/
interpreter.py
100 lines (73 loc) · 3.38 KB
/
interpreter.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
import os
from m2cgen import ast
from m2cgen.interpreters import utils, mixins
from m2cgen.interpreters.c.code_generator import CCodeGenerator
from m2cgen.interpreters.interpreter import ImperativeToCodeInterpreter
class CInterpreter(ImperativeToCodeInterpreter,
mixins.LinearAlgebraMixin):
supported_bin_vector_ops = {
ast.BinNumOpType.ADD: "add_vectors",
}
supported_bin_vector_num_ops = {
ast.BinNumOpType.MUL: "mul_vector_number",
}
abs_function_name = "fabs"
exponent_function_name = "exp"
logarithm_function_name = "log"
log1p_function_name = "log1p"
power_function_name = "pow"
sqrt_function_name = "sqrt"
tanh_function_name = "tanh"
def __init__(self, indent=4, function_name="score", *args, **kwargs):
self.function_name = function_name
cg = CCodeGenerator(indent=indent)
super(CInterpreter, self).__init__(cg, *args, **kwargs)
def interpret(self, expr):
self._cg.reset_state()
self._reset_reused_expr_cache()
args = [(True, self._feature_array_name)]
# C doesn't allow returning vectors, so if model returns vector we will
# have additional vector argument which we will populate at the end.
if expr.output_size > 1:
args += [(True, "output")]
with self._cg.function_definition(
name=self.function_name,
args=args,
is_scalar_output=expr.output_size == 1):
last_result = self._do_interpret(expr)
if expr.output_size > 1:
self._cg.add_assign_array_statement(
last_result, "output", expr.output_size)
else:
self._cg.add_return_statement(last_result)
if self.with_linear_algebra:
filename = os.path.join(
os.path.dirname(__file__), "linear_algebra.c")
self._cg.prepend_code_lines(utils.get_file_content(filename))
if self.with_vectors:
self._cg.add_dependency("<string.h>")
if self.with_math_module:
self._cg.add_dependency("<math.h>")
return self._cg.finalize_and_get_generated_code()
# Both methods supporting linear algebra do several things:
#
# 1. Call super method with extra parameters. Super method will return a
# string with a call to the respective linear algebra function;
# 2. Add variable declaration where the result of the operation will be
# stored;
# 3. Add code returned from super method to the result code;
# 4. Return name of the variable with current result.
def interpret_bin_vector_expr(self, expr, **kwargs):
var_name = self._cg.add_var_declaration(expr.output_size)
# Result: string like "addVectors(v1, v2, <size>, <var_name>)"
func_inv = super().interpret_bin_vector_expr(
expr, extra_func_args=[expr.output_size, var_name], **kwargs)
self._cg.add_code_line(func_inv + ";")
return var_name
def interpret_bin_vector_num_expr(self, expr, **kwargs):
var_name = self._cg.add_var_declaration(expr.output_size)
# Result: string like "mulVectorNumber(v1, num, <size>, <var_name>)"
func_inv = super().interpret_bin_vector_num_expr(
expr, extra_func_args=[expr.output_size, var_name], **kwargs)
self._cg.add_code_line(func_inv + ";")
return var_name