change texts in a range using sed 3

[huangwason]

Hello, guys, I am thinking about a script that can change element contents of a specified tag in xml file. For example:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE TS>
<TS version="2.0" language="en_GB">
<context>
<name>ezono::ccapp::bup::BiosUpgrade</name>
<source>Failed to update BIOS.</source>
<translation type="unfinished">huang </translation>
</message>
</context>
<context>
<name>ezono::ccapp::bup::BiosUpgrade</name>
<source>Failed to update BIOS.</source>
<translation> here is a test
</translation>
</context>
</TS>

I will change the element contents between the tag <translation> from lower case to upper case.

I intend to use Sed

sed -e '/<translation*>/ s/\(.*\)/\U\1/' -e '/<\/translation>/,$ s/\(.*\)/\U\1/' input.xml

some problems:
1) it changes the tag to upper case as well(<TRANSLATION>), how to figure it out only for those texts between tags
2) I use range by pattern of sed here
sed '/start/,/stop/ s/regular express/replaced string/'
the "start" partern I used is "<translation*>" which can handle tag begin with <translation>, can not handle tag such as <translation type="unfinished">
3) the above script is line oriented, that means it work if <translation> and </translation> in one line, but sometime, the begin and end of a tag are not always in one line.

How to solve the problem? Is Sed really good to solve this problem?

 

[PHV]

A starting point:

awk -F'>' 'BEGIN{RS="<"}/^translation/{$2=">"toupper($2)}NR>1{printf "<%s",$0}' input.xml

Hope This Helps, PH.
FAQ219-2884
FAQ181-2886

 

[huangwason]

Thank you very much PHV, it works fine. Here you use "toupper" function to change lower case to upper case, I used some ascii codes in the XML file, when change to upper case, these ascii codes should reserve, can you propose how to reserve ascii codes in this case?

 

[feherke]

Hi

I used some ascii codes in the XML file

ASCII codes ? Are you sure those are not character entities ? That would be logic in case of text stored in XML.

Anyway, please post some example.

Feherke.

http://free.rootshell.be/~feherke/

 

[huangwason]

Yes feherke, they are character entity, such like

<translation > THE MARKED PATIENT RECORD &APOS;%1&APOS; WAS NOT READABLE. </translation>

in this case, &APOS shouldn't be changed

 

[feherke]

Hi

Then instead of [tt]toupper()[/tt]ing the whole $2 field :
[ul]
[li][tt]split()[/tt] it up on /&/[/li]
[li]make the first piece [tt]toupper()[/tt][/li]
[li]loop over the rest of the pieces[ul]
[li]find the [tt]index()[/tt] of ";"[/li]
[li]copy the [tt]substr()[/tt] before the ";" unmodified[/li]
[li]make the [tt]substr()[/tt] after the ";" [tt]toupper()[/tt][/li][/ul][/li]
[/ul]
At least in my approach.

Feherke.

http://free.rootshell.be/~feherke/

 

[PHV]

awk -F'>' 'BEGIN{RS="<"}
/^translation/{$2=">"toupper($2);while(match($2,/&[A-Z]*;/)){x=substr($2,RSTART,RLENGTH);gsub(x,"\\"tolower(x),$2)}}
NR>1{printf "<%s",$0}
' input.xml

Hope This Helps, PH.
FAQ219-2884
FAQ181-2886

 

[huangwason]

Thanks your guys, I tired PHV's scripts, it works for most of my XML files, but

<message>
<location filename="../code/a_mes/ccapp_mes_qt_measurementapp.cpp" line="1824"/>
<source>Patients starting...</source>
<translation>Abriendo el &quot; Registro de Pacientes&quot; ...</translation>
</message>

 

[feherke]

Hi

It works for me if I change [tt][A-Z][/tt] with [tt][[:upper:]][/tt].

But I would not do it. It transforms the whole sting into uppercase, then transforms all character entities into lowercase. But character entities are case sensitive. There is a difference, for example in case of a-umlaut : &Auml; is Ä and &auml; is ä. But after the two transformations both will be lowercase.


Feherke.

http://free.rootshell.be/~feherke/

 

[huangwason]

thank you