print specific line from multiple files

Question

Here is the source code of a historical Unix goto.c: http://v6shell.org/history/goto.c

So we're trying to find a matching label recording the "goto label" call.

My problem:

if (getlin(line)) {
        write(1, "label not found\n", 16);
        return;
        }

So I would expect if getlin() returns true we should print out "label not found" and exit the program.

But look at this:

getlin(s)
char s[];
{
    int ch, i;

    i = 0;
l:
    if ((ch=getc())=='\0') return(1);
    if (ch!=':') {
        while(ch!='\n' && ch!='\0')
            ch = getc();
        goto l;
        }
    while ((ch=getc())==' ');
    while (ch!=' ' && ch!='\n' && ch!='\0') {
        s[i++] = ch;
        ch = getc();
        }
    while(ch != '\n')
        ch = getc();
    s[i] = '\0';
    return(0);
}

getlin() will return 0(true) if we found the label and 1(false) if we didn't.

But it should be the other way around or we would need to say:

if (**!** getlin(line)){}

What is wrong here?

Solution 1

 awk 'FNR==27 {print FILENAME, $0}' *.txt >output.txt

• FILENAME is built-in awk variable for current input file name
• FNR refer to line number of current file
• $0 means whole line

Solution 2

Another possibility is :

for i in * ; do echo -n $i" : "  ; head -n 27 "$i" | tail -n 1 ; done > output.txt

Solution 3

Using a for loop:

{ for i in *.txt; do echo "$i : $(sed -n '27p' "$i")"; done ;} >output.txt

The for loop may be expensive as you have 5000+ files but given the current hardwares should not be a problem.

---

Faster way, quitting sed after line 27 (thanks @Fiximan):

{ for i in *.txt; do echo "$i : $(sed -n '27p;q' "$i")"; done ;} >output.txt

Comments

[arkascha]

Not sure what this is about, but as a general rule of thumb: unix utilities traditionally return 0 (zero) to indicate success, everything else indicates an error (an error code). So I#d say the function implementation makes sense, the conditional is questionable.

[Joey]

Yeah that's the question. But it's working, so why does it work?

[FelixJN]

Would you need a '27p;q' to be faster or will sed quit automatically after hitting line 27?

[123]

@Fiximan yes that will be quicker

[heemayl]

@Fiximan That surely would..edited..thanks..

[Ravi Saini]

Thanx man this was much better for me.

[don_crissti]

@Fiximan - sed would quit so quick that it won't print anything...

[user1683793]

I set up a file with 500 million 100 byte records and for that case, this version works fastest for me at 1m2.582s followed by the sed version at 1m20.81s and awk at 2m14.756s

[user1485864]

Plus this is more flexible. If the logs g-zipped you can simply add in the loop the zcat command.