Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 1 | |
Georg Brandl | 0c77a82 | 2008-06-10 16:37:50 +0000 | [diff] [blame] | 2 | /* Bytes (String) object interface */ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 3 | |
Georg Brandl | 0c77a82 | 2008-06-10 16:37:50 +0000 | [diff] [blame] | 4 | #ifndef Py_BYTESOBJECT_H |
| 5 | #define Py_BYTESOBJECT_H |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 6 | #ifdef __cplusplus |
| 7 | extern "C" { |
| 8 | #endif |
| 9 | |
| 10 | #include <stdarg.h> |
| 11 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 12 | /* |
| 13 | Type PyBytesObject represents a character string. An extra zero byte is |
| 14 | reserved at the end to ensure it is zero-terminated, but a size is |
| 15 | present so strings with null bytes in them can be represented. This |
| 16 | is an immutable object type. |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 17 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 18 | There are functions to create new string objects, to test |
| 19 | an object for string-ness, and to get the |
| 20 | string value. The latter function returns a null pointer |
| 21 | if the object is not of the proper type. |
| 22 | There is a variant that takes an explicit size as well as a |
| 23 | variant that assumes a zero-terminated string. Note that none of the |
| 24 | functions should be applied to nil objects. |
| 25 | */ |
| 26 | |
| 27 | /* Caching the hash (ob_shash) saves recalculation of a string's hash value. |
| 28 | This significantly speeds up dict lookups. */ |
| 29 | |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 30 | #ifndef Py_LIMITED_API |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 31 | typedef struct { |
| 32 | PyObject_VAR_HEAD |
Benjamin Peterson | 8f67d08 | 2010-10-17 20:54:53 +0000 | [diff] [blame] | 33 | Py_hash_t ob_shash; |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 34 | char ob_sval[1]; |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 35 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 36 | /* Invariants: |
| 37 | * ob_sval contains space for 'ob_size+1' elements. |
| 38 | * ob_sval[ob_size] == 0. |
| 39 | * ob_shash is the hash of the string or -1 if not computed yet. |
| 40 | */ |
| 41 | } PyBytesObject; |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 42 | #endif |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 43 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 44 | PyAPI_DATA(PyTypeObject) PyBytes_Type; |
| 45 | PyAPI_DATA(PyTypeObject) PyBytesIter_Type; |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 46 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 47 | #define PyBytes_Check(op) \ |
| 48 | PyType_FastSubclass(Py_TYPE(op), Py_TPFLAGS_BYTES_SUBCLASS) |
| 49 | #define PyBytes_CheckExact(op) (Py_TYPE(op) == &PyBytes_Type) |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 50 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 51 | PyAPI_FUNC(PyObject *) PyBytes_FromStringAndSize(const char *, Py_ssize_t); |
| 52 | PyAPI_FUNC(PyObject *) PyBytes_FromString(const char *); |
Benjamin Peterson | c15a073 | 2008-08-26 16:46:47 +0000 | [diff] [blame] | 53 | PyAPI_FUNC(PyObject *) PyBytes_FromObject(PyObject *); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 54 | PyAPI_FUNC(PyObject *) PyBytes_FromFormatV(const char*, va_list) |
Serhiy Storchaka | 598ceae | 2017-11-28 17:56:10 +0200 | [diff] [blame] | 55 | Py_GCC_ATTRIBUTE((format(printf, 1, 0))); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 56 | PyAPI_FUNC(PyObject *) PyBytes_FromFormat(const char*, ...) |
Serhiy Storchaka | 598ceae | 2017-11-28 17:56:10 +0200 | [diff] [blame] | 57 | Py_GCC_ATTRIBUTE((format(printf, 1, 2))); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 58 | PyAPI_FUNC(Py_ssize_t) PyBytes_Size(PyObject *); |
| 59 | PyAPI_FUNC(char *) PyBytes_AsString(PyObject *); |
| 60 | PyAPI_FUNC(PyObject *) PyBytes_Repr(PyObject *, int); |
| 61 | PyAPI_FUNC(void) PyBytes_Concat(PyObject **, PyObject *); |
| 62 | PyAPI_FUNC(void) PyBytes_ConcatAndDel(PyObject **, PyObject *); |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 63 | #ifndef Py_LIMITED_API |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 64 | PyAPI_FUNC(int) _PyBytes_Resize(PyObject **, Py_ssize_t); |
Victor Stinner | 772b2b0 | 2015-10-14 09:56:53 +0200 | [diff] [blame] | 65 | PyAPI_FUNC(PyObject*) _PyBytes_FormatEx( |
| 66 | const char *format, |
| 67 | Py_ssize_t format_len, |
| 68 | PyObject *args, |
| 69 | int use_bytearray); |
Victor Stinner | 2bf8993 | 2015-10-14 11:25:33 +0200 | [diff] [blame] | 70 | PyAPI_FUNC(PyObject*) _PyBytes_FromHex( |
| 71 | PyObject *string, |
| 72 | int use_bytearray); |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 73 | #endif |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 74 | PyAPI_FUNC(PyObject *) PyBytes_DecodeEscape(const char *, Py_ssize_t, |
Serhiy Storchaka | 598ceae | 2017-11-28 17:56:10 +0200 | [diff] [blame] | 75 | const char *, Py_ssize_t, |
| 76 | const char *); |
Serhiy Storchaka | 34d0ac8 | 2016-12-27 14:57:39 +0200 | [diff] [blame] | 77 | #ifndef Py_LIMITED_API |
Eric V. Smith | 5646648 | 2016-10-31 14:46:26 -0400 | [diff] [blame] | 78 | /* Helper for PyBytes_DecodeEscape that detects invalid escape chars. */ |
| 79 | PyAPI_FUNC(PyObject *) _PyBytes_DecodeEscape(const char *, Py_ssize_t, |
| 80 | const char *, Py_ssize_t, |
| 81 | const char *, |
| 82 | const char **); |
Serhiy Storchaka | 34d0ac8 | 2016-12-27 14:57:39 +0200 | [diff] [blame] | 83 | #endif |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 84 | |
| 85 | /* Macro, trading safety for speed */ |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 86 | #ifndef Py_LIMITED_API |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 87 | #define PyBytes_AS_STRING(op) (assert(PyBytes_Check(op)), \ |
| 88 | (((PyBytesObject *)(op))->ob_sval)) |
| 89 | #define PyBytes_GET_SIZE(op) (assert(PyBytes_Check(op)),Py_SIZE(op)) |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 90 | #endif |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 91 | |
| 92 | /* _PyBytes_Join(sep, x) is like sep.join(x). sep must be PyBytesObject*, |
| 93 | x must be an iterable object. */ |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 94 | #ifndef Py_LIMITED_API |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 95 | PyAPI_FUNC(PyObject *) _PyBytes_Join(PyObject *sep, PyObject *x); |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 96 | #endif |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 97 | |
| 98 | /* Provides access to the internal data buffer and size of a string |
Martin Panter | 6245cb3 | 2016-04-15 02:14:19 +0000 | [diff] [blame] | 99 | object or the default encoded version of a Unicode object. Passing |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 100 | NULL as *len parameter will force the string buffer to be |
| 101 | 0-terminated (passing a string with embedded NULL characters will |
| 102 | cause an exception). */ |
| 103 | PyAPI_FUNC(int) PyBytes_AsStringAndSize( |
Antoine Pitrou | 9ed5f27 | 2013-08-13 20:18:52 +0200 | [diff] [blame] | 104 | PyObject *obj, /* string or Unicode object */ |
| 105 | char **s, /* pointer to buffer variable */ |
| 106 | Py_ssize_t *len /* pointer to length variable or NULL |
| 107 | (only possible for 0-terminated |
| 108 | strings) */ |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 109 | ); |
| 110 | |
| 111 | /* Using the current locale, insert the thousands grouping |
| 112 | into the string pointed to by buffer. For the argument descriptions, |
| 113 | see Objects/stringlib/localeutil.h */ |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 114 | #ifndef Py_LIMITED_API |
Eric Smith | 0923d1d | 2009-04-16 20:16:10 +0000 | [diff] [blame] | 115 | PyAPI_FUNC(Py_ssize_t) _PyBytes_InsertThousandsGroupingLocale(char *buffer, |
| 116 | Py_ssize_t n_buffer, |
| 117 | char *digits, |
| 118 | Py_ssize_t n_digits, |
| 119 | Py_ssize_t min_width); |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 120 | |
Eric Smith | a3b1ac8 | 2009-04-03 14:45:06 +0000 | [diff] [blame] | 121 | /* Using explicit passed-in values, insert the thousands grouping |
| 122 | into the string pointed to by buffer. For the argument descriptions, |
| 123 | see Objects/stringlib/localeutil.h */ |
Eric Smith | 0923d1d | 2009-04-16 20:16:10 +0000 | [diff] [blame] | 124 | PyAPI_FUNC(Py_ssize_t) _PyBytes_InsertThousandsGrouping(char *buffer, |
| 125 | Py_ssize_t n_buffer, |
| 126 | char *digits, |
| 127 | Py_ssize_t n_digits, |
| 128 | Py_ssize_t min_width, |
| 129 | const char *grouping, |
| 130 | const char *thousands_sep); |
Martin v. Löwis | 4d0d471 | 2010-12-03 20:14:31 +0000 | [diff] [blame] | 131 | #endif |
Eric Smith | a3b1ac8 | 2009-04-03 14:45:06 +0000 | [diff] [blame] | 132 | |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 133 | /* Flags used by string formatting */ |
| 134 | #define F_LJUST (1<<0) |
Serhiy Storchaka | 598ceae | 2017-11-28 17:56:10 +0200 | [diff] [blame] | 135 | #define F_SIGN (1<<1) |
Christian Heimes | 2c9c7a5 | 2008-05-26 13:42:13 +0000 | [diff] [blame] | 136 | #define F_BLANK (1<<2) |
Serhiy Storchaka | 598ceae | 2017-11-28 17:56:10 +0200 | [diff] [blame] | 137 | #define F_ALT (1<<3) |
| 138 | #define F_ZERO (1<<4) |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 139 | |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 140 | #ifndef Py_LIMITED_API |
Raymond Hettinger | 15f44ab | 2016-08-30 10:47:49 -0700 | [diff] [blame] | 141 | /* The _PyBytesWriter structure is big: it contains an embedded "stack buffer". |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 142 | A _PyBytesWriter variable must be declared at the end of variables in a |
| 143 | function to optimize the memory allocation on the stack. */ |
| 144 | typedef struct { |
Victor Stinner | 661aacc | 2015-10-14 09:41:48 +0200 | [diff] [blame] | 145 | /* bytes, bytearray or NULL (when the small buffer is used) */ |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 146 | PyObject *buffer; |
| 147 | |
Victor Stinner | 661aacc | 2015-10-14 09:41:48 +0200 | [diff] [blame] | 148 | /* Number of allocated size. */ |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 149 | Py_ssize_t allocated; |
| 150 | |
Victor Stinner | 53926a1 | 2015-10-09 12:37:03 +0200 | [diff] [blame] | 151 | /* Minimum number of allocated bytes, |
| 152 | incremented by _PyBytesWriter_Prepare() */ |
| 153 | Py_ssize_t min_size; |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 154 | |
Victor Stinner | 661aacc | 2015-10-14 09:41:48 +0200 | [diff] [blame] | 155 | /* If non-zero, use a bytearray instead of a bytes object for buffer. */ |
| 156 | int use_bytearray; |
| 157 | |
| 158 | /* If non-zero, overallocate the buffer (default: 0). |
| 159 | This flag must be zero if use_bytearray is non-zero. */ |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 160 | int overallocate; |
| 161 | |
| 162 | /* Stack buffer */ |
Victor Stinner | b3653a3 | 2015-10-09 03:38:24 +0200 | [diff] [blame] | 163 | int use_small_buffer; |
| 164 | char small_buffer[512]; |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 165 | } _PyBytesWriter; |
| 166 | |
| 167 | /* Initialize a bytes writer |
| 168 | |
| 169 | By default, the overallocation is disabled. Set the overallocate attribute |
| 170 | to control the allocation of the buffer. */ |
| 171 | PyAPI_FUNC(void) _PyBytesWriter_Init(_PyBytesWriter *writer); |
| 172 | |
| 173 | /* Get the buffer content and reset the writer. |
Victor Stinner | 661aacc | 2015-10-14 09:41:48 +0200 | [diff] [blame] | 174 | Return a bytes object, or a bytearray object if use_bytearray is non-zero. |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 175 | Raise an exception and return NULL on error. */ |
| 176 | PyAPI_FUNC(PyObject *) _PyBytesWriter_Finish(_PyBytesWriter *writer, |
Victor Stinner | c29e29b | 2015-10-12 13:12:54 +0200 | [diff] [blame] | 177 | void *str); |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 178 | |
| 179 | /* Deallocate memory of a writer (clear its internal buffer). */ |
| 180 | PyAPI_FUNC(void) _PyBytesWriter_Dealloc(_PyBytesWriter *writer); |
| 181 | |
| 182 | /* Allocate the buffer to write size bytes. |
| 183 | Return the pointer to the beginning of buffer data. |
| 184 | Raise an exception and return NULL on error. */ |
Victor Stinner | c29e29b | 2015-10-12 13:12:54 +0200 | [diff] [blame] | 185 | PyAPI_FUNC(void*) _PyBytesWriter_Alloc(_PyBytesWriter *writer, |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 186 | Py_ssize_t size); |
| 187 | |
Victor Stinner | c5c3ba4 | 2015-10-14 13:56:47 +0200 | [diff] [blame] | 188 | /* Ensure that the buffer is large enough to write *size* bytes. |
| 189 | Add size to the writer minimum size (min_size attribute). |
| 190 | |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 191 | str is the current pointer inside the buffer. |
| 192 | Return the updated current pointer inside the buffer. |
| 193 | Raise an exception and return NULL on error. */ |
Victor Stinner | c29e29b | 2015-10-12 13:12:54 +0200 | [diff] [blame] | 194 | PyAPI_FUNC(void*) _PyBytesWriter_Prepare(_PyBytesWriter *writer, |
| 195 | void *str, |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 196 | Py_ssize_t size); |
Victor Stinner | ce179bf | 2015-10-09 12:57:22 +0200 | [diff] [blame] | 197 | |
Victor Stinner | c5c3ba4 | 2015-10-14 13:56:47 +0200 | [diff] [blame] | 198 | /* Resize the buffer to make it larger. |
| 199 | The new buffer may be larger than size bytes because of overallocation. |
| 200 | Return the updated current pointer inside the buffer. |
| 201 | Raise an exception and return NULL on error. |
| 202 | |
| 203 | Note: size must be greater than the number of allocated bytes in the writer. |
| 204 | |
| 205 | This function doesn't use the writer minimum size (min_size attribute). |
| 206 | |
| 207 | See also _PyBytesWriter_Prepare(). |
| 208 | */ |
| 209 | PyAPI_FUNC(void*) _PyBytesWriter_Resize(_PyBytesWriter *writer, |
| 210 | void *str, |
| 211 | Py_ssize_t size); |
| 212 | |
Victor Stinner | ce179bf | 2015-10-09 12:57:22 +0200 | [diff] [blame] | 213 | /* Write bytes. |
| 214 | Raise an exception and return NULL on error. */ |
Victor Stinner | c29e29b | 2015-10-12 13:12:54 +0200 | [diff] [blame] | 215 | PyAPI_FUNC(void*) _PyBytesWriter_WriteBytes(_PyBytesWriter *writer, |
| 216 | void *str, |
| 217 | const void *bytes, |
Victor Stinner | ce179bf | 2015-10-09 12:57:22 +0200 | [diff] [blame] | 218 | Py_ssize_t size); |
Victor Stinner | 0016507 | 2015-10-09 01:53:21 +0200 | [diff] [blame] | 219 | #endif /* Py_LIMITED_API */ |
| 220 | |
Guido van Rossum | 4dfe8a1 | 2006-04-22 23:28:04 +0000 | [diff] [blame] | 221 | #ifdef __cplusplus |
| 222 | } |
| 223 | #endif |
Georg Brandl | 0c77a82 | 2008-06-10 16:37:50 +0000 | [diff] [blame] | 224 | #endif /* !Py_BYTESOBJECT_H */ |