Warning |
---|
This functionality is only applicable when building a QPL query manually, if you want custom operators for the parser, you will need to generate a new grammar in which case, please contact the development team |
PyQPL allows to include custom operators in your query and even add translation for these. There are 2 ways to make use of a custom operator, besides the translation process, one quick and simple and another one defining the operator with a class.
For both methods we need to make use of manual operator function (op)
Code Block | ||||
---|---|---|---|---|
| ||||
qpl.op(type_='MY_OPERATOR', operands='one', fields=['title'], boost=1.0) |
This method only consist on using the manual operator function (op), and adding the custom operator as the type, in this example GEO
Code Block | ||||
---|---|---|---|---|
| ||||
qpl = QPL(options=options)
qpl_query = qpl.and_(operands=[
qpl.or_(operands=[
qpl.phrase('San Jose'),
qpl.phrase('Costa Rica'),
qpl.not_(qpl.phrase('California'))
]),
qpl.op(type_='geo', operands=['9°54\'52.5"N', '84°05\'15.5"W'], fields='geo_point', boost=6)
])
print(qpl_query) |
The resulting qpl query will give this
Code Block |
---|
and(or(phrase(San Jose),phrase(Costa Rica),not(phrase(California))),geo_point:geo(9°54'52.5"N,84°05'15.5"W)^6) |
Note |
---|
From the important part is to know the operands(parameters) provided to the the function, since they could be anything we need to be extra careful when doing the translation |
This method is a more formal way to define the custom operator since it relays on a class to process, manipulate and optimize the query
The custom operator inherits from Operator Class, there for you can choose what to overwrite in order to fit your needs, in this case, we are validating that the operands are string, getting the cardinal coordinates and translating them into decimal coordinates
Tip |
---|
The operands are always expected to be Operators or string, so we need to transform the decimals into strings at the end |
Code Block | ||||
---|---|---|---|---|
| ||||
import re
from pyqpl.qpl import QPLOptions, QPL, Operator
from pyqpl.qpl.operator import OperatorOrStr
class GeoOP(Operator):
@staticmethod
def dms_to_decimal(dms_coordinate):
# Use regular expressions to extract degrees, minutes, seconds, and cardinal direction
pattern = r"(\d+)°(\d+)'([\d.]+)\"([NSWE])"
match = re.match(pattern, dms_coordinate)
if match:
degrees = int(match.group(1))
minutes = int(match.group(2))
seconds = float(match.group(3))
cardinal = match.group(4)
# Convert degrees, minutes, and seconds to decimal degrees
decimal = degrees + minutes / 60 + seconds / 3600
# Apply the cardinal direction to determine the sign of the decimal coordinate
if cardinal in ['S', 'W']:
decimal = -decimal
return decimal
else:
raise ValueError("Invalid coordinate format")
def add_operand(self, operand: OperatorOrStr):
if isinstance(operand, str):
op = self.dms_to_decimal(operand)
else:
raise ValueError('Invalid operand type, str expected')
super().add_operand(str(op))
def _optimize(self):
super()._optimize() |
Then when initializing the QPLOptions, we add the parameter custom_operators, this parameter accepts a dictionary, indicating the type of the operator and the class which will handle the type.
The rest of the example is the same but check the result
Code Block | ||||
---|---|---|---|---|
| ||||
options = QPLOptions(fields='content', custom_operators={
'geo': GeoOP
})
qpl = QPL(options=options)
qpl_query = qpl.and_(operands=[
qpl.or_(operands=[
qpl.phrase('San Jose'),
qpl.phrase('Costa Rica'),
qpl.not_(qpl.phrase('California'))
]),
qpl.op(type_='geo', operands=['9°54\'52.5"N', '84°05\'15.5"W'], fields='geo_poit', boost=6)
])
print(qpl_query) |
The result is the same qpl as with the simple method, but this time the coordinates where translated into decimal coordinates
Code Block |
---|
and(or(phrase(San Jose),phrase(Costa Rica),not(phrase(California))),geo_poit:geo(9.914583333333333,-84.08763888888889)^6) |
Translating a custom operator is the same as overwriting the transformation of a default operator, we will use the same steps as in Query Translation, but jus adjust the example to fit the ones used in this page
Code Block | ||||
---|---|---|---|---|
| ||||
import json
from pyqpl.translators import ElasticsearchTranslator
from pyqpl.qpl import OperatorType, Operator
def geo_point_query(operand: Operator):
field = operand.fields_or_defaults[0]
boost = float(field.boost)
if operand.boost != 1:
boost = operand.boost
return {
field.name: {
"type": "Point",
"coordinates": [float(operand.operands[0]), float(operand.operands[1])],
"boots": boost
}
}
translator = ElasticsearchTranslator(custom_or_overwrites={
'geo': geo_point_query
})
engine_query = translator.to_engine_query(qpl_query)
print(json.dumps(engine_query)) |
Code Block |
---|
{
"bool": {
"must": [
{
"bool": {
"should": [
{
"multi_match": {
"type": "phrase",
"query": "San Jose",
"fields": [
"content"
]
}
},
{
"multi_match": {
"type": "phrase",
"query": "Costa Rica",
"fields": [
"content"
]
}
},
{
"bool": {
"must_not": [
{
"multi_match": {
"type": "phrase",
"query": "California",
"fields": [
"content"
]
}
}
]
}
}
],
"minimum_should_match": 1
}
},
{
"geo_poit": {
"type": "Point",
"coordinates": [
9.914583333333333,
-84.08763888888889
],
"boots": 6.0
}
}
]
}
} |