C5 标准IO库：APUE 笔记

C5 ：标准IO库

　　在第三章中，所有IO函数都是围绕文件描述符展开，文件描述符用于后续IO操作。由于文件描述符相关的操作是不带缓冲的IO，需要操作者本人指定缓冲区分配、IO长度等，对设备环境要求一定的了解。

　　本章引入文件流概念，IO库将处理很多底层IO细节。

1 缓冲

　　提供缓冲的目的是尽量减少read和write的调用次数。标准IO提供了以下3种缓冲：

1.1 全缓冲

　　在填满标准IO缓冲区后再进行实际IO操作，标准IO函数通常使用malloc获取使用的缓冲区

1.2 行缓冲

　　在输入和输出中遇到换行符时，标准IO才执行IO操作。对于行缓冲有2个限制

　　标准IO库用来收集每一行的缓冲区长度固定，所以只要是填满了缓冲区，就会进行IO操作
　　只要是通过标准IO从一个不带缓冲的流或者一个行缓冲的流得到输入数据，那么就会flush行缓冲输出流

1.3 不带缓冲

　　标准库不对字符进行缓冲存储，标准错误流通常是不带缓冲的，这样就可以尽快将信息显示出来

　　可以使用setbuf函数打开或者关闭缓冲机制，为了带缓冲进行IO，参数buf必须指向一个长度为BUFSIZE的缓冲区，头文件stdio.h。成功返回-，出错返回-1。

1 /* If BUF is NULL, make STREAM unbuffered.
2    Else make it use buffer BUF, of size BUFSIZ.  */
3 extern void setbuf (FILE *__restrict __stream, char *__restrict __buf) __THROW;
4 
5 /* Make STREAM use buffering mode MODE.
6    If BUF is not NULL, use N bytes of it for buffering;
7    else allocate an internal buffer N bytes long.  */
8 extern int setvbuf (FILE *__restrict __stream, char *__restrict __buf,  int __modes, size_t __n) __THROW;

　　setvbuf的mode参数可以精确说明缓冲类型，mode参数实现为：

　　_IOFBF 全缓冲，_IOLBF 行缓冲，_IONBF 不带缓冲。

　　如果指定一个不带缓冲的流，则忽略buf和size参数。
　　如果指定全缓冲和行缓冲，则buf和size可选择指定一个缓冲区及其长度。
　　如果流是带缓冲的，而buf是NULL，则标准IO库自动为流分配合适长度的缓冲区，长度由BUFSIZE指定。

2 流

2.1 打开、关闭流

　　可以使用fopen等函数打开一个IO流，头文件stdio.h。成功返回文件指针，出错返回NULL。

 1 /* Open a file and create a new stream for it.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern FILE *fopen (const char *__restrict __filename,
 6             const char *__restrict __modes) __wur;
 7 /* Open a file, replacing an existing stream with it.
 8 
 9    This function is a possible cancellation point and therefore not
10    marked with __THROW.  */
11 extern FILE *freopen (const char *__restrict __filename,
12               const char *__restrict __modes,
13               FILE *__restrict __stream) __wur;

　　fopen函数打开路径为__filename的文件

　　freopen函数在一个指定的流上打开文件。如果该流已经打开，则先关闭该流。如果该流已经定向，则使用freopen清除该定向。

　　mode参数指定对该IO流的读写方式，ISO C规定mode参数可以右15种取值。

　　使用fclose可以关闭一个打开的流，头文件stdio.h。成功返回0，失败返回-1。

1 /* Close STREAM.
2 
3    This function is a possible cancellation point and therefore not
4    marked with __THROW.  */
5 extern int fclose (FILE *__stream);
6 /* Flush STREAM, or all streams if STREAM is NULL.

2.2 读、写流

2.2.1 每次一个字符的IO

　　下面函数用于一次读取一个字符，头文件stdio.h。成功返回下一个字符，达到文件尾或者出错，返回EOF（定义在libio.h）。

 1 /* Read a character from STREAM.
 2 
 3    These functions are possible cancellation points and therefore not
 4    marked with __THROW.  */
 5 extern int fgetc (FILE *__stream);
 6 extern int getc (FILE *__stream);
 7 
 8 /* Read a character from stdin.
 9 
10    This function is a possible cancellation point and therefore not
11    marked with __THROW.  */
12 extern int getchar (void);
13 
14 #ifndef EOF
15 # define EOF (-1)
16 #endif

　　前两个函数的区别是，getc可以被实现为宏，而fgetc不能实现为宏？

　　getchar等价于geic (stdin)。　　

　　这三个函数达到文件尾或者出错，都返回EOF。为区别文件尾或者出错，需要调用以下函数进行判断，头文件stdio.h。成功返回非0，失败返回0。可以作为bool型判断。

1 /* Clear the error and EOF indicators for STREAM.  */
2 extern void clearerr (FILE *__stream) __THROW;
3 /* Return the EOF indicator for STREAM.  */
4 extern int feof (FILE *__stream) __THROW __wur;
5 /* Return the error indicator for STREAM.  */
6 extern int ferror (FILE *__stream) __THROW __wur;

　　调用clearerr可以清除这两个标志。

　　也可以使用ubgetc将字符再写回流缓冲区，头文件stdio.h。成功返回c，失败返回EOF。

1 /* Push a character back onto the input buffer of STREAM.
2 
3    This function is a possible cancellation point and therefore not
4    marked with __THROW.  */
5 extern int ungetc (int __c, FILE *__stream);

　　对应的写缓冲区函数为：

 1 /* Write a character to STREAM.
 2 
 3    These functions are possible cancellation points and therefore not
 4    marked with __THROW.
 5 
 6    These functions is a possible cancellation point and therefore not
 7    marked with __THROW.  */
 8 extern int fputc (int __c, FILE *__stream);
 9 extern int putc (int __c, FILE *__stream);
10 
11 /* Write a character to stdout.
12 
13    This function is a possible cancellation point and therefore not
14    marked with __THROW.  */
15 extern int putchar (int __c);

2.2.2 每次一行IO

　　下面两个函数每次读写输出一行，头文件stdio.h。成功返回buf，到达文件尾或者出错，返回NULL。

 1 /* Get a newline-terminated string of finite length from STREAM.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern char *fgets (char *__restrict __s, int __n, FILE *__restrict __stream)
 6      __wur;
 7 
 8 /* Get a newline-terminated string from stdin, removing the newline.
 9 
10    This function is impossible to use safely.  It has been officially
11    removed from ISO C11 and ISO C++14, and we have also removed it
12    from the _GNU_SOURCE feature list.  It remains available when
13    explicitly using an old ISO C, Unix, or POSIX standard.
14 
15    This function is a possible cancellation point and therefore not
16    marked with __THROW.  */
17 extern char *gets (char *__s) __wur __attribute_deprecated__;
18 
19 
20 /* Write a string to STREAM.
21 
22    This function is a possible cancellation point and therefore not
23    marked with __THROW.  */
24 extern int fputs (const char *__restrict __s, FILE *__restrict __stream);
25 
26 /* Write a string, followed by a newline, to stdout.
27 
28    This function is a possible cancellation point and therefore not
29    marked with __THROW.  */
30 extern int puts (const char *__s);

　　注意以下两点：

　　应当屏蔽使用puts或者gets。这两个函数都是从标准输入或输出读写，但由于无法指定缓冲区长度，将会导致缓冲区溢出错误
　　程序退出处，显示调用exit(0)，将会冲洗任何未写的数据，然后关闭所有打开的流
　　读写一行都是以NULL为终止，但是缓冲区的NULL不会写入文件

3 二进制IO

　　上一节的流IO无法解决以下问题：当读取的数据是结构化数据（非字符串，可能读取到NULL，0），流IO将会截断。

　　为解决该问题，需要引入二进制IO函数fread和fwrite，头文件stdio.h。函数返回读或写的对象数。

 1 /* Read chunks of generic data from STREAM.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern size_t fread (void *__restrict __ptr, size_t __size,
 6              size_t __n, FILE *__restrict __stream) __wur;
 7 /* Write chunks of generic data to STREAM.
 8 
 9    This function is a possible cancellation point and therefore not
10    marked with __THROW.  */
11 extern size_t fwrite (const void *__restrict __ptr, size_t __size,
12               size_t __n, FILE *__restrict __s);

4 定位流（访问偏移）

　　可使用ftell获取当前文件访问偏移量、fseek对文件访问进行偏移、rewind将流设置到起始位置。

 1 /* Seek to a certain position on STREAM.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern int fseek (FILE *__stream, long int __off, int __whence);
 6 /* Return the current position of STREAM.
 7 
 8    This function is a possible cancellation point and therefore not
 9    marked with __THROW.  */
10 extern long int ftell (FILE *__stream) __wur;
11 /* Rewind to the beginning of STREAM.
12 
13    This function is a possible cancellation point and therefore not
14    marked with __THROW.  */
15 extern void rewind (FILE *__stream);

　　fseek的第三个参数whence与fseek的参数相同：

　　SEEK_SET指定从文件起始位置开始，SEEK_CUR指定从当前位置开始，SEEK_END指定从文件末尾

5 格式化IO

5.1 格式化输出

　　格式化输出由5个printf函数处理，头文件stdio.h。成功返回字符数，失败返回负值。

 1 /* Write formatted output to stdout.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern int printf (const char *__restrict __format, ...);
 6 
 7 /* Write formatted output to STREAM.
 8 
 9    This function is a possible cancellation point and therefore not
10    marked with __THROW.  */
11 extern int fprintf (FILE *__restrict __stream,
12             const char *__restrict __format, ...);
13 
14 extern int dprintf (int __fd, const char *__restrict __fmt, ...)
15      __attribute__ ((__format__ (__printf__, 2, 3)));
16 
17 /* Write formatted output to S.  */
18 extern int sprintf (char *__restrict __s,
19             const char *__restrict __format, ...) __THROWNL;
20 
21 /* Maximum chars of output to write in MAXLEN.  */
22 extern int snprintf (char *__restrict __s, size_t __maxlen,
23              const char *__restrict __format, ...)
24      __THROWNL __attribute__ ((__format__ (__printf__, 3, 4)));

　　printf将格式化数据写到标准输出

　　fprintf将格式化数据写到指定的流

　　dprintf写到指定的文件描述符

　　sprintf写到数组S

　　snprintf是sprintf的安全版本，指定了缓冲区的长度，防止缓冲区溢出

5.2 格式化输入

　　格式化输入由3个scanf函数处理，头文件stdio.h。成功返回输入的参数，失败或者到达文件末尾，返回EOF。

 1 /* Read formatted input from STREAM.
 2 
 3    This function is a possible cancellation point and therefore not
 4    marked with __THROW.  */
 5 extern int fscanf (FILE *__restrict __stream,
 6            const char *__restrict __format, ...) __wur;
 7 /* Read formatted input from stdin.
 8 
 9    This function is a possible cancellation point and therefore not
10    marked with __THROW.  */
11 extern int scanf (const char *__restrict __format, ...) __wur;
12 /* Read formatted input from S.  */
13 extern int sscanf (const char *__restrict __s,
14            const char *__restrict __format, ...) __THROW;

6 临时文件

　　可以使用tmpnam或者tmpfile创建临时二进制文件。头文件stdio.h。

1 /* Generate a temporary filename.  */
2 extern char *tmpnam (char *__s) __THROW __wur;
3 
4 /* Create a temporary file and open it read/write.
5 
6    This function is a possible cancellation point and therefore not
7    marked with __THROW.  */
8 extern FILE *tmpfile (void) __wur;

　　tmpnam生成一个和现有文件名不同的有效路径字符串，最多调用TMP_MAX次，定义在stdio.h。返回指向唯一路径名的指针。如果输入为NULL，则路径名存放在静态区；如果不是NULL，则应该指向长度为L_tmpnam个字符的数组。

　　tmpfile创建一个临时二进制文件，在关闭该文件或者程序结束时将自动删除该文件。可以使用tmpfile创建一个临时文件，然后立即unlink。

7 内存流

1 /* Create a new stream that refers to a memory buffer.  */
2 extern FILE *fmemopen (void *__s, size_t __len, const char *__modes)
3   __THROW __wur;
4 
5 /* Open a stream that writes into a malloc'd buffer that is expanded as
6    necessary.  *BUFLOC and *SIZELOC are updated with the buffer's location
7    and the number of characters written on fflush or fclose.  */
8 extern FILE *open_memstream (char **__bufloc, size_t *__sizeloc) __THROW __wur;