Linux diff与patch的深入分析

diff的输出格式分为传统格式和统一格式


1)diff的传统格式输出.

############################################
cat before.txt
输出:
This is a line to be deleted
This is a line that will be changed
This is a line that will be unchanged

cat after.txt
输出:
This is a line that has been changed
This is a line that will be unchanged
This is a line that has been added
############################################
diff before.txt after.txt
输 出:
1,2c1
< This is a line to be deleted
< This is a line that will be changed
---
> This is a line that has been changed
3a3
> This is a line that has been added
############################################

注释:
传统格式的输出
1,2c1是指替换第1个文件的第1,2行到第2个文件的第1行,这里的1,2是指第1个文件的第1,2行,c是替换的意思,最后的1是第2个文件的第1行
<号是指第1个文件更改或删除的行
---号是分割两个文件
>号是第2个文件中增加或删除的行
3a3是指将第2个文件的第3行插入到第一个文件的第3行
也就是说第1个文件的:
< This is a line to be deleted
< This is a line that will be changed
被替换成第2个文件的:
> This is a line that has been changed
由于第1个文件的第3行和第2个文件的第2行一致,所以不做修改.
由于第2个文件的第3行是第1个文件所不具有的,所以在第1个文件的最后一行增加:
> This is a line that has been added



2)patch命令的应用

用diff的传统格式输出:
#################################
diff before.txt after.txt >mypatch.txt
#################################

用patch修补before.txt文件,使before.txt和after.txt一致.
#################################
cat mypatch.txt |patch before.txt
输出:
patching file before.txt
#################################

比较两个文件，现在是一致的了.
#################################
cmp before.txt after.txt
#################################

用patch命令恢复before.txt.
#################################
patch -R before.txt 输出:
patching file before.txt
#################################

注:-R标记告诉patch在反向上应用区别或者撤销patch.

再比较两个文件,现在不一致了.
#################################
cmp before.txt after.txt
输出:
before.txt after.txt differ: byte 17, line 1
#################################



3)diff的统一格式输出.

#################################
diff -u before.txt after.txt |tee mypatch.diff
输出:
--- before.txt  2009-06-20 05:21:49.000000000 +0800
+++ after.txt   2009-06-20 04:03:16.000000000 +0800
@@ -1,3 +1,3 @@
-This is a line to be deleted
-This is a line that will be changed
+This is a line that has been changed
 This is a line that will be unchanged
+This is a line that has been added
#################################

注释:
diff -u选项是统一格式输出.
--- before.txt  2009-06-20 05:21:49.000000000 +0800
--- before.txt是指旧文件
+++ after.txt   2009-06-20 04:03:16.000000000 +0800
+++ after.txt是指新文件.
@@ -1,3 +1,3 @@
@@ -1,3是指第1个文件一共有3行,+1,3 @@是指第2个文件一共有3行.
-This is a line to be deleted
-This is a line that will be changed
是被删除的行
+This is a line that has been changed
是增加的行
 This is a line that will be unchanged
没有-号和+号是指该行不变,因为after.txt和before.txt都有这行.
+This is a line that has been added
是增加的行
diff的统一格式比较与输出是按顺序进行的.



4)diff命令在目录中的应用.

新建old和new目录,old目录包含了初始内容,new目录包含文件的最新版本.
##########################################
mkdir old new
echo "This is one. It's unchanged." | tee old/one new/one
echo "This is two. It will change." > old/two
echo "This is two. It changed.">new/two
echo "This is three. It's new" > new/three
##########################################

创建修补文件
##########################################
diff -Nur old/ new/ >mypatch.diff
##########################################
注:-r选项按照文件目录递归创建修补文件.
-u还是统一模式
-N是指当diff遇到一个只存在于两个树中的一个树中的文件时,默认情况下跳过文件并且打印一个警告到stderr.
这个行为可以通过-N选项来更改,这也导致了diff认为丢失的文件实际上是存在的,但它是空的.采用这种方式,
一个修补文件可以包括已经创建的文件.然后应用修补程序创建新的文件.

##########################################
more mypatch.diff
输出:
diff -Nur old/three new/three
--- old/three   1970-01-01 08:00:00.000000000 +0800
+++ new/three   2009-06-20 06:55:34.000000000 +0800
@@ -0,0 +1 @@
+This is three. It's new
diff -Nur old/two new/two
--- old/two     2009-06-20 06:55:08.000000000 +0800
+++ new/two     2009-06-20 06:55:21.000000000 +0800
@@ -1 +1 @@
-This is two. It will change.
+This is two. It changed.
##########################################
注释:
diff -Nur old/three new/three是指下面比较old/three new/three两个文件.
因为没有old/three文件,所以在old/three中增加+This is three. It's new
diff -Nur old/two new/two是指下面比较old/two new/two两个文件
因为old/two与new/two的第3行不一致,所以删除This is two. It will change.增加This is two. It changed.


打补丁到old目录,新建old/three以及更改old/two
##########################################
patch --dir old< mypatch.diff
ls -l     old/
输出:
one    three  two
##########################################

恢复old目录的内容,包括删除old/three,以及恢复old/two文件
##########################################
patch --dir old -R 输出:
ls -l old/
one  two 
##########################################



5)检查和合并更改

用vim突出显示单个字符的更改来表示区别.

##########################################
vim -d after.txt before.txt
##########################################


用gui工具gvimdiff来显示两个文件.

##########################################
gvimdiff after.txt before.txt
##########################################


新建文件orig.c

##########################################
vi orig.c
void foo(void)
{
    printf("This will be changed by me. \n");

    printf("This will be unchanged,\n");

    printf("This will be changed by you.\n");
}
##########################################


复制文件orig.c到me.c,更改第4行为printf("This was changed by me. \n");
##########################################
vi me.c
void foo(void)
{
    printf("This was changed by me. \n");

    printf("This will be unchanged,\n");

    printf("This will be changed by you.\n");
}
##########################################


复制文件orig.c到you.c,更改第7行为printf("This was changed by you.\n");
##########################################
vi you.c
void foo(void)
{
    printf("This will be changed by me. \n");

    printf("This will be unchanged,\n");

    printf("This was changed by you.\n");
}
##########################################

版本工具如cvs,subversion使用GNU合并工具称为diff3.
##########################################
diff3 me.c orig.c you.c
输出:
====1
1:3c
      printf("This was changed by me. \n");
2:3c
3:3c
      printf("This will be changed by me. \n");
====3
1:7c
2:7c
      printf("This will be changed by you.\n");
3:7c
      printf("This was changed by you.\n");

注:
在没有参数的情况下,diff3产生的输出说明了那行更改.
====1和====3指明造成同原始文件不同的是哪一个修改文件.
编号方式基于参数序列.
也就是第1个文件和第3个文件与原文件不同.
1:3c
      printf("This was changed by me. \n");
3:3c
      printf("This will be changed by me. \n");
1:3c表示第1个文件的第3行与3:3c表示的第3个文件的第3行不同.
为什么不显示与原文件的比较呢。因为第3个文件的第3行与源文件(第2个文件)相同.所以与哪个文件比较无所谓了.

2:7c
      printf("This will be changed by you.\n");
3:7c
      printf("This was changed by you.\n");
2:7c表示第2个文件的第7行与3:7c表示的第3个文件的第7行不同.


diff3会试图为我们进行合并.合并是在源文件的基础上,依据两个新文件进行修改
源文件是第二个文件,第一个文件和第三个文件可以互换，但他们必须有共同的祖先，就是第二个文件.

#######################################
diff3 -m me.c orig.c you.c |cat -n
输出:
     1    void foo(void)
     2    {
     3        printf("This was changed by me. \n");
     4   
     5        printf("This will be unchanged,\n");
     6   
     7        printf("This was changed by you.\n");
     8    }
########################################


为了测试更复杂的环境,新建一个文件orig.c.1
内容如下:
########################################
vi orig.c.1
void foo(void)
{
    printf("This will be changed by both of us.\n");
}
########################################


用diff3 -m再次比较输出,如下:
########################################
diff3 -m me.c orig.c.1 you.c
void foo(void)
{
<<<<<<< me.c
    printf("This was changed by me. \n");

    printf("This will be unchanged,\n");

    printf("This will be changed by you.\n");
||||||| orig.c.1
    printf("This will be changed by both of us.\n");
=======
    printf("This will be changed by me. \n");

    printf("This will be unchanged,\n");

    printf("This was changed by you.\n");
>>>>>>> you.c
}
########################################

注释:以上的格式,同cvs update，需要人工合并文件的格式是一致的.