def delete_doc_in_cloudsearch(doc_ids, cloudsearch_domain_name, validate=True, region='ap-southeast-2'): """ :param doc_ids: the doc id array :param cloudsearch_domain_name: the domain name !!! This method is in-efficient for validating. """ conn = boto.cloudsearch2.connect_to_region(region) domain = conn.describe_domains([cloudsearch_domain_name]) \ ['DescribeDomainsResponse']['DescribeDomainsResult']['DomainStatusList'] doc_service = CloudSearchDomainConnection( host=domain[0]['DocService']['Endpoint'], region=region) if len(doc_ids) > 0: batch = [] for id in doc_ids: batch.append({"id": str(id), "type": "delete"}) doc_service.upload_documents(json.dumps(batch), 'application/json') if validate: # we need some time for the cloudsearch to apply the deletion. MAX_RETRIES = 100 SLEEP_GAP = 10 for id in doc_ids: retry = 0 while retry < MAX_RETRIES: result = doc_service.search(query="(term field=_id '%s')" % id, query_parser='structured') if (result['hits']['found'] == 0): break retry += 1 sleep(SLEEP_GAP) if retry >= MAX_RETRIES: raise ValueError( "Intended to delete cloudsearch documents with id %s, but it still exists after deletion. " "Culprit document: %s" % (doc_ids, id))
class SearchConnection(object): def __init__(self, domain=None, endpoint=None): self.domain = domain self.endpoint = endpoint self.session = requests.Session() # Endpoint needs to be set before initializing CloudSearchDomainConnection if not endpoint: self.endpoint = domain.search_service_endpoint # Copy proxy settings from connection and check if request should be signed self.sign_request = False if self.domain and self.domain.layer1: if self.domain.layer1.use_proxy: self.session.proxies['http'] = self.domain.layer1.get_proxy_url_with_auth() self.sign_request = getattr(self.domain.layer1, 'sign_request', False) if self.sign_request: layer1 = self.domain.layer1 self.domain_connection = CloudSearchDomainConnection( host=self.endpoint, aws_access_key_id=layer1.aws_access_key_id, aws_secret_access_key=layer1.aws_secret_access_key, region=layer1.region, provider=layer1.provider ) def build_query(self, q=None, parser=None, fq=None, rank=None, return_fields=None, size=10, start=0, facet=None, highlight=None, sort=None, partial=None, options=None): return Query(q=q, parser=parser, fq=fq, expr=rank, return_fields=return_fields, size=size, start=start, facet=facet, highlight=highlight, sort=sort, partial=partial, options=options) def search(self, q=None, parser=None, fq=None, rank=None, return_fields=None, size=10, start=0, facet=None, highlight=None, sort=None, partial=None, options=None): """ Send a query to CloudSearch Each search query should use at least the q or bq argument to specify the search parameter. The other options are used to specify the criteria of the search. :type q: string :param q: A string to search the default search fields for. :type parser: string :param parser: The parser to use. 'simple', 'structured', 'lucene', 'dismax' :type fq: string :param fq: The filter query to use. :type sort: List of strings :param sort: A list of fields or rank expressions used to order the search results. Order is handled by adding 'desc' or 'asc' after the field name. ``['year desc', 'author asc']`` :type return_fields: List of strings :param return_fields: A list of fields which should be returned by the search. If this field is not specified, only IDs will be returned. ``['headline']`` :type size: int :param size: Number of search results to specify :type start: int :param start: Offset of the first search result to return (can be used for paging) :type facet: dict :param facet: Dictionary of fields for which facets should be returned The facet value is string of JSON options ``{'year': '{sort:"bucket", size:3}', 'genres': '{buckets:["Action","Adventure","Sci-Fi"]}'}`` :type highlight: dict :param highlight: Dictionary of fields for which highlights should be returned The facet value is string of JSON options ``{'genres': '{format:'text',max_phrases:2,pre_tag:'<b>',post_tag:'</b>'}'}`` :type partial: bool :param partial: Should partial results from a partioned service be returned if one or more index partitions are unreachable. :type options: str :param options: Options for the query parser specified in *parser*. Specified as a string in JSON format. ``{fields: ['title^5', 'description']}`` :rtype: :class:`boto.cloudsearch2.search.SearchResults` :return: Returns the results of this search The following examples all assume we have indexed a set of documents with fields: *author*, *date*, *headline* A simple search will look for documents whose default text search fields will contain the search word exactly: >>> search(q='Tim') # Return documents with the word Tim in them (but not Timothy) A simple search with more keywords will return documents whose default text search fields contain the search strings together or separately. >>> search(q='Tim apple') # Will match "tim" and "apple" More complex searches require the boolean search operator. Wildcard searches can be used to search for any words that start with the search string. >>> search(q="'Tim*'") # Return documents with words like Tim or Timothy) Search terms can also be combined. Allowed operators are "and", "or", "not", "field", "optional", "token", "phrase", or "filter" >>> search(q="(and 'Tim' (field author 'John Smith'))", parser='structured') Facets allow you to show classification information about the search results. For example, you can retrieve the authors who have written about Tim with a max of 3 >>> search(q='Tim', facet={'Author': '{sort:"bucket", size:3}'}) """ query = self.build_query(q=q, parser=parser, fq=fq, rank=rank, return_fields=return_fields, size=size, start=start, facet=facet, highlight=highlight, sort=sort, partial=partial, options=options) return self(query) def _search_with_auth(self, params): return self.domain_connection.search(params.pop("q", ""), **params) def _search_without_auth(self, params, api_version): url = "http://%s/%s/search" % (self.endpoint, api_version) resp = self.session.get(url, params=params) return {'body': resp.content.decode('utf-8'), 'status_code': resp.status_code} def __call__(self, query): """Make a call to CloudSearch :type query: :class:`boto.cloudsearch2.search.Query` :param query: A group of search criteria :rtype: :class:`boto.cloudsearch2.search.SearchResults` :return: search results """ api_version = '2013-01-01' if self.domain and self.domain.layer1: api_version = self.domain.layer1.APIVersion if self.sign_request: data = self._search_with_auth(query.to_domain_connection_params()) else: r = self._search_without_auth(query.to_params(), api_version) _body = r['body'] _status_code = r['status_code'] try: data = json.loads(_body) except ValueError: if _status_code == 403: msg = '' import re g = re.search('<html><body><h1>403 Forbidden</h1>([^<]+)<', _body) try: msg = ': %s' % (g.groups()[0].strip()) except AttributeError: pass raise SearchServiceException('Authentication error from Amazon%s' % msg) raise SearchServiceException("Got non-json response from Amazon. %s" % _body, query) if 'messages' in data and 'error' in data: for m in data['messages']: if m['severity'] == 'fatal': raise SearchServiceException("Error processing search %s " "=> %s" % (params, m['message']), query) elif 'error' in data: raise SearchServiceException("Unknown error processing search %s" % json.dumps(data), query) data['query'] = query data['search_service'] = self return SearchResults(**data) def get_all_paged(self, query, per_page): """Get a generator to iterate over all pages of search results :type query: :class:`boto.cloudsearch2.search.Query` :param query: A group of search criteria :type per_page: int :param per_page: Number of docs in each :class:`boto.cloudsearch2.search.SearchResults` object. :rtype: generator :return: Generator containing :class:`boto.cloudsearch2.search.SearchResults` """ query.update_size(per_page) page = 0 num_pages_needed = 0 while page <= num_pages_needed: results = self(query) num_pages_needed = results.num_pages_needed yield results query.start += query.real_size page += 1 def get_all_hits(self, query): """Get a generator to iterate over all search results Transparently handles the results paging from Cloudsearch search results so even if you have many thousands of results you can iterate over all results in a reasonably efficient manner. :type query: :class:`boto.cloudsearch2.search.Query` :param query: A group of search criteria :rtype: generator :return: All docs matching query """ page = 0 num_pages_needed = 0 while page <= num_pages_needed: results = self(query) num_pages_needed = results.num_pages_needed for doc in results: yield doc query.start += query.real_size page += 1 def get_num_hits(self, query): """Return the total number of hits for query :type query: :class:`boto.cloudsearch2.search.Query` :param query: a group of search criteria :rtype: int :return: Total number of hits for query """ query.update_size(1) return self(query).hits