How to change/convert the format of a timestamp?

PeterJones

You might want to search the forum for more about timestamps and/or captioning files (I am assuming this is an .srt or similar captioning file), because I know similar things have been asked.

I know there’s a conditional replace that @guy038 could bring to the party. I couldn’t get the syntax working correctly.
Without the conditionals, I can do it in n search/replace operations, where n depends on the possible minutes: if minutes is 60-99, then I can do it in 4 s/r (6x, 7x, 8x, 9x). If minutes can ever start at 100 or above, my method would need to increase n for every 10-minute group above the 9x.

assuming

that we only ever edit the minutes and append an hours segment when necessary (ie, don’t ever have to fix invalid seconds)
that 0:00.000 up to 59:59.999 will remain unchanged

The search/replace pairs (in regular expression mode) would be

(?<![:\d\.])(6)(\d{1}:\d{2}\.\d{3})
01:0$2
(?<![:\d\.])(7)(\d{1}:\d{2}\.\d{3})
01:1$2
(?<![:\d\.])(8)(\d{1}:\d{2}\.\d{3})
01:2$2
(?<![:\d\.])(9)(\d{1}:\d{2}\.\d{3})
01:3$2

If it goes higher, you can give it a try yourself, and paste what you tried if it didn’t work.

Honestly, if this were mine, I’d write up a Perl script to do it, completely outside of Notepad++. Others would choose Python or Lua. I could probably cobble it together in PythonScript, to have python make the changes, running inside Notepad++ editor window. But I don’t have time right now to do it, and I think I’ll wait until you’ve responded to the questions and clarifications, and maybe shown that you’ve tried to look for previous answers

Eko palypse

@Dana-Wright

sorry, not 100% clear for me - what is the format I should expect to convert.
I understand the result should be 01:18:14.269 --> 01:18:21.079
but which format is provided initially?
This 00:78:14.269 --> 00:78:21.079 or
that 78:14.269 --> 78:21.079?

Dana Wright

Sorry for the confusion. To hopefully clarify things, here’s an updated file with only timestamps that need to be converted. Please ignore the earlier file

Eko palypse

@Dana-Wright

I guess this python script should do the job

def change_format(m):
    parts = m.group(0).split(':')
    min = int(parts[0])
    if min > 60:
        real_minute = min % 60
        hours = min / 60
        return '{:02}:{:02}:{}'.format(hours, real_minute, parts[1])
    else:
        return '{}:{}'.format(*parts[:])
    
    
editor.rereplace('\d+:\d+\.\d+',change_format)

Dana Wright

@Eko-palypse thanks. I’ll give it a try.

Dana Wright

@Eko-palypse Worked like a charm! Thank you very much!

Now I need to analyze this and figure it out how it was done.

Eko palypse

@Dana-Wright

this is quite easy.
editor is a object exposed by pythonscript plugin which can manipulate the scintilla component used by notepad++.
rereplace is the method which allows to have the first parameter being a regular expression and
the second parameter being a function which is called for each match of the regex.

The function change_format gets the match object in variable m
As no regular expression matching group has been defined everything should be accessible in group 0.
parts = m.group(0).split(':')
split what has been reported in match object by colon and return a list in variable parts
min = int(parts[0])
take the first element in that list and convert it to an integer, save it in variable min
if min > 60:
if min greater than 60 we need to do something, if not else branch gets executed
real_minute = min % 60
real_minute will get the remainder of division by 60
hours = min / 60
hours the result of division by 60
return '{:02}:{:02}:{}'.format(hours, real_minute, parts[1])
return the new string.
:02 means that we want to have it in two digit format 1->01 but 10->10 …
That’s it.

Alan Kilborn

@Eko-palypse said:

That’s it.

Well…you didn’t explain the else part, and that’s where I have a question. I would do it this way and I don’t know why you did it differently:

return '{}:{}'.format(*parts)

I just had the thought that it would be nice if editor.rereplace() would take a replace function returning None as a signal to not do any replacement.

Eko palypse

you are absolutely correct - this survived the test when having hh::mm::ss::msec :-)
But python slicing is rescuing me :-D

I don’t understand the none return - I mean, why would you want to replace something with None
where None means don’t replace anything? Which case do you have in mind?

Alan Kilborn

@Eko-palypse

Maybe I misunderstood the intent of the return line I called out. I was thinking that it is simply putting back together the original text but I didn’t look at the OP’s data or problem description all that closely. I guess you would have returned m.group(0) if that were the case.

Anyway, a replace function could have some logic that in certain cases it would not change the original text. Returning None (or even, gasp, falling off the end of the function without returning anything) could be that signal. I certainly did not try editor.rereplace() in that manner, maybe it already works that way.

Eko palypse

@Alan-Kilborn

actually I’m trying to avoid function lookups as those are expensive, especially when it involves
Python->C->Python conversion. But I must admit, in this case I don’t think that I gain any performance improvement, it might be even slower. Let’s test it. Will come back.

Alan Kilborn

@Eko-palypse

Okay…so I didn’t follow any of that, but I look forward to the come back. :)

Eko palypse

@Alan-Kilborn

so it is still a little bit faster - to be honest, haven’t expected it.

looped 1000 times over the same text
16.7720000744 <-- return m.group(0)
16.6819999218 <- return ‘{}:{}’.format(*parts[:])

PeterJones

My two comments would be:

if min > 60: = if the data is 60:00.000, it wouldn’t change. Make it if min >=60:
If the OP (or someone else) has mixed data, or had partially changed them, and came back later and tried the same script, weird stuff will happen. I’d recommend: editor.rereplace('(?<![:\d])\d+:\d+\.\d+',change_format), which adds a negative lookbehind to not match if there’s a colon or another digit before the \d+

Eko palypse

@PeterJones

YES - this is a bug it should > 59 - omg.
About mixed data you are right but this is always the question what if it looked like
hh.mm.ss.msec …

PeterJones

@Eko-palypse ,

Indeed, there are always more formats that might exist. I’ve only seen colon-separated in .srt files, so I think that keeping it generic enough that it won’t mess up an existing .srt, even if it does have some with hours and some without.

BTW: I had forgotten why I included the [:\d] rather than just : in my negative lookbehind: without the \d in the character class, 1:15:00.000 (which shouldn’t match) would partially match on 5:00.000, which would be even worse.

And running a test with 1:15:00.000, even with your simpler expression, works correctly (ie, doesn’t try to change it) – ahh, that’s because the minutes are less than 60. I guess unless there’s a strange 1:65:00.000, yours won’t be a problem. I guess yours is generic enough.

Meta Chuh

@Eko-palypse

I’m trying to avoid function lookups as those are expensive …

yes, i’m a bit short on money too at the moment … and don’t even dare to give me an (s.h) for this comment 😉

Eko palypse

@Meta-Chuh

:-D - always reminds me of this

Meta Chuh

@Eko-palypse

singing: ahaaaa, ahahahaaa … all the things i could do … ;-)

Eko palypse

@Meta-Chuh

I don’t understand all of this but what I got makes me laughing … :-D