動作環境

GeForce GTX 1070 (8GB)
ASRock Z170M Pro4S [Intel Z170chipset]
Ubuntu 14.04 LTS desktop amd64
TensorFlow v0.11
cuDNN v5.1 for Linux
CUDA v8.0
Python 2.7.6
IPython 5.1.0 -- An enhanced Interactive Python.
gcc (Ubuntu 4.8.4-2ubuntu1~14.04.3) 4.8.4

C++のコードで生成する係数をJupyter/matplotlibで画像表示したい。
それにはCのfwrite()結果をPythonで読込むことになりそう。

実装してみた。

v0.1 (1つの値)

C (write)

writeTest.c

#include <stdio.h>
#include <stdlib.h>

int main(void){
	FILE *wfp;
	float pi = 3.141592;

	wfp = fopen("sample.bin","wb");
	fwrite(&pi, sizeof(float), 1, wfp);
	fclose(wfp);
}

Python (read)

参考 http://stackoverflow.com/questions/2146031/what-is-the-equivalent-of-fread-from-matlab-in-python

readTest.py

import numpy as np
import array

with open("sample.bin") as rfp:
	a = array.array("f") # f: typecode for float
	a.fromfile(rfp, 1)
	print (a)

上記のtypecodeは以下を参考にした。
https://docs.python.org/3.1/library/array.html

実行

sample.bin生成

$ gcc writeTest.c && ./a.out

sample.bin読込み

$ python readTest.py 
array('f', [3.141592025756836])

読めた。

v0.2 (3つの値)

C (write)

writeTest.c

#include <stdio.h>
#include <stdlib.h>

int main(void){
	FILE *wfp;
	float pi = 3.141592;
	float napier = 2.718;
	float avogadro1 = 6.022;

	wfp = fopen("sample.bin","wb");
	fwrite(&pi, sizeof(float), 1, wfp);
	fwrite(&napier, sizeof(float), 1, wfp);
	fwrite(&avogadro1, sizeof(float), 1, wfp);
	fclose(wfp);
}

Python (read)

readTest.py

import numpy as np
import array

with open("sample.bin","rb") as rfp:
	a = array.array("f") # f: typecode for float
	a.fromfile(rfp, 3)
	print (a)

実行

sample.bin生成

$ gcc writeTest.c && ./a.out

sample.bin読込み

$ python readTest.py 
array('f', [3.141592025756836, 2.7179999351501465, 6.021999835968018])

v0.3 (size_t)対応

(追記 2016/12/19)

size_t型の読込みで躓いていた。
v0.1 @ http://qiita.com/7of9/items/68673ac3239532064c28

size_t型

sizeofPrint.c

#include <stdio.h>
#include <stdlib.h>

int main(void) {
	printf("%ld\n", sizeof(size_t));
	printf("%ld\n", sizeof(int));
}

実行

$ gcc sizeofPrint.c && ./a.out 
8
4

こちらの環境ではsize_tは8。

C (write)

writeTest.c

#include <stdio.h>
#include <stdlib.h>

int main(void){
	FILE *wfp;
	size_t pi = 314;
	float napier = 2.718;
	float avogadro1 = 6.022;

	wfp = fopen("sample.bin","wb");
	fwrite(&pi, sizeof(size_t), 1, wfp);
	fwrite(&napier, sizeof(float), 1, wfp);
	fwrite(&avogadro1, sizeof(float), 1, wfp);
	fclose(wfp);
}

Python (read)

Python側でsize_t(sizeof()=8)の値を読むには[unsigned long]型だろう。

https://docs.python.org/3.1/library/array.html
によるとL

readTest.py

import numpy as np
import array

with open("sample.bin","rb") as rfp:
	szv = array.array("L") # L: typecodef for [unsigned long]
	szv.fromfile(rfp, 1)
	print (szv)
	flv = array.array("f") # f: typecode for float
	flv.fromfile(rfp, 2)
	print (flv)

実行

$ python readTest.py 
array('L', [314L])
array('f', [2.7179999351501465, 6.021999835968018])

v0.4 (double complex読込み)

(追記 2016/12/20)

double complex型を読込むにはどうするか。
double型でとりあえず読んでみた。

C (write)

アボガドロ数の実際の値とは違うがとりあえず。

writeTest.c

#include <stdio.h>
#include <stdlib.h>
#include <complex.h>

int main(void){
	FILE *wfp;
	size_t pi = 314;
	float napier = 2.718;
	double complex avogadro = 6.022 + 10.23 * I; 

	wfp = fopen("sample.bin","wb");
	fwrite(&pi, sizeof(size_t), 1, wfp);
	fwrite(&napier, sizeof(float), 1, wfp);
	fwrite(&avogadro, sizeof(double complex), 1, wfp);
	fclose(wfp);
}

Python (read)

readTest.py

import numpy as np
import array

with open("sample.bin","rb") as rfp:
	szv = array.array("L") # L: typecodef for [unsigned long]
	szv.fromfile(rfp, 1)
	print (szv)
	napier = array.array("f") # f: typecode for float
	napier.fromfile(rfp, 1)
	print (napier)
	avogadro = array.array("d") # d: typecode for double
	avogadro.fromfile(rfp, 2) # real and imaginary part of the complex number
	print (avogadro)

実行

$ gcc writeTest.c && ./a.out
$ python readTest.py 
array('L', [314L])
array('f', [2.7179999351501465])
array('d', [6.022, 10.23])

とりあえずdouble complexの実部と虚部の読込みはできた。
あとは２つの値を元にcomplex型にキャストするというような方法だろうか。