Unix search techniques.... 2

[areza123]

Using man I'm trying to find the Unix command line options to search a file name recursively. (the file may be present in directories inner to the current directory)

I execute...
man ls | grep "recur"

The output is...
list subdirectories recursively

Now I want to read the full sentence ? Or lets say 10 lines above this and 10 lines below this output. Is there some command ?

man ls | grep "recur" | head

and

man ls | grep "recur" | head

do not work. They give the same output.

Neither do...
head | man ls | grep "recur"
tail | man ls | grep "recur"

Help...

 

[jad]

find <dir> -name '<filename>'

eg. cd /etc ; find . -name 'hosts'

 

[areza123]

hey jad thanks for the quick response but I was looking for the search technique not the command.

Example:
When I execute man ls ...it gives me about 100 lines output....but I only need to see few lines in between say...


-R, --recursive
list subdirectories recursively

-s, --size
print size of each file, in blocks


Cheers

 

[Ygor]

For linux...

man ls | grep -C 10 &quot;recur&quot;

man grep for more options.

 

[johngiggs]

Ygor,

What does the -C do in Linux?

Thanks,

John

 

[jad]

umm actually it's not just for linux, its for use with GNU grep ... and it give you extra lines of context.

 

[Clairvoyant1332]

Another thing you can do is to just run &quot;man ls&quot; then when the &quot;more&quot; prompt comes up press the / key and enter your search string. This is identical to the search function in vi.

Dennis

 

[jad]

in 'man' you can type 'v' to enter vi mode on the man page

you will then probably want to:
:g/[_+-]^V^H/s///g

to get rid of the formatting ^V^H is Control-V followed by Control-H

 

[areza123]

Thanks for your help. I really appreciate. Now let me extend the problem statement to a more complicated but useful search technique....

man ls | grep -C 10 &quot;recur&quot;

works perfectly fine i.e. shows the 10 lines close to recur...the point here is that the output of man ls is one stream (or source...sorry I don't know if I'm using the right terminology)

Now consider the command:
grep -R &quot;recur&quot; *
This looks for the occurance of the word &quot;recur&quot; in all files (including files in child directories) i.e. there are multiple sources from where it picks up information. Now if I want to see 2 lines above and below each occurance of the keyword in each file. How do I acheive this

Example : I have 2 files say a.cc and b.cc

a.cc
====
cout << a;
cout << b;
cout << c;

b.cc
====
cout << d;
cout << b;
cout << f;

grep -R &quot;cout << b&quot; *

would yield the output
a.cc:cout << b;
b.cc:cout << b;

But what I am keen is too see is about 3 lines of source code above and below the occurance of &quot;cout << b&quot; in each of the files i.e. a.cc and b.cc ?

 

[tikual]

I am confused why you use -R instead of -C in this case?

 

[areza123]

Ok...actually your confusion created some new ideas with me which actually solved my problem.

The final solution to the above problem is

grep -R -C 1 &quot;ostream_iterator&quot; *

would search files listed in all subdirectories for the string &quot;ostream_iterator&quot; which is what I wanted.

Without the -R options the search was failing because I had only directories and not actual files as children of my current directory. My problem is solved but I hope your confusion is too!


Because I want to recurse inside subdirectories (select all files)

 

[areza123]

More Advances....

I want to modify my search pattern now
Assume I have the file a.txt

a.txt
-----
#ifdef USE_stub_in_nt_dll
...
#ifdef USE_stub_in_nt_dll
...
ifdef USE_stub_in_nt_dll
....
..
#ifdef USE_stub_in_nt_dll

If you observe there is only ONE USE_stub_in_nt_dll which is uncommented. (without #).
If this file contains about 1000 lines and I grep as

grep &quot;USE_stub_in_nt_dll&quot; *

I get the output
#ifdef USE_stub_in_nt_dll
#ifdef USE_stub_in_nt_dll
ifdef USE_stub_in_nt_dll
#ifdef USE_stub_in_nt_dll

But I only want to see the ONE line which is uncommented
i.e. ifdef USE_stub_in_nt_dll

So I issue the command
grep &quot;USE_stub_in_nt_dll&quot; * | grep -v &quot;#&quot;

Is this the correct way of doing it ?
I was wondering that there must be an option of doing it without the pipe ?

Note: grep -n -R -v &quot;# &quot;USE_stub_in_nt_dll&quot; *
will not work in this case. It outputs all the lines that do not have #

 

[tikual]

In your case, simply you can do like as this:

grep '^ifdef' filename

 

[Ygor]

Try...

grep -E '^[^#]*USE_stub_in_nt_dll'

Where..

-E = Extended regular-expression
^ = beginning of line
[^#]* = any number of characters other than #

See man regexp for more details.

 

[areza123]

How do I grep for more than 1 word in a line ?
If I have the following file:
a.txt
-----
abc ghi def
abc xyz
abc def ghi

I want to isolate only lines containing words abc & def
How do I acheive this ?

 

[Ygor]

I would use...

awk '/abc/&&/def/' a.txt

 

[PHV]

If the -w is not available for uoyr grep, try this:

grep -E '^abc[ \t]|[ \t]abc[ \t]|[ \t]abc$' a.txt |
  grep -E '^def[ \t]|[ \t]def[ \t]|[ \t]def$'

If -w is OK, try this:

grep -w 'abc' a.txt | grep -w 'def'

Hope This Help
PH.