Re: upper/lower case

here the first sed version for this topic, slightly modified.

Even with one single sed invocation, it takes about 5 times longer than
the puresed.sh

$ wc -l test.csv
36634 test.csv

$ time ./puresed.sh test.csv 3
.361014737

real    0m0.368s
user    0m0.803s
sys    0m0.069s

$ time ./only.sed.sh  test.csv 3
1.526779118

real    0m1.537s
user    0m1.494s
sys    0m0.044s

only.sed.sh
#!/bin/bash

COL=$2
((COL--))

sed -r '
 h
 s/^([^,]*,){'"$COL"'}([^,]*),.*/\2/
 s/[A-Z]+/\L\u\0/gi
 H
 x
 s/^(([^,]*,){'"$COL"'})[^,]*(,.*)\n(.*)/\1\4\3/
' $1 >fff


On 12/12/15 23:48, Daniel Goldman [email protected] [sed-users] wrote:
> So another variation! puresed, I like the name, is faster. However, I 
> would not call it "remarkable". 2:1 (what I see, based on multiple 
> runs), or even 4:1 (close to what you saw) is often not, IMO, because 
> the absolute difference is so small. I would suggest speed is often 
> overrated as a criterion. In this case, it seems not important:
>
> $ wc big.txt
>   100005  300015 6200301 big.txt
>
> $ time process.sh big.txt 3
> real    0m0.416s
> user    0m0.392s
> sys     0m0.008s
>
> $ time puresed.sh big.txt 3
> real    0m0.293s
> user    0m0.396s
> sys     0m0.020s
>
> IMO, the pedestrian cut-paste solution is better, not because it was my 
> suggestion, I'm here to learn and share, but because it works, it's 
> obvious what happens at each step, and it's fast. I wish it was shorter, 
> but we can't have everything.
>
> Daniel
>
> On 12/12/2015 12:53 PM, Thierry Blanc [email protected] [sed-users] 
> wrote:
>> this is a parameter driven pure sed script (apart from the oxymoron ;-) ).
>> The speed is remarkable. See below
>>
>> puresed.sh
>> #!/bin/bash
>>
>> COL=$2
>> ((COL--))
>>
>> MAXCOL=$(sed -n 's|,|\n|gp;q' $1 |wc -l)
>> ((MAXCOL--))
>>
>> let "COL2 = $MAXCOL - $COL"
>>
>> n1="$(printf "%${COL}s" |sed 's| |n;|g')"
>> n2="$(printf "%${COL2}s" |sed 's| |n;|g')"
>> N="$(printf "%${MAXCOL}s" |sed 's| |N;|g')"
>>
>> sed  's|,|\n|g' $1 |sed -r  "$n1;s|[A-Z]+|\L\u\0|g;$n2" |\
>> sed "$N;s|\n|,|g" > output
>>
>> ---
>>
>> process.sh
>> #!/bin/sh
>>
>> col_to_sed=3
>>
>> prev_col=`expr $col_to_sed - 1`
>> next_col=`expr $col_to_sed + 1`
>>
>> cut -d \;  -f -$prev_col     < $1 > cols_before.txt
>> cut -d \;  -f  $col_to_sed < $1 > cols_.txt
>> cut -d \;  -f  $next_col-    < $1 > cols_after.txt
>>
>> sed -e '1b;s/.*/\L&/;s/\<./\u&/g' < cols_.txt > cols_new.txt
>>
>> paste -d\;  cols_before.txt cols_new.txt cols_after.txt > output.csv
>> ---
>>
>>
>> $ ./puresed.sh test.csv 3
>> .354332867
>> $ ./puresed.sh test.csv 3
>> .347485474
>> $ ./puresed.sh test.csv 3
>> .345228981
>> $ ./puresed.sh test.csv 3
>> .398209122
>> $ ./process.sh test.csv 3
>> 1.250124104
>> $ ./process.sh test.csv 3
>> 1.242909844
>> $ ./process.sh test.csv 3
>> 1.236377148
>> $ ./process.sh test.csv 3
>> 1.240610077
>>
>> On 12/12/15 19:19, Daniel Goldman [email protected] [sed-users] wrote:
>>> Thanks for doing the speed test. Of course, another disadvantage of the
>>> while-loop script is that it hard codes (a b c d). To my understanding,
>>> the cut-paste script can be totally parameter-driven.
>>>
>>> Just double-checking: The post shows IFS=',' used for while-loop script.
>>> Was intended IFS=';' used in actual speed test?
>>>
>>> Daniel
>>>
>>> On 12/12/2015 12:57 AM, Thierry Blanc [sed-users] wrote:
>>>> speed test
>>>>
>>>> My concerns about a shell while loop were correct. The while-do loop is
>>>> much slower.
>>>>
>>>> 1.079455912 vs. .221387482
>>>>
>>>>
>>>>
>>>> while-loop script:
>>>>
>>>>       #!/bin/sh
>>>>
>>>>       START=$(date +%s.%N)
>>>>       while IFS=',' read a b c d
>>>>       do
>>>>       c="$(echo "$c" | sed -r 's/(.*)/\L\1/; s/\<(.)/\u\1/g' )"
>>>>       echo  "$a ; $b ;$c ; $d"
>>>>
>>>>       done <  $1 > $2
>>>>
>>>>       END=$(date +%s.%N)
>>>>       echo "$END - $START" | bc
>>>>
>>>>       #endofscript
>>>>
>>>> cut-paste script:
>>>>
>>>>       #!/bin/sh
>>>>
>>>>       START=$(date +%s.%N)
>>>>
>>>>       col_to_sed=3
>>>>
>>>>       prev_col=`expr $col_to_sed - 1`
>>>>       next_col=`expr $col_to_sed + 1`
>>>>
>>>>       cut -d \;  -f -$prev_col     < $1 > cols_before.txt
>>>>       cut -d \;  -f  $col_to_sed < $1 > cols_.txt
>>>>       cut -d \;  -f  $next_col-    < $1 > cols_after.txt
>>>>
>>>>       sed -e '1b;s/.*/\L&/;s/\<./\u&/g' < cols_.txt > cols_new.txt
>>>>
>>>>       paste -d\;  cols_before.txt cols_new.txt cols_after.txt > output.csv
>>>>
>>>>       END=$(date +%s.%N)
>>>>       echo "$END - $START" | bc
>>>>
>>>> #endofscript
>>>>
>>>>
>>> ------------------------------------
>>> Posted by: Daniel Goldman <[email protected]>
>>> ------------------------------------
>>>
>>
>>
>> ------------------------------------
>>
>> ------------------------------------
>>
>
> ------------------------------------
>
> ------------------------------------
>



------------------------------------

------------------------------------

-- 

------------------------------------

Yahoo Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/sed-users/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/sed-users/join
    (Yahoo! ID required)

<*> To change settings via email:
    [email protected] 
    [email protected]

<*> To unsubscribe from this group, send an email to:
    [email protected]

<*> Your use of Yahoo Groups is subject to:
    https://info.yahoo.com/legal/us/yahoo/utos/terms/