1"""sqlglot expressions - aggregate, window, and statistical functions."""
  2
  3from __future__ import annotations
  4
  5from sqlglot.expressions.core import Expression, Func, AggFunc, Binary
  6
  7
  8class AIAgg(Expression, AggFunc):
  9    arg_types = {"this": True, "expression": True}
 10    _sql_names = ["AI_AGG"]
 11
 12
 13class AISummarizeAgg(Expression, AggFunc):
 14    _sql_names = ["AI_SUMMARIZE_AGG"]
 15
 16
 17class AnyValue(Expression, AggFunc):
 18    pass
 19
 20
 21class ApproximateSimilarity(Expression, AggFunc):
 22    _sql_names = ["APPROXIMATE_SIMILARITY", "APPROXIMATE_JACCARD_INDEX"]
 23
 24
 25class ApproxPercentileAccumulate(Expression, AggFunc):
 26    pass
 27
 28
 29class ApproxPercentileCombine(Expression, AggFunc):
 30    pass
 31
 32
 33class ApproxPercentileEstimate(Expression, Func):
 34    arg_types = {"this": True, "percentile": True}
 35
 36
 37class ApproxQuantiles(Expression, AggFunc):
 38    arg_types = {"this": True, "expression": False}
 39
 40
 41class ApproxTopK(Expression, AggFunc):
 42    arg_types = {"this": True, "expression": False, "counters": False}
 43
 44
 45class ApproxTopKAccumulate(Expression, AggFunc):
 46    arg_types = {"this": True, "expression": False}
 47
 48
 49class ApproxTopKCombine(Expression, AggFunc):
 50    arg_types = {"this": True, "expression": False}
 51
 52
 53class ApproxTopKEstimate(Expression, Func):
 54    arg_types = {"this": True, "expression": False}
 55
 56
 57class ApproxTopSum(Expression, AggFunc):
 58    arg_types = {"this": True, "expression": True, "count": True}
 59
 60
 61class ArgMax(Expression, AggFunc):
 62    arg_types = {"this": True, "expression": True, "count": False}
 63    _sql_names = ["ARG_MAX", "ARGMAX", "MAX_BY"]
 64
 65
 66class ArgMin(Expression, AggFunc):
 67    arg_types = {"this": True, "expression": True, "count": False}
 68    _sql_names = ["ARG_MIN", "ARGMIN", "MIN_BY"]
 69
 70
 71class ArrayAgg(Expression, AggFunc):
 72    arg_types = {"this": True, "nulls_excluded": False}
 73
 74
 75class ArrayConcatAgg(Expression, AggFunc):
 76    pass
 77
 78
 79class ArrayUnionAgg(Expression, AggFunc):
 80    pass
 81
 82
 83class ArrayUniqueAgg(Expression, AggFunc):
 84    pass
 85
 86
 87class Avg(Expression, AggFunc):
 88    pass
 89
 90
 91class Corr(Expression, AggFunc, Binary):
 92    # Correlation divides by variance(column). If a column has 0 variance, the denominator
 93    # is 0 - some dialects return NaN (DuckDB) while others return NULL (Snowflake).
 94    # `null_on_zero_variance` is set to True at parse time for dialects that return NULL.
 95    arg_types = {"this": True, "expression": True, "null_on_zero_variance": False}
 96
 97
 98class Count(Expression, AggFunc):
 99    arg_types = {"this": False, "expressions": False, "big_int": False}
100    is_var_len_args = True
101
102
103class CountIf(Expression, AggFunc):
104    _sql_names = ["COUNT_IF", "COUNTIF"]
105    arg_types = {"this": True, "zero_on_all_null": False}
106
107
108class CovarPop(Expression, AggFunc):
109    arg_types = {"this": True, "expression": True}
110
111
112class CovarSamp(Expression, AggFunc):
113    arg_types = {"this": True, "expression": True}
114
115
116class CumeDist(Expression, AggFunc):
117    arg_types = {"expressions": False}
118    is_var_len_args = True
119
120
121class DenseRank(Expression, AggFunc):
122    arg_types = {"expressions": False}
123    is_var_len_args = True
124
125
126class First(Expression, AggFunc):
127    arg_types = {"this": True, "expression": False}
128
129
130class FirstValue(Expression, AggFunc):
131    pass
132
133
134class GroupConcat(Expression, AggFunc):
135    arg_types = {"this": True, "separator": False, "on_overflow": False}
136
137
138class Grouping(Expression, AggFunc):
139    arg_types = {"expressions": True}
140    is_var_len_args = True
141
142
143class GroupingId(Expression, AggFunc):
144    arg_types = {"expressions": False}
145    is_var_len_args = True
146
147
148class Kurtosis(Expression, AggFunc):
149    pass
150
151
152class Lag(Expression, AggFunc):
153    arg_types = {"this": True, "offset": False, "default": False}
154
155
156class Last(Expression, AggFunc):
157    arg_types = {"this": True, "expression": False}
158
159
160class LastValue(Expression, AggFunc):
161    pass
162
163
164class Lead(Expression, AggFunc):
165    arg_types = {"this": True, "offset": False, "default": False}
166
167
168class LogicalAnd(Expression, AggFunc):
169    _sql_names = ["LOGICAL_AND", "BOOL_AND", "BOOLAND_AGG"]
170
171
172class LogicalOr(Expression, AggFunc):
173    _sql_names = ["LOGICAL_OR", "BOOL_OR", "BOOLOR_AGG"]
174
175
176class Max(Expression, AggFunc):
177    arg_types = {"this": True, "expressions": False}
178    is_var_len_args = True
179
180
181class Median(Expression, AggFunc):
182    pass
183
184
185class Min(Expression, AggFunc):
186    arg_types = {"this": True, "expressions": False}
187    is_var_len_args = True
188
189
190class Minhash(Expression, AggFunc):
191    arg_types = {"this": True, "expressions": True}
192    is_var_len_args = True
193
194
195class MinhashCombine(Expression, AggFunc):
196    pass
197
198
199class Mode(Expression, AggFunc):
200    arg_types = {"this": False, "deterministic": False}
201
202
203class Ntile(Expression, AggFunc):
204    arg_types = {"this": False}
205
206
207class NthValue(Expression, AggFunc):
208    arg_types = {"this": True, "offset": True, "from_first": False}
209
210
211class ObjectAgg(Expression, AggFunc):
212    arg_types = {"this": True, "expression": False}
213
214
215class PercentileCont(Expression, AggFunc):
216    arg_types = {"this": True, "expression": False}
217
218
219class PercentileDisc(Expression, AggFunc):
220    arg_types = {"this": True, "expression": False}
221
222
223PERCENTILES = (PercentileCont, PercentileDisc)
224
225
226class PercentRank(Expression, AggFunc):
227    arg_types = {"expressions": False}
228    is_var_len_args = True
229
230
231class Quantile(Expression, AggFunc):
232    arg_types = {"this": True, "quantile": True}
233
234
235class ApproxQuantile(Quantile):
236    arg_types = {
237        "this": True,
238        "quantile": True,
239        "accuracy": False,
240        "weight": False,
241        "error_tolerance": False,
242    }
243
244
245class Rank(Expression, AggFunc):
246    arg_types = {"expressions": False}
247    is_var_len_args = True
248
249
250class RegrAvgx(Expression, AggFunc):
251    arg_types = {"this": True, "expression": True}
252
253
254class RegrAvgy(Expression, AggFunc):
255    arg_types = {"this": True, "expression": True}
256
257
258class RegrCount(Expression, AggFunc):
259    arg_types = {"this": True, "expression": False}
260
261
262class RegrIntercept(Expression, AggFunc):
263    arg_types = {"this": True, "expression": False}
264
265
266class RegrR2(Expression, AggFunc):
267    arg_types = {"this": True, "expression": False}
268
269
270class RegrSlope(Expression, AggFunc):
271    arg_types = {"this": True, "expression": False}
272
273
274class RegrSxx(Expression, AggFunc):
275    arg_types = {"this": True, "expression": True}
276
277
278class RegrSxy(Expression, AggFunc):
279    arg_types = {"this": True, "expression": True}
280
281
282class RegrSyy(Expression, AggFunc):
283    arg_types = {"this": True, "expression": True}
284
285
286class RegrValx(Expression, AggFunc):
287    arg_types = {"this": True, "expression": True}
288
289
290class RegrValy(Expression, AggFunc):
291    arg_types = {"this": True, "expression": True}
292
293
294class RowNumber(Expression, Func):
295    arg_types = {"this": False}
296
297
298class Skewness(Expression, AggFunc):
299    pass
300
301
302class Stddev(Expression, AggFunc):
303    _sql_names = ["STDDEV", "STDEV"]
304
305
306class StddevPop(Expression, AggFunc):
307    pass
308
309
310class StddevSamp(Expression, AggFunc):
311    pass
312
313
314class Sum(Expression, AggFunc):
315    pass
316
317
318class Variance(Expression, AggFunc):
319    _sql_names = ["VARIANCE", "VARIANCE_SAMP", "VAR_SAMP"]
320
321
322class VariancePop(Expression, AggFunc):
323    _sql_names = ["VARIANCE_POP", "VAR_POP"]