Shell Programming and Scripting

Hi, all,
I'm from the 8-bit micro days (Z-80, 6809, 6502, etc.) and used to program in assembly and machine code. But, that was 25 years ago and life happened. Now, I'm scripting for the hackintosh community and love every bit of it.

I'm parsing a DSDT compiler log for error/warning/remarks checking and would like to be able to replace the line number with spaces, because I need to retain the spacing. The positioning of the text is critical, as it points to the errors, as you'll see in the following example. The line number follows the dsdt_fixed.txt path. See the following random examples:

Code:

Debug/dsdt_fixed.txt   140:         Name (_T_0, Zero)
Remark   5111 -                              ^ Use of compiler reserved name (_T_0)

Debug/dsdt_fixed.txt 10965:             Acquire (MUTE, 0x03E8)
Warning  1105 -        Possible operator timeout is ignored ^ 

Debug/dsdt_fixed.txt 12151:             Store (FGET (Local0), Local1)
Warning  1093 -                                   ^ Called method may not always return a value

I'm removing the line number so I can do comparisons after DSDT editing. The editing may add/remove lines, so the line number will be pointing to different locations. I have all this figured out, but just think there's got to be a simpler way to accomplish this task.

A simple sed 's/[0-9]/ /g' won't work, as I may have digits in the rest of the line. Is there a way to limit the search/replace to just that line number or area of text, ignoring the rest of the DSDT code that follows on that line?

Here's the details:
"Debug/dsdt_fixed.txt" is a predictable string.
Line number appears to be in a fixed, flush right location, with padded spaces at left. (The compiler may determine the maximum number of lines in the code and set the number of spaces in advance.)

I was doing this in bash by breaking up the line, which works fairly well, but was relatively slow.

Code:

parse_dsdt_log(){
CM_LST=""
# Grep "Warning" and previous line; add "@" at beginning of line; strip newlines
# Replace "@" with newline; remove "--" divider added by grep -B1 command
CM_LST2=`echo "$TEMP" | grep -B1 "^${PARSE}" | sed 's/^'"$DSDT_PATH"'/@&/g' | tr -d '\n' | tr '@' '\n' | sed 's/--$//g'`

while read LINE; do

# Extract file path and line number
	NUM=`echo "$LINE" | sed 's/\('"$DSDT_PATH"' [0-9]*:\).*$/\1/'`

# Calculate number of characters in number and replace with spaces
	SPC=$( printf "%${#NUM}s" )
	SPC="${SPC// / }"
	LINE=`echo "$LINE" | sed 's/'"$DSDT_PATH"' [0-9]*: \(.*$\)/'"$SPC"' \1/'`

# Merge back lines
	CM_LST="$CM_LST
$LINE"

done<<<"$CM_LST2"
}

PARSE="Warning"
DSDT_PATH="Debug.*"
parse_dsdt_log

Then, I thought of this idea. It works fast fine, but seems amateurish:

Code:

PARSE="Warning"
PTH="Debug\/dsdt_fixed.txt "

CM_LST=`echo "$TEMP" | grep -B1 "^${PARSE}" | sed 's/^\('"$PTH"'\).:/\1  /;s/^\('"$PTH"'\)..:/\1   /'`
CM_LST=`echo "$CM_LST" | sed 's/^\('"$PTH"'\)...:/\1    /;s/^\('"$PTH"'\)....:/\1     /;s/^\('"$PTH"'\).....:/\1      /'`

EDIT: Okay, I need to change the order of the commands so that sed replaces the largest number first and works on down. Whoops.

Code:

PARSE="Warning"
PTH="Debug\/dsdt_fixed.txt "

CM_LST=`echo "$TEMP" | grep -B1 "^${PARSE}" | sed 's/^\('"$PTH"'\)[0-9][0-9][0-9][0-9][0-9]:/\1      /;s/^\('"$PTH"'\) [0-9][0-9][0-9][0-9]:/\1      /'`
CM_LST=`echo "$CM_LST" | sed 's/^\('"$PTH"'\)  [0-9][0-9][0-9]:/\1      /;s/^\('"$PTH"'\)   [0-9][0-9]:/\1      /;s/^\('"$PTH"'\)    [0-9]:/\1      /'`

THIRD EDIT <sigh>: Okay, I had to add the padding in front of the numbering, as well as the [0-9] regex, instead of "." Works correctly, now.
Surely, there's a better idea, right? Nothing wrong with it, except it just seems verbose. However, I'm sticking with the code, until I get a better solution.

best of wishes,
MAJ

---------- Post updated at 07:02 PM ---------- Previous update was at 06:04 PM ----------

While I'm still talking to myself , it seems the extended regex works, too.

Code:

PARSE="Warning"
PTH="Debug\/dsdt_fixed.txt "

CM_LST=`echo "$TEMP" | sed 's/^\('"$PTH"'\)[0-9]\{5\}:/\1      /;s/^\('"$PTH"'\) [0-9]\{4\}:/\1      /'`
CM_LST=`echo "$CM_LST" | sed 's/^\('"$PTH"'\)  [0-9]\{3\}:/\1      /;s/^\('"$PTH"'\)   [0-9]\{2\}:/\1      /;s/^\('"$PTH"'\)    [0-9]:/\1      /'`

Thanks for letting me think out loud. I was working on this off and on for the past two days. Nothing like stalling on something seemingly "simple." Going to bed, then thinking, "Oh, I got it!" Get up and find out, nope, that doesn't work. Go back to bed. Rinse, lather, repeat.

MAJ

EDIT: Sorry to litter this place up. But, I think was making things a bit more complicated that necessary. Once I determine the number of spaces set by the compiler, it's a simple line of code:

Code:

PARSE="Warning"
PTH="Debug\/dsdt_fixed.txt "

CM_LST=`echo "$CM_LST2" | sed 's/^\('"$PTH"'\)....../\1      /'`