lighttpd1.4/src/buffer.h

263 lines
9.5 KiB
C
Raw Normal View History

#ifndef _BUFFER_H_
#define _BUFFER_H_
#include "first.h"
struct tm; /* declaration */
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* generic string + binary data container; contains a terminating 0 in both
* cases
*
* used == 0 indicates a special "empty" state (unset config values); ptr
* might be NULL too then. otherwise an empty string has used == 1 (and ptr[0]
* == 0);
*
* copy/append functions will ensure used >= 1 (i.e. never leave it in the
* special empty state); only buffer_copy_buffer will copy the special empty
* state.
*/
typedef struct {
char *ptr;
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* "used" includes a terminating 0 */
uint32_t used;
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* size of allocated buffer at *ptr */
uint32_t size;
} buffer;
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* create new buffer; either empty or copy given data */
2019-10-09 03:31:34 +00:00
__attribute_returns_nonnull__
buffer* buffer_init(void);
2019-10-09 03:31:34 +00:00
__attribute_returns_nonnull__
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
buffer* buffer_init_buffer(const buffer *src); /* src can be NULL */
2019-10-09 03:31:34 +00:00
__attribute_returns_nonnull__
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
buffer* buffer_init_string(const char *str); /* str can be NULL */
void buffer_free(buffer *b); /* b can be NULL */
/* truncates to used == 0; frees large buffers, might keep smaller ones for reuse */
void buffer_reset(buffer *b); /* b can be NULL */
/* reset b. if NULL != b && NULL != src, move src content to b. reset src. */
void buffer_move(buffer * restrict b, buffer * restrict src);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* make sure buffer is large enough to store a string of given size
* and a terminating zero.
* sets b to an empty string, and may drop old content.
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
* @return b->ptr
*/
2019-10-09 03:31:34 +00:00
__attribute_returns_nonnull__
char* buffer_string_prepare_copy(buffer *b, size_t size);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* allocate buffer large enough to be able to append a string of given size
* if b was empty (used == 0) it will contain an empty string (used == 1)
* afterwards
* "used" data is preserved; if not empty buffer must contain a
* zero terminated string.
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
*/
2019-10-09 03:31:34 +00:00
__attribute_returns_nonnull__
char* buffer_string_prepare_append(buffer *b, size_t size);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* use after prepare_(copy,append) when you have written data to the buffer
* to increase the buffer length by size. also sets the terminating zero.
* requires enough space is present for the terminating zero (prepare with the
* same size to be sure).
*/
void buffer_commit(buffer *b, size_t size);
/* sets string length:
* - always stores a terminating zero to terminate the "new" string
* - does not modify the string data apart from terminating zero
* - reallocates the buffer iff needed
*/
void buffer_string_set_length(buffer *b, uint32_t len);
/* clear buffer
* - invalidate buffer contents
* - unsets used chars but does not modify existing ptr contents
* (b->ptr *is not* set to an empty, '\0'-terminated string "")
*/
static inline void buffer_clear(buffer *b);
void buffer_copy_string(buffer * restrict b, const char * restrict s);
void buffer_copy_string_len(buffer * restrict b, const char * restrict s, size_t s_len);
static inline void buffer_copy_buffer(buffer * restrict b, const buffer * restrict src);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
void buffer_append_string(buffer * restrict b, const char * restrict s);
void buffer_append_string_len(buffer * restrict b, const char * restrict s, size_t s_len);
static inline void buffer_append_string_buffer(buffer * restrict b, const buffer * restrict src);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
#define buffer_append_uint_hex(b,len) buffer_append_uint_hex_lc((b),(len))
void buffer_append_uint_hex_lc(buffer *b, uintmax_t len);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
void buffer_append_int(buffer *b, intmax_t val);
void buffer_append_strftime(buffer * restrict b, const char * restrict format, const struct tm * restrict tm);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* '-', log_10 (2^bits) = bits * log 2 / log 10 < bits * 0.31, terminating 0 */
#define LI_ITOSTRING_LENGTH (2 + (8 * sizeof(intmax_t) * 31 + 99) / 100)
size_t li_itostrn(char *buf, size_t buf_len, intmax_t val);
size_t li_utostrn(char *buf, size_t buf_len, uintmax_t val);
/* buf must be (at least) 2*s_len + 1 big. uses lower-case hex letters. */
#define li_tohex(buf,buf_len,s,s_len) li_tohex_lc((buf),(buf_len),(s),(s_len))
void li_tohex_lc(char * restrict buf, size_t buf_len, const char * restrict s, size_t s_len);
void li_tohex_uc(char * restrict buf, size_t buf_len, const char * restrict s, size_t s_len);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* NULL buffer or empty buffer (used == 0);
* unset "string" (buffer) config options are initialized to used == 0,
* while setting an empty string leads to used == 1
*/
__attribute_pure__
static inline int buffer_is_empty(const buffer *b);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* NULL buffer, empty buffer (used == 0) or empty string (used == 1) */
__attribute_pure__
static inline int buffer_string_is_empty(const buffer *b);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
__attribute_pure__
int buffer_eq_icase_ssn(const char * const a, const char * const b, const size_t len);
__attribute_pure__
int buffer_eq_icase_ss(const char * const a, const size_t alen, const char * const b, const size_t blen);
__attribute_pure__
int buffer_eq_icase_slen(const buffer * const b, const char * const s, const size_t slen);
__attribute_pure__
int buffer_eq_slen(const buffer * const b, const char * const s, const size_t slen);
__attribute_pure__
int buffer_is_equal(const buffer *a, const buffer *b);
__attribute_pure__
int buffer_is_equal_right_len(const buffer *a, const buffer *b, size_t len);
__attribute_pure__
int buffer_is_equal_string(const buffer *a, const char *s, size_t b_len);
__attribute_pure__
int buffer_is_equal_caseless_string(const buffer *a, const char *s, size_t b_len);
void buffer_substr_replace (buffer * restrict b, size_t offset, size_t len, const buffer * restrict replace);
2017-04-24 06:38:30 +00:00
void buffer_append_string_encoded_hex_lc(buffer * restrict b, const char * restrict s, size_t len);
void buffer_append_string_encoded_hex_uc(buffer * restrict b, const char * restrict s, size_t len);
typedef enum {
ENCODING_REL_URI, /* for coding a rel-uri (/with space/and%percent) nicely as part of a href */
ENCODING_REL_URI_PART, /* same as ENC_REL_URL plus coding / too as %2F */
ENCODING_HTML, /* & becomes &amp; and so on */
ENCODING_MINIMAL_XML /* minimal encoding for xml */
} buffer_encoding_t;
void buffer_append_string_encoded(buffer * restrict b, const char * restrict s, size_t s_len, buffer_encoding_t encoding);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* escape non-printable characters; simple escapes for \t, \r, \n; fallback to \xCC */
void buffer_append_string_c_escaped(buffer * restrict b, const char * restrict s, size_t s_len);
/* to upper case, replace non alpha-numerics with '_'; if is_http_header prefix with "HTTP_" unless s is "content-type" */
void buffer_copy_string_encoded_cgi_varnames(buffer * restrict b, const char * restrict s, size_t s_len, int is_http_header);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
void buffer_urldecode_path(buffer *url);
void buffer_urldecode_query(buffer *url);
int buffer_is_valid_UTF8(const buffer *b);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
void buffer_path_simplify(buffer *dest, buffer *src);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
void buffer_to_lower(buffer *b);
void buffer_to_upper(buffer *b);
/** deprecated */
char hex2int(unsigned char c);
char int2hex(char i);
__attribute_pure__
static inline int light_isdigit(int c);
static inline int light_isdigit(int c) {
return (c >= '0' && c <= '9');
}
__attribute_pure__
static inline int light_isxdigit(int c);
static inline int light_isxdigit(int c) {
return light_isdigit(c) || (c |= 32, c >= 'a' && c <= 'f');
}
__attribute_pure__
static inline int light_isalpha(int c);
static inline int light_isalpha(int c) {
return (c |= 32, c >= 'a' && c <= 'z');
}
__attribute_pure__
static inline int light_isalnum(int c);
static inline int light_isalnum(int c) {
return light_isdigit(c) || light_isalpha(c);
}
__attribute_pure__
static inline uint32_t buffer_string_length(const buffer *b); /* buffer string length without terminating 0 */
__attribute_pure__
static inline uint32_t buffer_string_space(const buffer *b); /* maximum length of string that can be stored without reallocating */
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
static inline void buffer_append_slash(buffer *b); /* append '/' no non-empty strings not ending in '/' */
void buffer_append_path_len(buffer * restrict b, const char * restrict a, size_t alen); /* join strings with '/', if '/' not present */
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
#define BUFFER_APPEND_STRING_CONST(x, y) \
buffer_append_string_len(x, y, sizeof(y) - 1)
#define BUFFER_COPY_STRING_CONST(x, y) \
buffer_copy_string_len(x, y, sizeof(y) - 1)
#define BUFFER_INTLEN_PTR(x) (x)->used ? (int)((x)->used - 1) : 0, (x)->ptr
#define CONST_LEN_STR(x) (uint32_t)sizeof(x)-1, x
#define CONST_STR_LEN(x) x, (uint32_t)sizeof(x) - 1
#define CONST_BUF_LEN(x) ((x) ? (x)->ptr : NULL), buffer_string_length(x)
#define LI_NORETURN __attribute_noreturn__
__attribute_cold__
void log_failed_assert(const char *filename, unsigned int line, const char *msg) LI_NORETURN;
#define force_assert(x) do { if (!(x)) log_failed_assert(__FILE__, __LINE__, "assertion failed: " #x); } while(0)
#define SEGFAULT() log_failed_assert(__FILE__, __LINE__, "aborted");
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
/* inline implementations */
static inline int buffer_is_empty(const buffer *b) {
return NULL == b || 0 == b->used;
}
static inline int buffer_string_is_empty(const buffer *b) {
return NULL == b || b->used < 2;
}
static inline uint32_t buffer_string_length(const buffer *b) {
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
return NULL != b && 0 != b->used ? b->used - 1 : 0;
}
static inline uint32_t buffer_string_space(const buffer *b) {
return NULL != b && b->size ? b->size - (b->used | (0 == b->used)) : 0;
}
static inline void buffer_copy_buffer(buffer * restrict b, const buffer * restrict src) {
buffer_copy_string_len(b, CONST_BUF_LEN(src));
}
static inline void buffer_append_string_buffer(buffer * restrict b, const buffer * restrict src) {
buffer_append_string_len(b, CONST_BUF_LEN(src));
}
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
static inline void buffer_append_slash(buffer *b) {
uint32_t len = buffer_string_length(b);
fix buffer, chunk and http_chunk API * remove unused structs and functions (buffer_array, read_buffer) * change return type from int to void for many functions, as the return value (indicating error/success) was never checked, and the function would only fail on programming errors and not on invalid input; changed functions to use force_assert instead of returning an error. * all "len" parameters now are the real size of the memory to be read. the length of strings is given always without the terminating 0. * the "buffer" struct still counts the terminating 0 in ->used, provide buffer_string_length() to get the length of a string in a buffer. unset config "strings" have used == 0, which is used in some places to distinguish unset values from "" (empty string) values. * most buffer usages should now use it as string container. * optimise some buffer copying by "moving" data to other buffers * use (u)intmax_t for generic int-to-string functions * remove unused enum values: UNUSED_CHUNK, ENCODING_UNSET * converted BUFFER_APPEND_SLASH to inline function (no macro feature needed) * refactor: create chunkqueue_steal: moving (partial) chunks into another queue * http_chunk: added separate function to terminate chunked body instead of magic handling in http_chunk_append_mem(). http_chunk_append_* now handle empty chunks, and never terminate the chunked body. From: Stefan Bühler <stbuehler@web.de> git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2975 152afb58-edef-0310-8abb-c4023f1b3aa9
2015-02-08 12:37:10 +00:00
if (len > 0 && '/' != b->ptr[len-1]) BUFFER_APPEND_STRING_CONST(b, "/");
}
static inline void buffer_clear(buffer *b) {
b->used = 0;
}
#endif