def test_index_dynamic_collections_basic_1(self): index = IndexQ(test_config['indexqbase'], 'testq') solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) if index._is_locked(): index._unlock() self.assertEqual(index.get_all_as_list(), []) # Set up mock for indexing temp = {} def mock(temp, coll, docs): temp[coll] = docs return True todo_file = index.add([ { 'type': '1', 'data': '1' }, { 'type': '1', 'data': '2' }, { 'type': '1', 'data': '3' }, { 'type': '2', 'data': '4' }, { 'type': '3', 'data': '5' }, ], finalize=True) runner_wrap = index._wrap_dynamic(partial(mock, temp), lambda x: x['type'], todo_file) self.assertTrue(runner_wrap) self.assertEqual(json.loads(temp['3']), [{"data": "5", "type": "3"}]) self.assertEqual(json.loads(temp['2']), [{'type': '2', 'data': '4'}]) self.assertEqual( sorted(json.loads(temp['1']), key=lambda x: x['data']), sorted([{ 'type': '1', 'data': '1' }, { 'type': '1', 'data': '2' }, { 'type': '1', 'data': '3' }], key=lambda x: x['data'])) self.assertFalse( index.get_all_as_list()) # Make sure item is completed
def test_index_bad_data(self): index = IndexQ(test_config['indexqbase'], 'testq') solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) if index._is_locked(): index._unlock() self.assertEqual(index.get_all_as_list(),[]) solr.delete_doc_by_id(test_config['SOLR_COLLECTION'],'*') todo_file = index.add({'date':'asd'}, finalize=True) self.assertEqual(index.get_all_as_list()[0],todo_file) with self.assertRaises(SolrError): index.index(solr,test_config['SOLR_COLLECTION']) self.assertEqual(index.get_all_as_list()[0],todo_file) self.assertFalse(index._is_locked())
def test_complete_compress_basic_re_indexing(self): log = logging.getLogger() solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) index = IndexQ(test_config['indexqbase'], 'testq', size = 1, log = log, compress=True) solr.delete_doc_by_id(test_config['SOLR_COLLECTION'], '*') for item in self.docs[1:10]: index.add(item, finalize=True) index.index(solr, test_config['SOLR_COLLECTION']) # At this point items are indexed and are moved into the done directory # Lets re-index them to make sure all json got properly encoded files = index.get_all_as_list('_done_dir') for f in index.get_all_as_list('_done_dir'): shutil.move(f, index._todo_dir) index.index(solr, test_config['SOLR_COLLECTION']) self.assertEqual(files, index.get_all_as_list('_done_dir'))
def test_ignore_fields(self): ''' Will export documents from Solr and put them into an IndexQ. ''' index = IndexQ(test_config['indexqbase'], 'test_reindexer', size=0) for dir in ['_todo_dir', '_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)] reindexer = Reindexer(source=self.solr, source_coll='source_coll', dest=index) for field in ['_version_', 'product_name_exact']: self.assertTrue(field in reindexer._ignore_fields)
def test_ignore_fields(self): """ Will export documents from Solr and put them into an IndexQ. """ index = IndexQ(test_config["indexqbase"], "test_reindexer", size=0) for dir in ["_todo_dir", "_done_dir"]: [os.remove(x) for x in index.get_all_as_list(dir=dir)] reindexer = Reindexer(source=self.solr, source_coll="source_coll", dest=index) for field in ["_version_", "product_name_exact"]: self.assertTrue(field in reindexer._ignore_fields)
def test_complete_compress_basic(self): log = logging.getLogger() index = IndexQ(test_config['indexqbase'], 'testq', size = 1, log = log, compress=True) for item in self.docs[1:10]: index.add(item, finalize=True) files = [] for item in index.get_all_as_list(): files.append(index.complete(item)) [self.assertTrue(os.path.exists(x)) for x in files]
def test_remove_copy_fields_from_data(self): index = IndexQ(test_config["indexqbase"], "test_reindexer", size=0) for dir in ["_todo_dir", "_done_dir"]: [os.remove(x) for x in index.get_all_as_list(dir=dir)] reindexer = Reindexer(source=self.solr, source_coll="source_coll", dest=index) reindexer.reindex() from_files = self.get_all_json_from_indexq(index) excluded_fields = reindexer._ignore_fields for doc in from_files: for field in excluded_fields: if field in doc: print(doc)
def test_remove_copy_fields_from_data(self): index = IndexQ(test_config['indexqbase'], 'test_reindexer', size=0) for dir in ['_todo_dir', '_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)] reindexer = Reindexer(source=self.solr, source_coll='source_coll', dest=index) reindexer.reindex() from_files = self.get_all_json_from_indexq(index) excluded_fields = reindexer._ignore_fields for doc in from_files: for field in excluded_fields: if field in doc: print(doc)
def test_complete_compress_basic_re_indexing(self): log = logging.getLogger() solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) index = IndexQ(test_config['indexqbase'], 'testq', size=1, log=log, compress=True) solr.delete_doc_by_id(test_config['SOLR_COLLECTION'], '*') for item in self.docs[1:10]: index.add(item, finalize=True) index.index(solr, test_config['SOLR_COLLECTION']) # At this point items are indexed and are moved into the done directory # Lets re-index them to make sure all json got properly encoded files = index.get_all_as_list('_done_dir') for f in index.get_all_as_list('_done_dir'): shutil.move(f, index._todo_dir) index.index(solr, test_config['SOLR_COLLECTION']) self.assertEqual(files, index.get_all_as_list('_done_dir'))
def test_complete_dir_rotate(self): log = logging.getLogger() rotate_func = lambda: '{}/{}/{}'.format(dt.now().year, dt.now().month, dt.now().day) index = IndexQ(test_config['indexqbase'], 'testq', size = 1, log = log, rotate_complete = rotate_func) dir_set = rotate_func() docs = self.rand_docs.get_docs(69) for item in self.docs[1:10]: index.add(item, finalize=True) files = [] for item in index.get_all_as_list(): files.append(index.complete(item)) [self.assertTrue(os.path.exists(x)) for x in files]
def test_complete_compress_basic(self): log = logging.getLogger() index = IndexQ(test_config['indexqbase'], 'testq', size=1, log=log, compress=True) for item in self.docs[1:10]: index.add(item, finalize=True) files = [] for item in index.get_all_as_list(): files.append(index.complete(item)) [self.assertTrue(os.path.exists(x)) for x in files]
def test_by_get_all_default_compression(self): size = 1 files = 2 index = IndexQ(test_config['indexqbase'], 'testq', size=size) buff = [] docs = [] for _ in range(files): doc = index.add(self.docs,finalize=True) docs.append(doc) sleep(1) index = IndexQ(test_config['indexqbase'], 'testq', mode='out') indexdocs = index.get_all_as_list() self.assertEqual(docs,indexdocs) [os.remove(doc) for doc in docs]
def test_solr_to_indexq(self): """ Will export documents from Solr and put them into an IndexQ. """ index = IndexQ(test_config["indexqbase"], "test_reindexer", size=0) for dir in ["_todo_dir", "_done_dir"]: [os.remove(x) for x in index.get_all_as_list(dir=dir)] self._index_docs(5000, self.colls[0]) reindexer = Reindexer(source=self.solr, source_coll="source_coll", dest=index) reindexer.reindex() from_files = self.get_all_json_from_indexq(index) from_solr = self.solr.query("source_coll", {"q": "*:*", "rows": 5000}).docs from_solr = reindexer._trim_fields(from_solr) self.assertEqual(sorted(from_files, key=lambda x: x["id"]), sorted(from_solr, key=lambda x: x["id"]))
def test_solr_to_indexq(self): ''' Will export documents from Solr and put them into an IndexQ. ''' index = IndexQ(test_config['indexqbase'], 'test_reindexer', size=0) for dir in ['_todo_dir', '_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)] self._index_docs(5000, self.colls[0]) reindexer = Reindexer(source=self.solr, source_coll='source_coll', dest=index) reindexer.reindex() from_files = self.get_all_json_from_indexq(index) from_solr = self.solr.query('source_coll', {'q': '*:*', 'rows': 5000}).docs from_solr = reindexer._trim_fields(from_solr) self.assertEqual(sorted(from_files, key=lambda x: x['id']), sorted(from_solr, key=lambda x: x['id']))
def test_by_get_all_default_compression(self): size = 1 files = 2 index = IndexQ(test_config['indexqbase'], 'testq', size=size) buff = [] docs = [] for _ in range(files): doc = index.add(self.docs, finalize=True) docs.append(doc) sleep(1) index = IndexQ(test_config['indexqbase'], 'testq', mode='out') indexdocs = index.get_all_as_list() self.assertEqual(docs, indexdocs) [os.remove(doc) for doc in docs]
def test_index_dynamic_collections_indexing_error_partial(self): index = IndexQ(test_config['indexqbase'], 'testq') solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) if index._is_locked(): index._unlock() self.assertEqual(index.get_all_as_list(), []) #Set up mock for indexing temp = {} def mock(temp, coll, docs): if json.loads(docs)[0]['type'] == '1': raise KeyError() else: temp[coll] = docs return True todo_file = index.add([ { 'type': '1', 'data': '1' }, { 'type': '1', 'data': '2' }, { 'type': '1', 'data': '3' }, { 'type': '2', 'data': '4' }, { 'type': '3', 'data': '5' }, ], finalize=True) runner_wrap = index._wrap_dynamic(partial(mock, temp), lambda x: x['type'], todo_file) self.assertFalse(runner_wrap)
def test_dequeue_100(self): size = 1 files = 100 rdocs = self.rand_docs.get_docs(500) index = IndexQ(test_config['indexqbase'], 'testq', size=size) buff = [] docs = [] for dir in ['_todo_dir','_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)] for _ in range(files): doc = index.add(rdocs,finalize=True) docs.append(doc) index = IndexQ(test_config['indexqbase'], 'testq') indexdocs = [] for x in index.get_todo_items(): indexdocs.append(x) self.assertEqual(docs,indexdocs) [os.remove(doc) for doc in docs]
def test_complete_dir_rotate(self): log = logging.getLogger() rotate_func = lambda: '{}/{}/{}'.format(dt.now().year, dt.now().month, dt.now().day) index = IndexQ(test_config['indexqbase'], 'testq', size=1, log=log, rotate_complete=rotate_func) dir_set = rotate_func() docs = self.rand_docs.get_docs(69) for item in self.docs[1:10]: index.add(item, finalize=True) files = [] for item in index.get_all_as_list(): files.append(index.complete(item)) [self.assertTrue(os.path.exists(x)) for x in files]
def test_thread_pool_high(self): ''' Index data using multiple threads. Verity that each thread ''' docs = self.rand_docs.get_docs(25000) index = IndexQ(test_config['indexqbase'], 'testq', size=.1, devel=True) for dir in ['_todo_dir', '_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)] threads = 25 with ThreadPool(threads) as p: p.map(index.add, docs) index.add(finalize=True) d = index.get_all_json_from_indexq() self.assertEqual(len(d), len(docs)) self.assertEqual(sorted(d, key=lambda x: x['id']), sorted(docs, key=lambda x: x['id']))
def test_index_dynamic_collections_func_basic_error_1(self): index = IndexQ(test_config['indexqbase'], 'testq') solr = SolrClient(test_config['SOLR_SERVER'], devel=True, auth=test_config['SOLR_CREDENTIALS']) if index._is_locked(): index._unlock() self.assertEqual(index.get_all_as_list(), []) #Set up mock for indexing temp = {} def mock(temp, coll, docs): temp[coll] = docs todo_file = index.add([ { 'type': '1', 'data': '1' }, { 'type': '1', 'data': '2' }, { 'type': '1', 'data': '3' }, { 'type': '2', 'data': '4' }, { 'type': '3', 'data': '5' }, ], finalize=True) with self.assertRaises(KeyError): index._wrap_dynamic(partial(mock, temp), lambda x: x['asdasdasd'], todo_file)
def setUp(self): index = IndexQ(test_config['indexqbase'], 'testq') for dir in ['_todo_dir', '_done_dir']: [os.remove(x) for x in index.get_all_as_list(dir=dir)]