lsup_rdf/cpython/py_graph.h

#ifndef _PY_GRAPH_MOD_H
#define _PY_GRAPH_MOD_H

#define PY_SSIZE_T_CLEAN

#include <Python.h>
#include <structmember.h>

#include "graph.h"
#include "codec/codec_nt.h"
#include "codec/codec_ttl.h"
#include "py_triple.h"


/*
 * Iterator helpers.
 */

/*
 * String iterator for encoder output.
 *
 * Yields one string (one or more lines) at a time.
 */
typedef struct {
    PyObject_HEAD
    void *it;
    const LSUP_Codec *codec;
    char *line;
} StringIteratorObject;


static void
StringIterator_dealloc (StringIteratorObject *it_obj)
{ it_obj->codec->encode_graph_done (it_obj->it); }


static PyObject *
StringIterator_next (StringIteratorObject *it_obj)
{
    LSUP_rc rc = it_obj->codec->encode_graph_iter (
            it_obj->it, &it_obj->line);
    if (rc != LSUP_OK) {
        if (rc != LSUP_END)
            PyErr_SetString (PyExc_ValueError, "Error encoding graph.");

        // If not an error, this raises StopIteration.
        return NULL;
    }

    return PyUnicode_FromString ((char *) it_obj->line);
}


/*
 * String iterator type.
 *
 * Objects of this type are never generated from Python code, rather from
 * Graph_encode, hence the type has no special new or init function.
 */
PyTypeObject StringIteratorType = {
    PyVarObject_HEAD_INIT(&PyType_Type, 0)
    .tp_name            = "graph.StringIterator",
    .tp_basicsize       = sizeof (StringIteratorObject),
    .tp_itemsize        = 0,
    .tp_flags           = Py_TPFLAGS_DEFAULT,
    .tp_dealloc         = (destructor) StringIterator_dealloc,
    .tp_iter            = PyObject_SelfIter,
    .tp_iternext        = (iternextfunc)StringIterator_next,
};


/*
 * Graph iterator.
 *
 * Yields one triple at a time.
 */
typedef struct {
    PyObject_HEAD
    LSUP_GraphIterator *it;
    LSUP_Triple *spo;
} GraphIteratorObject;


static void
GraphIterator_dealloc (GraphIteratorObject *it_obj)
{
    LSUP_graph_iter_free (it_obj->it);
    free (it_obj->spo);
}


static PyObject *
GraphIterator_next (GraphIteratorObject *it_obj)
{
    LSUP_rc rc = LSUP_graph_iter_next (it_obj->it, &it_obj->spo);
    if (rc != LSUP_OK) {
        if (rc != LSUP_END)
            PyErr_SetString (PyExc_ValueError, "Error encoding graph.");

        // If not an error, this raises StopIteration.
        return NULL;
    }

    return build_triple (it_obj->spo);
}


/*
 * Graph iterator type.
 */
PyTypeObject GraphIteratorType = {
    PyVarObject_HEAD_INIT(&PyType_Type, 0)
    .tp_name            = "graph.GraphIterator",
    .tp_basicsize       = sizeof (GraphIteratorObject),
    .tp_itemsize        = 0,
    .tp_flags           = Py_TPFLAGS_DEFAULT,
    .tp_dealloc         = (destructor) GraphIterator_dealloc,
    .tp_iter            = PyObject_SelfIter,
    .tp_iternext        = (iternextfunc) GraphIterator_next,
};


/*
 * Graph stuff.
 */

typedef struct {
    PyObject_HEAD
    LSUP_Graph *ob_struct;
} GraphObject;


static int
Graph_init (GraphObject *self, PyObject *args, PyObject *kwargs)
{
    unsigned char store_type;
    PyObject *uri_obj = NULL;
    LSUP_Term *uri = NULL, *src_uri = NULL;

    static char *kwlist[] = {"", "uri_obj", NULL};

    if (!PyArg_ParseTupleAndKeywords (
            args, kwargs, "b|O", kwlist, &store_type, &uri_obj))
        return -1;

    if (uri_obj) {
        if (!PyObject_TypeCheck (uri_obj, &TermType)) {
            PyErr_SetString (PyExc_TypeError, "uri is not a Term type.");
            return -1;
        }
        src_uri = ((TermObject *) uri_obj)->ob_struct;
        uri = LSUP_iriref_new (src_uri->data, LSUP_iriref_nsm (src_uri));
        if (! LSUP_IS_IRI (uri)) {
            PyErr_SetString (PyExc_TypeError, "uri is not a IRIREF type.");
            return -1;
        }

    } else uri = LSUP_iriref_new (NULL, NULL);

    // Set up the store if a function for that is defined.
    const LSUP_StoreInt *sif = LSUP_store_int (store_type);
    if (UNLIKELY (!sif)) {
            PyErr_SetString (
                    PyExc_TypeError,
                    "No interface defined for given store type.");
            return -1;
    }
    if (sif->setup_fn) {
        if (sif->setup_fn(NULL, false) < LSUP_OK) {
            PyErr_SetString (
                    PyExc_IOError, "Error initializing back end store.");
            return -1;
        }
    }

    // TODO Make store ID, nsm and initial size accessible.
    self->ob_struct = LSUP_graph_new (
            uri, (LSUP_StoreType) store_type, NULL, NULL, 0);
    if (!self->ob_struct) {
        PyErr_SetString (PyExc_ValueError, "Could not create graph.");
        return -1;
    }
    LSUP_Term *uri2 = LSUP_graph_uri (self->ob_struct);
    log_debug("Graph URI (%p): %s", uri2, uri2->data);

    return 0;
}

static void
Graph_dealloc (GraphObject *self)
{
    LSUP_graph_free (self->ob_struct);
    Py_TYPE (self)->tp_free ((PyObject *) self);
}


static PyObject *
Graph_get_uri (GraphObject *self, void *closure)
{
    LSUP_Term *uri = LSUP_graph_uri (self->ob_struct);
    log_debug("Graph URI address: %p", uri);
    log_debug("Graph URI: %s", uri->data);

    return PyUnicode_FromString (uri->data);
}


static int
Graph_set_uri (GraphObject *self, PyObject *value, void *closure)
{
    if (!PyObject_TypeCheck (value, &TermType)) {
        PyErr_SetString (PyExc_TypeError, "URI is not a Term type.");
        return -1;
    }
    LSUP_Term *gr_uri = ((TermObject*)value)->ob_struct;
    log_debug ("New graph URI: %s", (gr_uri->data));

    LSUP_rc rc = LSUP_graph_set_uri (self->ob_struct, LSUP_term_copy (gr_uri));

    return rc == LSUP_OK ? 0 : -1;
}


static PyGetSetDef Graph_getsetters[] = {
    {
        "uri", (getter) Graph_get_uri, (setter) Graph_set_uri,
        "Graph URI.", NULL
    },
    {NULL}
};


static int
Graph_copy_contents (GraphObject *self, GraphObject *dest)
{
    if (LSUP_graph_copy_contents (self->ob_struct, dest->ob_struct) < LSUP_OK)
    {
        PyErr_SetString (PyExc_ValueError, "Error copying graph contents.");
        return -1;
    }

    return 0;
};


static PyObject *
Graph_new_from_rdf (PyTypeObject *cls, PyObject *args)
{
    PyObject *buf, *fileno_fn, *fileno_obj;
    const char *type;
    if (! PyArg_ParseTuple (args, "Os", &buf, &type)) return NULL;

    // Get the file descriptor from the Python BufferedIO object.
    // FIXME This is not sure to be reliable. See
    // https://docs.python.org/3/library/io.html?highlight=io%20bufferedreader#io.IOBase.fileno
    if (! (fileno_fn = PyObject_GetAttrString (buf, "fileno"))) {
        PyErr_SetString (PyExc_TypeError, "Object has no fileno function.");
        return NULL;
    }
    PyObject* fileno_args = PyTuple_New(0);
    if (! (fileno_obj = PyObject_CallObject (fileno_fn, fileno_args))) {
        PyErr_SetString (PyExc_SystemError, "Error calling fileno function.");
        return NULL;
    }
    int fd = PyLong_AsSize_t (fileno_obj);

    /*
     * From the Linux man page:
     *
     * > The file descriptor is not dup'ed, and will be closed when the stream
     * > created by fdopen() is closed. The result of applying fdopen() to a
     * > shared memory object is undefined.
     *
     * Hence the `dup()`.
     */
    fd = dup (fd);
    FILE *fh = fdopen (fd, "r");

    GraphObject *res = (GraphObject *) cls->tp_alloc(cls, 0);
    if (!res) return PyErr_NoMemory();

    const LSUP_Codec *codec;
    if (strcmp(type, "nt") == 0) codec = &nt_codec;
    else if (strcmp (type, "ttl") == 0) codec = &ttl_codec;
    // TODO other codecs here.
    else {
            PyErr_SetString (PyExc_ValueError, "Unsupported codec.");
            return NULL;
    }

    size_t ct;
    char *err;
    codec->decode_graph (fh, &res->ob_struct, &ct, &err);
    fclose (fh);

    log_debug ("Decoded %lu triples.", ct);
    if (UNLIKELY (err)) {
        PyErr_SetString (PyExc_IOError, err);
        return NULL;
    }

    Py_INCREF (res);

    return (PyObject *) res;
}


/** @brief Build a triple pattern for lookup purposes.
 */
inline static int build_trp_pattern (PyObject *args, LSUP_Term *spo[])
{
    PyObject *s_obj, *p_obj, *o_obj;

    if (! (PyArg_ParseTuple (args, "OOO", &s_obj, &p_obj, &o_obj)))
        return -1;

    if (s_obj != Py_None && !PyObject_TypeCheck (s_obj, &TermType)) {
        PyErr_SetString (PyExc_TypeError, "Subject must be a term or None.");
        return -1;
    }
    if (p_obj != Py_None && !PyObject_TypeCheck (p_obj, &TermType)) {
        PyErr_SetString (PyExc_TypeError, "Predicate must be a term or None.");
        return -1;
    }
    if (o_obj != Py_None && !PyObject_TypeCheck (o_obj, &TermType)) {
        PyErr_SetString (PyExc_TypeError, "Object must be a term or None.");
        return -1;
    }

    spo[0] = s_obj != Py_None ? ((TermObject *)s_obj)->ob_struct : NULL;
    spo[1] = p_obj != Py_None ? ((TermObject *)p_obj)->ob_struct : NULL;
    spo[2] = o_obj != Py_None ? ((TermObject *)o_obj)->ob_struct : NULL;

    return 0;
}


static PyObject *
Graph_richcmp (PyObject *self, PyObject *other, int op)
{
    // Only equality and non-equality are supported.
    if (op != Py_EQ && op != Py_NE) Py_RETURN_NOTIMPLEMENTED;

    LSUP_Graph *t1 = ((GraphObject *) self)->ob_struct;
    LSUP_Graph *t2 = ((GraphObject *) other)->ob_struct;

    if (LSUP_graph_equals (t1, t2) ^ (op == Py_NE)) Py_RETURN_TRUE;
    Py_RETURN_FALSE;
 }


static inline PyObject *
Graph_bool_op (
        PyTypeObject *cls, LSUP_bool_op op, PyObject *gr1, PyObject *gr2)
{
    if (! PyObject_TypeCheck (gr1, cls) || ! PyObject_TypeCheck (gr2, cls))
        return NULL;

    GraphObject *res = (GraphObject *) cls->tp_alloc (cls, 0);
    if (!res) return NULL;

    LSUP_Graph *dest = LSUP_graph_new (
            NULL, LSUP_STORE_HTABLE, NULL, NULL, 0);
    if (!dest) {
        PyErr_SetString (PyExc_Exception, "Could not create destination graph.");
        return NULL;
    }

    LSUP_rc rc = LSUP_graph_bool_op (
            op, ((GraphObject *) gr1)->ob_struct,
            ((GraphObject *) gr2)->ob_struct, res->ob_struct);
    if (rc < LSUP_OK) {
        PyErr_SetString (PyExc_Exception, "Error performing boolean operation.");
        return NULL;
    }

    Py_INCREF(res);
    return (PyObject *) res;
}


static PyObject *
Graph_add (PyObject *self, PyObject *triples)
{
    // Triple may be any iterable.
    PyObject *iter = PyObject_GetIter (triples);
    if (! iter) {
        PyErr_SetString (
                PyExc_ValueError, "Triples object cannot be iterated.");
        return NULL;
    }

    PyObject *trp_obj;
    int rc = 0;
    size_t ct = 0;
    LSUP_GraphIterator *it = LSUP_graph_add_init (
            ((GraphObject *)self)->ob_struct);

    while ((trp_obj = PyIter_Next (iter))) {
        if (!PyObject_TypeCheck (trp_obj, &TripleType)) {
            PyErr_SetString (
                    PyExc_ValueError, "Object is not a triple.");
            rc = -1;
            goto finally;
        }

        log_trace ("Inserting triple #%lu", ct);

        LSUP_rc db_rc = LSUP_graph_add_iter (
                it, ((TripleObject *) trp_obj)->ob_struct);

        if (db_rc == LSUP_OK) {
            rc = LSUP_OK;
            ct++;
        } else if (UNLIKELY (db_rc < 0)) {
            PyErr_SetString (PyExc_ValueError, "Error while adding triples.");
            rc = -1;
            goto finally;
        }
        // If db_rc > 0, it's a no-op and the counter is not increased.
    }

finally:
    LSUP_graph_add_done (it);

    if (rc == LSUP_OK) return PyLong_FromSize_t (ct);
    return NULL;
}


static PyObject *Graph_remove (PyObject *self, PyObject *args)
{
    LSUP_rc rc;
    LSUP_Term *spo[3];

    rc = build_trp_pattern (args, spo);
    if (rc < 0) goto finally;

    size_t ct;
    rc = LSUP_graph_remove (
            ((GraphObject *)self)->ob_struct, spo[0], spo[1], spo[2], &ct);

    if (rc < 0) {
        // TODO implement strerror for more details.
        PyErr_SetString (PyExc_SystemError, "Error removing triples.");
        goto finally;
    }
    log_debug ("Removed %lu triples.", ct);

finally:
    if (rc < 0) return NULL;

    Py_RETURN_NONE;
}


static PyObject *Graph_lookup (PyObject *self, PyObject *args)
{
    LSUP_rc rc;
    GraphIteratorObject *it_obj = NULL;
    LSUP_Term *spo[3];

    rc = build_trp_pattern (args, spo);
    if (UNLIKELY (rc < 0)) goto finally;

    size_t ct;
    LSUP_GraphIterator *it = LSUP_graph_lookup (
            ((GraphObject *)self)->ob_struct, spo[0], spo[1], spo[2], &ct);
    if (UNLIKELY (!it)) {
        // TODO implement LSUP_strerror for more details.
        PyErr_SetString (PyExc_SystemError, "Error looking up triples.");
        rc = -1;
        goto finally;
    }
    log_debug ("Found %lu triples.", ct);

    // Initialize the generator object.
    it_obj = PyObject_New (
            GraphIteratorObject, &GraphIteratorType);
    if (UNLIKELY (!it_obj)) return PyErr_NoMemory();

    it_obj->it = it;
    it_obj->spo = TRP_DUMMY;

    Py_INCREF (it_obj);

finally:
    return (PyObject *)it_obj;
}


static PyObject *
Graph_encode (PyObject *self, PyObject *args)
{
    const char *type;

    if (! PyArg_ParseTuple (args, "s", &type)) return NULL;

    // Initialize the generator object.
    StringIteratorObject *it_obj = PyObject_New (
            StringIteratorObject, &StringIteratorType);
    if (!it_obj) return NULL;

    if (strcmp (type, "nt") == 0) it_obj->codec = &nt_codec;
    else if (strcmp (type, "ttl") == 0) it_obj->codec = &ttl_codec;
    // TODO other codecs here.
    else {
        PyErr_SetString (PyExc_ValueError, "Unsupported codec.");
        return NULL;
    }

    it_obj->it = it_obj->codec->encode_graph_init (
            ((GraphObject *)self)->ob_struct);

    it_obj->line = NULL;

    Py_INCREF (it_obj);
    return (PyObject *)it_obj;
}


static PyMethodDef Graph_methods[] = {
    {
        "copy", (PyCFunction) Graph_copy_contents, METH_CLASS | METH_VARARGS,
        "Copy the contents of a graph into another."
    },
    {
        "from_rdf", (PyCFunction) Graph_new_from_rdf,
        METH_CLASS | METH_VARARGS,
        "Create a graph from a RDF file."
    },
    {"add", (PyCFunction) Graph_add, METH_O, "Add triples to a graph."},
    {
        "remove", (PyCFunction) Graph_remove, METH_VARARGS,
        "Remove triples from a graph by matching a pattern."
    },
    {
        "lookup", (PyCFunction) Graph_lookup, METH_VARARGS,
        "Look triples in a graph by matching a pattern."
    },
    {
        "to_rdf", (PyCFunction) Graph_encode, METH_VARARGS,
        "Encode a graph into a RDF byte buffer."
    },
    {NULL},
};


static inline PyObject *Graph_bool_and (
        PyTypeObject *cls, PyObject *gr1, PyObject *gr2)
{ return Graph_bool_op (cls, LSUP_BOOL_INTERSECTION, gr1, gr2); }

static inline PyObject *Graph_bool_or (
        PyTypeObject *cls, PyObject *gr1, PyObject *gr2)
{ return Graph_bool_op (cls, LSUP_BOOL_UNION, gr1, gr2); }

static inline PyObject *Graph_bool_subtract (
        PyTypeObject *cls, PyObject *gr1, PyObject *gr2)
{ return Graph_bool_op (cls, LSUP_BOOL_SUBTRACTION, gr1, gr2); }

static inline PyObject *Graph_bool_xor (
        PyTypeObject *cls, PyObject *gr1, PyObject *gr2)
{ return Graph_bool_op (cls, LSUP_BOOL_XOR, gr1, gr2); }

static PyNumberMethods Graph_number_methods = {
    .nb_and = (binaryfunc) Graph_bool_and,
    .nb_or = (binaryfunc) Graph_bool_or,
    .nb_subtract = (binaryfunc) Graph_bool_subtract,
    .nb_xor = (binaryfunc) Graph_bool_xor,
};


static int
Graph_contains (PyObject *self, PyObject *value)
{
    if (!PyObject_TypeCheck (value, &TripleType)) {
        PyErr_SetString (PyExc_ValueError, "Error parsing input value.");
        return -1;
    }

    int rc = LSUP_graph_contains (
            ((GraphObject *) self)->ob_struct,
            ((TripleObject *) value)->ob_struct);

    return rc;
}


static Py_ssize_t
Graph_get_size (PyObject *self)
{ return LSUP_graph_size (((GraphObject *) self)->ob_struct); }


static PySequenceMethods Graph_seq_methods = {
    .sq_length = (lenfunc) Graph_get_size,
    .sq_contains = (objobjproc) Graph_contains,
};


PyTypeObject GraphType = {
    PyVarObject_HEAD_INIT(NULL, 0)
    .tp_name            = "graph.Graph",
    .tp_doc             = "RDF graph",
    .tp_basicsize       = sizeof (GraphObject),
    .tp_itemsize        = 0,
    .tp_flags           = Py_TPFLAGS_DEFAULT,
    .tp_new             = PyType_GenericNew,
    .tp_init            = (initproc) Graph_init,
    .tp_dealloc         = (destructor) Graph_dealloc,
    .tp_getset          = Graph_getsetters,
    .tp_methods         = Graph_methods,
    .tp_richcompare     = (richcmpfunc) Graph_richcmp,
    .tp_as_number       = &Graph_number_methods,
    .tp_as_sequence     = &Graph_seq_methods,
};

#endif