linux 文件读写

两种I/O文件函数

fopen
ANSI 标准文件I/O,基于低层次I/O

open
低层次I/O

ANSI I/O本质

DOS/WINDOWS平台，MS DOS 文件读写， 汇编语言编写

LINUX平台，unix/linux文件读写，C语言编写

文件类型FILE

包含一个指针

包含一个stream （C语言把文件看成stream
）

file open

r

Open text file for reading

r+

Open for reading and writing

w Truncate
file to zero length or create text file for writing

w+

The file is created if it does not exist,
otherwise it is truncated

a
Open for
appending (writing at end of file).

The stream is positioned at the end of the file.

fopen三种基本模式r,w,a只允许针对文本文
件

C语言允许最大同时打开16个文件

实际只允许同时打开13个文件，因为有另外三个标准文件stdout,stdin,stderr

文件读写的基本语句

1	#include <stdio.h> FILE *stream; 注意类型是大写
2	if((stream=fopen("testtmp","r+"))==NULL) { printf("open file error/n"); return; }
3	int ret; if ((ret=fclose(stream))==-1) printf("close file error/n");

顺序读文件(按字符)

while((c=fgetc(stream))!=
EOF
)
printf("%c",c);
每读一个字符，文件指针会（自动）移动一格

顺序读文件（按字符串）-------从一个文件依次读一行,直到读到NULL

char record[100],*re;

while ((re=fgets(record,100
,stream))!=NULL
)

printf("%s",record);

读一行，以回车换行作为读一行的尾部标志

从文件中fgets的字符串是读一行，即包括了回车

所以printf("%s")即可，不用"%s/n",因为字符串里面带了回车了

这也是个漏洞，字符串里带了回车

所以非常不利于字符串比对strcmp

scanf读文件，文件指针不移动，需要手动fseek移指针

fseek(stream,3,SEEK_SET); fscanf(stream,"%c",&c[0]); fseek(stream,11,SEEK_SET); fscanf(stream,"%c",&c[1]);

文件读写到文件末尾

文件指针会指到NULL
（而不是EOF，EOF不是地址，而是结尾字符
）

fput(),fgetc()会返回EOF

fgets()会返回NULL

linux和DOS对文件的结束判别是不同的

linux没有文件结束符
,是以目录项的文件长度做为文件结束的判别手段

DOS看0x1a文件结束符

linux 和DOS的换行符的区别

linux	DOS
0x0a 即 10 即’/n’ 即LF	CR，LF双码

读文件时, 不能用unsigned char c作为返回值，
因为EOF不在unsigned范围内

char c;

c=getc(stream);

fscanf读文本文件里面的数字，既可以直接用%c,也可以用%d

用%c(char)读，可以读出3,1	用%d(int)读，也可以读出3,1
fseek(stream,3,SEEK_SET); fscanf(stream,"%c",&c[0]); fseek(stream,11,SEEK_SET); fscanf(stream,"%c",&c[1]);	fseek(stream,ii,SEEK_SET); fscanf(stream,"%d",&ln[i].frame); fseek(stream,ii+8,SEEK_SET); fscanf(stream,"%d",&ln[i].line);

只不过这两个3,1,一个是实际值3,1,一个是'3','1'

几个常见特殊字符的整型数字

char	int
空格' '	32
TAB键	9
回车	10
文件结束EOF	-1
字符串结束符号	数字0（不是字符‘0’）

TAB字符处理要小心，经过到记事本copy/paste后，TAB键被转化成几个空格

for(;str[i]==' '||str[i]==' ';i++);

但经过到记事本copy/paste后，TAB键被转化成几个空格 所以系统总报warning: tmp.c:58: warning: comparison is always false due to limited range of data type tmp.c:59:27: warning: character constant too long for its type

fgets字符串指针改成字符串数组，消灭了Segmentation
fault错误

char *re,*rec; re=fgets(rec,100,srcstream); 出Segmentation fault错误

改成 char *re,rec[100]; re=fgets(rec,100,srcstream); 错误消失

一个普通的显示文件内容
的函数

printfile(FILE *stream,char *filename,int or) { int i,re; char rec[100]; FILE *tmpstream; if (or) { if((tmpstream=fopen(filename,"r"))==NULL) { printf("open file error/n"); return 0; } for(i=0;(re=fgets(rec,100,tmpstream))!=NULL;i++) printf("[%d]:%s",i,rec); 显示文件内容时，每行加行号 if ((re=fclose(tmpstream))==-1) printf("close file error/n"); } else { fseek(stream,0,SEEK_SET); for(i=0;(re=fgets(rec,100,stream))!=NULL;i++) printf("[%d]:%s",i,rec); } }

printfile(NULL,"testtmp.tmp",1 ); [0]:total 48 [1]:1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]:1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o

if((stream=fopen("testtmp","r+"))==NULL) { printf("open file error/n"); return 0; } printfile(stream ,NULL,0 ); [0]:total 48 [1]:-rw-rw-r-- 1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]: -rw-rw-r-- 1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o

按SED/AWK原理，读记录/行的函数——是从文件中取第recno行

int getrecord(char *rec,FILE *stream,int recno) { int i,ret; char *re; for(i=0;i<=recno;i++) re=fgets(rec,100,stream); 数行，并一边数一边读行 if(re==NULL) return 0; 如果所取行超过文件长度，则返回0 else return 1; }

getrecord(record,stream,4); 取记录/行，取文件第四行 printf("record is %s/n",record); printf("/n/n/n");

$ ./tmp record is -rw-rw-r-- 1 macg macg 0 Jan 16 02:56 testtmp

按SED/AWK原理，读字段的函数——从字符串中取第valueno字段

int getvalue(char *str,int valueno,char *value) { int i,j,vali; 数字段 for(i=0,vali=0;vali<valueno;vali++) { for(;str[i]==' '||str[i]==9;i++); for(;str[i]!=' '&&str[i]!=9&&str[i]!=0;i++); if(str[i]==0) return 0; } 取字段 for(;str[i]==' '||str[i]==9;i++); for(j=0;str[i]!=' '&&str[i]!=9&&str[i]!=0;i++,j++) { value[j]=str[i]; } value[j]=0; }

for(i=0;i<20;i++) { if(getvalue(record,i,val)) printf("no %d is %s/n",i,val); }

$ ./tmp record is -rw-rw-r-- 1 macg macg 0 Jan 16 02:56 testtmp 事先故意编辑过此行，加了几个TAB，连行尾也含TAB no 0 is -rw-rw-r-- no 1 is 1 no 2 is macg no 3 is macg no 4 is 0 no 5 is Jan no 6 is 16 no 7 is 02:56 no 8 is testtmp no 9 is

fputs(stren,streamwrite)
写入的串不带回车，fprintf(strea,"%s",str)写入的串也不带回车，所以，要写入文件回车,必须写入/n

fputs(stren,streamwrite); cat aaa.txt 008421aa

fprintf(streamwrite,"%s/n ",stren); cat aaa.txt 008421 aa

文件很难进行“修改写”,建议文件追加

最好的办法就是写到新文件
，或写到stdout再重定向入文件。然后把新文件覆盖旧文件

writefile(char *tmpfilename,char
*srcfilename,struct Filerecord *frs,int
frsi)-------修改文件的函数

实际是从 srcfile读出文件

把其中符合frs[j].recno的行，修改成frs[j].record

其他行不变，仍旧用srcfile读出的行

然后把这些行重新写入一个新文件tmpfile

writefile(char *tmpfilename,char *srcfilename,struct Filerecord *frs,int frsi o tmpfilename 新文件名 o srcfilename 旧文件名 o struct Filerecord *frs 放置要修改的行的结构体数组 struct Filerecord{ char record[100]; int recno; }; struct Filerecord fr[20]; o frsi 结构体数组的元素数量

writefile(char *tmpfilename,char *srcfilename,struct Filerecord *frs,int frsi) { int no,ret,i,j,bol; char *re,rec[100]; FILE *tmpstream,*srcstream; if((tmpstream=fopen(tmpfilename,"w+ "))==NULL) 写入一个新 { printf("open file error/n"); return 0; } if((srcstream=fopen(srcfilename,"r"))==NULL) { printf("open file error/n"); return 0; } for(no=0;(re=fgets(rec,100,srcstream))!=NULL;no++) { for(j=0,bol=1;j<frsi&&bol;j++) { if(no==frs[j].recno) { bol=0; strcpy(rec,frs[j].record); } } fputs(rec,tmpstream); } if ((ret=fclose(tmpstream))==-1) printf("close file error/n"); if ((ret=fclose(srcstream))==-1) printf("close file error/n"); }

例子：去掉文件中行的TAB键和起首空格或TAB 执行： 读出原文件每行，checkinvaild() ，看是否含有TAB，起首空格/TAB 如果确实有，就对此行处理，重新组合，去掉TAB，起首空格/TAB，然后写入struct person fr[20]数组 writefile("testtmp.tmp","testtmp",fr,fri); ------------------the sick file---------------------------------------- [0]:total 48 [1]:-rw-rw-r-- 1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]: -rw-rw-r-- 1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o --------------------modify line-------------------------------------- fri is 2,the new line is following: [1]row:1 macg macg 3301 Jan 16 02:16 file.c [3]row:1 macg macg 0 Jan 16 02:56 testtmp -------------------new file--------------------------------------- [0]:total 48 [1]:1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]:1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o

fseek(stream,offset,mode)的三个mode

SEEK_SET:从开头数第几个offset

SEEK_CUR: 从当前数第几个offset

SEEK_END: 从结尾倒数第几个offset

文件指针位置是从0(文件头)开始算的

fseek(stream,0,SEEK_SET)；

int ftell(FILE
*stream)

返回stream的当前指针位置

简单的取文件大小size的操作

fseek(stream,0,SEEK_END);先将指针指向文件尾部 ret=ftell(stream); 再获得指针所指的当前地址 这个指向文件尾部的指针地址就是文件大小

取文件大小的操作对读写文件很有用，因为读写文件很不好判定文件结尾。

文件例子

$ cat tmp.c #define DEBUG 0 #include <stdio.h> #include <string.h> #include <unistd.h> struct Filerecord{ char record[100]; int recno; }; main() { char directory[100]; #ifdef SYST getcwd(directory,100); printf("current directory is %s/n",directory); #endif filemanage(); } filemanage() { int i,j,fri,ret; char record[100],val[50],modi_rec[100]; FILE *stream; struct Filerecord fr[20]; if((stream=fopen("testtmp","r+"))==NULL) { printf("open file error/n"); return 0; } printf("------------------the sick file-----------------------/n"); printfile(stream,NULL,0); for(i=0,fri=0;getrecord(record,stream,i);i++) { if (DEBUG) printf("no %d row is:%s/n",i,record); if(ret=findinvalid(record)) { for(j=0;getvalue(record,j,val);j++) { if (DEBUG) printf("no %d's val is %s %d/n",j,val,val[0]); if(j==1) strcpy(modi_rec,val); if(j>1) { strcat(modi_rec," "); strcat(modi_rec,val); } } fr[fri].recno=i; strcpy(fr[fri].record,modi_rec); if (DEBUG) printf("fr[%d] is %d %s/n",fri,fr[fri].recno,fr[fri].record); fri++; } if (DEBUG) printf("---------------------------------/n"); } printf("--------------------modify line--------------------/n"); printf("fri is %d,the new line is following: /n",fri); for(i=0;i<fri;i++) printf("[%d]row:%s",fr[i].recno,fr[i].record); if ((ret=fclose(stream))==-1) { printf("close file error/n"); return 0; } writefile("testtmp.tmp","testtmp",fr,fri); printf("-------------------new file-----------------/n"); printfile(NULL,"testtmp.tmp",1); } int findinvalid(char *str) { int i,ret; i=0; if(str[i]==' ') { if (DEBUG) printf("str[%d] == %d/n",i,str[i]); return 32; } for(i=0;str[i]!=9&&str[i]!=0;i++) ; if(str[i]==9) { if (DEBUG) printf("str[%d] == %d/n",i,str[i]); return 9; } else { if (DEBUG) printf("str[%d] == %d/n",i,str[i]); return 0; } }

$ make gcc -g -c tmp.c tmp.c: In function 鈥榩rintfile鈥? tmp.c:158: warning: assignment makes integer from pointer without a cast tmp.c:158: warning: comparison between pointer and integer tmp.c:165: warning: assignment makes integer from pointer without a cast tmp.c:165: warning: comparison between pointer and integer gcc -o tmp tmp.o -g

$ ./tmp ------------------the sick file---------------------- [0]:total 48 [1]:-rw-rw-r-- 1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]: -rw-rw-r-- 1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o --------------------modify line-------------------- fri is 2,the new line is following: [1]row:1 macg macg 3301 Jan 16 02:16 file.c [3]row:1 macg macg 0 Jan 16 02:56 testtmp -------------------new file------------------------- [0]:total 48 [1]:1 macg macg 3301 Jan 16 02:16 file.c [2]:-rw-rw-r-- 1 macg macg 52 Jan 16 02:19 Makefile [3]:1 macg macg 0 Jan 16 02:56 testtmp [4]:this is a testthis is a testthis is a testthis is a testthis is a testth [5]:-rw-rw-r-- 1 macg macg 1428 Jan 16 02:56 tmp.o