| Pages in topic: [1 2] > |
| User | Thread poster: Nurzhan Nagashbekov Counting words in a txt file within quotation marks |
|---|
Nurzhan Nagashbekov 
Kazakhstan
Local time: 08:24
 Member (2008)
English to Kazakh
|
Hello fellow translators,
I have a txt file with software strings in it to be localized. It looks like:
#command some text "text to be localized" // comment
I want to count the words within quotation marks. Is there any way to do it, except manual counting?
I tried importing txt to MS Excel, but it seems the file is not correctly delimited. So, the words I need may appear on different columns.
Any help will be much appreciated.
| | | |
Philip Lees
Greece
Local time: 05:24
 Member (2008)
Greek to English
|
Give the file to somebody you know who uses the Perl programming language, and ask them to run this:
perl -i.bak -pe "s/^.+?\"//; s/\".+$//" yourfilename
That will remove everything from your file except the parts in quotes (the original file will be renamed as yourfilename.bak). You can then count the words in the new file.
This assumes that all the lines have the same format.
| | | |
Nurzhan Nagashbekov
Kazakhstan
Local time: 08:24
Member (2008)
English to Kazakh
TOPIC STARTER | | Some strings are different | Feb 9 |
Thanks Philip,
Unfortunately, some lines contain only comments and there are lines that contain #command... and "text" but no comments.
I was able to count the number of quotation marks in excel using countif function but it was useless, since there are lines with sentences in quotation marks.
| | | |
Amit Evron
Brazil
Local time: 23:24
 Partial member (2011)
Spanish to English
+ ...
|
If it's not confidential and if the file isn't too big, feel free to send it over and I'll write a quick perl script. Shouldn't take more than 5 minutes. Just send me a message through Proz and I'll reply with my e-mail address.
| | | |
Tony M
France
Local time: 04:24
Member
French to English
+ ...
|
Haven't tested it, but why not try this:
Select all your text and paste it into Word (etc.)
Do a 'replace all' on the " (careful to get the right character!), replacing with (say) Tab
Select all and convert text to table, using the character you replaced above (e.g. Tab) as the delimiter.
This should enable you to get a column that just has your text to be translated in, and you can take it from there
If you have any lines with no " " at all, they should just appear all in the first column.
Theoretically at least, you ought to be able to reverse the process at the end...
One proviso: one has to assume that each line does end with a Return character or similar; if necessary, you might need to go through and replace whatever the end-of-line delimiter is with something that will work in Word for the conversion to table.
| | | |
Philip Lees
Greece
Local time: 05:24
Member (2008)
Greek to English
|
Nurzhan Nagashbekov wrote:
Unfortunately, some lines contain only comments and there are lines that contain #command... and "text" but no comments.
|
|
I think my script should still work with a small modification (for the comment only lines), but as Amit has kindly offered to take it on I'm happy to hand over to him.
| | | |
Nurzhan Nagashbekov
Kazakhstan
Local time: 08:24
Member (2008)
English to Kazakh
TOPIC STARTER | | Thanks for suggestions! | Feb 9 |
Amit Evron wrote:
If it's not confidential and if the file isn't too big, feel free to send it over and I'll write a quick perl script. Shouldn't take more than 5 minutes. Just send me a message through Proz and I'll reply with my e-mail address. |
|
 It is confidential
| | | |
Nurzhan Nagashbekov
Kazakhstan
Local time: 08:24
Member (2008)
English to Kazakh
TOPIC STARTER |
Tony M wrote:
Haven't tested it, but why not try this:
Select all your text and paste it into Word (etc.)
Do a 'replace all' on the " ....
|
|
Thanks Tony, I will try your method.
| | | |
Jabberwock
Poland
Local time: 04:24
Member (2004)
English to Polish
|
I think the best option would be to use Okapi Rainbow, especially if you expect more such work form the client. Basically, it would allow you to extract the text you require (using regular expressions) and then calculate the wordcount.
Trados 2007 also has an option to import text based on regular expressions. You have to use a separate application Filter Settings for this. After the import you just analyze the resulting ttx file as usual.
I realize that having to learn regular expressions might seem daunting, but if you plan to translate such texts it will be a sensible investment of your time...
| | | |
FarkasAndras
Hungary
Local time: 04:24
English to Hungarian
+ ...
|
I fervently hope that you'll be using a CAT for this job. The localization of SW strings requires strict formatting consistency and there are a lot of repetitions etc., so it' really not the job you'd want to do by typing over the original.
Now, If you do use a CAT, just do the word count there.
Studio has the required capabilities (i.e. you can specify regex rules that separate the translatable text from the rest), and the Studio package also comes with a specialized sw localization tool (Passolo). Of course there are lots of other tools that'll work, too.
The more interesting question is: who is in charge of this project? Isn't there a PM/client who sorts these things out before you get involved?
| | | |
Philip Lees
Greece
Local time: 05:24
Member (2008)
Greek to English
|
I had a few minutes to spare, so I set up this:
http://quote.writewords.eu/
If you paste your text in the box and click Submit, it should return you only the stuff that's between quotes.
| | | |
FarkasAndras
Hungary
Local time: 04:24
English to Hungarian
+ ...
|
Philip Lees wrote:
Give the file to somebody you know who uses the Perl programming language, and ask them to run this:
perl -i.bak -pe "s/^.+?\"//; s/\".+$//" yourfilename
That will remove everything from your file except the parts in quotes (the original file will be renamed as yourfilename.bak). You can then count the words in the new file.
This assumes that all the lines have the same format.
|
|
It also assumes that there is only one pair of quotes in one line and that there are no escaped quotes inside quoted strings. It'll fail with lines like this:
StringID:4567267; text:"Press the \"Browse\" button to pick a file"; Button:"Browse"
And it doesn't skip lines that have no translatable content at all.
Also, .+? is better written as .* and the " may very well be the last character on the line so .+$// should be .*$//.
So, I'd rewrite your one-liner as:
perl -i.bak -pe "s/^.*\"(.*)\".*$/$1/" yourfilename
...but this still doesn't handle the problem cases I mentioned above.
You could do this (untested) to delete lines that don't contain any quoted string:
perl -i.bak -pe "next unless /\".*\"/; s/^.*\"(.*)\".*$/$1/" yourfilename
... but the bottom line is, it's still only usable if the input file is "simple". You could add negative lookahead/lookbehind to cater for escaped quotes inside the quoted strings etc. to make it work and then somehow adapt it for multiple strings per line, but it starts to get tricky there, and you need to see the input file (or know its spec) to take a reasonable stab at solving the problem.
[Edited at 2012-02-09 10:54 GMT]
|  |  |
| | | |
Nurzhan Nagashbekov
Kazakhstan
Local time: 08:24
Member (2008)
English to Kazakh
TOPIC STARTER | | Initial stage of the project | Feb 9 |
I am at the very beginning of the project and just wanted to know what is the wordcount for now. I will definitely try regex. Thanks!
| | | |
Philip Lees
Greece
Local time: 05:24
Member (2008)
Greek to English
|
FarkasAndras wrote:
It also assumes that there is only one pair of quotes in one line and that there are no escaped quotes inside quoted strings. It'll fail with lines like this:
StringID:4567267; text:"Press the \"Browse\" button to pick a file"; Button:"Browse"
|
|
Oh, sure, it breaks in lots of cases, as does the simpler match I used on the web version:
/"(.+?)"/
I am well aware of the pitfalls of text parsing, which is why I added the caveat about all lines having the same format as the example provided.
As this is not a Perl or a regex forum, I'll leave it at that.
| | | |
Ambrose Li
Canada
Local time: 22:24
Member (2011)
Chinese to English
+ ...
|
I think this single line of perl should suffice:
perl -nle 'print $1 if /#command\s+"([^"]*)"/'
This assumes that double quotation can’t occur inside the pair of double quotation marks that marks the string to be translated. Usually this is not the case and (assuming that " is escaped with a single backslash) the perl needed will more likely be
perl -nle 'print $1 if /#command\s+"((?:\\"|[^"])*)"/'
Of course, if escaping of quotation marks occurs but is not signalled by backslashes then the perl code needed will be different.
ETA: The above assumes that continuations don’t occur. If continuations do occur the above won’t work and one-liner solutions might not be sufficient…
[Edited at 2012-02-09 19:03 GMT]
| | | |
| Pages in topic: [1 2] > |
The translation workplace 
Translation news