now we're getting somewhere. all components are building. just need to clean up build process, test on osx/linux, and actually write the real glue code

This commit is contained in:
Thomas Fussell 2017-07-01 10:46:48 -04:00
parent fa2e66ff2d
commit 361078644e
10 changed files with 605 additions and 90 deletions

1
.gitignore vendored
View File

@ -13,3 +13,4 @@ node_modules/
.DS_Store
__pycache__/
Win32/
*.pyd

View File

@ -11,6 +11,7 @@ option(STATIC "Set to ON to build xlnt as a static library instead of a shared l
option(TESTS "Set to OFF to skip building test executable (in ./tests)" ON)
option(SAMPLES "Set to ON to build executable code samples (in ./samples)" OFF)
option(BENCHMARKS "Set to ON to build performance benchmarks (in ./benchmarks)" OFF)
option(ARROW "Set to ON to build Arrow conversion functions (in ./contrib/xlnt_arrow)" OFF)
# Platform specific options
if(NOT MSVC)
@ -30,4 +31,8 @@ if(TESTS)
add_subdirectory(tests)
endif()
if(ARROW)
add_subdirectory(contrib/xlnt_arrow)
endif()
add_subdirectory(source)

View File

@ -1,25 +0,0 @@
#include "xlnt-arrow.h"
void xlsx2arrow(std::istream &s, arrow::Table &table)
{
xlnt::streaming_workbook_reader reader;
reader.open(s);
reader.begin_worksheet();
int first_row = 0;
while (reader.has_cell())
{
auto cell = reader.read_cell();
if (first_row < 1)
{
first_row = cell.row();
}
}
reader.end_worksheet();
}
void arrow2xlsx(const arrow::Table &table, std::istream &s)
{
}

View File

@ -1,5 +0,0 @@
#include <iostream>
#include <arrow.h>
void xlsx2arrow(std::istream &s, arrow::Table &table);
void arrow2xlsx(const arrow::Table &table, std::istream &s);

View File

@ -0,0 +1,14 @@
cmake_minimum_required(VERSION 3.2)
project(xlnt_arrow)
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(XLNT_ARROW
${CMAKE_CURRENT_SOURCE_DIR}/xlnt_arrow.hpp
${CMAKE_CURRENT_SOURCE_DIR}/xlnt_arrow.cpp)
add_library(xlnt_arrow STATIC ${XLNT_ARROW})
target_include_directories(xlnt_arrow PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
target_include_directories(xlnt_arrow PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../../third-party/arrow/cpp/src)
target_include_directories(xlnt_arrow PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../../include)

View File

@ -0,0 +1,39 @@
#include <xlnt/xlnt.hpp>
#include <xlnt_arrow.hpp>
namespace xlnt {
namespace arrow {
void xlsx2arrow(std::istream &s, ::arrow::Table &table)
{
xlnt::streaming_workbook_reader reader;
reader.open(s);
reader.begin_worksheet();
int first_row = 0;
while (reader.has_cell())
{
auto cell = reader.read_cell();
if (first_row < 1)
{
first_row = cell.row();
}
if (cell.reference().row() % 1000 == 1)
{
std::cout << cell.reference().to_string() << std::endl;
}
}
reader.end_worksheet();
}
void arrow2xlsx(const ::arrow::Table &table, std::istream &s)
{
}
}
}

View File

@ -0,0 +1,11 @@
#include <iostream>
#include <arrow/api.h>
namespace xlnt {
namespace arrow {
void xlsx2arrow(std::istream &s, ::arrow::Table &table);
void arrow2xlsx(const ::arrow::Table &table, std::istream &s);
}
}

View File

@ -0,0 +1,477 @@
#ifndef BOOST_ADAPTBX_PYTHON_STREAMBUF_H
#define BOOST_ADAPTBX_PYTHON_STREAMBUF_H
#include <boost/python/object.hpp>
#include <boost/python/str.hpp>
#include <boost/python/extract.hpp>
#include <boost/optional.hpp>
#include <boost/utility/typed_in_place_factory.hpp>
#include <cassert>
#include <stdexcept>
#include <iostream>
namespace boost_adaptbx { namespace python {
namespace bp = boost::python;
/// A stream buffer getting data from and putting data into a Python file object
/** The aims are as follow:
- Given a C++ function acting on a standard stream, e.g.
\code
void read_inputs(std::istream& input) {
...
input >> something >> something_else;
}
\endcode
and given a piece of Python code which creates a file-like object,
to be able to pass this file object to that C++ function, e.g.
\code
import gzip
gzip_file_obj = gzip.GzipFile(...)
read_inputs(gzip_file_obj)
\endcode
and have the standard stream pull data from and put data into the Python
file object.
- When Python \c read_inputs() returns, the Python object is able to
continue reading or writing where the C++ code left off.
- Operations in C++ on mere files should be competitively fast compared
to the direct use of \c std::fstream.
\b Motivation
- the standard Python library offer of file-like objects (files,
compressed files and archives, network, ...) is far superior to the
offer of streams in the C++ standard library and Boost C++ libraries.
- i/o code involves a fair amount of text processing which is more
efficiently prototyped in Python but then one may need to rewrite
a time-critical part in C++, in as seamless a manner as possible.
\b Usage
This is 2-step:
- a trivial wrapper function
\code
using boost_adaptbx::python::streambuf;
void read_inputs_wrapper(streambuf& input)
{
streambuf::istream is(input);
read_inputs(is);
}
def("read_inputs", read_inputs_wrapper);
\endcode
which has to be written every time one wants a Python binding for
such a C++ function.
- the Python side
\code
from boost.python import streambuf
read_inputs(streambuf(python_file_obj=obj, buffer_size=1024))
\endcode
\c buffer_size is optional. See also: \c default_buffer_size
Note: references are to the C++ standard (the numbers between parentheses
at the end of references are margin markers).
*/
class streambuf : public std::basic_streambuf<char>
{
private:
typedef std::basic_streambuf<char> base_t;
public:
/* The syntax
using base_t::char_type;
would be nicer but Visual Studio C++ 8 chokes on it
*/
typedef base_t::char_type char_type;
typedef base_t::int_type int_type;
typedef base_t::pos_type pos_type;
typedef base_t::off_type off_type;
typedef base_t::traits_type traits_type;
// work around Visual C++ 7.1 problem
inline static int
traits_type_eof() { return traits_type::eof(); }
/// The default size of the read and write buffer.
/** They are respectively used to buffer data read from and data written to
the Python file object. It can be modified from Python.
*/
static std::size_t default_buffer_size;
/// Construct from a Python file object
/** if buffer_size is 0 the current default_buffer_size is used.
*/
streambuf(
bp::object& python_file_obj,
std::size_t buffer_size_=0)
:
py_read (getattr(python_file_obj, "read", bp::object())),
py_write(getattr(python_file_obj, "write", bp::object())),
py_seek (getattr(python_file_obj, "seek", bp::object())),
py_tell (getattr(python_file_obj, "tell", bp::object())),
buffer_size(buffer_size_ != 0 ? buffer_size_ : default_buffer_size),
write_buffer(0),
pos_of_read_buffer_end_in_py_file(0),
pos_of_write_buffer_end_in_py_file(buffer_size),
farthest_pptr(0)
{
assert(buffer_size != 0);
/* Some Python file objects (e.g. sys.stdout and sys.stdin)
have non-functional seek and tell. If so, assign None to
py_tell and py_seek.
*/
if (py_tell != bp::object()) {
try {
py_tell();
}
catch (bp::error_already_set&) {
py_tell = bp::object();
py_seek = bp::object();
/* Boost.Python does not do any Python exception handling whatsoever
So we need to catch it by hand like so.
*/
PyErr_Clear();
}
}
if (py_write != bp::object()) {
// C-like string to make debugging easier
write_buffer = new char[buffer_size + 1];
write_buffer[buffer_size] = '\0';
setp(write_buffer, write_buffer + buffer_size); // 27.5.2.4.5 (5)
farthest_pptr = pptr();
}
else {
// The first attempt at output will result in a call to overflow
setp(0, 0);
}
if (py_tell != bp::object()) {
off_type py_pos = bp::extract<off_type>(py_tell());
pos_of_read_buffer_end_in_py_file = py_pos;
pos_of_write_buffer_end_in_py_file = py_pos;
}
}
/// Mundane destructor freeing the allocated resources
virtual ~streambuf() {
if (write_buffer) delete[] write_buffer;
}
/// C.f. C++ standard section 27.5.2.4.3
/** It is essential to override this virtual function for the stream
member function readsome to work correctly (c.f. 27.6.1.3, alinea 30)
*/
virtual std::streamsize showmanyc() {
int_type const failure = traits_type::eof();
int_type status = underflow();
if (status == failure) return -1;
return egptr() - gptr();
}
/// C.f. C++ standard section 27.5.2.4.3
virtual int_type underflow() {
int_type const failure = traits_type::eof();
if (py_read == bp::object()) {
throw std::invalid_argument(
"That Python file object has no 'read' attribute");
}
read_buffer = py_read(buffer_size);
char *read_buffer_data;
bp::ssize_t py_n_read;
if (PyBytes_AsStringAndSize(read_buffer.ptr(),
&read_buffer_data, &py_n_read) == -1) {
setg(0, 0, 0);
throw std::invalid_argument(
"The method 'read' of the Python file object "
"did not return a string.");
}
off_type n_read = (off_type)py_n_read;
pos_of_read_buffer_end_in_py_file += n_read;
setg(read_buffer_data, read_buffer_data, read_buffer_data + n_read);
// ^^^27.5.2.3.1 (4)
if (n_read == 0) return failure;
return traits_type::to_int_type(read_buffer_data[0]);
}
/// C.f. C++ standard section 27.5.2.4.5
virtual int_type overflow(int_type c=traits_type_eof()) {
if (py_write == bp::object()) {
throw std::invalid_argument(
"That Python file object has no 'write' attribute");
}
farthest_pptr = std::max(farthest_pptr, pptr());
off_type n_written = (off_type)(farthest_pptr - pbase());
bp::str chunk(pbase(), farthest_pptr);
py_write(chunk);
if (!traits_type::eq_int_type(c, traits_type::eof())) {
py_write(traits_type::to_char_type(c));
n_written++;
}
if (n_written) {
pos_of_write_buffer_end_in_py_file += n_written;
setp(pbase(), epptr());
// ^^^ 27.5.2.4.5 (5)
farthest_pptr = pptr();
}
return traits_type::eq_int_type(
c, traits_type::eof()) ? traits_type::not_eof(c) : c;
}
/// Update the python file to reflect the state of this stream buffer
/** Empty the write buffer into the Python file object and set the seek
position of the latter accordingly (C++ standard section 27.5.2.4.2).
If there is no write buffer or it is empty, but there is a non-empty
read buffer, set the Python file object seek position to the
seek position in that read buffer.
*/
virtual int sync() {
int result = 0;
farthest_pptr = std::max(farthest_pptr, pptr());
if (farthest_pptr && farthest_pptr > pbase()) {
off_type delta = pptr() - farthest_pptr;
int_type status = overflow();
if (traits_type::eq_int_type(status, traits_type::eof())) result = -1;
if (py_seek != bp::object()) py_seek(delta, 1);
}
else if (gptr() && gptr() < egptr()) {
if (py_seek != bp::object()) py_seek(gptr() - egptr(), 1);
}
return result;
}
/// C.f. C++ standard section 27.5.2.4.2
/** This implementation is optimised to look whether the position is within
the buffers, so as to avoid calling Python seek or tell. It is
important for many applications that the overhead of calling into Python
is avoided as much as possible (e.g. parsers which may do a lot of
backtracking)
*/
virtual
pos_type seekoff(off_type off, std::ios_base::seekdir way,
std::ios_base::openmode which= std::ios_base::in
| std::ios_base::out)
{
/* In practice, "which" is either std::ios_base::in or out
since we end up here because either seekp or seekg was called
on the stream using this buffer. That simplifies the code
in a few places.
*/
int const failure = off_type(-1);
if (py_seek == bp::object()) {
throw std::invalid_argument(
"That Python file object has no 'seek' attribute");
}
// we need the read buffer to contain something!
if (which == std::ios_base::in && !gptr()) {
if (traits_type::eq_int_type(underflow(), traits_type::eof())) {
return failure;
}
}
// compute the whence parameter for Python seek
int whence;
switch (way) {
case std::ios_base::beg:
whence = 0;
break;
case std::ios_base::cur:
whence = 1;
break;
case std::ios_base::end:
whence = 2;
break;
default:
return failure;
}
// Let's have a go
boost::optional<off_type> result = seekoff_without_calling_python(
off, way, which);
if (!result) {
// we need to call Python
if (which == std::ios_base::out) overflow();
if (way == std::ios_base::cur) {
if (which == std::ios_base::in) off -= egptr() - gptr();
else if (which == std::ios_base::out) off += pptr() - pbase();
}
py_seek(off, whence);
result = off_type(bp::extract<off_type>(py_tell()));
if (which == std::ios_base::in) underflow();
}
return *result;
}
/// C.f. C++ standard section 27.5.2.4.2
virtual
pos_type seekpos(pos_type sp,
std::ios_base::openmode which= std::ios_base::in
| std::ios_base::out)
{
return streambuf::seekoff(sp, std::ios_base::beg, which);
}
private:
bp::object py_read, py_write, py_seek, py_tell;
std::size_t buffer_size;
/* This is actually a Python string and the actual read buffer is
its internal data, i.e. an array of characters. We use a Boost.Python
object so as to hold on it: as a result, the actual buffer can't
go away.
*/
bp::object read_buffer;
/* A mere array of char's allocated on the heap at construction time and
de-allocated only at destruction time.
*/
char *write_buffer;
off_type pos_of_read_buffer_end_in_py_file,
pos_of_write_buffer_end_in_py_file;
// the farthest place the buffer has been written into
char *farthest_pptr;
boost::optional<off_type> seekoff_without_calling_python(
off_type off,
std::ios_base::seekdir way,
std::ios_base::openmode which)
{
boost::optional<off_type> const failure;
// Buffer range and current position
off_type buf_begin, buf_end, buf_cur, upper_bound;
off_type pos_of_buffer_end_in_py_file;
if (which == std::ios_base::in) {
pos_of_buffer_end_in_py_file = pos_of_read_buffer_end_in_py_file;
buf_begin = reinterpret_cast<std::streamsize>(eback());
buf_cur = reinterpret_cast<std::streamsize>(gptr());
buf_end = reinterpret_cast<std::streamsize>(egptr());
upper_bound = buf_end;
}
else if (which == std::ios_base::out) {
pos_of_buffer_end_in_py_file = pos_of_write_buffer_end_in_py_file;
buf_begin = reinterpret_cast<std::streamsize>(pbase());
buf_cur = reinterpret_cast<std::streamsize>(pptr());
buf_end = reinterpret_cast<std::streamsize>(epptr());
farthest_pptr = std::max(farthest_pptr, pptr());
upper_bound = reinterpret_cast<std::streamsize>(farthest_pptr) + 1;
}
else {
throw std::runtime_error("unreachable");
}
// Sought position in "buffer coordinate"
off_type buf_sought;
if (way == std::ios_base::cur) {
buf_sought = buf_cur + off;
}
else if (way == std::ios_base::beg) {
buf_sought = buf_end + (off - pos_of_buffer_end_in_py_file);
}
else if (way == std::ios_base::end) {
return failure;
}
else {
throw std::runtime_error("unreachable");
}
// if the sought position is not in the buffer, give up
if (buf_sought < buf_begin || buf_sought >= upper_bound) return failure;
// we are in wonderland
if (which == std::ios_base::in) gbump(buf_sought - buf_cur);
else if (which == std::ios_base::out) pbump(buf_sought - buf_cur);
return pos_of_buffer_end_in_py_file + (buf_sought - buf_end);
}
public:
class istream : public std::istream
{
public:
istream(streambuf& buf) : std::istream(&buf)
{
exceptions(std::ios_base::badbit);
}
~istream() { if (this->good()) this->sync(); }
};
class ostream : public std::ostream
{
public:
ostream(streambuf& buf) : std::ostream(&buf)
{
exceptions(std::ios_base::badbit);
}
~ostream() { if (this->good()) this->flush(); }
};
};
std::size_t streambuf::default_buffer_size = 1024;
struct streambuf_capsule
{
streambuf python_streambuf;
streambuf_capsule(
bp::object& python_file_obj,
std::size_t buffer_size=0)
:
python_streambuf(python_file_obj, buffer_size)
{}
};
struct ostream : private streambuf_capsule, streambuf::ostream
{
ostream(
bp::object& python_file_obj,
std::size_t buffer_size=0)
:
streambuf_capsule(python_file_obj, buffer_size),
streambuf::ostream(python_streambuf)
{}
~ostream()
{
try {
if (this->good()) this->flush();
}
catch (bp::error_already_set&) {
PyErr_Clear();
throw std::runtime_error(
"Problem closing python ostream.\n"
" Known limitation: the error is unrecoverable. Sorry.\n"
" Suggestion for programmer: add ostream.flush() before"
" returning.");
}
}
};
}} // boost_adaptbx::python
#endif // GUARD

View File

@ -1,5 +1,19 @@
from distutils.core import setup, Extension
setup(
ext_modules=[Extension("xlntpyarrow", ["xlntpyarrow.cpp"])]
setup(name='xlntpyarrow',
include_dirs = [
'C:/Users/Thomas/Development/xlnt/contrib/xlnt_arrow',
'C:/Users/Thomas/Development/xlnt/contrib/xlntpyarrow',
'C:/Users/Thomas/Miniconda3/Library/include'],
ext_modules=[Extension("xlntpyarrow", ["xlntpyarrow.cpp"],
libraries = [
'arrow',
'xlnt_arrow',
'xlnt'
],
library_dirs = [
'C:/Users/Thomas/Miniconda3/Library/lib',
'C:/Users/Thomas/Development/xlnt/build/contrib/xlnt_arrow/Release',
'C:/Users/Thomas/Development/xlnt/build/source/Release'
])]
)

View File

@ -1,13 +1,23 @@
#include <iostream>
#include <memory>
#include <vector>
#include <xlnt_arrow.hpp>
#include <python_streambuf.hpp>
#include <Python.h>
class abc {
public:
static void def()
{
std::cout << "abc" << std::endl;
}
};
PyObject *xlsx2arrow(PyObject *file)
{
boost::python::handle<> boost_file_handle(file);
boost::python::object boost_file(boost_file_handle);
boost_adaptbx::python::streambuf buffer(boost_file);
std::istream stream(&buffer);
std::shared_ptr<arrow::Schema> schema;
std::vector<std::shared_ptr<arrow::Column>> columns;
arrow::Table table(schema, columns);
xlnt::arrow::xlsx2arrow(stream, table);
Py_RETURN_NONE;
}
extern "C" {
@ -18,27 +28,17 @@ PyDoc_STRVAR(xlntpyarrow_xlsx2arrow_doc, "xlsx2arrow(in_file)\
\
Returns an arrow table representing the given XLSX file object.");
PyObject *xlntpyarrow_xlsx2arrow(PyObject *self, PyObject *args, PyObject *kwargs) {
/* Shared references that do not need Py_DECREF before returning. */
PyObject *obj = NULL;
int number = 0;
PyObject *xlntpyarrow_xlsx2arrow(PyObject *self, PyObject *args, PyObject *kwargs)
{
PyObject *file = NULL;
static char* keywords[] = { "file", NULL };
abc::def();
/* Parse positional and keyword arguments */
static char* keywords[] = { "obj", "number", NULL };
if (!PyArg_ParseTupleAndKeywords(args, kwargs, "Oi", keywords, &obj, &number)) {
if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O", keywords, &file))
{
return NULL;
}
/* Function implementation starts here */
if (number < 0) {
PyErr_SetObject(PyExc_ValueError, obj);
return NULL; /* return NULL indicates error */
}
Py_RETURN_NONE;
return xlsx2arrow(file);
}
@ -49,64 +49,47 @@ PyDoc_STRVAR(xlntpyarrow_arrow2xlsx_doc, "arrow2xlsx(table, out_file)\
\
Writes the given arrow table to out_file as an XLSX file.");
PyObject *xlntpyarrow_arrow2xlsx(PyObject *self, PyObject *args, PyObject *kwargs) {
/* Shared references that do not need Py_DECREF before returning. */
PyObject *xlntpyarrow_arrow2xlsx(PyObject *self, PyObject *args, PyObject *kwargs)
{
PyObject *obj = NULL;
int number = 0;
static char* keywords[] = { "obj", NULL };
abc::def();
/* Parse positional and keyword arguments */
static char* keywords[] = { "obj", "number", NULL };
if (!PyArg_ParseTupleAndKeywords(args, kwargs, "Oi", keywords, &obj, &number)) {
if (!PyArg_ParseTupleAndKeywords(args, kwargs, "Oi", keywords, &obj))
{
return NULL;
}
/* Function implementation starts here */
if (number < 0) {
PyErr_SetObject(PyExc_ValueError, obj);
return NULL; /* return NULL indicates error */
}
Py_RETURN_NONE;
}
/*
* List of functions to add to xlntpyarrow in exec_xlntpyarrow().
*/
static PyMethodDef xlntpyarrow_functions[] = {
static PyMethodDef xlntpyarrow_functions[] =
{
{ "xlsx2arrow", (PyCFunction)xlntpyarrow_xlsx2arrow, METH_VARARGS | METH_KEYWORDS, xlntpyarrow_xlsx2arrow_doc },
{ "arrow2xlsx", (PyCFunction)xlntpyarrow_arrow2xlsx, METH_VARARGS | METH_KEYWORDS, xlntpyarrow_arrow2xlsx_doc },
{ NULL, NULL, 0, NULL } /* marks end of array */
};
/*
* Initialize xlntpyarrow. May be called multiple times, so avoid
* using static state.
*/
int exec_xlntpyarrow(PyObject *module) {
int exec_xlntpyarrow(PyObject *module)
{
PyModule_AddFunctions(module, xlntpyarrow_functions);
PyModule_AddStringConstant(module, "__author__", "Thomas Fussell");
PyModule_AddStringConstant(module, "__version__", "0.9.0");
PyModule_AddIntConstant(module, "year", 2017);
return 0; /* success */
return 0;
}
/*
* Documentation for xlntpyarrow.
*/
PyDoc_STRVAR(xlntpyarrow_doc, "The xlntpyarrow module");
static PyModuleDef_Slot xlntpyarrow_slots[] = {
static PyModuleDef_Slot xlntpyarrow_slots[] =
{
{ Py_mod_exec, exec_xlntpyarrow },
{ 0, NULL }
};
static PyModuleDef xlntpyarrow_def = {
static PyModuleDef xlntpyarrow_def =
{
PyModuleDef_HEAD_INIT,
"xlntpyarrow",
xlntpyarrow_doc,
@ -118,8 +101,9 @@ static PyModuleDef xlntpyarrow_def = {
NULL, /* m_free */
};
PyMODINIT_FUNC PyInit_xlntpyarrow() {
PyMODINIT_FUNC PyInit_xlntpyarrow()
{
return PyModuleDef_Init(&xlntpyarrow_def);
}
}
} // extern "C"