Linux 开发系列笔记 - 文件操作与目录操作

文件操作与目录操作

基本原理

标准C库与Linux系统文件IO对比

I0：Input / Output，输入输出，通常指文件在存储介质之间的传输。
标准C库IO函数与LinuxIO函数是调用与被调用的关系， 标准C库函数调用Linux函数进行文件IO，应用程序没有直接操作文件的权限，必须由操作系统代理完成。标准C库IO函数比LinuxIO函数更高级。但是，需要注意的是，标准C库IO函数效率高于Linux函数，标准C库为IO实现了缓冲区的功能，减少了效率低下的外部设备访问开销。

由于缓存区的存在，标准C库IO是非即时的，LinuxIO是即时的，例如在网络交互中需要使用后者，一般场景使用前者更优。

虚拟地址空间

在操作文件与内存中得到的所有地址都是虚拟地址，真实地址对于用户是不可见的，也是无需关心的。
MMU：Memory Management Unit，用于实现虚拟地址与真实地址之间的映射，从而完成CPU的内存管理请求。虚拟地址映射表存储在PCB中。

栈空间从高地址向低地址占用，堆空间从低地址向高地址占用。

文件描述符

文件描述符用于描述一个打开的文件，存储于进程PCB（Processing Control Block）中。
进程通过文件描述符访问一个打开的文件。
PCB中文件描述符列表的尺寸为1024，也就是说，一个进程最多同时打开1024个文件，关闭文件后可复用。
多次打开相同文件将占用多个文件描述符。

Linux 文件操作函数

打开文件

open
O_RDONLY、O_WRONLY、O_RDWR、O_APPEND、O_TRUNC、O_NONBLOCK、O_CREAT
UMASK：在后文的权限操作中，也会与UMASK操作后处理。

cpp 复制代码

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
// 	open a file
// 		pathname:
// 			the file path to open
// 		flags:
// 			must and exactly one:
// 				O_RDONLY 	for read only, 
// 				O_WRONLY 	for write only, 
// 				O_RDWR 		for read and write.
// 			optional and limitless:
// 				O_APPEND 	for append mode, 
// 				O_TRUNC		truncate file size to zero while open,
// 				O_NONBLOCK 	for non-block, 
// 				O_CREAT 	for creatable.
// 		mode:
// 			If O_CREATE flag is on, must follow with the key.
// 			Mode is a oct-based number and determine the limits of the new file.
// 			It includes three digits (such as 0664), for user - group - others.
//			And each digit can be extend to three binary-bits, for read - write - execute.
// 			The finally mode will be (mode^~UMASK).
// 			* UMASK:
// 				A mask on the mode to ensure the file didn't give the inappropriate permission. 
// 				Also you can change the value of UMASK, command "$ umask" to show the value.
// 				The default value is 0002. Because giving others write permission is inappropriate.
// 		return value:
// 			file descripor, or -1 for error
int open(const char *pathname, int flags);
int open(const char *pathname, int flags, mode_t mode);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("open");  // About more: $ man 3 perror

// About more
// $ man 2 open

关闭文件

close

cpp 复制代码

#include <unistd.h>
// 	close a file
// 	file will automatic close while process ended
// 		fd:
// 			file description
// 		return value:
// 			return 0 for success, -1 for error
int close(int fd);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("close");

// About more
// $ man 2 close

读取文件

read

cpp 复制代码

#include <unistd.h>
// 	read a file
// 		fd:
// 			file descriptor of a readable file
// 		buf:
// 			he container of data
// 		count:
// 			read size for each time
// 		return value:	
// 			real read size of data, or -1 for error
// 			Because the rest size maybe smaller than ideal. It become zero if read the end of file.
ssize_t read(int fd, void *buf, size_t count);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("read");

// About more
// $ man 2 read

写入文件

write

cpp 复制代码

#include <unistd.h>
// similar to read a file
ssize_t write(int fd, const void *buf, size_t count);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("write");

// About more
// $ man 2 write

文件指针

lseek
SEEK_SET、SEEK_CUR、SEEK_END

cpp 复制代码

#include <sys/types.h>
#include <unistd.h>
// 	reposition read / write file offset
// 		fd:
// 			file descriptor
// 		offset:
// 			relative offset to whence
// 		whence:
// 			SEEK_SET for BOF, SEEK_CUR for now file offset, SEEK_END for EOF
// 		return value:
// 			absolute file offset
off_t lseek(int fd, off_t offset, int whence);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("lseek");

// About more
// $ man 2 lseek

cpp 复制代码

int offset = lseek(fd, 0, SEEK_CUR);	// get now file offset
lseek(fd, offset, SEEK_SET);			// reposition to absolute offset
lseek(fd, 0, SEEK_END);					// reposition to the end of file

文件截断

truncate

cpp 复制代码

#include <unistd.h>
#include <sys/types.h>
// 	resize the file
// 		path:
// 			the file path
// 		length:
// 			new length of file
// 			append \x00 if become larger and cut down the rest tail if become smaller
// 		return value:
// 			return 0 for success, -1 for error
int truncate(const char *path, off_t length);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("truncate");

// About more
// $ man 2 truncate

Linux 文件属性函数

Linux中一切皆文件，你同样可以用下面这些函数操作目录等特殊文件。

返回文件属性

stat、lstat
常用属性：

st_mode：文件属性编码。

st_size：文件大小Bytes。

st_atime：访问时间。

st_mtime：修改时间。

st_ctime：修改时间，指属性修改。

cpp 复制代码

#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
// 	get file status
// 		pathname:
// 			the path of target file
// 		statbuf:
// 			the container to restore data
// 			struct stat {
// 			    dev_t     st_dev;         /* ID of device containing file */
// 			    ino_t     st_ino;         /* Inode number */
// 			    mode_t    st_mode;        /* File type and mode */
// 			    nlink_t   st_nlink;       /* Number of hard links */
// 			    uid_t     st_uid;         /* User ID of owner */
// 			    gid_t     st_gid;         /* Group ID of owner */
// 			    dev_t     st_rdev;        /* Device ID (if special file) */
// 			    off_t     st_size;        /* Total size, in bytes */
// 			    blksize_t st_blksize;     /* Block size for filesystem I/O */
// 			    blkcnt_t  st_blocks;      /* Number of 512B blocks allocated */
// 	
// 			    struct timespec st_atim;  /* Time of last access */
// 			    struct timespec st_mtim;  /* Time of last modification */
// 			    struct timespec st_ctim;  /* Time of last status change */
// 			
// 			    #define st_atime st_atim.tv_sec
// 			    #define st_mtime st_mtim.tv_sec
// 			    #define st_ctime st_ctim.tv_sec
// 			};
// 		return value:
// 			return 0 for success, -1 for error
int stat(const char *pathname, struct stat *statbuf);
// similar to stat
int lstat(const char *pathname, struct stat *statbuf);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("stat");

// About more
// $ man 2 stat

lstat与stat的区别：stat在访问软链接时，返回软链接属性；lstat在访问软链接时，返回指向文件属性。

cpp 复制代码

$ ln -s src.txt link.lnk  // 建立软链接 link.lnk -> src.txt

判断文件访问

access

cpp 复制代码

#include <unistd.h>
//  check permissions to the file for current process
// 		pathname:
// 			the file path
// 		mode:
// 			optional and limitless:
// 				F_OK for file exists, 
// 				R_OK for read permission for process, 
// 				W_OK for write permission for process, 
// 				X_OK for execute permission for process
// 		return value:
// 			return 0 for all permission is available, -1 for other case and error
int access(const char *pathname, int mode);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
perror("access");

// About more
// $ man 2 access

修改文件属性

chmod

cpp 复制代码

#include <sys/stat.h>
// 	change mode of file
// 		pathname:
// 			the file path
// 		mode:
// 			three digits of oct-based number
// 			UMASK work here
// 		return value:
// 			return 0 for success, -1 for error
int chmod(const char *pathname, mode_t mode);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
void perror("chmod");

// About more
// $ man 2 chmod

修改文件所有

chown

cpp 复制代码

#include <unistd.h>
// 	change owner of file
// 		pathname:
// 			the file path
// 		owner:
// 			the UID of owner
// 		group:
// 			the GID of group
// 		return 0 for success, -1 for error
int chown(const char *pathname, uid_t owner, gid_t group);

// if return value is -1, you can then call to print the error message
#include <stdio.h>
void perror("chown");

// About more
// $ man 2 chown

Linux 目录操作函数

切换工作目录

chdir

cpp 复制代码

#include <unistd.h>
int chdir(const char *path);
// $ man 2 chdir

查看工作目录

getcwd

cpp 复制代码

#include <unistd.h>
char *getcwd(char *buf, size_t size);
// $ man 2 getcwd

创建目录

mkdir

cpp 复制代码

#include <sys/stat.h>
#include <sys/types.h>
int mkdir(const char *pathname, mode_t mode);
// $ man 2 mkdir

重命名目录

rename

cpp 复制代码

#include <stdio.h>
// rename or move a file or director
int rename(const char *oldpath, const char *newpath);
// $ man 2 rename

移除目录

rmdir、unlink、remove

cpp 复制代码

#include <unistd.h>
// remove a director
int rmdir(const char *pathname);
// $ man 2 rmdir

cpp 复制代码

#include <unistd.h>
// delete a file
int unlink(const char *pathname);
// $ man 2 unlink

cpp 复制代码

#include <stdio.h>
// remove a file or director
int remove(const char *pathname);
// $ man 3 remove

遍历目录

opendir、readdir、seekdir、closedir
常用属性：

d_off：文件在当前目录的序号

d_name：文件名

d_reclen：文件名长度

cpp 复制代码

#include <sys/types.h>
#include <dirent.h>
// 	open a director
// 		name:
// 			the file path
// 		return value:
// 			a pointer to DIR, or NULL for error
DIR *opendir(const char *name);
// $ man 3 opendir

cpp 复制代码

#include <dirent.h>
// 	read an item in director
// 		dirp:
// 			a pointer to DIR return by "opendir"
// 		return value:
// 			an item in director and restore in struct dirent, NULL for EOF or other errors
// 			struct dirent {
// 			    ino_t          d_ino;       /* Inode number */
// 			    off_t          d_off;       /* Current position in the director stream */
// 			    unsigned short d_reclen;    /* Length of this record */
// 			    unsigned char  d_type;      /* Type of file; not supported by all filesystem types */
// 			    char           d_name[256]; /* Null-terminated filename */
// 			};
// 			It's similar to read a file, while read an item in director, the file seek will move to the next item. You can use "seekdir" to reset the file offset.
struct dirent *readdir(DIR *dirp);
// $ man 3 readdir

cpp 复制代码

#include <dirent.h>
// set the current position of pointer
// 		dirp:
// 			a pointer to DIR return by "opendir"
// 		loc:
// 			new position
void seekdir(DIR *dirp, long loc);
// $ man 3 seekdir

cpp 复制代码

#include <sys/types.h>
#include <dirent.h>
// 	close a director
// 		dirp:
// 			a pointer to DIR return by "opendir"
// 		return value:
// 			return 0 for success, -1 for error
int closedir(DIR *dirp);
// $ man 3 closedir

Linux 操作文件描述符

复制文件描述符

dup、dup2

cpp 复制代码

#include <unistd.h>
// 	duplicate a file descriptor
// 		oldfd:
// 			old file descriptor
// 		return value:
// 			new file descriptor, or -1 for error
int dup(int oldfd);
// 	duplicate a file descriptor
// 		oldfd:
// 			old file descriptor
// 		newfd:
// 			new file descriptor
// 		return value:
// 			newfd, or -1 for error
// 			If newfd is occupied, it will close the file first.
// 			It is same to reopen a same file in same flags.
int dup2(int oldfd, int newfd);
// $ man 2 dup

获取文件状态标记

fcntl：fcntl有五大功能，由cmd参数决定，这里介绍的获取与修改文件状态标记属于其中一种。这里仅介绍这一种。

cpp 复制代码

#include <unistd.h>
#include <fcntl.h>
// 	manipulate file descriptor: get file descriptor flags
// 		fd:
// 			file descriptor
// 		cmd:
// 			F_GETFD: get file descriptor flags
// 		return value:
// 			return file descriptor flags
int fcntl(int fd, int cmd);
// $ man 2 fcntl

修改文件状态标记

fcntl

cpp 复制代码

#include <unistd.h>
#include <fcntl.h>
// 	manipulate file descriptor: set file descriptor flags
// 		fd:
// 			file descriptor
// 		cmd:
// 			F_SETFD: get file descriptor flags
// 		flags:
// 			must and exactly one:
// 				O_RDONLY 	for read only, 
// 				O_WRONLY 	for write only, 
// 				O_RDWR 		for read and write
// 			optional and limitless:
// 				O_APPEND 	for append mode, 
// 				O_NONBLOCK 	for non-block, 
// 				O_CREATE 	for creatable.
int fcntl(int fd, int cmd, int flags);
// $ man 2 fcntl