def dispatch_task(self, collection, task): stage = task.stage payload = task.payload sync = task.context.get("sync", False) if stage.stage == OP_INDEX: index_many(stage, collection, sync=sync, **payload) if stage.stage == OP_LOAD_MAPPING: load_mapping(stage, collection, **payload) if stage.stage == OP_FLUSH_MAPPING: flush_mapping(stage, collection, sync=sync, **payload) if stage.stage == OP_REINGEST: reingest_collection(collection, job_id=stage.job.id, **payload) if stage.stage == OP_REINDEX: reindex_collection(collection, sync=sync, **payload) if stage.stage == OP_XREF: xref_collection(stage, collection) if stage.stage == OP_XREF_ITEM: xref_item(stage, collection, **payload) log.info("Task [%s]: %s (done)", task.job.dataset, stage.stage)
def handle(self, task): stage = task.stage payload = task.payload collection = Collection.by_foreign_id(task.job.dataset.name) if collection is None: log.error("Collection not found: %s", task.job.dataset) return sync = task.context.get('sync', False) if stage.stage == OP_INDEX: index_many(stage, collection, sync=sync, **payload) if stage.stage == OP_LOAD_MAPPING: load_mapping(stage, collection, **payload) if stage.stage == OP_FLUSH_MAPPING: flush_mapping(stage, collection, sync=sync, **payload) if stage.stage == OP_REINGEST: reingest_collection(collection, job_id=stage.job.id, **payload) if stage.stage == OP_REINDEX: reindex_collection(collection, sync=sync, **payload) if stage.stage == OP_XREF: xref_collection(stage, collection) if stage.stage == OP_XREF_ITEM: xref_item(stage, collection, **payload) log.info("Task [%s]: %s (done)", task.job.dataset, stage.stage)
def reingest_casefiles(index=False): """Re-ingest all the casefile collections.""" for collection in Collection.all_casefiles(): log.info("[%s] Starting to re-ingest", collection) reingest_collection(collection, index=index)
def reingest(foreign_id, index=False): """Process documents and database entities and index them.""" collection = get_collection(foreign_id) reingest_collection(collection, index=index)
def op_reingest_handler(collection, task): reingest_collection(collection, job_id=task.stage.job.id, **task.payload)