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" |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 15 | #include <stdbool.h> |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 16 | |
Andrew McNamara | 37d2bdf | 2005-01-10 12:22:48 +0000 | [diff] [blame] | 17 | |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 18 | typedef struct { |
| 19 | PyObject *error_obj; /* CSV exception */ |
| 20 | PyObject *dialects; /* Dialect registry */ |
| 21 | long field_limit; /* max parsed field size */ |
| 22 | } _csvstate; |
| 23 | |
| 24 | #define _csvstate(o) ((_csvstate *)PyModule_GetState(o)) |
| 25 | |
| 26 | static int |
| 27 | _csv_clear(PyObject *m) |
| 28 | { |
| 29 | Py_CLEAR(_csvstate(m)->error_obj); |
| 30 | Py_CLEAR(_csvstate(m)->dialects); |
| 31 | return 0; |
| 32 | } |
| 33 | |
| 34 | static int |
| 35 | _csv_traverse(PyObject *m, visitproc visit, void *arg) |
| 36 | { |
| 37 | Py_VISIT(_csvstate(m)->error_obj); |
| 38 | Py_VISIT(_csvstate(m)->dialects); |
| 39 | return 0; |
| 40 | } |
| 41 | |
| 42 | static void |
| 43 | _csv_free(void *m) |
| 44 | { |
| 45 | _csv_clear((PyObject *)m); |
| 46 | } |
| 47 | |
| 48 | static struct PyModuleDef _csvmodule; |
| 49 | |
| 50 | #define _csvstate_global ((_csvstate *)PyModule_GetState(PyState_FindModule(&_csvmodule))) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 51 | |
| 52 | typedef enum { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 53 | START_RECORD, START_FIELD, ESCAPED_CHAR, IN_FIELD, |
| 54 | IN_QUOTED_FIELD, ESCAPE_IN_QUOTED_FIELD, QUOTE_IN_QUOTED_FIELD, |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 55 | EAT_CRNL,AFTER_ESCAPED_CRNL |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 56 | } ParserState; |
| 57 | |
| 58 | typedef enum { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 59 | QUOTE_MINIMAL, QUOTE_ALL, QUOTE_NONNUMERIC, QUOTE_NONE |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 60 | } QuoteStyle; |
| 61 | |
| 62 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 63 | QuoteStyle style; |
Serhiy Storchaka | 2d06e84 | 2015-12-25 19:53:18 +0200 | [diff] [blame] | 64 | const char *name; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 65 | } StyleDesc; |
| 66 | |
Serhiy Storchaka | 2d06e84 | 2015-12-25 19:53:18 +0200 | [diff] [blame] | 67 | static const StyleDesc quote_styles[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 68 | { QUOTE_MINIMAL, "QUOTE_MINIMAL" }, |
| 69 | { QUOTE_ALL, "QUOTE_ALL" }, |
| 70 | { QUOTE_NONNUMERIC, "QUOTE_NONNUMERIC" }, |
| 71 | { QUOTE_NONE, "QUOTE_NONE" }, |
| 72 | { 0 } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 73 | }; |
| 74 | |
| 75 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 76 | PyObject_HEAD |
Guido van Rossum | 4626458 | 2007-08-06 19:32:18 +0000 | [diff] [blame] | 77 | |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 78 | char doublequote; /* is " represented by ""? */ |
| 79 | char skipinitialspace; /* ignore spaces following delimiter? */ |
| 80 | char strict; /* raise exception on bad CSV */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 81 | int quoting; /* style of quoting to write */ |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 82 | Py_UCS4 delimiter; /* field separator */ |
| 83 | Py_UCS4 quotechar; /* quote character */ |
| 84 | Py_UCS4 escapechar; /* escape character */ |
| 85 | PyObject *lineterminator; /* string to write between records */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 86 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 87 | } DialectObj; |
| 88 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 89 | static PyTypeObject Dialect_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 90 | |
| 91 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 92 | PyObject_HEAD |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 93 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 94 | PyObject *input_iter; /* iterate over this for input lines */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 95 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 96 | DialectObj *dialect; /* parsing dialect */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 97 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 98 | PyObject *fields; /* field list for current record */ |
| 99 | ParserState state; /* current CSV parse state */ |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 100 | Py_UCS4 *field; /* temporary buffer */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 101 | Py_ssize_t field_size; /* size of allocated buffer */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 102 | Py_ssize_t field_len; /* length of current field */ |
| 103 | int numeric_field; /* treat field as numeric */ |
| 104 | unsigned long line_num; /* Source-file line number */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 105 | } ReaderObj; |
| 106 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 107 | static PyTypeObject Reader_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 108 | |
Christian Heimes | 90aa764 | 2007-12-19 02:45:37 +0000 | [diff] [blame] | 109 | #define ReaderObject_Check(v) (Py_TYPE(v) == &Reader_Type) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 110 | |
| 111 | typedef struct { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 112 | PyObject_HEAD |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 113 | |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 114 | PyObject *write; /* write output lines to this file */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 115 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 116 | DialectObj *dialect; /* parsing dialect */ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 117 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 118 | Py_UCS4 *rec; /* buffer for parser.join */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 119 | Py_ssize_t rec_size; /* size of allocated record */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 120 | Py_ssize_t rec_len; /* length of record */ |
| 121 | int num_fields; /* number of fields in record */ |
Guido van Rossum | 4626458 | 2007-08-06 19:32:18 +0000 | [diff] [blame] | 122 | } WriterObj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 123 | |
Neal Norwitz | 227b533 | 2006-03-22 09:28:35 +0000 | [diff] [blame] | 124 | static PyTypeObject Writer_Type; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 125 | |
| 126 | /* |
| 127 | * DIALECT class |
| 128 | */ |
| 129 | |
| 130 | static PyObject * |
| 131 | get_dialect_from_registry(PyObject * name_obj) |
| 132 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 133 | PyObject *dialect_obj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 134 | |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 135 | dialect_obj = PyDict_GetItemWithError(_csvstate_global->dialects, name_obj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 136 | if (dialect_obj == NULL) { |
| 137 | if (!PyErr_Occurred()) |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 138 | PyErr_Format(_csvstate_global->error_obj, "unknown dialect"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 139 | } |
| 140 | else |
| 141 | Py_INCREF(dialect_obj); |
| 142 | return dialect_obj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 143 | } |
| 144 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 145 | static PyObject * |
| 146 | get_string(PyObject *str) |
| 147 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 148 | Py_XINCREF(str); |
| 149 | return str; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 150 | } |
| 151 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 152 | static PyObject * |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 153 | get_nullchar_as_None(Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 154 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 155 | if (c == '\0') { |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 156 | Py_RETURN_NONE; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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 * |
Serhiy Storchaka | d4f9cf5 | 2018-11-27 19:34:35 +0200 | [diff] [blame] | 163 | Dialect_get_lineterminator(DialectObj *self, void *Py_UNUSED(ignored)) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 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 * |
Serhiy Storchaka | d4f9cf5 | 2018-11-27 19:34:35 +0200 | [diff] [blame] | 169 | Dialect_get_delimiter(DialectObj *self, void *Py_UNUSED(ignored)) |
Guido van Rossum | a9769c2 | 2007-08-07 23:59:30 +0000 | [diff] [blame] | 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 * |
Serhiy Storchaka | d4f9cf5 | 2018-11-27 19:34:35 +0200 | [diff] [blame] | 175 | Dialect_get_escapechar(DialectObj *self, void *Py_UNUSED(ignored)) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 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 * |
Serhiy Storchaka | d4f9cf5 | 2018-11-27 19:34:35 +0200 | [diff] [blame] | 181 | Dialect_get_quotechar(DialectObj *self, void *Py_UNUSED(ignored)) |
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 * |
Serhiy Storchaka | d4f9cf5 | 2018-11-27 19:34:35 +0200 | [diff] [blame] | 187 | Dialect_get_quoting(DialectObj *self, void *Py_UNUSED(ignored)) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 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 |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 193 | _set_bool(const char *name, char *target, PyObject *src, bool 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; |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 201 | *target = (char)b; |
Antoine Pitrou | 6f430e4 | 2012-08-15 23:18:25 +0200 | [diff] [blame] | 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 { |
Victor Stinner | 7a6dbb7 | 2016-10-19 16:00:37 +0200 | [diff] [blame] | 212 | int value; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 213 | if (!PyLong_CheckExact(src)) { |
| 214 | PyErr_Format(PyExc_TypeError, |
| 215 | "\"%s\" must be an integer", name); |
| 216 | return -1; |
| 217 | } |
Victor Stinner | 7a6dbb7 | 2016-10-19 16:00:37 +0200 | [diff] [blame] | 218 | value = _PyLong_AsInt(src); |
| 219 | if (value == -1 && PyErr_Occurred()) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 220 | return -1; |
| 221 | } |
Victor Stinner | 7a6dbb7 | 2016-10-19 16:00:37 +0200 | [diff] [blame] | 222 | *target = value; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 223 | } |
| 224 | return 0; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 225 | } |
| 226 | |
| 227 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 228 | _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] | 229 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 230 | if (src == NULL) |
| 231 | *target = dflt; |
| 232 | else { |
| 233 | *target = '\0'; |
| 234 | if (src != Py_None) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 235 | Py_ssize_t len; |
Serhiy Storchaka | cac23a5 | 2013-12-19 16:27:18 +0200 | [diff] [blame] | 236 | if (!PyUnicode_Check(src)) { |
| 237 | PyErr_Format(PyExc_TypeError, |
| 238 | "\"%s\" must be string, not %.200s", name, |
| 239 | src->ob_type->tp_name); |
| 240 | return -1; |
| 241 | } |
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, |
Berker Peksag | 0f41acb | 2014-07-27 23:22:34 +0300 | [diff] [blame] | 245 | "\"%s\" must be a 1-character string", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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_INCREF(src); |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 274 | Py_XSETREF(*target, src); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 275 | } |
| 276 | } |
| 277 | return 0; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 278 | } |
| 279 | |
| 280 | static int |
| 281 | dialect_check_quoting(int quoting) |
| 282 | { |
Serhiy Storchaka | 2d06e84 | 2015-12-25 19:53:18 +0200 | [diff] [blame] | 283 | const StyleDesc *qs; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 284 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 285 | for (qs = quote_styles; qs->name; qs++) { |
Victor Stinner | 706768c | 2014-08-16 01:03:39 +0200 | [diff] [blame] | 286 | if ((int)qs->style == quoting) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 287 | return 0; |
| 288 | } |
| 289 | PyErr_Format(PyExc_TypeError, "bad \"quoting\" value"); |
| 290 | return -1; |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 291 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 292 | |
| 293 | #define D_OFF(x) offsetof(DialectObj, x) |
| 294 | |
| 295 | static struct PyMemberDef Dialect_memberlist[] = { |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 296 | { "skipinitialspace", T_BOOL, D_OFF(skipinitialspace), READONLY }, |
| 297 | { "doublequote", T_BOOL, D_OFF(doublequote), READONLY }, |
| 298 | { "strict", T_BOOL, D_OFF(strict), READONLY }, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 299 | { NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 300 | }; |
| 301 | |
| 302 | static PyGetSetDef Dialect_getsetlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 303 | { "delimiter", (getter)Dialect_get_delimiter}, |
| 304 | { "escapechar", (getter)Dialect_get_escapechar}, |
| 305 | { "lineterminator", (getter)Dialect_get_lineterminator}, |
| 306 | { "quotechar", (getter)Dialect_get_quotechar}, |
| 307 | { "quoting", (getter)Dialect_get_quoting}, |
| 308 | {NULL}, |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 309 | }; |
| 310 | |
| 311 | static void |
| 312 | Dialect_dealloc(DialectObj *self) |
| 313 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 314 | Py_XDECREF(self->lineterminator); |
| 315 | Py_TYPE(self)->tp_free((PyObject *)self); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 316 | } |
| 317 | |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 318 | static char *dialect_kws[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 319 | "dialect", |
| 320 | "delimiter", |
| 321 | "doublequote", |
| 322 | "escapechar", |
| 323 | "lineterminator", |
| 324 | "quotechar", |
| 325 | "quoting", |
| 326 | "skipinitialspace", |
| 327 | "strict", |
| 328 | NULL |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 329 | }; |
| 330 | |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 331 | static PyObject * |
| 332 | dialect_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 333 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 334 | DialectObj *self; |
| 335 | PyObject *ret = NULL; |
| 336 | PyObject *dialect = NULL; |
| 337 | PyObject *delimiter = NULL; |
| 338 | PyObject *doublequote = NULL; |
| 339 | PyObject *escapechar = NULL; |
| 340 | PyObject *lineterminator = NULL; |
| 341 | PyObject *quotechar = NULL; |
| 342 | PyObject *quoting = NULL; |
| 343 | PyObject *skipinitialspace = NULL; |
| 344 | PyObject *strict = NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 345 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 346 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, |
| 347 | "|OOOOOOOOO", dialect_kws, |
| 348 | &dialect, |
| 349 | &delimiter, |
| 350 | &doublequote, |
| 351 | &escapechar, |
| 352 | &lineterminator, |
| 353 | "echar, |
| 354 | "ing, |
| 355 | &skipinitialspace, |
| 356 | &strict)) |
| 357 | return NULL; |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 358 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 359 | if (dialect != NULL) { |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 360 | if (PyUnicode_Check(dialect)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 361 | dialect = get_dialect_from_registry(dialect); |
| 362 | if (dialect == NULL) |
| 363 | return NULL; |
| 364 | } |
| 365 | else |
| 366 | Py_INCREF(dialect); |
| 367 | /* Can we reuse this instance? */ |
| 368 | if (PyObject_TypeCheck(dialect, &Dialect_Type) && |
Serhiy Storchaka | 0b3ec19 | 2017-03-23 17:53:47 +0200 | [diff] [blame] | 369 | delimiter == NULL && |
| 370 | doublequote == NULL && |
| 371 | escapechar == NULL && |
| 372 | lineterminator == NULL && |
| 373 | quotechar == NULL && |
| 374 | quoting == NULL && |
| 375 | skipinitialspace == NULL && |
| 376 | strict == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 377 | return dialect; |
| 378 | } |
Andrew McNamara | 29bf4e4 | 2005-01-11 04:49:53 +0000 | [diff] [blame] | 379 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 380 | self = (DialectObj *)type->tp_alloc(type, 0); |
| 381 | if (self == NULL) { |
| 382 | Py_XDECREF(dialect); |
| 383 | return NULL; |
| 384 | } |
| 385 | self->lineterminator = NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 386 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 387 | Py_XINCREF(delimiter); |
| 388 | Py_XINCREF(doublequote); |
| 389 | Py_XINCREF(escapechar); |
| 390 | Py_XINCREF(lineterminator); |
| 391 | Py_XINCREF(quotechar); |
| 392 | Py_XINCREF(quoting); |
| 393 | Py_XINCREF(skipinitialspace); |
| 394 | Py_XINCREF(strict); |
| 395 | if (dialect != NULL) { |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 396 | #define DIALECT_GETATTR(v, n) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 397 | if (v == NULL) \ |
| 398 | v = PyObject_GetAttrString(dialect, n) |
| 399 | DIALECT_GETATTR(delimiter, "delimiter"); |
| 400 | DIALECT_GETATTR(doublequote, "doublequote"); |
| 401 | DIALECT_GETATTR(escapechar, "escapechar"); |
| 402 | DIALECT_GETATTR(lineterminator, "lineterminator"); |
| 403 | DIALECT_GETATTR(quotechar, "quotechar"); |
| 404 | DIALECT_GETATTR(quoting, "quoting"); |
| 405 | DIALECT_GETATTR(skipinitialspace, "skipinitialspace"); |
| 406 | DIALECT_GETATTR(strict, "strict"); |
| 407 | PyErr_Clear(); |
| 408 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 409 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 410 | /* check types and convert to C values */ |
Andrew McNamara | 1196cf1 | 2005-01-07 04:42:45 +0000 | [diff] [blame] | 411 | #define DIASET(meth, name, target, src, dflt) \ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 412 | if (meth(name, target, src, dflt)) \ |
| 413 | goto err |
| 414 | DIASET(_set_char, "delimiter", &self->delimiter, delimiter, ','); |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 415 | DIASET(_set_bool, "doublequote", &self->doublequote, doublequote, true); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 416 | DIASET(_set_char, "escapechar", &self->escapechar, escapechar, 0); |
| 417 | DIASET(_set_str, "lineterminator", &self->lineterminator, lineterminator, "\r\n"); |
| 418 | DIASET(_set_char, "quotechar", &self->quotechar, quotechar, '"'); |
| 419 | DIASET(_set_int, "quoting", &self->quoting, quoting, QUOTE_MINIMAL); |
Serhiy Storchaka | 323748a | 2018-07-26 13:21:09 +0300 | [diff] [blame] | 420 | DIASET(_set_bool, "skipinitialspace", &self->skipinitialspace, skipinitialspace, false); |
| 421 | DIASET(_set_bool, "strict", &self->strict, strict, false); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 422 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 423 | /* validate options */ |
| 424 | if (dialect_check_quoting(self->quoting)) |
| 425 | goto err; |
| 426 | if (self->delimiter == 0) { |
Serhiy Storchaka | cac23a5 | 2013-12-19 16:27:18 +0200 | [diff] [blame] | 427 | PyErr_SetString(PyExc_TypeError, |
Berker Peksag | 0f41acb | 2014-07-27 23:22:34 +0300 | [diff] [blame] | 428 | "\"delimiter\" must be a 1-character string"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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 | { |
Victor Stinner | 6412f49 | 2016-08-23 00:21:34 +0200 | [diff] [blame] | 515 | PyObject *type = (PyObject *)&Dialect_Type; |
| 516 | if (dialect_inst) { |
| 517 | return _PyObject_FastCallDict(type, &dialect_inst, 1, kwargs); |
| 518 | } |
| 519 | else { |
| 520 | return _PyObject_FastCallDict(type, NULL, 0, kwargs); |
| 521 | } |
Andrew McNamara | 91b9746 | 2005-01-11 01:07:23 +0000 | [diff] [blame] | 522 | } |
| 523 | |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 524 | /* |
| 525 | * READER |
| 526 | */ |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 527 | static int |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 528 | parse_save_field(ReaderObj *self) |
| 529 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 530 | PyObject *field; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 531 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 532 | field = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, |
| 533 | (void *) self->field, self->field_len); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 534 | if (field == NULL) |
| 535 | return -1; |
| 536 | self->field_len = 0; |
| 537 | if (self->numeric_field) { |
| 538 | PyObject *tmp; |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 539 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 540 | self->numeric_field = 0; |
| 541 | tmp = PyNumber_Float(field); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 542 | Py_DECREF(field); |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 543 | if (tmp == NULL) |
| 544 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 545 | field = tmp; |
| 546 | } |
Victor Stinner | b80b378 | 2013-11-14 21:29:34 +0100 | [diff] [blame] | 547 | if (PyList_Append(self->fields, field) < 0) { |
| 548 | Py_DECREF(field); |
| 549 | return -1; |
| 550 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 551 | Py_DECREF(field); |
| 552 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 553 | } |
| 554 | |
| 555 | static int |
| 556 | parse_grow_buff(ReaderObj *self) |
| 557 | { |
Sergey Fedoseev | 67b9cc8 | 2018-08-16 09:27:50 +0500 | [diff] [blame] | 558 | assert((size_t)self->field_size <= PY_SSIZE_T_MAX / sizeof(Py_UCS4)); |
| 559 | |
| 560 | Py_ssize_t field_size_new = self->field_size ? 2 * self->field_size : 4096; |
| 561 | Py_UCS4 *field_new = self->field; |
| 562 | PyMem_Resize(field_new, Py_UCS4, field_size_new); |
| 563 | if (field_new == NULL) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 564 | PyErr_NoMemory(); |
| 565 | return 0; |
| 566 | } |
Sergey Fedoseev | 67b9cc8 | 2018-08-16 09:27:50 +0500 | [diff] [blame] | 567 | self->field = field_new; |
| 568 | self->field_size = field_size_new; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 569 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 570 | } |
| 571 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 572 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 573 | parse_add_char(ReaderObj *self, Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 574 | { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 575 | if (self->field_len >= _csvstate_global->field_limit) { |
| 576 | PyErr_Format(_csvstate_global->error_obj, "field larger than field limit (%ld)", |
| 577 | _csvstate_global->field_limit); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 578 | return -1; |
| 579 | } |
| 580 | if (self->field_len == self->field_size && !parse_grow_buff(self)) |
| 581 | return -1; |
| 582 | self->field[self->field_len++] = c; |
| 583 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 584 | } |
| 585 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 586 | static int |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 587 | parse_process_char(ReaderObj *self, Py_UCS4 c) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 588 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 589 | DialectObj *dialect = self->dialect; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 590 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 591 | switch (self->state) { |
| 592 | case START_RECORD: |
| 593 | /* start of record */ |
| 594 | if (c == '\0') |
| 595 | /* empty line - return [] */ |
| 596 | break; |
| 597 | else if (c == '\n' || c == '\r') { |
| 598 | self->state = EAT_CRNL; |
| 599 | break; |
| 600 | } |
| 601 | /* normal character - handle as START_FIELD */ |
| 602 | self->state = START_FIELD; |
| 603 | /* fallthru */ |
| 604 | case START_FIELD: |
| 605 | /* expecting field */ |
| 606 | if (c == '\n' || c == '\r' || c == '\0') { |
| 607 | /* save empty field - return [fields] */ |
| 608 | if (parse_save_field(self) < 0) |
| 609 | return -1; |
| 610 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 611 | } |
| 612 | else if (c == dialect->quotechar && |
| 613 | dialect->quoting != QUOTE_NONE) { |
| 614 | /* start quoted field */ |
| 615 | self->state = IN_QUOTED_FIELD; |
| 616 | } |
| 617 | else if (c == dialect->escapechar) { |
| 618 | /* possible escaped character */ |
| 619 | self->state = ESCAPED_CHAR; |
| 620 | } |
| 621 | else if (c == ' ' && dialect->skipinitialspace) |
| 622 | /* ignore space at start of field */ |
| 623 | ; |
| 624 | else if (c == dialect->delimiter) { |
| 625 | /* save empty field */ |
| 626 | if (parse_save_field(self) < 0) |
| 627 | return -1; |
| 628 | } |
| 629 | else { |
| 630 | /* begin new unquoted field */ |
| 631 | if (dialect->quoting == QUOTE_NONNUMERIC) |
| 632 | self->numeric_field = 1; |
| 633 | if (parse_add_char(self, c) < 0) |
| 634 | return -1; |
| 635 | self->state = IN_FIELD; |
| 636 | } |
| 637 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 638 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 639 | case ESCAPED_CHAR: |
R David Murray | 9a7d376 | 2013-03-20 00:15:20 -0400 | [diff] [blame] | 640 | if (c == '\n' || c=='\r') { |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 641 | if (parse_add_char(self, c) < 0) |
| 642 | return -1; |
| 643 | self->state = AFTER_ESCAPED_CRNL; |
| 644 | break; |
| 645 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 646 | if (c == '\0') |
| 647 | c = '\n'; |
| 648 | if (parse_add_char(self, c) < 0) |
| 649 | return -1; |
| 650 | self->state = IN_FIELD; |
| 651 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 652 | |
R David Murray | c7c42ef | 2013-03-19 22:41:47 -0400 | [diff] [blame] | 653 | case AFTER_ESCAPED_CRNL: |
| 654 | if (c == '\0') |
| 655 | break; |
| 656 | /*fallthru*/ |
| 657 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 658 | case IN_FIELD: |
| 659 | /* in unquoted field */ |
| 660 | if (c == '\n' || c == '\r' || c == '\0') { |
| 661 | /* end of line - return [fields] */ |
| 662 | if (parse_save_field(self) < 0) |
| 663 | return -1; |
| 664 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 665 | } |
| 666 | else if (c == dialect->escapechar) { |
| 667 | /* possible escaped character */ |
| 668 | self->state = ESCAPED_CHAR; |
| 669 | } |
| 670 | else if (c == dialect->delimiter) { |
| 671 | /* save field - wait for new field */ |
| 672 | if (parse_save_field(self) < 0) |
| 673 | return -1; |
| 674 | self->state = START_FIELD; |
| 675 | } |
| 676 | else { |
| 677 | /* normal character - save in field */ |
| 678 | if (parse_add_char(self, c) < 0) |
| 679 | return -1; |
| 680 | } |
| 681 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 682 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 683 | case IN_QUOTED_FIELD: |
| 684 | /* in quoted field */ |
| 685 | if (c == '\0') |
| 686 | ; |
| 687 | else if (c == dialect->escapechar) { |
| 688 | /* Possible escape character */ |
| 689 | self->state = ESCAPE_IN_QUOTED_FIELD; |
| 690 | } |
| 691 | else if (c == dialect->quotechar && |
| 692 | dialect->quoting != QUOTE_NONE) { |
| 693 | if (dialect->doublequote) { |
| 694 | /* doublequote; " represented by "" */ |
| 695 | self->state = QUOTE_IN_QUOTED_FIELD; |
| 696 | } |
| 697 | else { |
| 698 | /* end of quote part of field */ |
| 699 | self->state = IN_FIELD; |
| 700 | } |
| 701 | } |
| 702 | else { |
| 703 | /* normal character - save in field */ |
| 704 | if (parse_add_char(self, c) < 0) |
| 705 | return -1; |
| 706 | } |
| 707 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 708 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 709 | case ESCAPE_IN_QUOTED_FIELD: |
| 710 | if (c == '\0') |
| 711 | c = '\n'; |
| 712 | if (parse_add_char(self, c) < 0) |
| 713 | return -1; |
| 714 | self->state = IN_QUOTED_FIELD; |
| 715 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 716 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 717 | case QUOTE_IN_QUOTED_FIELD: |
Serhiy Storchaka | 6a7b3a7 | 2016-04-17 08:32:47 +0300 | [diff] [blame] | 718 | /* doublequote - seen a quote in a quoted field */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 719 | if (dialect->quoting != QUOTE_NONE && |
| 720 | c == dialect->quotechar) { |
| 721 | /* save "" as " */ |
| 722 | if (parse_add_char(self, c) < 0) |
| 723 | return -1; |
| 724 | self->state = IN_QUOTED_FIELD; |
| 725 | } |
| 726 | else if (c == dialect->delimiter) { |
| 727 | /* save field - wait for new field */ |
| 728 | if (parse_save_field(self) < 0) |
| 729 | return -1; |
| 730 | self->state = START_FIELD; |
| 731 | } |
| 732 | else if (c == '\n' || c == '\r' || c == '\0') { |
| 733 | /* end of line - return [fields] */ |
| 734 | if (parse_save_field(self) < 0) |
| 735 | return -1; |
| 736 | self->state = (c == '\0' ? START_RECORD : EAT_CRNL); |
| 737 | } |
| 738 | else if (!dialect->strict) { |
| 739 | if (parse_add_char(self, c) < 0) |
| 740 | return -1; |
| 741 | self->state = IN_FIELD; |
| 742 | } |
| 743 | else { |
| 744 | /* illegal */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 745 | PyErr_Format(_csvstate_global->error_obj, "'%c' expected after '%c'", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 746 | dialect->delimiter, |
| 747 | dialect->quotechar); |
| 748 | return -1; |
| 749 | } |
| 750 | break; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 751 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 752 | case EAT_CRNL: |
| 753 | if (c == '\n' || c == '\r') |
| 754 | ; |
| 755 | else if (c == '\0') |
| 756 | self->state = START_RECORD; |
| 757 | else { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 758 | 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] | 759 | return -1; |
| 760 | } |
| 761 | break; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 762 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 763 | } |
| 764 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 765 | } |
| 766 | |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 767 | static int |
| 768 | parse_reset(ReaderObj *self) |
| 769 | { |
Serhiy Storchaka | 4884271 | 2016-04-06 09:45:48 +0300 | [diff] [blame] | 770 | Py_XSETREF(self->fields, PyList_New(0)); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 771 | if (self->fields == NULL) |
| 772 | return -1; |
| 773 | self->field_len = 0; |
| 774 | self->state = START_RECORD; |
| 775 | self->numeric_field = 0; |
| 776 | return 0; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 777 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 778 | |
| 779 | static PyObject * |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 780 | Reader_iternext(ReaderObj *self) |
| 781 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 782 | PyObject *fields = NULL; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 783 | Py_UCS4 c; |
| 784 | Py_ssize_t pos, linelen; |
| 785 | unsigned int kind; |
| 786 | void *data; |
| 787 | PyObject *lineobj; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 788 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 789 | if (parse_reset(self) < 0) |
| 790 | return NULL; |
| 791 | do { |
| 792 | lineobj = PyIter_Next(self->input_iter); |
| 793 | if (lineobj == NULL) { |
| 794 | /* End of input OR exception */ |
Senthil Kumaran | 67b7b98 | 2012-09-25 02:30:27 -0700 | [diff] [blame] | 795 | if (!PyErr_Occurred() && (self->field_len != 0 || |
| 796 | self->state == IN_QUOTED_FIELD)) { |
| 797 | if (self->dialect->strict) |
Senthil Kumaran | 49d1302 | 2012-09-25 02:37:20 -0700 | [diff] [blame] | 798 | PyErr_SetString(_csvstate_global->error_obj, |
| 799 | "unexpected end of data"); |
Senthil Kumaran | 67b7b98 | 2012-09-25 02:30:27 -0700 | [diff] [blame] | 800 | else if (parse_save_field(self) >= 0) |
| 801 | break; |
| 802 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 803 | return NULL; |
| 804 | } |
| 805 | if (!PyUnicode_Check(lineobj)) { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 806 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 807 | "iterator should return strings, " |
| 808 | "not %.200s " |
| 809 | "(did you open the file in text mode?)", |
| 810 | lineobj->ob_type->tp_name |
| 811 | ); |
| 812 | Py_DECREF(lineobj); |
| 813 | return NULL; |
| 814 | } |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 815 | if (PyUnicode_READY(lineobj) == -1) { |
| 816 | Py_DECREF(lineobj); |
| 817 | return NULL; |
| 818 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 819 | ++self->line_num; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 820 | kind = PyUnicode_KIND(lineobj); |
| 821 | data = PyUnicode_DATA(lineobj); |
| 822 | pos = 0; |
| 823 | linelen = PyUnicode_GET_LENGTH(lineobj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 824 | while (linelen--) { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 825 | c = PyUnicode_READ(kind, data, pos); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 826 | if (c == '\0') { |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 827 | Py_DECREF(lineobj); |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 828 | PyErr_Format(_csvstate_global->error_obj, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 829 | "line contains NULL byte"); |
| 830 | goto err; |
| 831 | } |
| 832 | if (parse_process_char(self, c) < 0) { |
| 833 | Py_DECREF(lineobj); |
| 834 | goto err; |
| 835 | } |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 836 | pos++; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 837 | } |
| 838 | Py_DECREF(lineobj); |
| 839 | if (parse_process_char(self, 0) < 0) |
| 840 | goto err; |
| 841 | } while (self->state != START_RECORD); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 842 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 843 | fields = self->fields; |
| 844 | self->fields = NULL; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 845 | err: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 846 | return fields; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 847 | } |
| 848 | |
| 849 | static void |
| 850 | Reader_dealloc(ReaderObj *self) |
| 851 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 852 | PyObject_GC_UnTrack(self); |
| 853 | Py_XDECREF(self->dialect); |
| 854 | Py_XDECREF(self->input_iter); |
| 855 | Py_XDECREF(self->fields); |
| 856 | if (self->field != NULL) |
| 857 | PyMem_Free(self->field); |
| 858 | PyObject_GC_Del(self); |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 859 | } |
| 860 | |
| 861 | static int |
| 862 | Reader_traverse(ReaderObj *self, visitproc visit, void *arg) |
| 863 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 864 | Py_VISIT(self->dialect); |
| 865 | Py_VISIT(self->input_iter); |
| 866 | Py_VISIT(self->fields); |
| 867 | return 0; |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 868 | } |
| 869 | |
| 870 | static int |
| 871 | Reader_clear(ReaderObj *self) |
| 872 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 873 | Py_CLEAR(self->dialect); |
| 874 | Py_CLEAR(self->input_iter); |
| 875 | Py_CLEAR(self->fields); |
| 876 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 877 | } |
| 878 | |
| 879 | PyDoc_STRVAR(Reader_Type_doc, |
| 880 | "CSV reader\n" |
| 881 | "\n" |
| 882 | "Reader objects are responsible for reading and parsing tabular data\n" |
| 883 | "in CSV format.\n" |
| 884 | ); |
| 885 | |
| 886 | static struct PyMethodDef Reader_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 887 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 888 | }; |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 889 | #define R_OFF(x) offsetof(ReaderObj, x) |
| 890 | |
| 891 | static struct PyMemberDef Reader_memberlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 892 | { "dialect", T_OBJECT, R_OFF(dialect), READONLY }, |
| 893 | { "line_num", T_ULONG, R_OFF(line_num), READONLY }, |
| 894 | { NULL } |
Andrew McNamara | f69d94f | 2005-01-13 11:30:54 +0000 | [diff] [blame] | 895 | }; |
| 896 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 897 | |
| 898 | static PyTypeObject Reader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 899 | PyVarObject_HEAD_INIT(NULL, 0) |
| 900 | "_csv.reader", /*tp_name*/ |
| 901 | sizeof(ReaderObj), /*tp_basicsize*/ |
| 902 | 0, /*tp_itemsize*/ |
| 903 | /* methods */ |
| 904 | (destructor)Reader_dealloc, /*tp_dealloc*/ |
| 905 | (printfunc)0, /*tp_print*/ |
| 906 | (getattrfunc)0, /*tp_getattr*/ |
| 907 | (setattrfunc)0, /*tp_setattr*/ |
| 908 | 0, /*tp_reserved*/ |
| 909 | (reprfunc)0, /*tp_repr*/ |
| 910 | 0, /*tp_as_number*/ |
| 911 | 0, /*tp_as_sequence*/ |
| 912 | 0, /*tp_as_mapping*/ |
| 913 | (hashfunc)0, /*tp_hash*/ |
| 914 | (ternaryfunc)0, /*tp_call*/ |
| 915 | (reprfunc)0, /*tp_str*/ |
| 916 | 0, /*tp_getattro*/ |
| 917 | 0, /*tp_setattro*/ |
| 918 | 0, /*tp_as_buffer*/ |
| 919 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | |
| 920 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
| 921 | Reader_Type_doc, /*tp_doc*/ |
| 922 | (traverseproc)Reader_traverse, /*tp_traverse*/ |
| 923 | (inquiry)Reader_clear, /*tp_clear*/ |
| 924 | 0, /*tp_richcompare*/ |
| 925 | 0, /*tp_weaklistoffset*/ |
| 926 | PyObject_SelfIter, /*tp_iter*/ |
| 927 | (getiterfunc)Reader_iternext, /*tp_iternext*/ |
| 928 | Reader_methods, /*tp_methods*/ |
| 929 | Reader_memberlist, /*tp_members*/ |
| 930 | 0, /*tp_getset*/ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 931 | |
| 932 | }; |
| 933 | |
| 934 | static PyObject * |
| 935 | csv_reader(PyObject *module, PyObject *args, PyObject *keyword_args) |
| 936 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 937 | PyObject * iterator, * dialect = NULL; |
| 938 | ReaderObj * self = PyObject_GC_New(ReaderObj, &Reader_Type); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 939 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 940 | if (!self) |
| 941 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 942 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 943 | self->dialect = NULL; |
| 944 | self->fields = NULL; |
| 945 | self->input_iter = NULL; |
| 946 | self->field = NULL; |
| 947 | self->field_size = 0; |
| 948 | self->line_num = 0; |
Andrew McNamara | 0f0599d | 2005-01-12 09:45:18 +0000 | [diff] [blame] | 949 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 950 | if (parse_reset(self) < 0) { |
| 951 | Py_DECREF(self); |
| 952 | return NULL; |
| 953 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 954 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 955 | if (!PyArg_UnpackTuple(args, "", 1, 2, &iterator, &dialect)) { |
| 956 | Py_DECREF(self); |
| 957 | return NULL; |
| 958 | } |
| 959 | self->input_iter = PyObject_GetIter(iterator); |
| 960 | if (self->input_iter == NULL) { |
| 961 | PyErr_SetString(PyExc_TypeError, |
| 962 | "argument 1 must be an iterator"); |
| 963 | Py_DECREF(self); |
| 964 | return NULL; |
| 965 | } |
| 966 | self->dialect = (DialectObj *)_call_dialect(dialect, keyword_args); |
| 967 | if (self->dialect == NULL) { |
| 968 | Py_DECREF(self); |
| 969 | return NULL; |
| 970 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 971 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 972 | PyObject_GC_Track(self); |
| 973 | return (PyObject *)self; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 974 | } |
| 975 | |
| 976 | /* |
| 977 | * WRITER |
| 978 | */ |
| 979 | /* ---------------------------------------------------------------- */ |
| 980 | static void |
| 981 | join_reset(WriterObj *self) |
| 982 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 983 | self->rec_len = 0; |
| 984 | self->num_fields = 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 985 | } |
| 986 | |
| 987 | #define MEM_INCR 32768 |
| 988 | |
| 989 | /* Calculate new record length or append field to record. Return new |
| 990 | * record length. |
| 991 | */ |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 992 | static Py_ssize_t |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 993 | join_append_data(WriterObj *self, unsigned int field_kind, void *field_data, |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 994 | Py_ssize_t field_len, int *quoted, |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 995 | int copy_phase) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 996 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 997 | DialectObj *dialect = self->dialect; |
| 998 | int i; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 999 | Py_ssize_t rec_len; |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1000 | |
Benjamin Peterson | 6e01d90 | 2016-08-13 17:17:06 -0700 | [diff] [blame] | 1001 | #define INCLEN \ |
| 1002 | do {\ |
| 1003 | if (!copy_phase && rec_len == PY_SSIZE_T_MAX) { \ |
| 1004 | goto overflow; \ |
| 1005 | } \ |
| 1006 | rec_len++; \ |
| 1007 | } while(0) |
| 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;\ |
Benjamin Peterson | 6e01d90 | 2016-08-13 17:17:06 -0700 | [diff] [blame] | 1013 | INCLEN;\ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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 (*quoted) { |
| 1065 | if (copy_phase) |
| 1066 | ADDCH(dialect->quotechar); |
Benjamin Peterson | 6e01d90 | 2016-08-13 17:17:06 -0700 | [diff] [blame] | 1067 | else { |
| 1068 | INCLEN; /* starting quote */ |
| 1069 | INCLEN; /* ending quote */ |
| 1070 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1071 | } |
| 1072 | return rec_len; |
Benjamin Peterson | 6e01d90 | 2016-08-13 17:17:06 -0700 | [diff] [blame] | 1073 | |
| 1074 | overflow: |
| 1075 | PyErr_NoMemory(); |
| 1076 | return -1; |
Andrew McNamara | c89f284 | 2005-01-12 07:44:42 +0000 | [diff] [blame] | 1077 | #undef ADDCH |
Benjamin Peterson | 6e01d90 | 2016-08-13 17:17:06 -0700 | [diff] [blame] | 1078 | #undef INCLEN |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1079 | } |
| 1080 | |
| 1081 | static int |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1082 | join_check_rec_size(WriterObj *self, Py_ssize_t rec_len) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1083 | { |
Sergey Fedoseev | 67b9cc8 | 2018-08-16 09:27:50 +0500 | [diff] [blame] | 1084 | assert(rec_len >= 0); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 1085 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1086 | if (rec_len > self->rec_size) { |
Sergey Fedoseev | 67b9cc8 | 2018-08-16 09:27:50 +0500 | [diff] [blame] | 1087 | size_t rec_size_new = (size_t)(rec_len / MEM_INCR + 1) * MEM_INCR; |
| 1088 | Py_UCS4 *rec_new = self->rec; |
| 1089 | PyMem_Resize(rec_new, Py_UCS4, rec_size_new); |
| 1090 | if (rec_new == NULL) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1091 | PyErr_NoMemory(); |
| 1092 | return 0; |
| 1093 | } |
Sergey Fedoseev | 67b9cc8 | 2018-08-16 09:27:50 +0500 | [diff] [blame] | 1094 | self->rec = rec_new; |
| 1095 | self->rec_size = (Py_ssize_t)rec_size_new; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1096 | } |
| 1097 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1098 | } |
| 1099 | |
| 1100 | static int |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1101 | join_append(WriterObj *self, PyObject *field, int quoted) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1102 | { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1103 | unsigned int field_kind = -1; |
| 1104 | void *field_data = NULL; |
| 1105 | Py_ssize_t field_len = 0; |
Antoine Pitrou | 4045575 | 2010-08-15 18:51:10 +0000 | [diff] [blame] | 1106 | Py_ssize_t rec_len; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1107 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1108 | if (field != NULL) { |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1109 | if (PyUnicode_READY(field) == -1) |
| 1110 | return 0; |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1111 | field_kind = PyUnicode_KIND(field); |
| 1112 | field_data = PyUnicode_DATA(field); |
| 1113 | field_len = PyUnicode_GET_LENGTH(field); |
| 1114 | } |
| 1115 | rec_len = join_append_data(self, field_kind, field_data, field_len, |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1116 | "ed, 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1117 | if (rec_len < 0) |
| 1118 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1119 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1120 | /* grow record buffer if necessary */ |
| 1121 | if (!join_check_rec_size(self, rec_len)) |
| 1122 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1123 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1124 | self->rec_len = join_append_data(self, field_kind, field_data, field_len, |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1125 | "ed, 1); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1126 | self->num_fields++; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1127 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1128 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1129 | } |
| 1130 | |
| 1131 | static int |
| 1132 | join_append_lineterminator(WriterObj *self) |
| 1133 | { |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1134 | Py_ssize_t terminator_len, i; |
| 1135 | unsigned int term_kind; |
| 1136 | void *term_data; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1137 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1138 | terminator_len = PyUnicode_GET_LENGTH(self->dialect->lineterminator); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1139 | if (terminator_len == -1) |
| 1140 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1141 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1142 | /* grow record buffer if necessary */ |
| 1143 | if (!join_check_rec_size(self, self->rec_len + terminator_len)) |
| 1144 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1145 | |
Antoine Pitrou | 77ea640 | 2011-10-07 04:26:55 +0200 | [diff] [blame] | 1146 | term_kind = PyUnicode_KIND(self->dialect->lineterminator); |
| 1147 | term_data = PyUnicode_DATA(self->dialect->lineterminator); |
| 1148 | for (i = 0; i < terminator_len; i++) |
| 1149 | 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] | 1150 | self->rec_len += terminator_len; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1151 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1152 | return 1; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1153 | } |
| 1154 | |
| 1155 | PyDoc_STRVAR(csv_writerow_doc, |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1156 | "writerow(iterable)\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1157 | "\n" |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1158 | "Construct and write a CSV record from an iterable of fields. Non-string\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1159 | "elements will be converted to string."); |
| 1160 | |
| 1161 | static PyObject * |
| 1162 | csv_writerow(WriterObj *self, PyObject *seq) |
| 1163 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1164 | DialectObj *dialect = self->dialect; |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1165 | PyObject *iter, *field, *line, *result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1166 | |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1167 | iter = PyObject_GetIter(seq); |
| 1168 | if (iter == NULL) |
| 1169 | return PyErr_Format(_csvstate_global->error_obj, |
| 1170 | "iterable expected, not %.200s", |
| 1171 | seq->ob_type->tp_name); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1172 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1173 | /* Join all fields in internal buffer. |
| 1174 | */ |
| 1175 | join_reset(self); |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1176 | while ((field = PyIter_Next(iter))) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1177 | int append_ok; |
| 1178 | int quoted; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1179 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1180 | switch (dialect->quoting) { |
| 1181 | case QUOTE_NONNUMERIC: |
| 1182 | quoted = !PyNumber_Check(field); |
| 1183 | break; |
| 1184 | case QUOTE_ALL: |
| 1185 | quoted = 1; |
| 1186 | break; |
| 1187 | default: |
| 1188 | quoted = 0; |
| 1189 | break; |
| 1190 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1191 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1192 | if (PyUnicode_Check(field)) { |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1193 | append_ok = join_append(self, field, quoted); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1194 | Py_DECREF(field); |
| 1195 | } |
| 1196 | else if (field == Py_None) { |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1197 | append_ok = join_append(self, NULL, quoted); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1198 | Py_DECREF(field); |
| 1199 | } |
| 1200 | else { |
| 1201 | PyObject *str; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1202 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1203 | str = PyObject_Str(field); |
| 1204 | Py_DECREF(field); |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1205 | if (str == NULL) { |
| 1206 | Py_DECREF(iter); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1207 | return NULL; |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1208 | } |
| 1209 | append_ok = join_append(self, str, quoted); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1210 | Py_DECREF(str); |
| 1211 | } |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1212 | if (!append_ok) { |
| 1213 | Py_DECREF(iter); |
| 1214 | return NULL; |
| 1215 | } |
| 1216 | } |
| 1217 | Py_DECREF(iter); |
| 1218 | if (PyErr_Occurred()) |
| 1219 | return NULL; |
| 1220 | |
Licht Takeuchi | 2001900 | 2017-12-12 18:57:06 +0900 | [diff] [blame] | 1221 | if (self->num_fields > 0 && self->rec_len == 0) { |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1222 | if (dialect->quoting == QUOTE_NONE) { |
| 1223 | PyErr_Format(_csvstate_global->error_obj, |
| 1224 | "single empty field record must be quoted"); |
| 1225 | return NULL; |
| 1226 | } |
| 1227 | self->num_fields--; |
| 1228 | if (!join_append(self, NULL, 1)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1229 | return NULL; |
| 1230 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1231 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1232 | /* Add line terminator. |
| 1233 | */ |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1234 | if (!join_append_lineterminator(self)) { |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1235 | return NULL; |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1236 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1237 | |
Antoine Pitrou | 72ca65d | 2011-10-07 04:35:30 +0200 | [diff] [blame] | 1238 | line = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, |
| 1239 | (void *) self->rec, self->rec_len); |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1240 | if (line == NULL) { |
Antoine Pitrou | 72ca65d | 2011-10-07 04:35:30 +0200 | [diff] [blame] | 1241 | return NULL; |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1242 | } |
| 1243 | result = PyObject_CallFunctionObjArgs(self->write, line, NULL); |
Antoine Pitrou | 72ca65d | 2011-10-07 04:35:30 +0200 | [diff] [blame] | 1244 | Py_DECREF(line); |
| 1245 | return result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1246 | } |
| 1247 | |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1248 | PyDoc_STRVAR(csv_writerows_doc, |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1249 | "writerows(iterable of iterables)\n" |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1250 | "\n" |
Serhiy Storchaka | 7901b48 | 2015-03-30 09:09:54 +0300 | [diff] [blame] | 1251 | "Construct and write a series of iterables to a csv file. Non-string\n" |
Skip Montanaro | 860fc0b | 2003-04-12 18:57:52 +0000 | [diff] [blame] | 1252 | "elements will be converted to string."); |
| 1253 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1254 | static PyObject * |
| 1255 | csv_writerows(WriterObj *self, PyObject *seqseq) |
| 1256 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1257 | PyObject *row_iter, *row_obj, *result; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1258 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1259 | row_iter = PyObject_GetIter(seqseq); |
| 1260 | if (row_iter == NULL) { |
| 1261 | PyErr_SetString(PyExc_TypeError, |
| 1262 | "writerows() argument must be iterable"); |
| 1263 | return NULL; |
| 1264 | } |
| 1265 | while ((row_obj = PyIter_Next(row_iter))) { |
| 1266 | result = csv_writerow(self, row_obj); |
| 1267 | Py_DECREF(row_obj); |
| 1268 | if (!result) { |
| 1269 | Py_DECREF(row_iter); |
| 1270 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1271 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1272 | else |
| 1273 | Py_DECREF(result); |
| 1274 | } |
| 1275 | Py_DECREF(row_iter); |
| 1276 | if (PyErr_Occurred()) |
| 1277 | return NULL; |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 1278 | Py_RETURN_NONE; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1279 | } |
| 1280 | |
| 1281 | static struct PyMethodDef Writer_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1282 | { "writerow", (PyCFunction)csv_writerow, METH_O, csv_writerow_doc}, |
| 1283 | { "writerows", (PyCFunction)csv_writerows, METH_O, csv_writerows_doc}, |
| 1284 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1285 | }; |
| 1286 | |
| 1287 | #define W_OFF(x) offsetof(WriterObj, x) |
| 1288 | |
| 1289 | static struct PyMemberDef Writer_memberlist[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1290 | { "dialect", T_OBJECT, W_OFF(dialect), READONLY }, |
| 1291 | { NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1292 | }; |
| 1293 | |
| 1294 | static void |
| 1295 | Writer_dealloc(WriterObj *self) |
| 1296 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1297 | PyObject_GC_UnTrack(self); |
| 1298 | Py_XDECREF(self->dialect); |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1299 | Py_XDECREF(self->write); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1300 | if (self->rec != NULL) |
| 1301 | PyMem_Free(self->rec); |
| 1302 | PyObject_GC_Del(self); |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 1303 | } |
| 1304 | |
| 1305 | static int |
| 1306 | Writer_traverse(WriterObj *self, visitproc visit, void *arg) |
| 1307 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1308 | Py_VISIT(self->dialect); |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1309 | Py_VISIT(self->write); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1310 | return 0; |
Jeremy Hylton | 42a8aed | 2003-04-14 02:20:55 +0000 | [diff] [blame] | 1311 | } |
| 1312 | |
| 1313 | static int |
| 1314 | Writer_clear(WriterObj *self) |
| 1315 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1316 | Py_CLEAR(self->dialect); |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1317 | Py_CLEAR(self->write); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1318 | return 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1319 | } |
| 1320 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1321 | PyDoc_STRVAR(Writer_Type_doc, |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1322 | "CSV writer\n" |
| 1323 | "\n" |
| 1324 | "Writer objects are responsible for generating tabular data\n" |
| 1325 | "in CSV format from sequence input.\n" |
| 1326 | ); |
| 1327 | |
| 1328 | static PyTypeObject Writer_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1329 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1330 | "_csv.writer", /*tp_name*/ |
| 1331 | sizeof(WriterObj), /*tp_basicsize*/ |
| 1332 | 0, /*tp_itemsize*/ |
| 1333 | /* methods */ |
| 1334 | (destructor)Writer_dealloc, /*tp_dealloc*/ |
| 1335 | (printfunc)0, /*tp_print*/ |
| 1336 | (getattrfunc)0, /*tp_getattr*/ |
| 1337 | (setattrfunc)0, /*tp_setattr*/ |
| 1338 | 0, /*tp_reserved*/ |
| 1339 | (reprfunc)0, /*tp_repr*/ |
| 1340 | 0, /*tp_as_number*/ |
| 1341 | 0, /*tp_as_sequence*/ |
| 1342 | 0, /*tp_as_mapping*/ |
| 1343 | (hashfunc)0, /*tp_hash*/ |
| 1344 | (ternaryfunc)0, /*tp_call*/ |
| 1345 | (reprfunc)0, /*tp_str*/ |
| 1346 | 0, /*tp_getattro*/ |
| 1347 | 0, /*tp_setattro*/ |
| 1348 | 0, /*tp_as_buffer*/ |
| 1349 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | |
| 1350 | Py_TPFLAGS_HAVE_GC, /*tp_flags*/ |
| 1351 | Writer_Type_doc, |
| 1352 | (traverseproc)Writer_traverse, /*tp_traverse*/ |
| 1353 | (inquiry)Writer_clear, /*tp_clear*/ |
| 1354 | 0, /*tp_richcompare*/ |
| 1355 | 0, /*tp_weaklistoffset*/ |
| 1356 | (getiterfunc)0, /*tp_iter*/ |
| 1357 | (getiterfunc)0, /*tp_iternext*/ |
| 1358 | Writer_methods, /*tp_methods*/ |
| 1359 | Writer_memberlist, /*tp_members*/ |
| 1360 | 0, /*tp_getset*/ |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1361 | }; |
| 1362 | |
| 1363 | static PyObject * |
| 1364 | csv_writer(PyObject *module, PyObject *args, PyObject *keyword_args) |
| 1365 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1366 | PyObject * output_file, * dialect = NULL; |
| 1367 | WriterObj * self = PyObject_GC_New(WriterObj, &Writer_Type); |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1368 | _Py_IDENTIFIER(write); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1369 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1370 | if (!self) |
| 1371 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1372 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1373 | self->dialect = NULL; |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1374 | self->write = NULL; |
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 | self->rec = NULL; |
| 1377 | self->rec_size = 0; |
| 1378 | self->rec_len = 0; |
| 1379 | self->num_fields = 0; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1380 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1381 | if (!PyArg_UnpackTuple(args, "", 1, 2, &output_file, &dialect)) { |
| 1382 | Py_DECREF(self); |
| 1383 | return NULL; |
| 1384 | } |
Rémi Lapeyre | 2bc158f | 2019-05-14 15:45:14 +0200 | [diff] [blame] | 1385 | self->write = _PyObject_GetAttrId(output_file, &PyId_write); |
| 1386 | if (self->write == NULL || !PyCallable_Check(self->write)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1387 | PyErr_SetString(PyExc_TypeError, |
| 1388 | "argument 1 must have a \"write\" method"); |
| 1389 | Py_DECREF(self); |
| 1390 | return NULL; |
| 1391 | } |
| 1392 | self->dialect = (DialectObj *)_call_dialect(dialect, keyword_args); |
| 1393 | if (self->dialect == NULL) { |
| 1394 | Py_DECREF(self); |
| 1395 | return NULL; |
| 1396 | } |
| 1397 | PyObject_GC_Track(self); |
| 1398 | return (PyObject *)self; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1399 | } |
| 1400 | |
| 1401 | /* |
| 1402 | * DIALECT REGISTRY |
| 1403 | */ |
| 1404 | static PyObject * |
| 1405 | csv_list_dialects(PyObject *module, PyObject *args) |
| 1406 | { |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1407 | return PyDict_Keys(_csvstate_global->dialects); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1408 | } |
| 1409 | |
| 1410 | static PyObject * |
Andrew McNamara | 8662597 | 2005-01-11 01:28:33 +0000 | [diff] [blame] | 1411 | csv_register_dialect(PyObject *module, PyObject *args, PyObject *kwargs) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1412 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1413 | PyObject *name_obj, *dialect_obj = NULL; |
| 1414 | PyObject *dialect; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1415 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1416 | if (!PyArg_UnpackTuple(args, "", 1, 2, &name_obj, &dialect_obj)) |
| 1417 | return NULL; |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1418 | if (!PyUnicode_Check(name_obj)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1419 | PyErr_SetString(PyExc_TypeError, |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1420 | "dialect name must be a string"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1421 | return NULL; |
| 1422 | } |
Stefan Krah | e6996ed | 2012-11-02 14:44:20 +0100 | [diff] [blame] | 1423 | if (PyUnicode_READY(name_obj) == -1) |
| 1424 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1425 | dialect = _call_dialect(dialect_obj, kwargs); |
| 1426 | if (dialect == NULL) |
| 1427 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1428 | if (PyDict_SetItem(_csvstate_global->dialects, name_obj, dialect) < 0) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1429 | Py_DECREF(dialect); |
| 1430 | return NULL; |
| 1431 | } |
| 1432 | Py_DECREF(dialect); |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 1433 | Py_RETURN_NONE; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1434 | } |
| 1435 | |
| 1436 | static PyObject * |
Skip Montanaro | 577c7a7 | 2003-04-12 19:17:14 +0000 | [diff] [blame] | 1437 | csv_unregister_dialect(PyObject *module, PyObject *name_obj) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1438 | { |
Serhiy Storchaka | a24107b | 2019-02-25 17:59:46 +0200 | [diff] [blame] | 1439 | if (PyDict_DelItem(_csvstate_global->dialects, name_obj) < 0) { |
| 1440 | if (PyErr_ExceptionMatches(PyExc_KeyError)) { |
| 1441 | PyErr_Format(_csvstate_global->error_obj, "unknown dialect"); |
| 1442 | } |
| 1443 | return NULL; |
| 1444 | } |
Serhiy Storchaka | 228b12e | 2017-01-23 09:47:21 +0200 | [diff] [blame] | 1445 | Py_RETURN_NONE; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1446 | } |
| 1447 | |
| 1448 | static PyObject * |
Skip Montanaro | 577c7a7 | 2003-04-12 19:17:14 +0000 | [diff] [blame] | 1449 | csv_get_dialect(PyObject *module, PyObject *name_obj) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1450 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1451 | return get_dialect_from_registry(name_obj); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1452 | } |
| 1453 | |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1454 | static PyObject * |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1455 | csv_field_size_limit(PyObject *module, PyObject *args) |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1456 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1457 | PyObject *new_limit = NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1458 | long old_limit = _csvstate_global->field_limit; |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1459 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1460 | if (!PyArg_UnpackTuple(args, "field_size_limit", 0, 1, &new_limit)) |
| 1461 | return NULL; |
| 1462 | if (new_limit != NULL) { |
| 1463 | if (!PyLong_CheckExact(new_limit)) { |
| 1464 | PyErr_Format(PyExc_TypeError, |
| 1465 | "limit must be an integer"); |
| 1466 | return NULL; |
| 1467 | } |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1468 | _csvstate_global->field_limit = PyLong_AsLong(new_limit); |
| 1469 | if (_csvstate_global->field_limit == -1 && PyErr_Occurred()) { |
| 1470 | _csvstate_global->field_limit = old_limit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1471 | return NULL; |
| 1472 | } |
| 1473 | } |
| 1474 | return PyLong_FromLong(old_limit); |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1475 | } |
| 1476 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1477 | /* |
| 1478 | * MODULE |
| 1479 | */ |
| 1480 | |
| 1481 | PyDoc_STRVAR(csv_module_doc, |
| 1482 | "CSV parsing and writing.\n" |
| 1483 | "\n" |
| 1484 | "This module provides classes that assist in the reading and writing\n" |
| 1485 | "of Comma Separated Value (CSV) files, and implements the interface\n" |
| 1486 | "described by PEP 305. Although many CSV files are simple to parse,\n" |
| 1487 | "the format is not formally defined by a stable specification and\n" |
| 1488 | "is subtle enough that parsing lines of a CSV file with something\n" |
| 1489 | "like line.split(\",\") is bound to fail. The module supports three\n" |
| 1490 | "basic APIs: reading, writing, and registration of dialects.\n" |
| 1491 | "\n" |
| 1492 | "\n" |
| 1493 | "DIALECT REGISTRATION:\n" |
| 1494 | "\n" |
| 1495 | "Readers and writers support a dialect argument, which is a convenient\n" |
| 1496 | "handle on a group of settings. When the dialect argument is a string,\n" |
| 1497 | "it identifies one of the dialects previously registered with the module.\n" |
| 1498 | "If it is a class or instance, the attributes of the argument are used as\n" |
| 1499 | "the settings for the reader or writer:\n" |
| 1500 | "\n" |
| 1501 | " class excel:\n" |
| 1502 | " delimiter = ','\n" |
| 1503 | " quotechar = '\"'\n" |
| 1504 | " escapechar = None\n" |
| 1505 | " doublequote = True\n" |
| 1506 | " skipinitialspace = False\n" |
Johannes Gijsbers | 8d3b9dd | 2004-08-15 12:23:10 +0000 | [diff] [blame] | 1507 | " lineterminator = '\\r\\n'\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1508 | " quoting = QUOTE_MINIMAL\n" |
| 1509 | "\n" |
| 1510 | "SETTINGS:\n" |
| 1511 | "\n" |
oldk | aa0735f | 2018-02-02 16:52:55 +0800 | [diff] [blame] | 1512 | " * quotechar - specifies a one-character string to use as the\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1513 | " quoting character. It defaults to '\"'.\n" |
oldk | aa0735f | 2018-02-02 16:52:55 +0800 | [diff] [blame] | 1514 | " * delimiter - specifies a one-character string to use as the\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1515 | " field separator. It defaults to ','.\n" |
| 1516 | " * skipinitialspace - specifies how to interpret whitespace which\n" |
| 1517 | " immediately follows a delimiter. It defaults to False, which\n" |
| 1518 | " means that whitespace immediately following a delimiter is part\n" |
| 1519 | " of the following field.\n" |
oldk | aa0735f | 2018-02-02 16:52:55 +0800 | [diff] [blame] | 1520 | " * lineterminator - specifies the character sequence which should\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1521 | " terminate rows.\n" |
| 1522 | " * quoting - controls when quotes should be generated by the writer.\n" |
| 1523 | " It can take on any of the following module constants:\n" |
| 1524 | "\n" |
| 1525 | " csv.QUOTE_MINIMAL means only when required, for example, when a\n" |
| 1526 | " field contains either the quotechar or the delimiter\n" |
| 1527 | " csv.QUOTE_ALL means that quotes are always placed around fields.\n" |
| 1528 | " csv.QUOTE_NONNUMERIC means that quotes are always placed around\n" |
Skip Montanaro | 148eb6a | 2003-12-02 18:57:47 +0000 | [diff] [blame] | 1529 | " fields which do not parse as integers or floating point\n" |
| 1530 | " numbers.\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1531 | " csv.QUOTE_NONE means that quotes are never placed around fields.\n" |
oldk | aa0735f | 2018-02-02 16:52:55 +0800 | [diff] [blame] | 1532 | " * escapechar - specifies a one-character string used to escape\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1533 | " the delimiter when quoting is set to QUOTE_NONE.\n" |
| 1534 | " * doublequote - controls the handling of quotes inside fields. When\n" |
| 1535 | " True, two consecutive quotes are interpreted as one during read,\n" |
| 1536 | " and when writing, each quote character embedded in the data is\n" |
| 1537 | " written as two quotes\n"); |
| 1538 | |
| 1539 | PyDoc_STRVAR(csv_reader_doc, |
| 1540 | " csv_reader = reader(iterable [, dialect='excel']\n" |
| 1541 | " [optional keyword args])\n" |
| 1542 | " for row in csv_reader:\n" |
| 1543 | " process(row)\n" |
| 1544 | "\n" |
| 1545 | "The \"iterable\" argument can be any object that returns a line\n" |
| 1546 | "of input for each iteration, such as a file object or a list. The\n" |
| 1547 | "optional \"dialect\" parameter is discussed below. The function\n" |
| 1548 | "also accepts optional keyword arguments which override settings\n" |
| 1549 | "provided by the dialect.\n" |
| 1550 | "\n" |
| 1551 | "The returned object is an iterator. Each iteration returns a row\n" |
Berker Peksag | e2382c5 | 2015-10-02 19:25:32 +0300 | [diff] [blame] | 1552 | "of the CSV file (which can span multiple input lines).\n"); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1553 | |
| 1554 | PyDoc_STRVAR(csv_writer_doc, |
| 1555 | " csv_writer = csv.writer(fileobj [, dialect='excel']\n" |
| 1556 | " [optional keyword args])\n" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1557 | " for row in sequence:\n" |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1558 | " csv_writer.writerow(row)\n" |
| 1559 | "\n" |
| 1560 | " [or]\n" |
| 1561 | "\n" |
| 1562 | " csv_writer = csv.writer(fileobj [, dialect='excel']\n" |
| 1563 | " [optional keyword args])\n" |
| 1564 | " csv_writer.writerows(rows)\n" |
| 1565 | "\n" |
| 1566 | "The \"fileobj\" argument can be any object that supports the file API.\n"); |
| 1567 | |
| 1568 | PyDoc_STRVAR(csv_list_dialects_doc, |
| 1569 | "Return a list of all know dialect names.\n" |
| 1570 | " names = csv.list_dialects()"); |
| 1571 | |
| 1572 | PyDoc_STRVAR(csv_get_dialect_doc, |
| 1573 | "Return the dialect instance associated with name.\n" |
| 1574 | " dialect = csv.get_dialect(name)"); |
| 1575 | |
| 1576 | PyDoc_STRVAR(csv_register_dialect_doc, |
| 1577 | "Create a mapping from a string name to a dialect class.\n" |
Berker Peksag | 12b50ce | 2015-06-05 15:17:51 +0300 | [diff] [blame] | 1578 | " dialect = csv.register_dialect(name[, dialect[, **fmtparams]])"); |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1579 | |
| 1580 | PyDoc_STRVAR(csv_unregister_dialect_doc, |
| 1581 | "Delete the name/dialect mapping associated with a string name.\n" |
| 1582 | " csv.unregister_dialect(name)"); |
| 1583 | |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1584 | PyDoc_STRVAR(csv_field_size_limit_doc, |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1585 | "Sets an upper limit on parsed fields.\n" |
Andrew McNamara | 31d8896 | 2005-01-12 03:45:10 +0000 | [diff] [blame] | 1586 | " csv.field_size_limit([limit])\n" |
Andrew McNamara | e4d05c4 | 2005-01-11 07:32:02 +0000 | [diff] [blame] | 1587 | "\n" |
| 1588 | "Returns old limit. If limit is not given, no new limit is set and\n" |
| 1589 | "the old limit is returned"); |
| 1590 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1591 | static struct PyMethodDef csv_methods[] = { |
Serhiy Storchaka | 62be742 | 2018-11-27 13:27:31 +0200 | [diff] [blame] | 1592 | { "reader", (PyCFunction)(void(*)(void))csv_reader, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1593 | METH_VARARGS | METH_KEYWORDS, csv_reader_doc}, |
Serhiy Storchaka | 62be742 | 2018-11-27 13:27:31 +0200 | [diff] [blame] | 1594 | { "writer", (PyCFunction)(void(*)(void))csv_writer, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1595 | METH_VARARGS | METH_KEYWORDS, csv_writer_doc}, |
| 1596 | { "list_dialects", (PyCFunction)csv_list_dialects, |
| 1597 | METH_NOARGS, csv_list_dialects_doc}, |
Serhiy Storchaka | 62be742 | 2018-11-27 13:27:31 +0200 | [diff] [blame] | 1598 | { "register_dialect", (PyCFunction)(void(*)(void))csv_register_dialect, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1599 | METH_VARARGS | METH_KEYWORDS, csv_register_dialect_doc}, |
| 1600 | { "unregister_dialect", (PyCFunction)csv_unregister_dialect, |
| 1601 | METH_O, csv_unregister_dialect_doc}, |
| 1602 | { "get_dialect", (PyCFunction)csv_get_dialect, |
| 1603 | METH_O, csv_get_dialect_doc}, |
| 1604 | { "field_size_limit", (PyCFunction)csv_field_size_limit, |
| 1605 | METH_VARARGS, csv_field_size_limit_doc}, |
| 1606 | { NULL, NULL } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1607 | }; |
| 1608 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1609 | static struct PyModuleDef _csvmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1610 | PyModuleDef_HEAD_INIT, |
| 1611 | "_csv", |
| 1612 | csv_module_doc, |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1613 | sizeof(_csvstate), |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1614 | csv_methods, |
| 1615 | NULL, |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1616 | _csv_traverse, |
| 1617 | _csv_clear, |
| 1618 | _csv_free |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1619 | }; |
| 1620 | |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1621 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1622 | PyInit__csv(void) |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1623 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1624 | PyObject *module; |
Serhiy Storchaka | 2d06e84 | 2015-12-25 19:53:18 +0200 | [diff] [blame] | 1625 | const StyleDesc *style; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1626 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1627 | if (PyType_Ready(&Dialect_Type) < 0) |
| 1628 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1629 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1630 | if (PyType_Ready(&Reader_Type) < 0) |
| 1631 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1632 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1633 | if (PyType_Ready(&Writer_Type) < 0) |
| 1634 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1635 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1636 | /* Create the module and add the functions */ |
| 1637 | module = PyModule_Create(&_csvmodule); |
| 1638 | if (module == NULL) |
| 1639 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1640 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1641 | /* Add version to the module. */ |
| 1642 | if (PyModule_AddStringConstant(module, "__version__", |
| 1643 | MODULE_VERSION) == -1) |
| 1644 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1645 | |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1646 | /* Set the field limit */ |
| 1647 | _csvstate(module)->field_limit = 128 * 1024; |
| 1648 | /* Do I still need to add this var to the Module Dict? */ |
| 1649 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1650 | /* Add _dialects dictionary */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1651 | _csvstate(module)->dialects = PyDict_New(); |
| 1652 | if (_csvstate(module)->dialects == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1653 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1654 | Py_INCREF(_csvstate(module)->dialects); |
| 1655 | if (PyModule_AddObject(module, "_dialects", _csvstate(module)->dialects)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1656 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1657 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1658 | /* Add quote styles into dictionary */ |
| 1659 | for (style = quote_styles; style->name; style++) { |
| 1660 | if (PyModule_AddIntConstant(module, style->name, |
| 1661 | style->style) == -1) |
| 1662 | return NULL; |
| 1663 | } |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1664 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1665 | /* Add the Dialect type */ |
| 1666 | Py_INCREF(&Dialect_Type); |
| 1667 | if (PyModule_AddObject(module, "Dialect", (PyObject *)&Dialect_Type)) |
| 1668 | return NULL; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1669 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1670 | /* Add the CSV exception object to the module. */ |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1671 | _csvstate(module)->error_obj = PyErr_NewException("_csv.Error", NULL, NULL); |
| 1672 | if (_csvstate(module)->error_obj == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1673 | return NULL; |
Antoine Pitrou | e7672d3 | 2012-05-16 11:33:08 +0200 | [diff] [blame] | 1674 | Py_INCREF(_csvstate(module)->error_obj); |
| 1675 | PyModule_AddObject(module, "Error", _csvstate(module)->error_obj); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1676 | return module; |
Skip Montanaro | b4a0417 | 2003-03-20 23:29:12 +0000 | [diff] [blame] | 1677 | } |