Stripping a subset of text from a file

[solman66]

alright...
I have file A
I have file B which is a subset of A
i want to strip file B out of file A
i.e. A - B = C

This sounds like it should be easy and i feel i'm missing something....
file B is multiple lines and has some special characters...
I'm not ready to put this in a perl script just yet... i just have a feeling that there is a simpler way

tried

cat A |grep -v "`cat B`" > C

as well as a similar sed commmand
I escaped the special chars in B for that as well....
i think they fail because of the multiple lines....

I have used

 diff -bi A B | sed -e 's/^&#91;0-9&#93;*,&#91;0-9&#93;*&#91;a-z&#93;*&#91;0-9&#93;*$//g' | sed -e 's/^< //g' | sed -e 's/^> //g'| sed -e 's/---//g' > C

And this works.... however i don't like it because if B is not in A, then i get a mess....

Real world application:
Taking over a webmaster position and the previous webmaster used all static HTML with a long header.... i want to strip the header out and keep the "meat" of the page.... unfortunately not every single one of the headers are exactly the same... there are about 3-4 different versions (hence the reason my diff script would make more of a mess than its worth)

any suggestions?

[jim mcnamara]

sort -o A.srt A
sort -o B.srt B
comm -1 A.srt B.srt   > FileA_minus_FileB

[solman66]

Thanks for the suggestion...
but sorting will do no good...
for one, i need the text to stay in order.
for two, comm looks basically like a diff which is going would leave me in a similar situation as the other script i posted

[jim mcnamara]

comm in fact does do what you want.
But since you dont like it try some C code:

#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <stdlib.h>
#include <stdio.h>
#include <sys/stat.h>
#include <string.h>

#define ck&#40;x&#41; if&#40;&#40;x&#41;==NULL&#41;\
&#123;perror&#40;"Error"&#41;; exit&#40;EXIT_FAILURE&#41;;&#125;
/* read nbyte from a file  - can read whole file */
ssize_t readall&#40;int fd, void *buf, size_t nbyte&#41;&#123;
     ssize_t nread = 0,
                   n=0;

     do &#123;
         if &#40;&#40;n = read&#40;fd, &&#40;&#40;char *&#41;buf&#41;&#91;nread&#93;, nbyte - nread&#41;&#41; == -1&#41; &#123;
             if &#40;errno == EINTR&#41;
                 continue;
             else
                 return &#40;-1&#41;;
         &#125;
         if &#40;n == 0&#41;
             return nread;
         nread += n;
     &#125; while &#40;nread < nbyte&#41;;
     return nread;
&#125;

 /* get file size */
size_t file_size&#40;FILE *in&#41;
&#123;
	 struct stat st;
	 if&#40;fstat&#40;fileno&#40;in&#41;, &st&#41; == &#40;-1&#41;&#41;
	 &#123;
	 	perror&#40;"stat error"&#41;;
	 	exit&#40;EXIT_FAILURE&#41;;
	 &#125;	
     return st.st_size;
&#125;
/* argv&#91;1&#93; = FileA  argv&#91;2&#93; = FileB*/
int main&#40;int argc, char *argv&#91;1&#93;&#41;
&#123;
	char *buf=NULL;
	FILE *in=fopen&#40;argv&#91;1&#93;,"r"&#41;;    /* open afile for read */
	FILE *in1=fopen&#40;argv&#91;2&#93;,"r"&#41;;
	size_t filebytes=0;
	char record&#91;256&#93;=&#123;0x0&#125;;

	ck&#40;in&#41;;                       /* check file errors */
	ck&#40;in1&#41;;
	filebytes=file_size&#40;in&#41;;      /* get size of buf we need */
	if&#40;filebytes&#41;                 /* do we have a file with data in it? */
	&#123;
		ck&#40;buf=malloc&#40;filebytes+1&#41; &#41;;	/* create storage */
		memset&#40;buf,0x0, filebytes+1&#41;;   /* init storage */
		if&#40; readall&#40;fileno&#40;in&#41;,buf,filebytes&#41;>0 &#41; /* read entire file */
		&#123;
			ck&#40;fprintf&#40;stdout,"%s",buf&#41;&#41;; /* print whole file */
		&#125;
		else                         /* complain about errors reading file */
		&#123;
			ck&#40;fprintf&#40;stderr,"file read error\n"&#41; &#41;;
			exit&#40;EXIT_FAILURE&#41;;
	    &#125;
	    while&#40;fgets&#40;record,sizeof&#40;record&#41;,in&#41;!=NULL&#41;
	    &#123;
	    	if&#40;	strstr&#40;buf,record&#41;==NULL&#41; /* not found */
	    	&#123;
	    		ck&#40;fprintf&#40;stdout,"%s",record&#41;&#41;;
	    	&#125;
	    &#125;
	    free&#40;buf&#41;;                 /* release the buffer */
	&#125;
	if&#40;! fclose&#40;in&#41;	&#41;              /* close file with error check */
    &#123;
    	return 0;                  /* normal return */
    &#125;
	ck&#40;fprintf&#40;stderr, "filesystem error\n"&#41; &#41;;
	return EXIT_FAILURE;           /* file close error - return */
&#125;

[solman66]

sort -o A.srt A
sort -o B.srt B
comm -1 A.srt B.srt   > FileA_minus_FileB

This leaves me with FileA_minus_FileB = B.srt....
not saying i don't believe you, just saying it doesn't work for me.
Thanks for the C code too... may have to try it out someday...
As it turns out i just wrote a perl script to:
read file A, put the full contents into a single string, using a special character string to represent newlines
read file B, put full contents into a single string, using the same special character string for newlines
use a regex replace of string_of_A compared to string_of_B
write file C using regex replace newline_string with newline in string_of_A

[Loridan]

alines = Array.new
clines = Array.new
#store each line in file a to an array
File.open(/path/filea).each{|line| alines.push(line)}
#check each line of file b against all lines in a
File.open(/path/fileb).each{|line|
if not alines.include?(line)
clines.push(line)
end
}
#write the resulting lines into file c
filec = File.new(/path/filec)
clines.each{|line| filec.puts(line)}

I hope that you find the code interesting...