Skip Montanaro | a16b21f | 2003-03-23 14:32:54 +0000 | [diff] [blame] | 1 | /* csv module */ |
| 2 | |
| 3 | /* |
| 4 | |
| 5 | This module provides the low-level underpinnings of a CSV reading/writing |
| 6 | module. Users should not use this module directly, but import the csv.py |
| 7 | module instead. |
| 8 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 9 | */ |
| 10 | |
Skip Montanaro | 7b01a83 | 2003-04-12 19:23:46 +0000 | [diff] [blame] | 11 | #define MODULE_VERSION "1.0" |
| 12 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 13 | #include "Python.h" |
| 14 | #include "structmember.h" |
| 15 | |
Andrew McNamara | 37d2bdf | 2005-01-10 12:22:48 +0000 | [diff] [blame] | 16 | |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 17 | typedef struct { |
| 18 | PyObject *error_obj; /* CSV exception */ |
| 19 | PyObject *dialects; /* Dialect registry */ |
| 20 | long field_limit; /* max parsed field size */ |
| 21 | } _csvstate; |
| 22 | |
| 23 | #define _csvstate(o) ((_csvstate *)PyModule_GetState(o)) |
| 24 | |
| 25 | static int |
| 26 | _csv_clear(PyObject *m) |
| 27 | { |
| 28 | Py_CLEAR(_csvstate(m)->error_obj); |
| 29 | Py_CLEAR(_csvstate(m)->dialects); |
| 30 | return 0; |
| 31 | } |
| 32 | |
| 33 | static int |
| 34 | _csv_traverse(PyObject *m, visitproc visit, void *arg) |
| 35 | { |
| 36 | Py_VISIT(_csvstate(m)->error_obj); |
| 37 | Py_VISIT(_csvstate(m)->dialects); |
| 38 | return 0; |
| 39 | } |
| 40 | |
| 41 | static void |
| 42 | _csv_free(void *m) |
| 43 | { |
| 44 | _csv_clear((PyObject *)m); |
| 45 | } |
| 46 | |
| 47 | static struct PyModuleDef _csvmodule; |
| 48 | |
| 49 | #define _csvstate_global ((_csvstate *)PyModule_GetState(PyState_FindModule(&_csvmodule))) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 50 | |
| 51 | typedef enum { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 52 | START_RECORD, START_FIELD, ESCAPED_CHAR, IN_FIELD, |
| 53 | IN_QUOTED_FIELD, ESCAPE_IN_QUOTED_FIELD, QUOTE_IN_QUOTED_FIELD, |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 54 | EAT_CRNL,AFTER_ESCAPED_CRNL |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 55 | } ParserState; |
| 56 | |
| 57 | typedef enum { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 58 | QUOTE_MINIMAL, QUOTE_ALL, QUOTE_NONNUMERIC, QUOTE_NONE |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 59 | } QuoteStyle; |
| 60 | |
| 61 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 62 | QuoteStyle style; |
| 63 | char *name; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 64 | } StyleDesc; |
| 65 | |
| 66 | static StyleDesc quote_styles[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 67 | { QUOTE_MINIMAL, "QUOTE_MINIMAL" }, |
| 68 | { QUOTE_ALL, "QUOTE_ALL" }, |
| 69 | { QUOTE_NONNUMERIC, "QUOTE_NONNUMERIC" }, |
| 70 | { QUOTE_NONE, "QUOTE_NONE" }, |
| 71 | { 0 } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 72 | }; |
| 73 | |
| 74 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 75 | PyObject_HEAD |
Guido van Rossum | 4626458 | 2007-08-06 19:32:18 +0000 | [diff] [blame] | 76 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 77 | int doublequote; /* is " represented by ""? */ |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 78 | Py_UCS4 delimiter; /* field separator */ |
| 79 | Py_UCS4 quotechar; /* quote character */ |
| 80 | Py_UCS4 escapechar; /* escape character */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 81 | int skipinitialspace; /* ignore spaces following delimiter? */ |
| 82 | PyObject *lineterminator; /* string to write between records */ |
| 83 | int quoting; /* style of quoting to write */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 84 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 85 | int strict; /* raise exception on bad CSV */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 86 | } DialectObj; |
| 87 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 88 | static PyTypeObject Dialect_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 89 | |
| 90 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 91 | PyObject_HEAD |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 92 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 93 | PyObject *input_iter; /* iterate over this for input lines */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 94 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 95 | DialectObj *dialect; /* parsing dialect */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 96 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 97 | PyObject *fields; /* field list for current record */ |
| 98 | ParserState state; /* current CSV parse state */ |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 99 | Py_UCS4 *field; /* temporary buffer */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 100 | Py_ssize_t field_size; /* size of allocated buffer */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 101 | Py_ssize_t field_len; /* length of current field */ |
| 102 | int numeric_field; /* treat field as numeric */ |
| 103 | unsigned long line_num; /* Source-file line number */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 104 | } ReaderObj; |
| 105 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 106 | static PyTypeObject Reader_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 107 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 108 | #define ReaderObject_Check(v) (Py_TYPE(v) == &Reader_Type) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 109 | |
| 110 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 111 | PyObject_HEAD |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 112 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 113 | PyObject *writeline; /* write output lines to this file */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 114 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 115 | DialectObj *dialect; /* parsing dialect */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 116 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 117 | Py_UCS4 *rec; /* buffer for parser.join */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 118 | Py_ssize_t rec_size; /* size of allocated record */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 119 | Py_ssize_t rec_len; /* length of record */ |
| 120 | int num_fields; /* number of fields in record */ |
Guido van Rossum | 4626458 | 2007-08-06 19:32:18 +0000 | [diff] [blame] | 121 | } WriterObj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 122 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 123 | static PyTypeObject Writer_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 124 | |
| 125 | /* |
| 126 | * DIALECT class |
| 127 | */ |
| 128 | |
| 129 | static PyObject * |
| 130 | get_dialect_from_registry(PyObject * name_obj) |
| 131 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 132 | PyObject *dialect_obj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 133 | |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 134 | dialect_obj = PyDict_GetItem(_csvstate_global->dialects, name_obj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 135 | if (dialect_obj == NULL) { |
| 136 | if (!PyErr_Occurred()) |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 137 | PyErr_Format(_csvstate_global->error_obj, "unknown dialect"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 138 | } |
| 139 | else |
| 140 | Py_INCREF(dialect_obj); |
| 141 | return dialect_obj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 142 | } |
| 143 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 144 | static PyObject * |
| 145 | get_string(PyObject *str) |
| 146 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 147 | Py_XINCREF(str); |
| 148 | return str; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 149 | } |
| 150 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 151 | static PyObject * |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 152 | get_nullchar_as_None(Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 153 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 154 | if (c == '\0') { |
| 155 | Py_INCREF(Py_None); |
| 156 | return Py_None; |
| 157 | } |
| 158 | else |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 159 | return PyUnicode_FromOrdinal(c); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 160 | } |
| 161 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 162 | static PyObject * |
| 163 | Dialect_get_lineterminator(DialectObj *self) |
| 164 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 165 | return get_string(self->lineterminator); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 166 | } |
| 167 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 168 | static PyObject * |
Guido van Rossum | a9769c2 | 2007-08-07 23:59:30 +0000 | [diff] [blame] | 169 | Dialect_get_delimiter(DialectObj *self) |
| 170 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 171 | return get_nullchar_as_None(self->delimiter); |
Guido van Rossum | a9769c2 | 2007-08-07 23:59:30 +0000 | [diff] [blame] | 172 | } |
| 173 | |
| 174 | static PyObject * |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 175 | Dialect_get_escapechar(DialectObj *self) |
| 176 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 177 | return get_nullchar_as_None(self->escapechar); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 178 | } |
| 179 | |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 180 | static PyObject * |
| 181 | Dialect_get_quotechar(DialectObj *self) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 182 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 183 | return get_nullchar_as_None(self->quotechar); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 184 | } |
| 185 | |
| 186 | static PyObject * |
| 187 | Dialect_get_quoting(DialectObj *self) |
| 188 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 189 | return PyLong_FromLong(self->quoting); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 190 | } |
| 191 | |
| 192 | static int |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 193 | _set_bool(const char *name, int *target, PyObject *src, int dflt) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 194 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 195 | if (src == NULL) |
| 196 | *target = dflt; |
Antoine Pitrou | 6f430e4 | 2012-08-15 23:18:25 +0200 | [diff] [blame] | 197 | else { |
| 198 | int b = PyObject_IsTrue(src); |
| 199 | if (b < 0) |
| 200 | return -1; |
| 201 | *target = b; |
| 202 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 203 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 204 | } |
| 205 | |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 206 | static int |
| 207 | _set_int(const char *name, int *target, PyObject *src, int dflt) |
| 208 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 209 | if (src == NULL) |
| 210 | *target = dflt; |
| 211 | else { |
| 212 | long value; |
| 213 | if (!PyLong_CheckExact(src)) { |
| 214 | PyErr_Format(PyExc_TypeError, |
| 215 | "\"%s\" must be an integer", name); |
| 216 | return -1; |
| 217 | } |
| 218 | value = PyLong_AsLong(src); |
| 219 | if (value == -1 && PyErr_Occurred()) |
| 220 | return -1; |
Martin v. Löwis | d1a1d1e | 2007-12-04 22:10:37 +0000 | [diff] [blame] | 221 | #if SIZEOF_LONG > SIZEOF_INT |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 222 | if (value > INT_MAX || value < INT_MIN) { |
| 223 | PyErr_Format(PyExc_ValueError, |
| 224 | "integer out of range for \"%s\"", name); |
| 225 | return -1; |
| 226 | } |
Martin v. Löwis | d1a1d1e | 2007-12-04 22:10:37 +0000 | [diff] [blame] | 227 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 228 | *target = (int)value; |
| 229 | } |
| 230 | return 0; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 231 | } |
| 232 | |
| 233 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 234 | _set_char(const char *name, Py_UCS4 *target, PyObject *src, Py_UCS4 dflt) |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 235 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 236 | if (src == NULL) |
| 237 | *target = dflt; |
| 238 | else { |
| 239 | *target = '\0'; |
| 240 | if (src != Py_None) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 241 | Py_ssize_t len; |
Victor Stinner | 9e30aa5 | 2011-11-21 02:49:52 +0100 | [diff] [blame] | 242 | len = PyUnicode_GetLength(src); |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 243 | if (len > 1) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 244 | PyErr_Format(PyExc_TypeError, |
| 245 | "\"%s\" must be an 1-character string", |
| 246 | name); |
| 247 | return -1; |
| 248 | } |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 249 | /* PyUnicode_READY() is called in PyUnicode_GetLength() */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 250 | if (len > 0) |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 251 | *target = PyUnicode_READ_CHAR(src, 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 252 | } |
| 253 | } |
| 254 | return 0; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 255 | } |
| 256 | |
| 257 | static int |
| 258 | _set_str(const char *name, PyObject **target, PyObject *src, const char *dflt) |
| 259 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 260 | if (src == NULL) |
| 261 | *target = PyUnicode_DecodeASCII(dflt, strlen(dflt), NULL); |
| 262 | else { |
| 263 | if (src == Py_None) |
| 264 | *target = NULL; |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 265 | else if (!PyUnicode_Check(src)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 266 | PyErr_Format(PyExc_TypeError, |
| 267 | "\"%s\" must be a string", name); |
| 268 | return -1; |
| 269 | } |
| 270 | else { |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 271 | if (PyUnicode_READY(src) == -1) |
| 272 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 273 | Py_XDECREF(*target); |
| 274 | Py_INCREF(src); |
| 275 | *target = src; |
| 276 | } |
| 277 | } |
| 278 | return 0; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 279 | } |
| 280 | |
| 281 | static int |
| 282 | dialect_check_quoting(int quoting) |
| 283 | { |
Victor Stinner | 4fe519b | 2010-11-09 09:40:16 +0000 | [diff] [blame] | 284 | StyleDesc *qs; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 285 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 286 | for (qs = quote_styles; qs->name; qs++) { |
| 287 | if (qs->style == quoting) |
| 288 | return 0; |
| 289 | } |
| 290 | PyErr_Format(PyExc_TypeError, "bad \"quoting\" value"); |
| 291 | return -1; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 292 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 293 | |
| 294 | #define D_OFF(x) offsetof(DialectObj, x) |
| 295 | |
| 296 | static struct PyMemberDef Dialect_memberlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 297 | { "skipinitialspace", T_INT, D_OFF(skipinitialspace), READONLY }, |
| 298 | { "doublequote", T_INT, D_OFF(doublequote), READONLY }, |
| 299 | { "strict", T_INT, D_OFF(strict), READONLY }, |
| 300 | { NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 301 | }; |
| 302 | |
| 303 | static PyGetSetDef Dialect_getsetlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 304 | { "delimiter", (getter)Dialect_get_delimiter}, |
| 305 | { "escapechar", (getter)Dialect_get_escapechar}, |
| 306 | { "lineterminator", (getter)Dialect_get_lineterminator}, |
| 307 | { "quotechar", (getter)Dialect_get_quotechar}, |
| 308 | { "quoting", (getter)Dialect_get_quoting}, |
| 309 | {NULL}, |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 310 | }; |
| 311 | |
| 312 | static void |
| 313 | Dialect_dealloc(DialectObj *self) |
| 314 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 315 | Py_XDECREF(self->lineterminator); |
| 316 | Py_TYPE(self)->tp_free((PyObject *)self); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 317 | } |
| 318 | |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 319 | static char *dialect_kws[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 320 | "dialect", |
| 321 | "delimiter", |
| 322 | "doublequote", |
| 323 | "escapechar", |
| 324 | "lineterminator", |
| 325 | "quotechar", |
| 326 | "quoting", |
| 327 | "skipinitialspace", |
| 328 | "strict", |
| 329 | NULL |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 330 | }; |
| 331 | |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 332 | static PyObject * |
| 333 | dialect_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 334 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 335 | DialectObj *self; |
| 336 | PyObject *ret = NULL; |
| 337 | PyObject *dialect = NULL; |
| 338 | PyObject *delimiter = NULL; |
| 339 | PyObject *doublequote = NULL; |
| 340 | PyObject *escapechar = NULL; |
| 341 | PyObject *lineterminator = NULL; |
| 342 | PyObject *quotechar = NULL; |
| 343 | PyObject *quoting = NULL; |
| 344 | PyObject *skipinitialspace = NULL; |
| 345 | PyObject *strict = NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 346 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 347 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, |
| 348 | "|OOOOOOOOO", dialect_kws, |
| 349 | &dialect, |
| 350 | &delimiter, |
| 351 | &doublequote, |
| 352 | &escapechar, |
| 353 | &lineterminator, |
| 354 | "echar, |
| 355 | "ing, |
| 356 | &skipinitialspace, |
| 357 | &strict)) |
| 358 | return NULL; |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 359 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 360 | if (dialect != NULL) { |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 361 | if (PyUnicode_Check(dialect)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 362 | dialect = get_dialect_from_registry(dialect); |
| 363 | if (dialect == NULL) |
| 364 | return NULL; |
| 365 | } |
| 366 | else |
| 367 | Py_INCREF(dialect); |
| 368 | /* Can we reuse this instance? */ |
| 369 | if (PyObject_TypeCheck(dialect, &Dialect_Type) && |
| 370 | delimiter == 0 && |
| 371 | doublequote == 0 && |
| 372 | escapechar == 0 && |
| 373 | lineterminator == 0 && |
| 374 | quotechar == 0 && |
| 375 | quoting == 0 && |
| 376 | skipinitialspace == 0 && |
| 377 | strict == 0) |
| 378 | return dialect; |
| 379 | } |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 380 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 381 | self = (DialectObj *)type->tp_alloc(type, 0); |
| 382 | if (self == NULL) { |
| 383 | Py_XDECREF(dialect); |
| 384 | return NULL; |
| 385 | } |
| 386 | self->lineterminator = NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 387 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 388 | Py_XINCREF(delimiter); |
| 389 | Py_XINCREF(doublequote); |
| 390 | Py_XINCREF(escapechar); |
| 391 | Py_XINCREF(lineterminator); |
| 392 | Py_XINCREF(quotechar); |
| 393 | Py_XINCREF(quoting); |
| 394 | Py_XINCREF(skipinitialspace); |
| 395 | Py_XINCREF(strict); |
| 396 | if (dialect != NULL) { |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 397 | #define DIALECT_GETATTR(v, n) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 398 | if (v == NULL) \ |
| 399 | v = PyObject_GetAttrString(dialect, n) |
| 400 | DIALECT_GETATTR(delimiter, "delimiter"); |
| 401 | DIALECT_GETATTR(doublequote, "doublequote"); |
| 402 | DIALECT_GETATTR(escapechar, "escapechar"); |
| 403 | DIALECT_GETATTR(lineterminator, "lineterminator"); |
| 404 | DIALECT_GETATTR(quotechar, "quotechar"); |
| 405 | DIALECT_GETATTR(quoting, "quoting"); |
| 406 | DIALECT_GETATTR(skipinitialspace, "skipinitialspace"); |
| 407 | DIALECT_GETATTR(strict, "strict"); |
| 408 | PyErr_Clear(); |
| 409 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 410 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 411 | /* check types and convert to C values */ |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 412 | #define DIASET(meth, name, target, src, dflt) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 413 | if (meth(name, target, src, dflt)) \ |
| 414 | goto err |
| 415 | DIASET(_set_char, "delimiter", &self->delimiter, delimiter, ','); |
| 416 | DIASET(_set_bool, "doublequote", &self->doublequote, doublequote, 1); |
| 417 | DIASET(_set_char, "escapechar", &self->escapechar, escapechar, 0); |
| 418 | DIASET(_set_str, "lineterminator", &self->lineterminator, lineterminator, "\r\n"); |
| 419 | DIASET(_set_char, "quotechar", &self->quotechar, quotechar, '"'); |
| 420 | DIASET(_set_int, "quoting", &self->quoting, quoting, QUOTE_MINIMAL); |
| 421 | DIASET(_set_bool, "skipinitialspace", &self->skipinitialspace, skipinitialspace, 0); |
| 422 | DIASET(_set_bool, "strict", &self->strict, strict, 0); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 423 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 424 | /* validate options */ |
| 425 | if (dialect_check_quoting(self->quoting)) |
| 426 | goto err; |
| 427 | if (self->delimiter == 0) { |
| 428 | PyErr_SetString(PyExc_TypeError, "delimiter must be set"); |
| 429 | goto err; |
| 430 | } |
| 431 | if (quotechar == Py_None && quoting == NULL) |
| 432 | self->quoting = QUOTE_NONE; |
| 433 | if (self->quoting != QUOTE_NONE && self->quotechar == 0) { |
| 434 | PyErr_SetString(PyExc_TypeError, |
| 435 | "quotechar must be set if quoting enabled"); |
| 436 | goto err; |
| 437 | } |
| 438 | if (self->lineterminator == 0) { |
| 439 | PyErr_SetString(PyExc_TypeError, "lineterminator must be set"); |
| 440 | goto err; |
| 441 | } |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 442 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 443 | ret = (PyObject *)self; |
| 444 | Py_INCREF(self); |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 445 | err: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 446 | Py_XDECREF(self); |
| 447 | Py_XDECREF(dialect); |
| 448 | Py_XDECREF(delimiter); |
| 449 | Py_XDECREF(doublequote); |
| 450 | Py_XDECREF(escapechar); |
| 451 | Py_XDECREF(lineterminator); |
| 452 | Py_XDECREF(quotechar); |
| 453 | Py_XDECREF(quoting); |
| 454 | Py_XDECREF(skipinitialspace); |
| 455 | Py_XDECREF(strict); |
| 456 | return ret; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 457 | } |
| 458 | |
| 459 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 460 | PyDoc_STRVAR(Dialect_Type_doc, |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 461 | "CSV dialect\n" |
| 462 | "\n" |
| 463 | "The Dialect type records CSV parsing and generation options.\n"); |
| 464 | |
| 465 | static PyTypeObject Dialect_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 466 | PyVarObject_HEAD_INIT(NULL, 0) |
| 467 | "_csv.Dialect", /* tp_name */ |
| 468 | sizeof(DialectObj), /* tp_basicsize */ |
| 469 | 0, /* tp_itemsize */ |
| 470 | /* methods */ |
| 471 | (destructor)Dialect_dealloc, /* tp_dealloc */ |
| 472 | (printfunc)0, /* tp_print */ |
| 473 | (getattrfunc)0, /* tp_getattr */ |
| 474 | (setattrfunc)0, /* tp_setattr */ |
| 475 | 0, /* tp_reserved */ |
| 476 | (reprfunc)0, /* tp_repr */ |
| 477 | 0, /* tp_as_number */ |
| 478 | 0, /* tp_as_sequence */ |
| 479 | 0, /* tp_as_mapping */ |
| 480 | (hashfunc)0, /* tp_hash */ |
| 481 | (ternaryfunc)0, /* tp_call */ |
| 482 | (reprfunc)0, /* tp_str */ |
| 483 | 0, /* tp_getattro */ |
| 484 | 0, /* tp_setattro */ |
| 485 | 0, /* tp_as_buffer */ |
| 486 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 487 | Dialect_Type_doc, /* tp_doc */ |
| 488 | 0, /* tp_traverse */ |
| 489 | 0, /* tp_clear */ |
| 490 | 0, /* tp_richcompare */ |
| 491 | 0, /* tp_weaklistoffset */ |
| 492 | 0, /* tp_iter */ |
| 493 | 0, /* tp_iternext */ |
| 494 | 0, /* tp_methods */ |
| 495 | Dialect_memberlist, /* tp_members */ |
| 496 | Dialect_getsetlist, /* tp_getset */ |
| 497 | 0, /* tp_base */ |
| 498 | 0, /* tp_dict */ |
| 499 | 0, /* tp_descr_get */ |
| 500 | 0, /* tp_descr_set */ |
| 501 | 0, /* tp_dictoffset */ |
| 502 | 0, /* tp_init */ |
| 503 | 0, /* tp_alloc */ |
| 504 | dialect_new, /* tp_new */ |
| 505 | 0, /* tp_free */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 506 | }; |
| 507 | |
Andrew McNamara | 91b9746 | 2005-01-11 01:07:23 +0000 | [diff] [blame] | 508 | /* |
| 509 | * Return an instance of the dialect type, given a Python instance or kwarg |
| 510 | * description of the dialect |
| 511 | */ |
| 512 | static PyObject * |
| 513 | _call_dialect(PyObject *dialect_inst, PyObject *kwargs) |
| 514 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 515 | PyObject *ctor_args; |
| 516 | PyObject *dialect; |
Andrew McNamara | 91b9746 | 2005-01-11 01:07:23 +0000 | [diff] [blame] | 517 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 518 | ctor_args = Py_BuildValue(dialect_inst ? "(O)" : "()", dialect_inst); |
| 519 | if (ctor_args == NULL) |
| 520 | return NULL; |
| 521 | dialect = PyObject_Call((PyObject *)&Dialect_Type, ctor_args, kwargs); |
| 522 | Py_DECREF(ctor_args); |
| 523 | return dialect; |
Andrew McNamara | 91b9746 | 2005-01-11 01:07:23 +0000 | [diff] [blame] | 524 | } |
| 525 | |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 526 | /* |
| 527 | * READER |
| 528 | */ |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 529 | static int |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 530 | parse_save_field(ReaderObj *self) |
| 531 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 532 | PyObject *field; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 533 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 534 | field = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, |
| 535 | (void *) self->field, self->field_len); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 536 | if (field == NULL) |
| 537 | return -1; |
| 538 | self->field_len = 0; |
| 539 | if (self->numeric_field) { |
| 540 | PyObject *tmp; |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 541 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 542 | self->numeric_field = 0; |
| 543 | tmp = PyNumber_Float(field); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 544 | Py_DECREF(field); |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 545 | if (tmp == NULL) |
| 546 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 547 | field = tmp; |
| 548 | } |
| 549 | PyList_Append(self->fields, field); |
| 550 | Py_DECREF(field); |
| 551 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 552 | } |
| 553 | |
| 554 | static int |
| 555 | parse_grow_buff(ReaderObj *self) |
| 556 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 557 | if (self->field_size == 0) { |
| 558 | self->field_size = 4096; |
| 559 | if (self->field != NULL) |
| 560 | PyMem_Free(self->field); |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 561 | self->field = PyMem_New(Py_UCS4, self->field_size); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 562 | } |
| 563 | else { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 564 | Py_UCS4 *field = self->field; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 565 | if (self->field_size > PY_SSIZE_T_MAX / 2) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 566 | PyErr_NoMemory(); |
| 567 | return 0; |
| 568 | } |
| 569 | self->field_size *= 2; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 570 | self->field = PyMem_Resize(field, Py_UCS4, self->field_size); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 571 | } |
| 572 | if (self->field == NULL) { |
| 573 | PyErr_NoMemory(); |
| 574 | return 0; |
| 575 | } |
| 576 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 577 | } |
| 578 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 579 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 580 | parse_add_char(ReaderObj *self, Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 581 | { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 582 | if (self->field_len >= _csvstate_global->field_limit) { |
| 583 | PyErr_Format(_csvstate_global->error_obj, "field larger than field limit (%ld)", |
| 584 | _csvstate_global->field_limit); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 585 | return -1; |
| 586 | } |
| 587 | if (self->field_len == self->field_size && !parse_grow_buff(self)) |
| 588 | return -1; |
| 589 | self->field[self->field_len++] = c; |
| 590 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 591 | } |
| 592 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 593 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 594 | parse_process_char(ReaderObj *self, Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 595 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 596 | DialectObj *dialect = self->dialect; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 597 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 598 | switch (self->state) { |
| 599 | case START_RECORD: |
| 600 | /* start of record */ |
| 601 | if (c == '\0') |
| 602 | /* empty line - return [] */ |
| 603 | break; |
| 604 | else if (c == '\n' || c == '\r') { |
| 605 | self->state = EAT_CRNL; |
| 606 | break; |
| 607 | } |
| 608 | /* normal character - handle as START_FIELD */ |
| 609 | self->state = START_FIELD; |
| 610 | /* fallthru */ |
| 611 | case START_FIELD: |
| 612 | /* expecting field */ |
| 613 | if (c == '\n' || c == '\r' || c == '\0') { |
| 614 | /* save empty field - return [fields] */ |
| 615 | if (parse_save_field(self) < 0) |
| 616 | return -1; |
| 617 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 618 | } |
| 619 | else if (c == dialect->quotechar && |
| 620 | dialect->quoting != QUOTE_NONE) { |
| 621 | /* start quoted field */ |
| 622 | self->state = IN_QUOTED_FIELD; |
| 623 | } |
| 624 | else if (c == dialect->escapechar) { |
| 625 | /* possible escaped character */ |
| 626 | self->state = ESCAPED_CHAR; |
| 627 | } |
| 628 | else if (c == ' ' && dialect->skipinitialspace) |
| 629 | /* ignore space at start of field */ |
| 630 | ; |
| 631 | else if (c == dialect->delimiter) { |
| 632 | /* save empty field */ |
| 633 | if (parse_save_field(self) < 0) |
| 634 | return -1; |
| 635 | } |
| 636 | else { |
| 637 | /* begin new unquoted field */ |
| 638 | if (dialect->quoting == QUOTE_NONNUMERIC) |
| 639 | self->numeric_field = 1; |
| 640 | if (parse_add_char(self, c) < 0) |
| 641 | return -1; |
| 642 | self->state = IN_FIELD; |
| 643 | } |
| 644 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 645 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 646 | case ESCAPED_CHAR: |
R David Murray | 9a7d376 | 2013-03-20 00:15:20 -0400 | [diff] [blame] | 647 | if (c == '\n' || c=='\r') { |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 648 | if (parse_add_char(self, c) < 0) |
| 649 | return -1; |
| 650 | self->state = AFTER_ESCAPED_CRNL; |
| 651 | break; |
| 652 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 653 | if (c == '\0') |
| 654 | c = '\n'; |
| 655 | if (parse_add_char(self, c) < 0) |
| 656 | return -1; |
| 657 | self->state = IN_FIELD; |
| 658 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 659 | |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 660 | case AFTER_ESCAPED_CRNL: |
| 661 | if (c == '\0') |
| 662 | break; |
| 663 | /*fallthru*/ |
| 664 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 665 | case IN_FIELD: |
| 666 | /* in unquoted field */ |
| 667 | if (c == '\n' || c == '\r' || c == '\0') { |
| 668 | /* end of line - return [fields] */ |
| 669 | if (parse_save_field(self) < 0) |
| 670 | return -1; |
| 671 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 672 | } |
| 673 | else if (c == dialect->escapechar) { |
| 674 | /* possible escaped character */ |
| 675 | self->state = ESCAPED_CHAR; |
| 676 | } |
| 677 | else if (c == dialect->delimiter) { |
| 678 | /* save field - wait for new field */ |
| 679 | if (parse_save_field(self) < 0) |
| 680 | return -1; |
| 681 | self->state = START_FIELD; |
| 682 | } |
| 683 | else { |
| 684 | /* normal character - save in field */ |
| 685 | if (parse_add_char(self, c) < 0) |
| 686 | return -1; |
| 687 | } |
| 688 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 689 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 690 | case IN_QUOTED_FIELD: |
| 691 | /* in quoted field */ |
| 692 | if (c == '\0') |
| 693 | ; |
| 694 | else if (c == dialect->escapechar) { |
| 695 | /* Possible escape character */ |
| 696 | self->state = ESCAPE_IN_QUOTED_FIELD; |
| 697 | } |
| 698 | else if (c == dialect->quotechar && |
| 699 | dialect->quoting != QUOTE_NONE) { |
| 700 | if (dialect->doublequote) { |
| 701 | /* doublequote; " represented by "" */ |
| 702 | self->state = QUOTE_IN_QUOTED_FIELD; |
| 703 | } |
| 704 | else { |
| 705 | /* end of quote part of field */ |
| 706 | self->state = IN_FIELD; |
| 707 | } |
| 708 | } |
| 709 | else { |
| 710 | /* normal character - save in field */ |
| 711 | if (parse_add_char(self, c) < 0) |
| 712 | return -1; |
| 713 | } |
| 714 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 715 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 716 | case ESCAPE_IN_QUOTED_FIELD: |
| 717 | if (c == '\0') |
| 718 | c = '\n'; |
| 719 | if (parse_add_char(self, c) < 0) |
| 720 | return -1; |
| 721 | self->state = IN_QUOTED_FIELD; |
| 722 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 723 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 724 | case QUOTE_IN_QUOTED_FIELD: |
| 725 | /* doublequote - seen a quote in an quoted field */ |
| 726 | if (dialect->quoting != QUOTE_NONE && |
| 727 | c == dialect->quotechar) { |
| 728 | /* save "" as " */ |
| 729 | if (parse_add_char(self, c) < 0) |
| 730 | return -1; |
| 731 | self->state = IN_QUOTED_FIELD; |
| 732 | } |
| 733 | else if (c == dialect->delimiter) { |
| 734 | /* save field - wait for new field */ |
| 735 | if (parse_save_field(self) < 0) |
| 736 | return -1; |
| 737 | self->state = START_FIELD; |
| 738 | } |
| 739 | else if (c == '\n' || c == '\r' || c == '\0') { |
| 740 | /* end of line - return [fields] */ |
| 741 | if (parse_save_field(self) < 0) |
| 742 | return -1; |
| 743 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 744 | } |
| 745 | else if (!dialect->strict) { |
| 746 | if (parse_add_char(self, c) < 0) |
| 747 | return -1; |
| 748 | self->state = IN_FIELD; |
| 749 | } |
| 750 | else { |
| 751 | /* illegal */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 752 | PyErr_Format(_csvstate_global->error_obj, "'%c' expected after '%c'", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 753 | dialect->delimiter, |
| 754 | dialect->quotechar); |
| 755 | return -1; |
| 756 | } |
| 757 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 758 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 759 | case EAT_CRNL: |
| 760 | if (c == '\n' || c == '\r') |
| 761 | ; |
| 762 | else if (c == '\0') |
| 763 | self->state = START_RECORD; |
| 764 | else { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 765 | PyErr_Format(_csvstate_global->error_obj, "new-line character seen in unquoted field - do you need to open the file in universal-newline mode?"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 766 | return -1; |
| 767 | } |
| 768 | break; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 769 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 770 | } |
| 771 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 772 | } |
| 773 | |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 774 | static int |
| 775 | parse_reset(ReaderObj *self) |
| 776 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 777 | Py_XDECREF(self->fields); |
| 778 | self->fields = PyList_New(0); |
| 779 | if (self->fields == NULL) |
| 780 | return -1; |
| 781 | self->field_len = 0; |
| 782 | self->state = START_RECORD; |
| 783 | self->numeric_field = 0; |
| 784 | return 0; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 785 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 786 | |
| 787 | static PyObject * |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 788 | Reader_iternext(ReaderObj *self) |
| 789 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 790 | PyObject *fields = NULL; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 791 | Py_UCS4 c; |
| 792 | Py_ssize_t pos, linelen; |
| 793 | unsigned int kind; |
| 794 | void *data; |
| 795 | PyObject *lineobj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 796 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 797 | if (parse_reset(self) < 0) |
| 798 | return NULL; |
| 799 | do { |
| 800 | lineobj = PyIter_Next(self->input_iter); |
| 801 | if (lineobj == NULL) { |
| 802 | /* End of input OR exception */ |
Senthil Kumaran | 67b7b98 | 2012-09-25 02:30:27 -0700 | [diff] [blame] | 803 | if (!PyErr_Occurred() && (self->field_len != 0 || |
| 804 | self->state == IN_QUOTED_FIELD)) { |
| 805 | if (self->dialect->strict) |
Senthil Kumaran | 49d1302 | 2012-09-25 02:37:20 -0700 | [diff] [blame] | 806 | PyErr_SetString(_csvstate_global->error_obj, |
| 807 | "unexpected end of data"); |
Senthil Kumaran | 67b7b98 | 2012-09-25 02:30:27 -0700 | [diff] [blame] | 808 | else if (parse_save_field(self) >= 0) |
| 809 | break; |
| 810 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 811 | return NULL; |
| 812 | } |
| 813 | if (!PyUnicode_Check(lineobj)) { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 814 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 815 | "iterator should return strings, " |
| 816 | "not %.200s " |
| 817 | "(did you open the file in text mode?)", |
| 818 | lineobj->ob_type->tp_name |
| 819 | ); |
| 820 | Py_DECREF(lineobj); |
| 821 | return NULL; |
| 822 | } |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 823 | if (PyUnicode_READY(lineobj) == -1) { |
| 824 | Py_DECREF(lineobj); |
| 825 | return NULL; |
| 826 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 827 | ++self->line_num; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 828 | kind = PyUnicode_KIND(lineobj); |
| 829 | data = PyUnicode_DATA(lineobj); |
| 830 | pos = 0; |
| 831 | linelen = PyUnicode_GET_LENGTH(lineobj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 832 | while (linelen--) { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 833 | c = PyUnicode_READ(kind, data, pos); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 834 | if (c == '\0') { |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 835 | Py_DECREF(lineobj); |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 836 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 837 | "line contains NULL byte"); |
| 838 | goto err; |
| 839 | } |
| 840 | if (parse_process_char(self, c) < 0) { |
| 841 | Py_DECREF(lineobj); |
| 842 | goto err; |
| 843 | } |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 844 | pos++; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 845 | } |
| 846 | Py_DECREF(lineobj); |
| 847 | if (parse_process_char(self, 0) < 0) |
| 848 | goto err; |
| 849 | } while (self->state != START_RECORD); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 850 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 851 | fields = self->fields; |
| 852 | self->fields = NULL; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 853 | err: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 854 | return fields; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 855 | } |
| 856 | |
| 857 | static void |
| 858 | Reader_dealloc(ReaderObj *self) |
| 859 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 860 | PyObject_GC_UnTrack(self); |
| 861 | Py_XDECREF(self->dialect); |
| 862 | Py_XDECREF(self->input_iter); |
| 863 | Py_XDECREF(self->fields); |
| 864 | if (self->field != NULL) |
| 865 | PyMem_Free(self->field); |
| 866 | PyObject_GC_Del(self); |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 867 | } |
| 868 | |
| 869 | static int |
| 870 | Reader_traverse(ReaderObj *self, visitproc visit, void *arg) |
| 871 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 872 | Py_VISIT(self->dialect); |
| 873 | Py_VISIT(self->input_iter); |
| 874 | Py_VISIT(self->fields); |
| 875 | return 0; |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 876 | } |
| 877 | |
| 878 | static int |
| 879 | Reader_clear(ReaderObj *self) |
| 880 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 881 | Py_CLEAR(self->dialect); |
| 882 | Py_CLEAR(self->input_iter); |
| 883 | Py_CLEAR(self->fields); |
| 884 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 885 | } |
| 886 | |
| 887 | PyDoc_STRVAR(Reader_Type_doc, |
| 888 | "CSV reader\n" |
| 889 | "\n" |
| 890 | "Reader objects are responsible for reading and parsing tabular data\n" |
| 891 | "in CSV format.\n" |
| 892 | ); |
| 893 | |
| 894 | static struct PyMethodDef Reader_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 895 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 896 | }; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 897 | #define R_OFF(x) offsetof(ReaderObj, x) |
| 898 | |
| 899 | static struct PyMemberDef Reader_memberlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 900 | { "dialect", T_OBJECT, R_OFF(dialect), READONLY }, |
| 901 | { "line_num", T_ULONG, R_OFF(line_num), READONLY }, |
| 902 | { NULL } |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 903 | }; |
| 904 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 905 | |
| 906 | static PyTypeObject Reader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 907 | PyVarObject_HEAD_INIT(NULL, 0) |
| 908 | "_csv.reader", /*tp_name*/ |
| 909 | sizeof(ReaderObj), /*tp_basicsize*/ |
| 910 | 0, /*tp_itemsize*/ |
| 911 | /* methods */ |
| 912 | (destructor)Reader_dealloc, /*tp_dealloc*/ |
| 913 | (printfunc)0, /*tp_print*/ |
| 914 | (getattrfunc)0, /*tp_getattr*/ |
| 915 | (setattrfunc)0, /*tp_setattr*/ |
| 916 | 0, /*tp_reserved*/ |
| 917 | (reprfunc)0, /*tp_repr*/ |
| 918 | 0, /*tp_as_number*/ |
| 919 | 0, /*tp_as_sequence*/ |
| 920 | 0, /*tp_as_mapping*/ |
| 921 | (hashfunc)0, /*tp_hash*/ |
| 922 | (ternaryfunc)0, /*tp_call*/ |
| 923 | (reprfunc)0, /*tp_str*/ |
| 924 | 0, /*tp_getattro*/ |
| 925 | 0, /*tp_setattro*/ |
| 926 | 0, /*tp_as_buffer*/ |
| 927 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | |
| 928 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
| 929 | Reader_Type_doc, /*tp_doc*/ |
| 930 | (traverseproc)Reader_traverse, /*tp_traverse*/ |
| 931 | (inquiry)Reader_clear, /*tp_clear*/ |
| 932 | 0, /*tp_richcompare*/ |
| 933 | 0, /*tp_weaklistoffset*/ |
| 934 | PyObject_SelfIter, /*tp_iter*/ |
| 935 | (getiterfunc)Reader_iternext, /*tp_iternext*/ |
| 936 | Reader_methods, /*tp_methods*/ |
| 937 | Reader_memberlist, /*tp_members*/ |
| 938 | 0, /*tp_getset*/ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 939 | |
| 940 | }; |
| 941 | |
| 942 | static PyObject * |
| 943 | csv_reader(PyObject *module, PyObject *args, PyObject *keyword_args) |
| 944 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 945 | PyObject * iterator, * dialect = NULL; |
| 946 | ReaderObj * self = PyObject_GC_New(ReaderObj, &Reader_Type); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 947 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 948 | if (!self) |
| 949 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 950 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 951 | self->dialect = NULL; |
| 952 | self->fields = NULL; |
| 953 | self->input_iter = NULL; |
| 954 | self->field = NULL; |
| 955 | self->field_size = 0; |
| 956 | self->line_num = 0; |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 957 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 958 | if (parse_reset(self) < 0) { |
| 959 | Py_DECREF(self); |
| 960 | return NULL; |
| 961 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 962 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 963 | if (!PyArg_UnpackTuple(args, "", 1, 2, &iterator, &dialect)) { |
| 964 | Py_DECREF(self); |
| 965 | return NULL; |
| 966 | } |
| 967 | self->input_iter = PyObject_GetIter(iterator); |
| 968 | if (self->input_iter == NULL) { |
| 969 | PyErr_SetString(PyExc_TypeError, |
| 970 | "argument 1 must be an iterator"); |
| 971 | Py_DECREF(self); |
| 972 | return NULL; |
| 973 | } |
| 974 | self->dialect = (DialectObj *)_call_dialect(dialect, keyword_args); |
| 975 | if (self->dialect == NULL) { |
| 976 | Py_DECREF(self); |
| 977 | return NULL; |
| 978 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 979 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 980 | PyObject_GC_Track(self); |
| 981 | return (PyObject *)self; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 982 | } |
| 983 | |
| 984 | /* |
| 985 | * WRITER |
| 986 | */ |
| 987 | /* ---------------------------------------------------------------- */ |
| 988 | static void |
| 989 | join_reset(WriterObj *self) |
| 990 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 991 | self->rec_len = 0; |
| 992 | self->num_fields = 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 993 | } |
| 994 | |
| 995 | #define MEM_INCR 32768 |
| 996 | |
| 997 | /* Calculate new record length or append field to record. Return new |
| 998 | * record length. |
| 999 | */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1000 | static Py_ssize_t |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1001 | join_append_data(WriterObj *self, unsigned int field_kind, void *field_data, |
| 1002 | Py_ssize_t field_len, int quote_empty, int *quoted, |
| 1003 | int copy_phase) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1004 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1005 | DialectObj *dialect = self->dialect; |
| 1006 | int i; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1007 | Py_ssize_t rec_len; |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1008 | |
| 1009 | #define ADDCH(c) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1010 | do {\ |
| 1011 | if (copy_phase) \ |
| 1012 | self->rec[rec_len] = c;\ |
| 1013 | rec_len++;\ |
| 1014 | } while(0) |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1015 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1016 | rec_len = self->rec_len; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1017 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1018 | /* If this is not the first field we need a field separator */ |
| 1019 | if (self->num_fields > 0) |
| 1020 | ADDCH(dialect->delimiter); |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1021 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1022 | /* Handle preceding quote */ |
| 1023 | if (copy_phase && *quoted) |
| 1024 | ADDCH(dialect->quotechar); |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1025 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1026 | /* Copy/count field data */ |
| 1027 | /* If field is null just pass over */ |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1028 | for (i = 0; field_data && (i < field_len); i++) { |
| 1029 | Py_UCS4 c = PyUnicode_READ(field_kind, field_data, i); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1030 | int want_escape = 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1031 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1032 | if (c == dialect->delimiter || |
| 1033 | c == dialect->escapechar || |
| 1034 | c == dialect->quotechar || |
Martin v. Löwis | 5f4f4c5 | 2011-11-01 18:42:23 +0100 | [diff] [blame] | 1035 | PyUnicode_FindChar( |
| 1036 | dialect->lineterminator, c, 0, |
| 1037 | PyUnicode_GET_LENGTH(dialect->lineterminator), 1) >= 0) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1038 | if (dialect->quoting == QUOTE_NONE) |
| 1039 | want_escape = 1; |
| 1040 | else { |
| 1041 | if (c == dialect->quotechar) { |
| 1042 | if (dialect->doublequote) |
| 1043 | ADDCH(dialect->quotechar); |
| 1044 | else |
| 1045 | want_escape = 1; |
| 1046 | } |
| 1047 | if (!want_escape) |
| 1048 | *quoted = 1; |
| 1049 | } |
| 1050 | if (want_escape) { |
| 1051 | if (!dialect->escapechar) { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1052 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1053 | "need to escape, but no escapechar set"); |
| 1054 | return -1; |
| 1055 | } |
| 1056 | ADDCH(dialect->escapechar); |
| 1057 | } |
| 1058 | } |
| 1059 | /* Copy field character into record buffer. |
| 1060 | */ |
| 1061 | ADDCH(c); |
| 1062 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1063 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1064 | /* If field is empty check if it needs to be quoted. |
| 1065 | */ |
| 1066 | if (i == 0 && quote_empty) { |
| 1067 | if (dialect->quoting == QUOTE_NONE) { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1068 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1069 | "single empty field record must be quoted"); |
| 1070 | return -1; |
| 1071 | } |
| 1072 | else |
| 1073 | *quoted = 1; |
| 1074 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1075 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1076 | if (*quoted) { |
| 1077 | if (copy_phase) |
| 1078 | ADDCH(dialect->quotechar); |
| 1079 | else |
| 1080 | rec_len += 2; |
| 1081 | } |
| 1082 | return rec_len; |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1083 | #undef ADDCH |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1084 | } |
| 1085 | |
| 1086 | static int |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1087 | join_check_rec_size(WriterObj *self, Py_ssize_t rec_len) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1088 | { |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 1089 | |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1090 | if (rec_len < 0 || rec_len > PY_SSIZE_T_MAX - MEM_INCR) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1091 | PyErr_NoMemory(); |
| 1092 | return 0; |
| 1093 | } |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 1094 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1095 | if (rec_len > self->rec_size) { |
| 1096 | if (self->rec_size == 0) { |
| 1097 | self->rec_size = (rec_len / MEM_INCR + 1) * MEM_INCR; |
| 1098 | if (self->rec != NULL) |
| 1099 | PyMem_Free(self->rec); |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1100 | self->rec = PyMem_New(Py_UCS4, self->rec_size); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1101 | } |
| 1102 | else { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1103 | Py_UCS4* old_rec = self->rec; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1104 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1105 | self->rec_size = (rec_len / MEM_INCR + 1) * MEM_INCR; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1106 | self->rec = PyMem_Resize(old_rec, Py_UCS4, self->rec_size); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1107 | if (self->rec == NULL) |
| 1108 | PyMem_Free(old_rec); |
| 1109 | } |
| 1110 | if (self->rec == NULL) { |
| 1111 | PyErr_NoMemory(); |
| 1112 | return 0; |
| 1113 | } |
| 1114 | } |
| 1115 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1116 | } |
| 1117 | |
| 1118 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1119 | join_append(WriterObj *self, PyObject *field, int *quoted, int quote_empty) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1120 | { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1121 | unsigned int field_kind = -1; |
| 1122 | void *field_data = NULL; |
| 1123 | Py_ssize_t field_len = 0; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1124 | Py_ssize_t rec_len; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1125 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1126 | if (field != NULL) { |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1127 | if (PyUnicode_READY(field) == -1) |
| 1128 | return 0; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1129 | field_kind = PyUnicode_KIND(field); |
| 1130 | field_data = PyUnicode_DATA(field); |
| 1131 | field_len = PyUnicode_GET_LENGTH(field); |
| 1132 | } |
| 1133 | rec_len = join_append_data(self, field_kind, field_data, field_len, |
| 1134 | quote_empty, quoted, 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1135 | if (rec_len < 0) |
| 1136 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1137 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1138 | /* grow record buffer if necessary */ |
| 1139 | if (!join_check_rec_size(self, rec_len)) |
| 1140 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1141 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1142 | self->rec_len = join_append_data(self, field_kind, field_data, field_len, |
| 1143 | quote_empty, quoted, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1144 | self->num_fields++; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1145 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1146 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1147 | } |
| 1148 | |
| 1149 | static int |
| 1150 | join_append_lineterminator(WriterObj *self) |
| 1151 | { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1152 | Py_ssize_t terminator_len, i; |
| 1153 | unsigned int term_kind; |
| 1154 | void *term_data; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1155 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1156 | terminator_len = PyUnicode_GET_LENGTH(self->dialect->lineterminator); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1157 | if (terminator_len == -1) |
| 1158 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1159 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1160 | /* grow record buffer if necessary */ |
| 1161 | if (!join_check_rec_size(self, self->rec_len + terminator_len)) |
| 1162 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1163 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1164 | term_kind = PyUnicode_KIND(self->dialect->lineterminator); |
| 1165 | term_data = PyUnicode_DATA(self->dialect->lineterminator); |
| 1166 | for (i = 0; i < terminator_len; i++) |
| 1167 | self->rec[self->rec_len + i] = PyUnicode_READ(term_kind, term_data, i); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1168 | self->rec_len += terminator_len; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1169 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1170 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1171 | } |
| 1172 | |
| 1173 | PyDoc_STRVAR(csv_writerow_doc, |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1174 | "writerow(sequence)\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1175 | "\n" |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1176 | "Construct and write a CSV record from a sequence of fields. Non-string\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1177 | "elements will be converted to string."); |
| 1178 | |
| 1179 | static PyObject * |
| 1180 | csv_writerow(WriterObj *self, PyObject *seq) |
| 1181 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1182 | DialectObj *dialect = self->dialect; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1183 | Py_ssize_t len, i; |
Antoine Pitrou | 72ca65d | 2011-10-07 04:35:30 +0200 | [diff] [blame] | 1184 | PyObject *line, *result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1185 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1186 | if (!PySequence_Check(seq)) |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1187 | return PyErr_Format(_csvstate_global->error_obj, "sequence expected"); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1188 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1189 | len = PySequence_Length(seq); |
| 1190 | if (len < 0) |
| 1191 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1192 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1193 | /* Join all fields in internal buffer. |
| 1194 | */ |
| 1195 | join_reset(self); |
| 1196 | for (i = 0; i < len; i++) { |
| 1197 | PyObject *field; |
| 1198 | int append_ok; |
| 1199 | int quoted; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1200 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1201 | field = PySequence_GetItem(seq, i); |
| 1202 | if (field == NULL) |
| 1203 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1204 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1205 | switch (dialect->quoting) { |
| 1206 | case QUOTE_NONNUMERIC: |
| 1207 | quoted = !PyNumber_Check(field); |
| 1208 | break; |
| 1209 | case QUOTE_ALL: |
| 1210 | quoted = 1; |
| 1211 | break; |
| 1212 | default: |
| 1213 | quoted = 0; |
| 1214 | break; |
| 1215 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1216 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1217 | if (PyUnicode_Check(field)) { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1218 | append_ok = join_append(self, field, "ed, len == 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1219 | Py_DECREF(field); |
| 1220 | } |
| 1221 | else if (field == Py_None) { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1222 | append_ok = join_append(self, NULL, "ed, len == 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1223 | Py_DECREF(field); |
| 1224 | } |
| 1225 | else { |
| 1226 | PyObject *str; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1227 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1228 | str = PyObject_Str(field); |
| 1229 | Py_DECREF(field); |
| 1230 | if (str == NULL) |
| 1231 | return NULL; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1232 | append_ok = join_append(self, str, "ed, len == 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1233 | Py_DECREF(str); |
| 1234 | } |
| 1235 | if (!append_ok) |
| 1236 | return NULL; |
| 1237 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1238 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1239 | /* Add line terminator. |
| 1240 | */ |
| 1241 | if (!join_append_lineterminator(self)) |
| 1242 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1243 | |
Antoine Pitrou | 72ca65d | 2011-10-07 04:35:30 +0200 | [diff] [blame] | 1244 | line = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, |
| 1245 | (void *) self->rec, self->rec_len); |
| 1246 | if (line == NULL) |
| 1247 | return NULL; |
| 1248 | result = PyObject_CallFunctionObjArgs(self->writeline, line, NULL); |
| 1249 | Py_DECREF(line); |
| 1250 | return result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1251 | } |
| 1252 | |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1253 | PyDoc_STRVAR(csv_writerows_doc, |
| 1254 | "writerows(sequence of sequences)\n" |
| 1255 | "\n" |
| 1256 | "Construct and write a series of sequences to a csv file. Non-string\n" |
| 1257 | "elements will be converted to string."); |
| 1258 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1259 | static PyObject * |
| 1260 | csv_writerows(WriterObj *self, PyObject *seqseq) |
| 1261 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1262 | PyObject *row_iter, *row_obj, *result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1263 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1264 | row_iter = PyObject_GetIter(seqseq); |
| 1265 | if (row_iter == NULL) { |
| 1266 | PyErr_SetString(PyExc_TypeError, |
| 1267 | "writerows() argument must be iterable"); |
| 1268 | return NULL; |
| 1269 | } |
| 1270 | while ((row_obj = PyIter_Next(row_iter))) { |
| 1271 | result = csv_writerow(self, row_obj); |
| 1272 | Py_DECREF(row_obj); |
| 1273 | if (!result) { |
| 1274 | Py_DECREF(row_iter); |
| 1275 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1276 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1277 | else |
| 1278 | Py_DECREF(result); |
| 1279 | } |
| 1280 | Py_DECREF(row_iter); |
| 1281 | if (PyErr_Occurred()) |
| 1282 | return NULL; |
| 1283 | Py_INCREF(Py_None); |
| 1284 | return Py_None; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1285 | } |
| 1286 | |
| 1287 | static struct PyMethodDef Writer_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1288 | { "writerow", (PyCFunction)csv_writerow, METH_O, csv_writerow_doc}, |
| 1289 | { "writerows", (PyCFunction)csv_writerows, METH_O, csv_writerows_doc}, |
| 1290 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1291 | }; |
| 1292 | |
| 1293 | #define W_OFF(x) offsetof(WriterObj, x) |
| 1294 | |
| 1295 | static struct PyMemberDef Writer_memberlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1296 | { "dialect", T_OBJECT, W_OFF(dialect), READONLY }, |
| 1297 | { NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1298 | }; |
| 1299 | |
| 1300 | static void |
| 1301 | Writer_dealloc(WriterObj *self) |
| 1302 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1303 | PyObject_GC_UnTrack(self); |
| 1304 | Py_XDECREF(self->dialect); |
| 1305 | Py_XDECREF(self->writeline); |
| 1306 | if (self->rec != NULL) |
| 1307 | PyMem_Free(self->rec); |
| 1308 | PyObject_GC_Del(self); |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 1309 | } |
| 1310 | |
| 1311 | static int |
| 1312 | Writer_traverse(WriterObj *self, visitproc visit, void *arg) |
| 1313 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1314 | Py_VISIT(self->dialect); |
| 1315 | Py_VISIT(self->writeline); |
| 1316 | return 0; |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 1317 | } |
| 1318 | |
| 1319 | static int |
| 1320 | Writer_clear(WriterObj *self) |
| 1321 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1322 | Py_CLEAR(self->dialect); |
| 1323 | Py_CLEAR(self->writeline); |
| 1324 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1325 | } |
| 1326 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1327 | PyDoc_STRVAR(Writer_Type_doc, |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1328 | "CSV writer\n" |
| 1329 | "\n" |
| 1330 | "Writer objects are responsible for generating tabular data\n" |
| 1331 | "in CSV format from sequence input.\n" |
| 1332 | ); |
| 1333 | |
| 1334 | static PyTypeObject Writer_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1335 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1336 | "_csv.writer", /*tp_name*/ |
| 1337 | sizeof(WriterObj), /*tp_basicsize*/ |
| 1338 | 0, /*tp_itemsize*/ |
| 1339 | /* methods */ |
| 1340 | (destructor)Writer_dealloc, /*tp_dealloc*/ |
| 1341 | (printfunc)0, /*tp_print*/ |
| 1342 | (getattrfunc)0, /*tp_getattr*/ |
| 1343 | (setattrfunc)0, /*tp_setattr*/ |
| 1344 | 0, /*tp_reserved*/ |
| 1345 | (reprfunc)0, /*tp_repr*/ |
| 1346 | 0, /*tp_as_number*/ |
| 1347 | 0, /*tp_as_sequence*/ |
| 1348 | 0, /*tp_as_mapping*/ |
| 1349 | (hashfunc)0, /*tp_hash*/ |
| 1350 | (ternaryfunc)0, /*tp_call*/ |
| 1351 | (reprfunc)0, /*tp_str*/ |
| 1352 | 0, /*tp_getattro*/ |
| 1353 | 0, /*tp_setattro*/ |
| 1354 | 0, /*tp_as_buffer*/ |
| 1355 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | |
| 1356 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
| 1357 | Writer_Type_doc, |
| 1358 | (traverseproc)Writer_traverse, /*tp_traverse*/ |
| 1359 | (inquiry)Writer_clear, /*tp_clear*/ |
| 1360 | 0, /*tp_richcompare*/ |
| 1361 | 0, /*tp_weaklistoffset*/ |
| 1362 | (getiterfunc)0, /*tp_iter*/ |
| 1363 | (getiterfunc)0, /*tp_iternext*/ |
| 1364 | Writer_methods, /*tp_methods*/ |
| 1365 | Writer_memberlist, /*tp_members*/ |
| 1366 | 0, /*tp_getset*/ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1367 | }; |
| 1368 | |
| 1369 | static PyObject * |
| 1370 | csv_writer(PyObject *module, PyObject *args, PyObject *keyword_args) |
| 1371 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1372 | PyObject * output_file, * dialect = NULL; |
| 1373 | WriterObj * self = PyObject_GC_New(WriterObj, &Writer_Type); |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1374 | _Py_IDENTIFIER(write); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1375 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1376 | if (!self) |
| 1377 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1378 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1379 | self->dialect = NULL; |
| 1380 | self->writeline = NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1381 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1382 | self->rec = NULL; |
| 1383 | self->rec_size = 0; |
| 1384 | self->rec_len = 0; |
| 1385 | self->num_fields = 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1386 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1387 | if (!PyArg_UnpackTuple(args, "", 1, 2, &output_file, &dialect)) { |
| 1388 | Py_DECREF(self); |
| 1389 | return NULL; |
| 1390 | } |
Martin v. Löwis | 1ee1b6f | 2011-10-10 18:11:30 +0200 | [diff] [blame] | 1391 | self->writeline = _PyObject_GetAttrId(output_file, &PyId_write); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1392 | if (self->writeline == NULL || !PyCallable_Check(self->writeline)) { |
| 1393 | PyErr_SetString(PyExc_TypeError, |
| 1394 | "argument 1 must have a \"write\" method"); |
| 1395 | Py_DECREF(self); |
| 1396 | return NULL; |
| 1397 | } |
| 1398 | self->dialect = (DialectObj *)_call_dialect(dialect, keyword_args); |
| 1399 | if (self->dialect == NULL) { |
| 1400 | Py_DECREF(self); |
| 1401 | return NULL; |
| 1402 | } |
| 1403 | PyObject_GC_Track(self); |
| 1404 | return (PyObject *)self; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1405 | } |
| 1406 | |
| 1407 | /* |
| 1408 | * DIALECT REGISTRY |
| 1409 | */ |
| 1410 | static PyObject * |
| 1411 | csv_list_dialects(PyObject *module, PyObject *args) |
| 1412 | { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1413 | return PyDict_Keys(_csvstate_global->dialects); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1414 | } |
| 1415 | |
| 1416 | static PyObject * |
Andrew McNamara | 8662597 | 2005-01-11 01:28:33 +0000 | [diff] [blame] | 1417 | csv_register_dialect(PyObject *module, PyObject *args, PyObject *kwargs) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1418 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1419 | PyObject *name_obj, *dialect_obj = NULL; |
| 1420 | PyObject *dialect; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1421 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1422 | if (!PyArg_UnpackTuple(args, "", 1, 2, &name_obj, &dialect_obj)) |
| 1423 | return NULL; |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1424 | if (!PyUnicode_Check(name_obj)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1425 | PyErr_SetString(PyExc_TypeError, |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1426 | "dialect name must be a string"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1427 | return NULL; |
| 1428 | } |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1429 | if (PyUnicode_READY(name_obj) == -1) |
| 1430 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1431 | dialect = _call_dialect(dialect_obj, kwargs); |
| 1432 | if (dialect == NULL) |
| 1433 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1434 | if (PyDict_SetItem(_csvstate_global->dialects, name_obj, dialect) < 0) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1435 | Py_DECREF(dialect); |
| 1436 | return NULL; |
| 1437 | } |
| 1438 | Py_DECREF(dialect); |
| 1439 | Py_INCREF(Py_None); |
| 1440 | return Py_None; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1441 | } |
| 1442 | |
| 1443 | static PyObject * |
Skip Montanaro | 577c7a7 | 2003-04-12 19:17:14 +0000 | [diff] [blame] | 1444 | csv_unregister_dialect(PyObject *module, PyObject *name_obj) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1445 | { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1446 | if (PyDict_DelItem(_csvstate_global->dialects, name_obj) < 0) |
| 1447 | return PyErr_Format(_csvstate_global->error_obj, "unknown dialect"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1448 | Py_INCREF(Py_None); |
| 1449 | return Py_None; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1450 | } |
| 1451 | |
| 1452 | static PyObject * |
Skip Montanaro | 577c7a7 | 2003-04-12 19:17:14 +0000 | [diff] [blame] | 1453 | csv_get_dialect(PyObject *module, PyObject *name_obj) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1454 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1455 | return get_dialect_from_registry(name_obj); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1456 | } |
| 1457 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1458 | static PyObject * |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1459 | csv_field_size_limit(PyObject *module, PyObject *args) |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1460 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1461 | PyObject *new_limit = NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1462 | long old_limit = _csvstate_global->field_limit; |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1463 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1464 | if (!PyArg_UnpackTuple(args, "field_size_limit", 0, 1, &new_limit)) |
| 1465 | return NULL; |
| 1466 | if (new_limit != NULL) { |
| 1467 | if (!PyLong_CheckExact(new_limit)) { |
| 1468 | PyErr_Format(PyExc_TypeError, |
| 1469 | "limit must be an integer"); |
| 1470 | return NULL; |
| 1471 | } |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1472 | _csvstate_global->field_limit = PyLong_AsLong(new_limit); |
| 1473 | if (_csvstate_global->field_limit == -1 && PyErr_Occurred()) { |
| 1474 | _csvstate_global->field_limit = old_limit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1475 | return NULL; |
| 1476 | } |
| 1477 | } |
| 1478 | return PyLong_FromLong(old_limit); |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1479 | } |
| 1480 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1481 | /* |
| 1482 | * MODULE |
| 1483 | */ |
| 1484 | |
| 1485 | PyDoc_STRVAR(csv_module_doc, |
| 1486 | "CSV parsing and writing.\n" |
| 1487 | "\n" |
| 1488 | "This module provides classes that assist in the reading and writing\n" |
| 1489 | "of Comma Separated Value (CSV) files, and implements the interface\n" |
| 1490 | "described by PEP 305. Although many CSV files are simple to parse,\n" |
| 1491 | "the format is not formally defined by a stable specification and\n" |
| 1492 | "is subtle enough that parsing lines of a CSV file with something\n" |
| 1493 | "like line.split(\",\") is bound to fail. The module supports three\n" |
| 1494 | "basic APIs: reading, writing, and registration of dialects.\n" |
| 1495 | "\n" |
| 1496 | "\n" |
| 1497 | "DIALECT REGISTRATION:\n" |
| 1498 | "\n" |
| 1499 | "Readers and writers support a dialect argument, which is a convenient\n" |
| 1500 | "handle on a group of settings. When the dialect argument is a string,\n" |
| 1501 | "it identifies one of the dialects previously registered with the module.\n" |
| 1502 | "If it is a class or instance, the attributes of the argument are used as\n" |
| 1503 | "the settings for the reader or writer:\n" |
| 1504 | "\n" |
| 1505 | " class excel:\n" |
| 1506 | " delimiter = ','\n" |
| 1507 | " quotechar = '\"'\n" |
| 1508 | " escapechar = None\n" |
| 1509 | " doublequote = True\n" |
| 1510 | " skipinitialspace = False\n" |
Johannes Gijsbers | 8d3b9dd | 2004-08-15 12:23:10 +0000 | [diff] [blame] | 1511 | " lineterminator = '\\r\\n'\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1512 | " quoting = QUOTE_MINIMAL\n" |
| 1513 | "\n" |
| 1514 | "SETTINGS:\n" |
| 1515 | "\n" |
| 1516 | " * quotechar - specifies a one-character string to use as the \n" |
| 1517 | " quoting character. It defaults to '\"'.\n" |
| 1518 | " * delimiter - specifies a one-character string to use as the \n" |
| 1519 | " field separator. It defaults to ','.\n" |
| 1520 | " * skipinitialspace - specifies how to interpret whitespace which\n" |
| 1521 | " immediately follows a delimiter. It defaults to False, which\n" |
| 1522 | " means that whitespace immediately following a delimiter is part\n" |
| 1523 | " of the following field.\n" |
| 1524 | " * lineterminator - specifies the character sequence which should \n" |
| 1525 | " terminate rows.\n" |
| 1526 | " * quoting - controls when quotes should be generated by the writer.\n" |
| 1527 | " It can take on any of the following module constants:\n" |
| 1528 | "\n" |
| 1529 | " csv.QUOTE_MINIMAL means only when required, for example, when a\n" |
| 1530 | " field contains either the quotechar or the delimiter\n" |
| 1531 | " csv.QUOTE_ALL means that quotes are always placed around fields.\n" |
| 1532 | " csv.QUOTE_NONNUMERIC means that quotes are always placed around\n" |
Skip Montanaro | 148eb6a | 2003-12-02 18:57:47 +0000 | [diff] [blame] | 1533 | " fields which do not parse as integers or floating point\n" |
| 1534 | " numbers.\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1535 | " csv.QUOTE_NONE means that quotes are never placed around fields.\n" |
| 1536 | " * escapechar - specifies a one-character string used to escape \n" |
| 1537 | " the delimiter when quoting is set to QUOTE_NONE.\n" |
| 1538 | " * doublequote - controls the handling of quotes inside fields. When\n" |
| 1539 | " True, two consecutive quotes are interpreted as one during read,\n" |
| 1540 | " and when writing, each quote character embedded in the data is\n" |
| 1541 | " written as two quotes\n"); |
| 1542 | |
| 1543 | PyDoc_STRVAR(csv_reader_doc, |
| 1544 | " csv_reader = reader(iterable [, dialect='excel']\n" |
| 1545 | " [optional keyword args])\n" |
| 1546 | " for row in csv_reader:\n" |
| 1547 | " process(row)\n" |
| 1548 | "\n" |
| 1549 | "The \"iterable\" argument can be any object that returns a line\n" |
| 1550 | "of input for each iteration, such as a file object or a list. The\n" |
| 1551 | "optional \"dialect\" parameter is discussed below. The function\n" |
| 1552 | "also accepts optional keyword arguments which override settings\n" |
| 1553 | "provided by the dialect.\n" |
| 1554 | "\n" |
| 1555 | "The returned object is an iterator. Each iteration returns a row\n" |
Johannes Gijsbers | 8d3b9dd | 2004-08-15 12:23:10 +0000 | [diff] [blame] | 1556 | "of the CSV file (which can span multiple input lines):\n"); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1557 | |
| 1558 | PyDoc_STRVAR(csv_writer_doc, |
| 1559 | " csv_writer = csv.writer(fileobj [, dialect='excel']\n" |
| 1560 | " [optional keyword args])\n" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1561 | " for row in sequence:\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1562 | " csv_writer.writerow(row)\n" |
| 1563 | "\n" |
| 1564 | " [or]\n" |
| 1565 | "\n" |
| 1566 | " csv_writer = csv.writer(fileobj [, dialect='excel']\n" |
| 1567 | " [optional keyword args])\n" |
| 1568 | " csv_writer.writerows(rows)\n" |
| 1569 | "\n" |
| 1570 | "The \"fileobj\" argument can be any object that supports the file API.\n"); |
| 1571 | |
| 1572 | PyDoc_STRVAR(csv_list_dialects_doc, |
| 1573 | "Return a list of all know dialect names.\n" |
| 1574 | " names = csv.list_dialects()"); |
| 1575 | |
| 1576 | PyDoc_STRVAR(csv_get_dialect_doc, |
| 1577 | "Return the dialect instance associated with name.\n" |
| 1578 | " dialect = csv.get_dialect(name)"); |
| 1579 | |
| 1580 | PyDoc_STRVAR(csv_register_dialect_doc, |
| 1581 | "Create a mapping from a string name to a dialect class.\n" |
| 1582 | " dialect = csv.register_dialect(name, dialect)"); |
| 1583 | |
| 1584 | PyDoc_STRVAR(csv_unregister_dialect_doc, |
| 1585 | "Delete the name/dialect mapping associated with a string name.\n" |
| 1586 | " csv.unregister_dialect(name)"); |
| 1587 | |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1588 | PyDoc_STRVAR(csv_field_size_limit_doc, |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1589 | "Sets an upper limit on parsed fields.\n" |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1590 | " csv.field_size_limit([limit])\n" |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1591 | "\n" |
| 1592 | "Returns old limit. If limit is not given, no new limit is set and\n" |
| 1593 | "the old limit is returned"); |
| 1594 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1595 | static struct PyMethodDef csv_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1596 | { "reader", (PyCFunction)csv_reader, |
| 1597 | METH_VARARGS | METH_KEYWORDS, csv_reader_doc}, |
| 1598 | { "writer", (PyCFunction)csv_writer, |
| 1599 | METH_VARARGS | METH_KEYWORDS, csv_writer_doc}, |
| 1600 | { "list_dialects", (PyCFunction)csv_list_dialects, |
| 1601 | METH_NOARGS, csv_list_dialects_doc}, |
| 1602 | { "register_dialect", (PyCFunction)csv_register_dialect, |
| 1603 | METH_VARARGS | METH_KEYWORDS, csv_register_dialect_doc}, |
| 1604 | { "unregister_dialect", (PyCFunction)csv_unregister_dialect, |
| 1605 | METH_O, csv_unregister_dialect_doc}, |
| 1606 | { "get_dialect", (PyCFunction)csv_get_dialect, |
| 1607 | METH_O, csv_get_dialect_doc}, |
| 1608 | { "field_size_limit", (PyCFunction)csv_field_size_limit, |
| 1609 | METH_VARARGS, csv_field_size_limit_doc}, |
| 1610 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1611 | }; |
| 1612 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1613 | static struct PyModuleDef _csvmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1614 | PyModuleDef_HEAD_INIT, |
| 1615 | "_csv", |
| 1616 | csv_module_doc, |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1617 | sizeof(_csvstate), |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1618 | csv_methods, |
| 1619 | NULL, |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1620 | _csv_traverse, |
| 1621 | _csv_clear, |
| 1622 | _csv_free |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1623 | }; |
| 1624 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1625 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1626 | PyInit__csv(void) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1627 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1628 | PyObject *module; |
| 1629 | StyleDesc *style; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1630 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1631 | if (PyType_Ready(&Dialect_Type) < 0) |
| 1632 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1633 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1634 | if (PyType_Ready(&Reader_Type) < 0) |
| 1635 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1636 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1637 | if (PyType_Ready(&Writer_Type) < 0) |
| 1638 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1639 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1640 | /* Create the module and add the functions */ |
| 1641 | module = PyModule_Create(&_csvmodule); |
| 1642 | if (module == NULL) |
| 1643 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1644 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1645 | /* Add version to the module. */ |
| 1646 | if (PyModule_AddStringConstant(module, "__version__", |
| 1647 | MODULE_VERSION) == -1) |
| 1648 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1649 | |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1650 | /* Set the field limit */ |
| 1651 | _csvstate(module)->field_limit = 128 * 1024; |
| 1652 | /* Do I still need to add this var to the Module Dict? */ |
| 1653 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1654 | /* Add _dialects dictionary */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1655 | _csvstate(module)->dialects = PyDict_New(); |
| 1656 | if (_csvstate(module)->dialects == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1657 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1658 | Py_INCREF(_csvstate(module)->dialects); |
| 1659 | if (PyModule_AddObject(module, "_dialects", _csvstate(module)->dialects)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1660 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1661 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1662 | /* Add quote styles into dictionary */ |
| 1663 | for (style = quote_styles; style->name; style++) { |
| 1664 | if (PyModule_AddIntConstant(module, style->name, |
| 1665 | style->style) == -1) |
| 1666 | return NULL; |
| 1667 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1668 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1669 | /* Add the Dialect type */ |
| 1670 | Py_INCREF(&Dialect_Type); |
| 1671 | if (PyModule_AddObject(module, "Dialect", (PyObject *)&Dialect_Type)) |
| 1672 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1673 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1674 | /* Add the CSV exception object to the module. */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1675 | _csvstate(module)->error_obj = PyErr_NewException("_csv.Error", NULL, NULL); |
| 1676 | if (_csvstate(module)->error_obj == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1677 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1678 | Py_INCREF(_csvstate(module)->error_obj); |
| 1679 | PyModule_AddObject(module, "Error", _csvstate(module)->error_obj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1680 | return module; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1681 | } |