dulwich/_objects.c

   1 /*
   2  * Copyright (C) 2009 Jelmer Vernooij <jelmer@samba.org>
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License
   6  * as published by the Free Software Foundation; version 2
   7  * of the License or (at your option) a later version of the License.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program; if not, write to the Free Software
  16  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
  17  * MA  02110-1301, USA.
  18  */
  19
  20 #include <Python.h>
  21 #include <stdlib.h>
  22 #include <sys/stat.h>
  23
  24 #if (PY_VERSION_HEX < 0x02050000)
  25 typedef int Py_ssize_t;
  26 #endif
  27
  28 #if defined(__MINGW32_VERSION) || defined(__APPLE__)
  29 size_t rep_strnlen(char *text, size_t maxlen);
  30 size_t rep_strnlen(char *text, size_t maxlen)
  31 {
  32         const char *last = memchr(text, '\0', maxlen);
  33         return last ? (size_t) (last - text) : maxlen;
  34 }
  35 #define strnlen rep_strnlen
  36 #endif
  37
  38 #define bytehex(x) (((x)<0xa)?('0'+(x)):('a'-0xa+(x)))
  39
  40 static PyObject *tree_entry_cls;
  41 static PyObject *object_format_exception_cls;
  42
  43 static PyObject *sha_to_pyhex(const unsigned char *sha)
  44 {
  45         char hexsha[41];
  46         int i;
  47         for (i = 0; i < 20; i++) {
  48                 hexsha[i*2] = bytehex((sha[i] & 0xF0) >> 4);
  49                 hexsha[i*2+1] = bytehex(sha[i] & 0x0F);
  50         }
  51
  52         return PyString_FromStringAndSize(hexsha, 40);
  53 }
  54
  55 static PyObject *py_parse_tree(PyObject *self, PyObject *args, PyObject *kw)
  56 {
  57         char *text, *start, *end;
  58         int len, namelen, strict;
  59         PyObject *ret, *item, *name, *sha, *py_strict = NULL;
  60         static char *kwlist[] = {"text", "strict", NULL};
  61
  62         if (!PyArg_ParseTupleAndKeywords(args, kw, "s#|O", kwlist,
  63                                          &text, &len, &py_strict))
  64                 return NULL;
  65         strict = py_strict ?  PyObject_IsTrue(py_strict) : 0;
  66         /* TODO: currently this returns a list; if memory usage is a concern,
  67          * consider rewriting as a custom iterator object */
  68         ret = PyList_New(0);
  69         if (ret == NULL) {
  70                 return NULL;
  71         }
  72         start = text;
  73         end = text + len;
  74         while (text < end) {
  75                 long mode;
  76                 if (strict && text[0] == '0') {
  77                         PyErr_SetString(object_format_exception_cls,
  78                                         "Illegal leading zero on mode");
  79                         Py_DECREF(ret);
  80                         return NULL;
  81                 }
  82                 mode = strtol(text, &text, 8);
  83                 if (*text != ' ') {
  84                         PyErr_SetString(PyExc_ValueError, "Expected space");
  85                         Py_DECREF(ret);
  86                         return NULL;
  87                 }
  88                 text++;
  89                 namelen = strnlen(text, len - (text - start));
  90                 name = PyString_FromStringAndSize(text, namelen);
  91                 if (name == NULL) {
  92                         Py_DECREF(ret);
  93                         return NULL;
  94                 }
  95                 if (text + namelen + 20 >= end) {
  96                         PyErr_SetString(PyExc_ValueError, "SHA truncated");
  97                         Py_DECREF(ret);
  98                         Py_DECREF(name);
  99                         return NULL;
 100                 }
 101                 sha = sha_to_pyhex((unsigned char *)text+namelen+1);
 102                 if (sha == NULL) {
 103                         Py_DECREF(ret);
 104                         Py_DECREF(name);
 105                         return NULL;
 106                 }
 107                 item = Py_BuildValue("(NlN)", name, mode, sha);
 108                 if (item == NULL) {
 109                         Py_DECREF(ret);
 110                         Py_DECREF(sha);
 111                         Py_DECREF(name);
 112                         return NULL;
 113                 }
 114                 if (PyList_Append(ret, item) == -1) {
 115                         Py_DECREF(ret);
 116                         Py_DECREF(item);
 117                         return NULL;
 118                 }
 119                 Py_DECREF(item);
 120                 text += namelen+21;
 121         }
 122         return ret;
 123 }
 124
 125 struct tree_item {
 126         const char *name;
 127         int mode;
 128         PyObject *tuple;
 129 };
 130
 131 int cmp_tree_item(const void *_a, const void *_b)
 132 {
 133         const struct tree_item *a = _a, *b = _b;
 134         const char *remain_a, *remain_b;
 135         int ret, common;
 136         if (strlen(a->name) > strlen(b->name)) {
 137                 common = strlen(b->name);
 138                 remain_a = a->name + common;
 139                 remain_b = (S_ISDIR(b->mode)?"/":"");
 140         } else if (strlen(b->name) > strlen(a->name)) {
 141                 common = strlen(a->name);
 142                 remain_a = (S_ISDIR(a->mode)?"/":"");
 143                 remain_b = b->name + common;
 144         } else { /* strlen(a->name) == strlen(b->name) */
 145                 common = 0;
 146                 remain_a = a->name;
 147                 remain_b = b->name;
 148         }
 149         ret = strncmp(a->name, b->name, common);
 150         if (ret != 0)
 151                 return ret;
 152         return strcmp(remain_a, remain_b);
 153 }
 154
 155 int cmp_tree_item_name_order(const void *_a, const void *_b) {
 156         const struct tree_item *a = _a, *b = _b;
 157         return strcmp(a->name, b->name);
 158 }
 159
 160 static PyObject *py_sorted_tree_items(PyObject *self, PyObject *args)
 161 {
 162         struct tree_item *qsort_entries = NULL;
 163         int name_order, num_entries, n = 0, i;
 164         PyObject *entries, *py_name_order, *ret, *key, *value, *py_mode, *py_sha;
 165         Py_ssize_t pos = 0;
 166         int (*cmp)(const void *, const void *);
 167
 168         if (!PyArg_ParseTuple(args, "OO", &entries, &py_name_order))
 169                 goto error;
 170
 171         if (!PyDict_Check(entries)) {
 172                 PyErr_SetString(PyExc_TypeError, "Argument not a dictionary");
 173                 goto error;
 174         }
 175
 176         name_order = PyObject_IsTrue(py_name_order);
 177         if (name_order == -1)
 178                 goto error;
 179         cmp = name_order ? cmp_tree_item_name_order : cmp_tree_item;
 180
 181         num_entries = PyDict_Size(entries);
 182         if (PyErr_Occurred())
 183                 goto error;
 184         qsort_entries = PyMem_New(struct tree_item, num_entries);
 185         if (!qsort_entries) {
 186                 PyErr_NoMemory();
 187                 goto error;
 188         }
 189
 190         while (PyDict_Next(entries, &pos, &key, &value)) {
 191                 if (!PyString_Check(key)) {
 192                         PyErr_SetString(PyExc_TypeError, "Name is not a string");
 193                         goto error;
 194                 }
 195
 196                 if (PyTuple_Size(value) != 2) {
 197                         PyErr_SetString(PyExc_ValueError, "Tuple has invalid size");
 198                         goto error;
 199                 }
 200
 201                 py_mode = PyTuple_GET_ITEM(value, 0);
 202                 if (!PyInt_Check(py_mode)) {
 203                         PyErr_SetString(PyExc_TypeError, "Mode is not an integral type");
 204                         goto error;
 205                 }
 206
 207                 py_sha = PyTuple_GET_ITEM(value, 1);
 208                 if (!PyString_Check(py_sha)) {
 209                         PyErr_SetString(PyExc_TypeError, "SHA is not a string");
 210                         goto error;
 211                 }
 212                 qsort_entries[n].name = PyString_AS_STRING(key);
 213                 qsort_entries[n].mode = PyInt_AS_LONG(py_mode);
 214
 215                 qsort_entries[n].tuple = PyObject_CallFunctionObjArgs(
 216                                 tree_entry_cls, key, py_mode, py_sha, NULL);
 217                 if (qsort_entries[n].tuple == NULL)
 218                         goto error;
 219                 n++;
 220         }
 221
 222         qsort(qsort_entries, num_entries, sizeof(struct tree_item), cmp);
 223
 224         ret = PyList_New(num_entries);
 225         if (ret == NULL) {
 226                 PyErr_NoMemory();
 227                 goto error;
 228         }
 229
 230         for (i = 0; i < num_entries; i++) {
 231                 PyList_SET_ITEM(ret, i, qsort_entries[i].tuple);
 232         }
 233         PyMem_Free(qsort_entries);
 234         return ret;
 235
 236 error:
 237         for (i = 0; i < n; i++) {
 238                 Py_XDECREF(qsort_entries[i].tuple);
 239         }
 240         PyMem_Free(qsort_entries);
 241         return NULL;
 242 }
 243
 244 static PyMethodDef py_objects_methods[] = {
 245         { "parse_tree", (PyCFunction)py_parse_tree, METH_VARARGS | METH_KEYWORDS,
 246           NULL },
 247         { "sorted_tree_items", py_sorted_tree_items, METH_VARARGS, NULL },
 248         { NULL, NULL, 0, NULL }
 249 };
 250
 251 PyMODINIT_FUNC
 252 init_objects(void)
 253 {
 254         PyObject *m, *objects_mod, *errors_mod;
 255
 256         m = Py_InitModule3("_objects", py_objects_methods, NULL);
 257         if (m == NULL)
 258                 return;
 259
 260
 261         errors_mod = PyImport_ImportModule("dulwich.errors");
 262         if (errors_mod == NULL)
 263                 return;
 264
 265         object_format_exception_cls = PyObject_GetAttrString(
 266                 errors_mod, "ObjectFormatException");
 267         Py_DECREF(errors_mod);
 268         if (object_format_exception_cls == NULL)
 269                 return;
 270
 271         /* This is a circular import but should be safe since this module is
 272          * imported at at the very bottom of objects.py. */
 273         objects_mod = PyImport_ImportModule("dulwich.objects");
 274         if (objects_mod == NULL)
 275                 return;
 276
 277         tree_entry_cls = PyObject_GetAttrString(objects_mod, "TreeEntry");
 278         Py_DECREF(objects_mod);
 279         if (tree_entry_cls == NULL)
 280                 return;
 281 }