[Mplayer-cvslog] CVS: main/help help_diff.sh,1.6,1.7

[Diego Biurrun]

Diego Biurrun CVS schrieb:
> Modified Files:
> 	help_diff.sh 
> Log Message:
> shorter, sweeter, more portable, better calling syntax
> 
> --- help_diff.sh	3 Jan 2003 12:14:34 -0000	1.6
> +++ help_diff.sh	17 Aug 2003 01:20:48 -0000	1.7
> @@ -1,28 +1,19 @@
>  #!/bin/sh
>  
> -# This script walks through the master (stdin) help/message file, and
> -# prints (stdout) only those messages which are missing from the help
> -# file given as parameter ($1).
> +# This script compares a translated help file to the master file and prints
> +# out any missing messages.  Needs the language code as parameter ($1).
>  #
> -# Example: help_diff.sh help_mp-hu.h < help_mp-en.h > missing.h
> +# Example: help_diff.sh hu
>  
> -curr=""
> +exec < help_mp-en.h
>  
> -while read -r line; do
> -	if echo "$line" | grep -q '^#define'; then
> +while read line; do
> +	if echo "$line" | grep '^#define' > /dev/null 2>&1; then
>  		curr=`echo "$line" | cut -d ' ' -f 2`
> -		if grep -q "^#define $curr " $1; then
> -			curr=""
> +		if grep "^#define $curr" help_mp-$1.h > /dev/null 2>&1; then
> +			true

: should be better here

> +		else
> +			echo "$line"
>  		fi
> -	else
> -		if [ -z "$line" ]; then
> -			curr=""
> -		fi
> -	fi
> -
> -	if [ -n "$curr" ]; then
> -		echo "$line"
>  	fi
>  done
> -
> -

Yes, it runs on Solaris ... ;-)

And even in the Cygwin ash... (it does not like read -r)

As Rich just noticed this does not exactly match the previous version of 
the script, since this one only outputs one line of each untranslated 
message, while the old one printed multiline untranslated messages to 
the screen.  I'm not completely sure if this is a disadvantage or not. 
Seeing the whole untranslated message is nice to directly translate it, 
of course, but you have to open the file for editing anyway and check in 
the master file at which point you have to insert the newly translated 
message..  So maybe some translator has an opinion on this, otherwise I 
am tempted to remain with the new version.

Another point is that this does not solve the general performance 
bottleneck of grepping through the whole master file for each message. 
Rich came up with some sed/regexp wizardry to solve this:

sed "`sed '/^#define *[^ ][^ ]* .*$/!d;s@^#define *\([^ ][^ ]*\) 
.*$@/^#define *\1 /d@' < help_mp-$1.h ; echo '/^#define/!d'`" < help_mp-en.h

This version is way faster and should probably committed instead if we 
stay with the new version.  Or maybe the sed voodoo can be extended to 
match the old behavior.  I'll see if I manage to do it and else prod 
Rich ;-)

Diego