def test_matrix_ir_parses(self):
hl.index_bgen(resource('example.8bits.bgen'),
reference_genome=hail.get_reference('GRCh37'),
contig_recoding={'01': '1'})
collect_sig = ir.AggSignature('Collect', [], None, [hl.tint32])
collect = ir.MakeStruct([('x',
ir.ApplyAggOp([ir.I32(0)], [], None,
collect_sig,
hl.tarray(hl.tint32)))])
matrix_read = ir.MatrixRead(
resource('backward_compatability/1.0.0/matrix_table/0.hmt'), False,
False)
table_read = ir.TableRead(
resource('backward_compatability/1.0.0/table/0.ht'), False, None)
matrix_irs = [
ir.MatrixUnionRows(ir.MatrixRange(5, 5, 1),
ir.MatrixRange(5, 5, 1)),
ir.UnlocalizeEntries(ir.LocalizeEntries(matrix_read, '__entries'),
ir.MatrixColsTable(matrix_read), '__entries'),
ir.MatrixAggregateRowsByKey(matrix_read, collect, collect),
ir.MatrixAggregateColsByKey(matrix_read, collect, collect),
ir.MatrixRange(1, 1, 10),
ir.MatrixImportVCF([resource('sample.vcf')], False, False,
None, None, False, ['GT'],
hail.get_reference('GRCh37'), {}, True, False),
ir.MatrixImportBGEN([resource('example.8bits.bgen')], ['GP'],
resource('example.sample'), {}, 10, 1,
['varid'], None),
ir.MatrixFilterRows(matrix_read, ir.FalseIR()),
ir.MatrixFilterCols(matrix_read, ir.FalseIR()),
ir.MatrixFilterEntries(matrix_read, ir.FalseIR()),
ir.MatrixChooseCols(matrix_read, [1, 0]),
ir.MatrixMapCols(matrix_read, ir.MakeStruct([('x', ir.I64(20))]),
['x']),
ir.MatrixKeyRowsBy(matrix_read, ['row_i64'], False),
ir.MatrixMapRows(ir.MatrixKeyRowsBy(matrix_read, []),
ir.MakeStruct([('x', ir.I64(20))])),
ir.MatrixMapEntries(matrix_read,
ir.MakeStruct([('x', ir.I64(20))])),
ir.MatrixMapGlobals(matrix_read,
ir.MakeStruct([('x', ir.I64(20))])),
ir.TableToMatrixTable(table_read, ['f32', 'i64'], ['m', 'astruct'],
['aset'], ['mset'], 100),
ir.MatrixCollectColsByKey(matrix_read),
ir.MatrixExplodeRows(matrix_read, ['row_aset']),
ir.MatrixExplodeCols(matrix_read, ['col_aset']),
ir.MatrixAnnotateRowsTable(matrix_read, table_read, '__foo', None),
ir.MatrixAnnotateColsTable(matrix_read, table_read, '__foo'),
]
for x in matrix_irs:
try:
Env.hail().expr.Parser.parse_matrix_ir(str(x))
except Exception as e:
raise ValueError(str(x)) from e
def value_irs(self):
b = ir.TrueIR()
c = ir.Ref('c')
i = ir.I32(5)
j = ir.I32(7)
st = ir.Str('Hail')
a = ir.Ref('a')
aa = ir.Ref('aa')
da = ir.Ref('da')
v = ir.Ref('v')
s = ir.Ref('s')
t = ir.Ref('t')
call = ir.Ref('call')
collect_sig = ir.AggSignature('Collect', [], None, [hl.tint32])
call_stats_sig = ir.AggSignature('CallStats', [], [hl.tint32], [hl.tcall])
call_stats_type = hl.tstruct(AC=hl.tarray(hl.tint32),
AF=hl.tarray(hl.tfloat64),
AN=hl.tint32,
homozygote_count=hl.tarray(hl.tint32))
hist_sig = ir.AggSignature(
'Histogram', [hl.tfloat64, hl.tfloat64, hl.tint32], None, [hl.tfloat64])
hist_type = hl.tstruct(bin_edges=hl.tarray(hl.tfloat64),
bin_freq=hl.tarray(hl.tint64),
n_smaller=hl.tint64,
n_larger=hl.tint64)
take_by_sig = ir.AggSignature('TakeBy', [hl.tint32], None, [hl.tfloat64, hl.tfloat64])
take_by_type = hl.tarray(hl.tfloat64)
value_irs = [
i, ir.I64(5), ir.F32(3.14), ir.F64(3.14), s, ir.TrueIR(), ir.FalseIR(), ir.Void(),
ir.Cast(i, hl.tfloat64),
ir.NA(hl.tint32),
ir.IsNA(i),
ir.If(b, i, j),
ir.Let('v', i, v),
ir.Ref('x'),
ir.ApplyBinaryOp('+', i, j),
ir.ApplyUnaryOp('-', i),
ir.ApplyComparisonOp('EQ', i, j),
ir.MakeArray([i, ir.NA(hl.tint32), ir.I32(-3)], hl.tarray(hl.tint32)),
ir.ArrayRef(a, i),
ir.ArrayLen(a),
ir.ArrayRange(ir.I32(0), ir.I32(5), ir.I32(1)),
ir.ArraySort(a, b, False),
ir.ToSet(a),
ir.ToDict(da),
ir.ToArray(a),
ir.LowerBoundOnOrderedCollection(a, i, True),
ir.GroupByKey(da),
ir.ArrayMap(a, 'v', v),
ir.ArrayFilter(a, 'v', v),
ir.ArrayFlatMap(aa, 'v', v),
ir.ArrayFold(a, ir.I32(0), 'x', 'v', v),
ir.ArrayScan(a, ir.I32(0), 'x', 'v', v),
ir.ArrayFor(a, 'v', ir.Void()),
ir.AggFilter(ir.TrueIR(), ir.I32(0)),
ir.AggExplode(ir.ArrayRange(ir.I32(0), ir.I32(2), ir.I32(1)), 'x', ir.I32(0)),
ir.AggGroupBy(ir.TrueIR(), ir.I32(0)),
ir.ApplyAggOp([], None, [ir.I32(0)], collect_sig),
ir.ApplyScanOp([], None, [ir.I32(0)], collect_sig),
ir.ApplyAggOp([ir.F64(-5.0), ir.F64(5.0), ir.I32(100)], None, [ir.F64(-2.11)], hist_sig),
ir.ApplyAggOp([], [ir.I32(2)], [call], call_stats_sig),
ir.ApplyAggOp([ir.I32(10)], None, [ir.F64(-2.11), ir.F64(-2.11)], take_by_sig),
ir.InitOp(ir.I32(0), [ir.I32(2)], call_stats_sig),
ir.SeqOp(ir.I32(0), [i], collect_sig),
ir.SeqOp(ir.I32(0), [ir.F64(-2.11), ir.I32(17)], take_by_sig),
ir.Begin([ir.Void()]),
ir.MakeStruct([('x', i)]),
ir.SelectFields(s, ['x', 'z']),
ir.InsertFields(s, [('x', i)]),
ir.GetField(s, 'x'),
ir.MakeTuple([i, b]),
ir.GetTupleElement(t, 1),
ir.StringSlice(st, ir.I32(1), ir.I32(2)),
ir.StringLength(st),
ir.In(2, hl.tfloat64),
ir.Die('mumblefoo', hl.tfloat64),
ir.Apply('&&', b, c),
ir.Apply('toFloat64', i),
ir.Uniroot('x', ir.F64(3.14), ir.F64(-5.0), ir.F64(5.0)),
ir.Literal(hl.tarray(hl.tint32), [1, 2, None]),
]
return value_irs
def value_irs(self):
b = ir.TrueIR()
c = ir.Ref('c', hl.tbool)
i = ir.I32(5)
j = ir.I32(7)
st = ir.Str('Hail')
a = ir.Ref('a', hl.tarray(hl.tint32))
aa = ir.Ref('aa', hl.tarray(hl.tarray(hl.tint32)))
da = ir.Ref('da', hl.tarray(hl.ttuple(hl.tint32, hl.tstr)))
v = ir.Ref('v', hl.tint32)
s = ir.Ref('s', hl.tstruct(x=hl.tint32, y=hl.tint64, z=hl.tfloat64))
t = ir.Ref('t', hl.ttuple(hl.tint32, hl.tint64, hl.tfloat64))
call = ir.Ref('call', hl.tcall)
collect_sig = ir.AggSignature('Collect', [], None, [hl.tint32])
call_stats_sig = ir.AggSignature('CallStats', [], [hl.tint32],
[hl.tcall])
hist_sig = ir.AggSignature('Histogram',
[hl.tfloat64, hl.tfloat64, hl.tint32], None,
[hl.tfloat64])
take_by_sig = ir.AggSignature('TakeBy', [hl.tint32], None,
[hl.tfloat64, hl.tfloat64])
value_irs = [
i,
ir.I64(5),
ir.F32(3.14),
ir.F64(3.14), s,
ir.TrueIR(),
ir.FalseIR(),
ir.Void(),
ir.Cast(i, hl.tfloat64),
ir.NA(hl.tint32),
ir.IsNA(i),
ir.If(b, i, j),
ir.Let('v', i, v),
ir.Ref('x', hl.tint32),
ir.ApplyBinaryOp('+', i, j),
ir.ApplyUnaryOp('-', i),
ir.ApplyComparisonOp('EQ', i, j),
ir.MakeArray([i, ir.NA(hl.tint32), ir.I32(-3)],
hl.tarray(hl.tint32)),
ir.ArrayRef(a, i),
ir.ArrayLen(a),
ir.ArrayRange(ir.I32(0), ir.I32(5), ir.I32(1)),
ir.ArraySort(a, b, False),
ir.ToSet(a),
ir.ToDict(da),
ir.ToArray(a),
ir.LowerBoundOnOrderedCollection(a, i, True),
ir.GroupByKey(da),
ir.ArrayMap(a, 'v', v),
ir.ArrayFilter(a, 'v', v),
ir.ArrayFlatMap(aa, 'v', v),
ir.ArrayFold(a, ir.I32(0), 'x', 'v', v),
ir.ArrayFor(a, 'v', ir.Void()),
ir.ApplyAggOp(ir.I32(0), [], None, collect_sig),
ir.ApplyScanOp(ir.I32(0), [], None, collect_sig),
ir.ApplyAggOp(ir.F64(-2.11),
[ir.F64(-5.0),
ir.F64(5.0), ir.I32(100)], None, hist_sig),
ir.ApplyAggOp(call, [], [ir.I32(2)], call_stats_sig),
ir.ApplyAggOp(ir.F64(-2.11), [ir.I32(10)], None, take_by_sig),
ir.InitOp(ir.I32(0), [ir.I32(2)], call_stats_sig),
ir.SeqOp(ir.I32(0), [i], collect_sig),
ir.SeqOp(ir.I32(0), [ir.F64(-2.11), ir.I32(17)], take_by_sig),
ir.Begin([ir.Void()]),
ir.MakeStruct([('x', i)]),
ir.SelectFields(s, ['x', 'z']),
ir.InsertFields(s, [('x', i)]),
ir.GetField(s, 'x'),
ir.MakeTuple([i, b]),
ir.GetTupleElement(t, 1),
ir.StringSlice(st, ir.I32(1), ir.I32(2)),
ir.StringLength(st),
ir.In(2, hl.tfloat64),
ir.Die('mumblefoo', hl.tfloat64),
ir.Apply('&&', b, c),
ir.Apply('toFloat64', i),
ir.Apply('isDefined', s),
ir.Uniroot('x', ir.F64(3.14), ir.F64(-5.0), ir.F64(5.0))
]
return value_irs