bin/stdc: Improve output formatting

Remove leading whitespace and collapse multi-line declarations into a single line using (g)awk.

Signed-off-by Oskari Pirhonen <[email protected]>

---

Here's a snippet for comparison. I've manually removed some of the formatting repeats from C99 and C11 output for the sake of making this message shorter.

Original:

    $ ./stdc c89 '[[:alpha:]]*scanf'
             int fscanf(FILE *stream, const char *format, ...);
             int scanf(const char *format, ...);
             int sscanf(const char *s, const char *format, ...);

    $ ./stdc c99 '[[:alpha:]]*scanf'
          int fscanf(FILE * restrict stream,
               const char * restrict format, ...);
          int scanf(const char * restrict format, ...);
            int fwscanf(FILE * restrict stream,
                 const wchar_t * restrict format, ...);
            int wscanf(const wchar_t * restrict format, ...);

    $ ./stdc c11 '[[:alpha:]]*scanf'
          int fscanf(FILE * restrict stream,
               const char * restrict format, ...);
          int scanf(const char * restrict format, ...);
            int vsscanf(const char * restrict s,
                 const char * restrict format, va_list arg);

New:

    $ ./stdc c89 '[[:alpha:]]*scanf'
    int fscanf(FILE *stream, const char *format, ...);
    int scanf(const char *format, ...);
    int sscanf(const char *s, const char *format, ...);

    $ ./stdc c99 '[[:alpha:]]*scanf'
    int fscanf(FILE * restrict stream, const char * restrict format, ...);
    int scanf(const char * restrict format, ...);
    int fwscanf(FILE * restrict stream, const wchar_t * restrict format, ...);
    int wscanf(const wchar_t * restrict format, ...);

    $ ./stdc c11 '[[:alpha:]]*scanf'
    int fscanf(FILE * restrict stream, const char * restrict format, ...);
    int scanf(const char * restrict format, ...);
    int vsscanf(const char * restrict s, const char * restrict format, va_list arg);

 bin/stdc | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

bin/stdc +8 -1

index b685df8..8c07956 100755
@@ -14,7 +14,14 @@ err()
 
 grep_proto()
 {
-	pcre2grep -M "(?s)\b$1 *\([[:alnum:]*,._\s\(\)-]*\);$";
+	pcre2grep -M "(?s)\b$1 *\([[:alnum:]*,._\s\(\)-]*\);$" \
+	| awk -e 'BEGIN { RS=";\n"; ORS=RS; }
+		{
+			gsub(/\n/, " ");
+			sub(/^ +/, "");
+			gsub(/ +/, " ");
+			print;
+		}';
 }
 
 libc_summ_c89()

Hi Oskari,

Quoted 2 lines · Oskari Pirhonen

Remove leading whitespace and collapse multi-line declarations into a single line using (g)awk.

I can't reak awk(1) :(

But I like the idea. I implemented the same using sed(1) after your suggestion. Does the below patch look good to you?

Cheers,
Alex

P.S.: I forgot about writing a man page. I'll start now.

Hi Alex,

Quoted 2 lines · Alejandro Colomar

Hi Oskari,

Quoted 6 lines · Oskari Pirhonen

Quoted 2 lines

Remove leading whitespace and collapse multi-line declarations into a single line using (g)awk.

I can't reak awk(1) :(

Awww man, but I even left the optional semicolons in...

Quoted 4 lines

But I like the idea. I implemented the same using sed(1) after your suggestion. Does the below patch look good to you?

I actually had an earlier version with sed(1), but it used looping/branching to handle the multi-line bits, so I figured it was a bit ugly and didn't send it. I didn't think to try -z.

It seems to do the same thing, so LGTM.

Quoted 6 lines

Cheers,
Alex

P.S.: I forgot about writing a man page. I'll start now.

I was about to say "and license file and appropriate blurb" but then I saw your commit. I've got some suggestions for the man page, so I'll send some patches sometime soon.

- Oskari

Since you said you can't read awk, then just to satisfy your curiosity,

here's what was going on:

    BEGIN {
        RS = ";\n"
        ORS = RS
    }

This block is run at the start before any records are processed.

The default Record Separator is "\n", but here we set it to ";\n". In (g)awk, a value of RS that is >1 char is actually a regex, but we only need to match a literal string. The Output Record Separator is by default also "\n".

    {
        gsub(/\n/, " ")
        sub(/^ +/, "")
        gsub(/ +/, " ")
        print
    }

This block is run on all records, since it doesn't have any patterns for contitional execution attached to it.

gsub() does an in-place global regex replace on a string, similar to the s/regex/replace/g you're familiar with. It takes an optional third arg, but if it's left out then it has an implicit $0 which means the entire current record. sub() is like gsub(), but only does the first match, similar to s/regex/replace/.

print, without any args, prints the current record followed by ORS. This was set earlier because the RS is consumed from the input as it is being processed record by record, but we want to keep the output looking intact.

Hopefully not so bad after all.

Awk is pretty nice IMO (and gawk in
particular), and I would recommend checking it out if you find yourself borded one day :)