diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-05-13 16:02:41 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-05-13 16:02:41 -0700 |
commit | aef7b788326313a44e47549af98fc93690b34661 (patch) | |
tree | 6909d1a9c2ae5c8abe7145b77cb883d83af12a1e /fatcat_scholar/search.py | |
parent | 70efe84e5c413dfb8b752faf3b4a5ce31e3d6bf2 (diff) | |
download | fatcat-scholar-aef7b788326313a44e47549af98fc93690b34661.tar.gz fatcat-scholar-aef7b788326313a44e47549af98fc93690b34661.zip |
skeleton of basic search, using covid19 index
Diffstat (limited to 'fatcat_scholar/search.py')
-rw-r--r-- | fatcat_scholar/search.py | 139 |
1 files changed, 139 insertions, 0 deletions
diff --git a/fatcat_scholar/search.py b/fatcat_scholar/search.py new file mode 100644 index 0000000..2373245 --- /dev/null +++ b/fatcat_scholar/search.py @@ -0,0 +1,139 @@ + +""" +Helpers to make elasticsearch queries. +""" + +import json +import datetime + +import elasticsearch +from elasticsearch_dsl import Search, Q +from dynaconf import settings + + +def generic_search_execute(search, limit=25, offset=0, deep_page_limit=2000): + + # Sanity checks + if limit > 100: + limit = 100 + if offset < 0: + offset = 0 + if offset > deep_page_limit: + # Avoid deep paging problem. + offset = deep_page_limit + + search = search[int(offset):int(offset)+int(limit)] + + try: + resp = search.execute() + except elasticsearch.exceptions.RequestError as e: + # this is a "user" error + print("elasticsearch 400: " + str(e.info)) + #flash("Search query failed to parse; you might need to use quotes.<p><code>{}: {}</code>".format(e.error, e.info['error']['root_cause'][0]['reason'])) + # XXX: abort(e.status_code) + raise Exception() + except elasticsearch.exceptions.TransportError as e: + # all other errors + print("elasticsearch non-200 status code: {}".format(e.info)) + # XXX: abort(e.status_code) + raise Exception() + + # convert from objects to python dicts + results = [] + for h in resp: + r = h._d_ + #print(json.dumps(h.meta._d_, indent=2)) + r['_highlights'] = [] + if 'highlight' in dir(h.meta): + highlights = h.meta.highlight._d_ + for k in highlights: + r['_highlights'] += highlights[k] + results.append(r) + + for h in results: + # Handle surrogate strings that elasticsearch returns sometimes, + # probably due to mangled data processing in some pipeline. + # "Crimes against Unicode"; production workaround + for key in h: + if type(h[key]) is str: + h[key] = h[key].encode('utf8', 'ignore').decode('utf8') + + return { + "count_returned": len(results), + "count_found": int(resp.hits.total), + "results": results, + "offset": offset, + "limit": limit, + "deep_page_limit": deep_page_limit, + "query_time_ms": int(resp.took), + } + +def do_fulltext_search(q, limit=25, offset=0, filter_time=None, filter_type=None): + + # Convert raw DOIs to DOI queries + if len(q.split()) == 1 and q.startswith("10.") and q.count("/") >= 1: + q = 'doi:"{}"'.format(q) + + es_client = elasticsearch.Elasticsearch(settings.ELASTICSEARCH_BACKEND) + search = Search(using=es_client, index=settings.ELASTICSEARCH_FULLTEXT_INDEX) + + # type filters + if filter_type == "papers": + search = search.filter("terms", release_type=[ "article-journal", "paper-conference", "chapter", ]) + elif filter_type == "reports": + search = search.filter("terms", release_type=[ "report", "standard", ]) + elif filter_type == "datasets": + search = search.filter("terms", release_type=[ "dataset", "software", ]) + elif filter_type == "everything" or filter_type == None: + pass + else: + # XXX: abort(400) + raise Exception() + + # time filters + if filter_time == "past_week": + week_ago_date = str(datetime.date.today() - datetime.timedelta(days=7)) + search = search.filter("range", release_date=dict(gte=week_ago_date)) + elif filter_time == "this_year": + search = search.filter("term", release_year=datetime.date.today().year) + elif filter_time == "since_2000": + search = search.filter("range", release_year=dict(gte=2000)) + elif filter_time == "before_1925": + search = search.filter("range", release_year=dict(lte=1924)) + elif filter_time == "all_time" or filter_time == None: + pass + else: + # XXX: abort(400) + raise Exception() + + search = search.query( + 'query_string', + query=q, + default_operator="AND", + analyze_wildcard=True, + lenient=True, + fields=[ + "everything", + "abstract", + "fulltext.body", + "fulltext.annex", + ], + ) + search = search.highlight( + "abstract", + "fulltext.body", + "fulltext.annex", + number_of_fragments=3, + fragment_size=150, + ) + + resp = generic_search_execute(search, offset=offset) + + for h in resp['results']: + # Ensure 'contrib_names' is a list, not a single string + if type(h['contrib_names']) is not list: + h['contrib_names'] = [h['contrib_names'], ] + h['contrib_names'] = [name.encode('utf8', 'ignore').decode('utf8') for name in h['contrib_names']] + + resp["query"] = { "q": q } + return resp |