Hello,

short story: I need a way to take a .txt that has a float per line, like this:
1.45
1.23
5.67
4.23
...

and arrange it so they increase in value, like this:

1.23
1.45
4.23
5.67

How can I do this automagically and with ease?

thank you very much

long story:

I am building a patch to create and show subtitles for audio files.

I devised a simple system to help me manually log the exact time in s.ms when a /show next subtitle/ event should happen.

I do this with a simple timer abstraction; I start it, time runs, and then I keep pressing bang when I want to log a /show next subtitle/ event. I log the time in s.ms into a textfile object.

This way I obtain a .txt with floats that represent the /next sub/ events in s.ms
1.23
1.45
4.23
5.67

BUT if I miss an event, and I try to add that one without deleting the rest, I get something like this

1.23
1.45
4.23
5.67
2.34 <---- shouldn't be here but before 4.23

And this is why I am looking for a way to arrange the floats automagically, so I can easily go on adding forgotten /next sub/ events, and I only have to edit the file directly to delete them.

Thank you very much

Hi,

Do you have to do this in Pd or would it suit you to have a separate application which you could execute every time you want to order your text file? This type of thing is trivial to implement in most programming languages (e.g.: http://en.wikipedia.org/wiki/Bubble_sort ), but as far as I can see it won't be easy in Pd. Maybe I am completely wrong here, and I'd be glad if that's the case!

Take care,
Gilberto

Hi,

I thought a bit more about your problem, and I found a way of doing what you want. I created a subpatch (can be transformed into an abstraction if you plan to use it often) which goes through a text file and outputs the messages according to the correct time stamp. The format of the textfile should be <time in seconds> <subtitle>;:

0.00 first phrase, an example of subtitle;
1.37 second phrase;
2.99 another phrase;
120.00 yet another phrase, which will be displayed at the exact 2 minutes mark!;

Please only use two digits after the dot in the seconds (so the precision is 0.01 seconds, or 10 ms). If your text file contains something like 1.037, it will be ignored.

You can put the phrases all out of order and it still will work out. So this textfile below will output the same thing as the one above:

120.00 yet another phrase, which will be displayed at the exact 2 minutes mark!;
0.00 first phrase, an example of subtitle;
2.99 another phrase;
1.37 second phrase;

Please note that my patch does NOT fix the order of your input file, it simply outputs it in the correct order! So if you open your .txt file, the phrases will still be out of order.

Here is the patch: eventPlayer.pd
And here is a test textfile: text-file-example.txt

If you find any problems using it, just let me know. I hope this helps.

Take care,
Gilberto

You beat me to it again! jeje. Thank you very much Gilberto.

It would be great if you could explain a bit your patch. I think that I can adapt it to store again the arranged output into the original .txt

I myself was trying to understand and implement a sorting algorithm within PD (something I never did anywhere!)

I was attemtping to build a patch that did this:

get original list
compare second and first elements of the list
if second is bigger than first (2nd>1st=1) store them like that in a temp list
else (2nd>1st=0) swap their order
write them like that in the original list
repeat with third and second elements of list
write in original list
then fourth and third, etc
when reaching last position in list start over

do this until you get, within a single interation of the list, a consecutive ammount of "1"s in the ">" comparison that matches the length of the list (which would mean all items are in place)

I did not finish this, but I think it would be able to put the list in order. Although if lowest number is at the last position of the list, the patch will have to iterate through the whole list as many times as the length of the list, to bring the number to the first position.

This is because the patch only moves an item one position per iteration.

I can also foresee a possible problem. Since in every iteration the patch goes up one position in the list, but actually stores a pair of floats, it will probably have repeated numbers on the arranged list (i.e. 12233445) So I should have a filter when storing the result of the comparisons in the arranged list.

Intuition tells me it may be enough to just store the first number of each pair, in every comparison, and not both of them.

PD is fun !

abrazo

Hello again, I am starting to understand the patch.

One difference with my setup is that I do not have a txt file with the format "time + sub" I only have "time" which is used to fire subs in another txt file.

I create the txt with the times for each "next sub" event with the help of pd, but I create the txt with the subs manually.

In order to get a txt file in the format your patch needs I would need to build a patch that would merge the txt with times & the txt with subs... or do that manually.

So now I want to understand your patch so I can hack it to get the arranged times (and store them in a text file) instead of the subs associated to the arranged times.

Cheers!

Hi ferr,

The algorithm you described above is the bubble sort algorithm. I also tried to implement it, but things got very complicated very fast and that's why I tried a different solution.

You say you have a time.txt and a sub.txt. So you mean that the subs are in order, but the time.txt is not, right? In that case, you just need to adapt a little thing in that patch:

This way, it will output a bang every time a number of your time.txt matches the internal chronometer, then you can use this bang to move to the next line in another [msgfile] which has your subs.txt loaded. Does this make sense to you?

So you will need a file for the time, something like this:

0.00;
12.00;
5.45;
4.97;

And then another subs.txt like:

first subtitle;
second subtitle;
third subtitle;
etc...;

Let me know how it goes then.

PD is fun !

Ha, you tell me about it!

Take care,
Gilberto