - console.log(`!! On-finish triggered !!`); - - const transformedInsertTriples = await transformLandingZoneGraph(fetch, endpoint); - - console.log(`Transformed ${transformedInsertTriples.length} triples`); - - await batchedDbUpdate( - muAuthSudo.updateSudo, - TARGET_GRAPH, - transformedInsertTriples, - { 'mu-call-scope-id': MU_CALL_SCOPE_ID_INITIAL_SYNC }, - endpoint, - BATCH_SIZE, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION - ); -} - -module.exports = { - dispatch, - onFinishInitialIngest -}; diff --git a/config/loket-consumer/leidinggevenden-dispatching/util.js b/config/loket-consumer/leidinggevenden-dispatching/util.js deleted file mode 100644 index 5eced4a..0000000 --- a/config/loket-consumer/leidinggevenden-dispatching/util.js +++ /dev/null @@ -1,168 +0,0 @@ -const { - MAX_REASONING_RETRY_ATTEMPTS, - SLEEP_TIME_AFTER_FAILED_REASONING_OPERATION, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION, - TARGET_GRAPH, - LANDING_ZONE_GRAPH, - BATCH_SIZE, -} = require('./config'); - -async function batchedDbUpdate( - muUpdate, - graph, - triples, - extraHeaders, - endpoint, - batchSize, - maxAttempts, - sleepBetweenBatches = 1000, - sleepTimeOnFail = 1000, - operation = 'INSERT' -) { - - for (let i = 0; i < triples.length; i += batchSize) { - console.log(`Inserting triples in batch: ${i}-${i + batchSize}`); - - const batch = triples.slice(i, i + batchSize).join('\n'); - - const insertCall = async () => { - await muUpdate(` -${operation} DATA { -GRAPH <${graph}> { -${batch} -} -} -`, extraHeaders, endpoint); - }; - - await operationWithRetry(insertCall, 0, maxAttempts, sleepTimeOnFail); - - console.log(`Sleeping before next query execution: ${sleepBetweenBatches}`); - await new Promise(r => setTimeout(r, sleepBetweenBatches)); - } -} - -async function operationWithRetry(callback, - attempt, - maxAttempts, - sleepTimeOnFail) { - try { - if (typeof callback === "function") - return await callback(); - else // Catch error from promise - not how I would do it normally, but allows re use of existing code. - return await callback; - } - catch (e) { - console.log(`Operation failed for ${callback.toString()}, attempt: ${attempt} of ${maxAttempts}`); - console.log(`Error: ${e}`); - console.log(`Sleeping ${sleepTimeOnFail} ms`); - - if (attempt >= maxAttempts) { - console.log(`Max attempts reached for ${callback.toString()}, giving up`); - throw e; - } - - await new Promise(r => setTimeout(r, sleepTimeOnFail)); - return operationWithRetry(callback, ++attempt, maxAttempts, sleepTimeOnFail); - } -} - -/** - * Splits an array into two parts, a part that passes and a part that fails a predicate function. - * Credits: https://github.com/benjay10 - * @public - * @function partition - * @param {Array} arr - Array to be partitioned - * @param {Function} fn - Function that accepts single argument: an element of the array, and should return a truthy or falsy value. - * @returns {Object} Object that contains keys passes and fails, each representing an array with elemets that pass or fail the predicate respectively - */ -function partition(arr, fn) { - let passes = [], fails = []; - arr.forEach((item) => (fn(item) ? passes : fails).push(item)); - return { passes, fails }; -} - -/** - * Send triples to reasoning service for conversion - * - */ -async function transformTriples(fetch, triples, mapping) { - return await operationWithRetry(mainConversion(fetch, triples, mapping), 0, - MAX_REASONING_RETRY_ATTEMPTS, SLEEP_TIME_AFTER_FAILED_REASONING_OPERATION); -} - -async function mainConversion(fetch, triples, mapping) { - let formdata = new URLSearchParams(); - formdata.append("data", triples); - - let requestOptions = { - method: 'POST', - body: formdata, - redirect: 'follow' - }; - - const response = await fetch(`http://reasoner/reason/dl2op/${mapping}`, requestOptions) - return await response.text(); -} - -async function transformStatements(fetch, triples, mapping = 'main') { - console.log(`Received ${JSON.stringify(triples)} to transform`); - const transformedTriples = await transformTriples(fetch, triples.join('\n'), mapping); - statements = transformedTriples ? transformedTriples.replace(/\n{2,}/g, '').split('\n') : []; - console.log(`CONVERSION: FROM ${triples.length} triples to ${statements.length}`); - return statements; -} - -async function deleteFromTargetGraph(lib, statements) { - console.log(`Deleting ${statements.length} statements from target graph`); - console.log(`Statements: ${JSON.stringify(statements)}`) - await batchedDbUpdate( - lib.muAuthSudo.updateSudo, - TARGET_GRAPH, - statements, - {}, - process.env.MU_SPARQL_ENDPOINT, - BATCH_SIZE, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION, - 'DELETE'); -} - -async function insertIntoTargetGraph(lib, statements) { - console.log(`Inserting ${statements.length} statements into target graph`); - console.log(`Statements: ${JSON.stringify(statements)}`) - - await batchedDbUpdate( - lib.muAuthSudo.updateSudo, - TARGET_GRAPH, - statements, - {}, - process.env.MU_SPARQL_ENDPOINT, - BATCH_SIZE, - MAX_DB_RETRY_ATTEMPTS, - SLEEP_BETWEEN_BATCHES, - SLEEP_TIME_AFTER_FAILED_DB_OPERATION, - 'INSERT'); -} - -async function transformLandingZoneGraph(fetch, endpoint, mapping = 'main') { - console.log(`Transforming landing zone graph: ${LANDING_ZONE_GRAPH}`); - - const response = await fetch(`http://reasoner/reason/dl2op/${mapping}?data=${encodeURIComponent(`${endpoint}?default-graph-uri=&query=CONSTRUCT+%7B%0D%0A%3Fs+%3Fp+%3Fo%0D%0A%7D+WHERE+%7B%0D%0A+GRAPH+<${LANDING_ZONE_GRAPH}>+%7B%0D%0A%3Fs+%3Fp+%3Fo%0D%0A%7D%0D%0A%7D&should-sponge=&format=text%2Fplain&timeout=0&run=+Run+Query`)}`); - const text = await response.text(); - const statements = text.replace(/\n{2,}/g, '').split('\n'); - - return statements; -} - -module.exports = { - batchedDbUpdate, - partition, - transformStatements, - transformLandingZoneGraph, - deleteFromTargetGraph, - insertIntoTargetGraph -}; diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/data-cleanup/phone.rq b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/data-cleanup/phone.rq new file mode 100644 index 0000000..7e761c1 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/data-cleanup/phone.rq @@ -0,0 +1,21 @@ +PREFIX schema: + +CONSTRUCT { + ?s schema:telephone ?cleanedPhoneNumber . +} WHERE { + ?s schema:telephone ?phoneNumber . + + # Remove all non digits + BIND(REPLACE(?phoneNumber, """[^+\\d]""", "") AS ?scrapeResult) + + # Replace leading 00 with + + BIND(REPLACE(?scrapeResult, """^00""", "+") AS ?replacedZeroZero) + + BIND(REPLACE(?replacedZeroZero, """^0""", "+32") AS ?defaultBe) + + BIND(REPLACE(?defaultBe, """\\+320""", "+32") AS ?extendedDefaultBe) + + BIND(REPLACE(?extendedDefaultBe, """^320""", "+32") AS ?finalResult) + + BIND(IF(REGEX(?finalResult, """^\\+((32\\d{8})|(324\\d{8})|((?!32)\\d+))$"""), CONCAT("tel:", ?finalResult), ?phoneNumber) AS ?cleanedPhoneNumber) +} diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/Agent.rq b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/Agent.rq new file mode 100644 index 0000000..66d5eb2 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/Agent.rq @@ -0,0 +1,15 @@ +PREFIX dc_terms: +PREFIX lblodlg: +PREFIX person: + +CONSTRUCT { + ?s a dc_terms:Agent . +} WHERE { + ?s a ?type . + + VALUES ?type { + lblodlg:Functionaris + person:Person + } + +} diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/AgentInPositie.rq b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/AgentInPositie.rq new file mode 100644 index 0000000..e3a97b3 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inheritance/AgentInPositie.rq @@ -0,0 +1,11 @@ +PREFIX ch: +PREFIX lblodlg: + +CONSTRUCT { + ?s a ch:AgentInPositie . +} WHERE { + ?s a ?type . + VALUES ?type { + lblodlg:Functionaris + } +} diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inverse/holds-to-held-by.rq b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inverse/holds-to-held-by.rq new file mode 100644 index 0000000..0b373ed --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/inverse/holds-to-held-by.rq @@ -0,0 +1,8 @@ +PREFIX org: + +CONSTRUCT { + ?o org:heldBy ?s. +} +WHERE { + ?s org:holds ?o. +} diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/address.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/address.sparql new file mode 100644 index 0000000..41f7523 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/address.sparql @@ -0,0 +1,10 @@ +CONSTRUCT { + ?s ?p ?o. +} WHERE { + ?s a ; + ?p ?o. + + FILTER NOT EXISTS { + ?s ?p "" . + } +} \ No newline at end of file diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/bestuursfunctie.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/bestuursfunctie.sparql new file mode 100644 index 0000000..238cd07 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/bestuursfunctie.sparql @@ -0,0 +1,7 @@ +CONSTRUCT { + ?s ?contactpoint. +} +WHERE { + ?s a . + ?s ?contactpoint. +} \ No newline at end of file diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/contactpoint.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/contactpoint.sparql new file mode 100644 index 0000000..d87a2ef --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/contactpoint.sparql @@ -0,0 +1,11 @@ +CONSTRUCT { + ?s ?p ?o. +} WHERE { + ?s a ; + ?p ?o. + FILTER (?p NOT IN ()) + + FILTER NOT EXISTS { + ?s ?p "" . + } +} \ No newline at end of file diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/functionaris.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/functionaris.sparql new file mode 100644 index 0000000..e753a79 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/functionaris.sparql @@ -0,0 +1,10 @@ +CONSTRUCT { + ?s ?p ?o. +} WHERE { + ?s a ; + ?p ?o. + + FILTER NOT EXISTS { + ?s ?p "" . + } +} \ No newline at end of file diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/person.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/person.sparql new file mode 100644 index 0000000..02b21ea --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/pass/person.sparql @@ -0,0 +1,16 @@ +CONSTRUCT { + ?s ?p ?o. +} WHERE { + ?s a ; + ?p ?o. + + FILTER (?p IN ( + , + , + + )) + + FILTER NOT EXISTS { + ?s ?p "" . + } +} \ No newline at end of file diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-foaf-name.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-foaf-name.sparql new file mode 100644 index 0000000..0347ec9 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-foaf-name.sparql @@ -0,0 +1,21 @@ +PREFIX foaf: +PREFIX persoon: +PREFIX persoon_loket: +PREFIX person: + +# DL and OP both use persoon:gebruikteVoornaam but with different semantics (!) + +# First name: persoon:gebruikteVoornaam => foaf:givenName +# Alternatieve voornaam: foaf:name => persoon:gebruikteVoornaam + +# We do both in one query with an OPTIONAL because they need to happen simultaneously +# not to end up in a loop + +# Part 1 + +CONSTRUCT { + ?s persoon:gebruikteVoornaam ?name . +} WHERE { + ?s a person:Person; + foaf:name ?name . +} diff --git a/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-gebruikte-voornaam.sparql b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-gebruikte-voornaam.sparql new file mode 100644 index 0000000..87c6463 --- /dev/null +++ b/config/loket-consumer/leidinggevenden/mapping/queries/dl2op/predicate/person-gebruikte-voornaam.sparql @@ -0,0 +1,21 @@ +PREFIX foaf: +PREFIX persoon: +PREFIX persoon_loket: +PREFIX person: + +# DL and OP both use persoon:gebruikteVoornaam but with different semantics (!) + +# First name: persoon:gebruikteVoornaam => foaf:givenName +# Alternatieve voornaam: foaf:name => persoon:gebruikteVoornaam + +# We do both in one query with an OPTIONAL because they need to happen simultaneously +# not to end up in a loop + +# Part 2 + +CONSTRUCT { + ?s foaf:givenName ?gebruikteVoornaam . +} WHERE { + ?s a person:Person ; + persoon_loket:gebruikteVoornaam ?gebruikteVoornaam . +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151927-flush-existing-leidinggevenden-data.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151927-flush-existing-leidinggevenden-data.sparql new file mode 100644 index 0000000..e93d26b --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151927-flush-existing-leidinggevenden-data.sparql @@ -0,0 +1,25 @@ +PREFIX foaf: + +DELETE { + GRAPH { + ?s a ; + ?p ?o. + } +} WHERE { + + VALUES ?p { + + + foaf:givenName + foaf:familyName + + } + + ?functionaris a ; + ?s. + + GRAPH { + ?s a ; + ?p ?o. + } +} \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151928-flush-existing-leidinggevenden-data.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151928-flush-existing-leidinggevenden-data.sparql new file mode 100644 index 0000000..d8c4d26 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151928-flush-existing-leidinggevenden-data.sparql @@ -0,0 +1,25 @@ +PREFIX foaf: + +DELETE { + GRAPH { + ?s a ; + ?p ?o. + } + +} WHERE { + + VALUES ?p { + + + + + + + + } + + GRAPH { + ?s a ; + ?p ?o. + } +} \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151929-flush-existing-leidinggevenden-data.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151929-flush-existing-leidinggevenden-data.sparql new file mode 100644 index 0000000..455cadb --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151929-flush-existing-leidinggevenden-data.sparql @@ -0,0 +1,37 @@ +PREFIX foaf: + +DELETE { + GRAPH { + ?s a ; + ?p ?o. + } + +} WHERE { + + VALUES ?p { + + + + + + + + + + + + + + } + + ?bestuursfunctie a ; + ?contactpoint. + + ?contactpoint a ; + ?s. + + GRAPH { + ?s a ; + ?p ?o. + } +} \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151930-flush-existing-leidinggevenden-data.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151930-flush-existing-leidinggevenden-data.sparql new file mode 100644 index 0000000..9fb0993 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151930-flush-existing-leidinggevenden-data.sparql @@ -0,0 +1,28 @@ +PREFIX foaf: + +DELETE { + GRAPH { + ?s a ; + ?p ?o. + } + +} WHERE { + + VALUES ?p { + + + + + + + + } + + ?bestuursfunctie a ; + ?s. + + GRAPH { + ?s a ; + ?p ?o. + } +} \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151931-flush-existing-leidinggevenden-data.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151931-flush-existing-leidinggevenden-data.sparql new file mode 100644 index 0000000..a60160b --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151931-flush-existing-leidinggevenden-data.sparql @@ -0,0 +1,13 @@ +PREFIX foaf: + +DELETE { + GRAPH { + ?bestuursfunctie ?contactpoint. + } + +} WHERE { + GRAPH { + ?bestuursfunctie a ; + ?contactpoint. + } +} \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151947-flush-landing-graph.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151947-flush-landing-graph.sparql new file mode 100644 index 0000000..2848ff2 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113151947-flush-landing-graph.sparql @@ -0,0 +1 @@ +CLEAR GRAPH \ No newline at end of file diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152654-flush-leidinggevenden-consumer-containers.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152654-flush-leidinggevenden-consumer-containers.sparql new file mode 100644 index 0000000..c92f096 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152654-flush-leidinggevenden-consumer-containers.sparql @@ -0,0 +1,37 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?container ?containerP ?containerO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + task:operation ?operation; + dct:created ?created. + + FILTER(?created < "2024-01-01T00:00:00"^^xsd:dateTime) + + ?task dct:isPartOf ?job; + task:resultsContainer ?container. + + ?container a nfo:DataContainer; + ?containerP ?containerO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152655-flush-leidinggevenden-consumer-tasks.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152655-flush-leidinggevenden-consumer-tasks.sparql new file mode 100644 index 0000000..94475f6 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152655-flush-leidinggevenden-consumer-tasks.sparql @@ -0,0 +1,34 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?task ?taskP ?taskO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + dct:created ?created; + task:operation ?operation. + + FILTER(?created < "2024-01-01T00:00:00"^^xsd:dateTime) + + ?task dct:isPartOf ?job; + ?taskP ?taskO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152656-flush-leidinggevenden-consumer-jobs.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152656-flush-leidinggevenden-consumer-jobs.sparql new file mode 100644 index 0000000..52ef4d4 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2023/20250113152656-flush-leidinggevenden-consumer-jobs.sparql @@ -0,0 +1,32 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?job ?jobP ?jobO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + dct:created ?created; + task:operation ?operation; + ?jobP ?jobO. + + FILTER(?created < "2024-01-01T00:00:00"^^xsd:dateTime) + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152657-flush-leidinggevenden-consumer-containers.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152657-flush-leidinggevenden-consumer-containers.sparql new file mode 100644 index 0000000..842e847 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152657-flush-leidinggevenden-consumer-containers.sparql @@ -0,0 +1,37 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?container ?containerP ?containerO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + task:operation ?operation; + dct:created ?created. + + FILTER(?created < "2025-01-01T00:00:00"^^xsd:dateTime) + + ?task dct:isPartOf ?job; + task:resultsContainer ?container. + + ?container a nfo:DataContainer; + ?containerP ?containerO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152658-flush-leidinggevenden-consumer-tasks.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152658-flush-leidinggevenden-consumer-tasks.sparql new file mode 100644 index 0000000..7847429 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152658-flush-leidinggevenden-consumer-tasks.sparql @@ -0,0 +1,34 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?task ?taskP ?taskO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + dct:created ?created; + task:operation ?operation. + + FILTER(?created < "2025-01-01T00:00:00"^^xsd:dateTime) + + ?task dct:isPartOf ?job; + ?taskP ?taskO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152659-flush-leidinggevenden-consumer-jobs.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152659-flush-leidinggevenden-consumer-jobs.sparql new file mode 100644 index 0000000..4f33643 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2024/20250113152659-flush-leidinggevenden-consumer-jobs.sparql @@ -0,0 +1,32 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?job ?jobP ?jobO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + dct:created ?created; + task:operation ?operation; + ?jobP ?jobO. + + FILTER(?created < "2025-01-01T00:00:00"^^xsd:dateTime) + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152660-flush-leidinggevenden-consumer-containers.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152660-flush-leidinggevenden-consumer-containers.sparql new file mode 100644 index 0000000..4d90259 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152660-flush-leidinggevenden-consumer-containers.sparql @@ -0,0 +1,34 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?container ?containerP ?containerO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + task:operation ?operation. + + ?task dct:isPartOf ?job; + task:resultsContainer ?container. + + ?container a nfo:DataContainer; + ?containerP ?containerO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152661-flush-leidinggevenden-consumer-tasks.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152661-flush-leidinggevenden-consumer-tasks.sparql new file mode 100644 index 0000000..1564173 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152661-flush-leidinggevenden-consumer-tasks.sparql @@ -0,0 +1,31 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?task ?taskP ?taskO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + task:operation ?operation. + + ?task dct:isPartOf ?job; + ?taskP ?taskO. + } +} diff --git a/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152662-flush-leidinggevenden-consumer-jobs.sparql b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152662-flush-leidinggevenden-consumer-jobs.sparql new file mode 100644 index 0000000..32ae341 --- /dev/null +++ b/config/migrations/2025/20250113000000-update-leidinggevenden-consumer/20250113152654-flush-old-jobs/2025/20250113152662-flush-leidinggevenden-consumer-jobs.sparql @@ -0,0 +1,29 @@ +PREFIX mu: +PREFIX task: +PREFIX dct: +PREFIX prov: +PREFIX nie: +PREFIX ext: +PREFIX oslc: +PREFIX cogs: +PREFIX adms: +PREFIX nfo: +PREFIX dbpedia: + + +DELETE { + GRAPH ?g { + ?job ?jobP ?jobO. + } +} WHERE { + VALUES ?operation { + + + } + + GRAPH ?g { + ?job a ; + task:operation ?operation; + ?jobP ?jobO. + } +} diff --git a/docker-compose.yml b/docker-compose.yml index 4928d01..5914f08 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -401,24 +401,26 @@ services: # LOKET LEIDINGGEVENDEN CONSUMERS ################################################################################ leidinggevenden-consumer: - image: lblod/delta-consumer:0.0.24 + image: lblod/delta-consumer:0.1.0 environment: DCR_SERVICE_NAME: "leidinggevenden-consumer" DCR_SYNC_BASE_URL: "https://loket.lblod.info" # replace with link to Loket API + DCR_SYNC_FILES_PATH: "/sync/leidinggevenden/files" DCR_SYNC_DATASET_SUBJECT: "http://data.lblod.info/datasets/delta-producer/dumps/LeidinggevendenCacheGraphDump" DCR_INITIAL_SYNC_JOB_OPERATION: "http://redpencil.data.gift/id/jobs/concept/JobOperation/deltas/consumer/initialSync/leidinggevenden" DCR_DELTA_SYNC_JOB_OPERATION: "http://redpencil.data.gift/id/jobs/concept/JobOperation/deltas/consumer/deltaSync/leidinggevenden" DCR_JOB_CREATOR_URI: "http://data.lblod.info/services/id/leidinggevenden-consumer" DCR_DISABLE_INITIAL_SYNC: "true" - BYPASS_MU_AUTH_FOR_EXPENSIVE_QUERIES: "true" - DCR_SYNC_FILES_PATH: "/sync/leidinggevenden/files" DCR_KEEP_DELTA_FILES: "true" - BATCH_SIZE: 1000 - DCR_ENABLE_DELTA_CONTEXT: "true" - DCR_LANDING_ZONE_DATABASE: "triplestore" + DCR_DELTA_JOBS_RETENTION_PERIOD: "30" + DCR_ENABLE_TRIPLE_REMAPPING: "true" DCR_LANDING_ZONE_GRAPH: "http://mu.semte.ch/graphs/landing-zone/leidinggevenden" + DCR_REMAPPING_GRAPH: "http://mu.semte.ch/graphs/administrative-unit" + DCR_BATCH_SIZE: 1000 + SUDO_QUERY_RETRY_FOR_HTTP_STATUS_CODES: "404,500,503" + SUDO_QUERY_RETRY: "true" volumes: - - ./config/loket-consumer/leidinggevenden-dispatching:/config/triples-dispatching/custom-dispatching + - ./config/loket-consumer/leidinggevenden/mapping:/config/mapping - ./data/files/consumer-files/leidinggevenden:/consumer-files/ restart: always labels: