def suggest_query_full_example(mapping, input_doc, query_text, analyzer=None):
"""
no mapping provided
"""
INDEX = "pytest"
es = create_index(INDEX, mapping=mapping, analyzer=analyzer)
# update_index_mapping(INDEX, es, mapping)
insert_single_doc(INDEX, es, input_doc)
query_index_mapping(INDEX, es)
if analyzer is None:
body = {"text": query_text, "suggest-1": {"term": {"field": "tags"}}}
else:
analyzer_name = get_analyzer_name(analyzer)
query_analyzer(INDEX, es, analyzer_name, query_text)
body = {
"text": query_text,
"suggest-1": {"term": {"field": "tags", "analyzer": analyzer_name}},
}
res = suggest_query(INDEX, es, body=body)
res = res["suggest-1"]
for r in res:
inp = r["text"]
opt = [o["text"] for o in r["options"]]
print(f"input text : {inp}\noptions :")
pp.pprint(opt)
print()
delete_index(INDEX, es)
return opt
def analyze_text(txt, analyzer):
INDEX = "pytest"
input_doc = {"tags": [txt]}
delete_index(INDEX)
es = create_index(INDEX, analyzer=analyzer)
insert_single_doc(INDEX, es, input_doc)
res = query_analyzer(INDEX, es, get_analyzer_name(analyzer), txt)
delete_index(INDEX, es)
tokens = [e["token"] for e in res["tokens"]]
return tokens
def suggest_phrase_query_full_example(mapping, input_doc, query_text, analyzer=None):
INDEX = "pytest"
es = create_index(INDEX, mapping=mapping, analyzer=analyzer)
insert_single_doc(INDEX, es, input_doc)
query_index_mapping(INDEX, es)
if analyzer is None:
body = {
"text": query_text,
"suggest-1": {
"phrase": {
"field": "tags",
"size": 5,
"gram_size": 3,
"direct_generator": [{"field": "tags", "suggest_mode": "always"}],
"highlight": {"pre_tag": "<em>", "post_tag": "</em>"},
}
},
}
else:
analyzer_name = get_analyzer_name(analyzer)
query_analyzer(INDEX, es, analyzer_name, query_text)
body = {
"text": query_text,
"suggest-1": {
"phrase": {
"field": "tags",
"analyzer": analyzer_name,
"size": 5,
"gram_size": 3,
"direct_generator": [{"field": "tags", "suggest_mode": "always"}],
"highlight": {"pre_tag": "<em>", "post_tag": "</em>"},
}
},
}
res = suggest_query(INDEX, es, body=body)
res = res["suggest-1"]
for r in res:
inp = r["text"]
opt = [o["text"] for o in r["options"]]
print(f"input text : {inp}\noptions :")
pp.pprint(opt)
print()
delete_index(INDEX, es)
return opt
def suggest_phrase_query_full_example_mimacom(input_doc,
query_text,
body=None,
mapping=None,
analyzer=None,
fuzziness=None):
# https://blog.mimacom.com/autocomplete-elasticsearch-part3/
INDEX = "pytest"
if mapping is None:
mapping = {"tags": {"type": "completion"}}
es = create_index(INDEX, mapping=mapping, analyzer=analyzer)
insert_single_doc(INDEX, es, input_doc)
query_index_mapping(INDEX, es)
if body is None:
body = {
"suggest-1": {
"prefix": query_text,
"completion": {
"field": "tags"
},
}
}
if not (fuzziness is None):
for k in body.keys():
d = body[k]["completion"]
d.update({"fuzzy": {"fuzziness": 1}})
body[k]["completion"] = d
res = suggest_query(INDEX, es, body=body)
# pp.pprint(res)
opt = {}
for k, v in res.items():
for r in v:
inp = r["text"]
opt[k] = [o["text"] for o in r["options"]]
print(f"input text : {inp}\noptions :")
pp.pprint(opt)
print()
delete_index(INDEX, es)
return opt
def query_disease_example(input_doc,
query_text,
body=None,
mapping=None,
analyzer=None,
fuzziness=None):
INDEX = "pytest"
if mapping is None:
mapping = {"diseases": {"type": "text"}}
es = create_index(INDEX, mapping=mapping, analyzer=analyzer)
insert_single_doc(INDEX, es, input_doc)
query_index_mapping(INDEX, es)
if body is None:
body = {
"query": {
"bool": {
"must": [{
"match": {
"diseases": {
"query": query_text,
"fuzziness": 0 if fuzziness is None else 1,
}
}
}]
}
}
}
res = query_index(INDEX, es, body)
delete_index(INDEX, es)
try:
return res[0]["_source"]["diseases"]
except IndexError:
return []
inp = r["text"]
opt[k] = [o["text"] for o in r["options"]]
print(f"input text : {inp}\noptions :")
pp.pprint(opt)
print()
delete_index(INDEX, es)
return opt
################################################################################
if __name__ == "__main__":
if False:
INDEX = "pytest"
delete_index(INDEX)
input_doc = {"tags": ["Bactéries", "Appareil génital féminin"]}
query_text = "appar"
suggest_phrase_query_full_example_mimacom(input_doc, query_text)
if False:
INDEX = "pytest"
delete_index(INDEX)
input_doc = {"tags": ["Bactéries", "Appareil génital féminin"]}
query_text = "apareil gén"
suggest_phrase_query_full_example_mimacom(input_doc,
query_text,
fuzziness=1)
if False: # accent
# old version of ES !!
return opt
################################################################################
if __name__ == "__main__":
if False:
suggest_query_full_example(
{"tags": {"type": "keyword"}}, {"tags": ["Bactéries"]}, "Bactéri"
)
if False:
print(get_analyzer_name(FR_ANALYZER))
if False: # test analyzer
INDEX = "pytest"
mapping = {"tags": {"type": "keyword"}}
input_doc = {"tags": ["Bactéries", "bact", "Appareil génital féminin"]}
delete_index(INDEX)
for analyzer in [
STD_ANALYZER,
FR_ANALYZER,
TRI_GRAM_ANALYZER,
REVERSE_ANALYZER,
ACCENT_ANALYZER,
]:
for query_text in input_doc["tags"]:
es = create_index(INDEX, mapping=mapping, analyzer=analyzer)
insert_single_doc(INDEX, es, input_doc)
query_analyzer(INDEX, es, get_analyzer_name(analyzer), query_text)
delete_index(INDEX, es)