static PyMethodDef tessexMethods[] = {{"recognize", (PyCFunction)tessex_recognize,METH_VARARGS|METH_KEYWORDS,"recognize text in an image."},{NULL, NULL, 0, NULL}
};PyMODINIT_FUNC inittessex(void) {(void) Py_InitModule("tessex", tessexMethods);
}

static PyObject* tessex_recognize(PyObject* self, PyObject* args, PyObject* kw){

　　static const char* kwlist[]={"data", "w", "h", "channels", "step", NULL};

    PyStringObject *data;int w=0;int h=0;int channels=0;int step=0;if (!PyArg_ParseTupleAndKeywords(args, kw, "Siiii", (char**)(kwlist),&data, &w, &h, &channels, &step)) {PyErr_SetString(PyExc_Exception, "Tessex: Failed to parse arguments.");return NULL;}

    tesseract::TessBaseAPI *api = new tesseract::TessBaseAPI();if (api->Init(NULL, "eng")) {PyErr_SetString(PyExc_Exception, "Tessex: Failed to initialize tesseract API.");return NULL;}

    api->SetImage((const unsigned char*)PyString_AS_STRING(data),w, h, channels, step);api->Recognize(0);

    PyObject* l = PyList_New(0);tesseract::ResultIterator* it = api->GetIterator();it->Begin();while(1) {char* utf8_text;int left, top, right, bottom;int confidence = 0;utf8_text = it->GetUTF8Text(tesseract::RIL_TEXTLINE);if (utf8_text == NULL)break;confidence = it->Confidence(tesseract::RIL_TEXTLINE);it->BoundingBox(tesseract::RIL_TEXTLINE, &left, &top, &right, &bottom);PyObject* t = Py_BuildValue("(siiiii)", utf8_text,left, top, right, bottom, confidence);PyList_Append(l, t);delete []utf8_text;it->Next(tesseract::RIL_TEXTLINE);}api->End();PyObject* o = Py_BuildValue("O", l);return o;

#include "Python.h"
#include <tesseract/baseapi.h>
#include <tesseract/resultiterator.h>

from distutils.core import setup, Extensiontessenigma = Extension ('tessex', sources=['tessex.cpp'],include_dirs = ['/usr/local/include'],libraries=[ 'tesseract'], library_dirs=['/usr/local/lib']
)setup (name='tessex', version='1.0',description='This is a tesseract extensiion.',ext_modules = [tessex])

import tesseximport cv

cv_img = cv.LoadImage(path, cv.CV_LOAD_IMAGE_COLOR)

lines = tessex.recognize(data=cv_img.tostring(),w=cv_img.width,h=cv_img.height,channels=cv_img.nChannels, step=cv_img.width * cv_img.nChannels * cv_img.depth / 8)for line in lines:line_text, left, top, right, bottom, confidence = line