Hi Friends,
I am streaming fov, pitch, yaw, roll, position and other parameters in the form of ascii strings, from the GoProVr Video Player. I am parsing the strings using a series of [route] which then get fed into [float2ascii] and [json-decode]. This works reasonably well until I run into situations where a string changes from let's say 2 characters to 3 or more. When this happens I need to use [unpack] for the max number of characters. The independent characters are combined to a single number using various [expr], while [spigot] and [switch] allow me to choose a different [expr] and number atom, based on the number of characters being generated.

The problem comes when I try to move on from one parameter (fov) to let's say (pitch), because I can no longer serialize using [route], as the values can not be known in advance. I wouldn't mind being able to truncate the pitch and yaw values, as well.

Here's an some example data using [bytes2any]:
print: "fov":90
print: "id":"ked"
print: "pitch":0.036522697657346725
print: "position":45600
print: "roll":0
print: "state":2
print: "url":"file:///Users/laptopolist/Desktop/UltraFlight.mp4"
print: "yaw":-0.09542924165725708

I've also included an abstraction that demonstrates what I'm doing.
fov-example.pd

Thanks, Matt

Okay, I just realized I can replace the [pd fov_decode] subpatch with [bytes2any]. That cleans things up a fair bit, but doesn't solve the problem of needing to continue serializing [route]. It seems logical to me that there would be a way of going directly from [udpreceive] to the parameter I want to decode, instead of needing to cascade from fov to pitch to yaw, etc... Is there an object I'm not familiar with that will allow this?

Matt

You could the data into a list using [zexy/symbol2list], using ": as the separator:

This should give you a list that you can route.

Another option would be to convert the symbol to ASCII as you have done, drip the list routing out undesired characters, collect it and then convert it to a symbol again:

You might also (if in Linux) trying using the [shell] object. Because them you have access to sed, awk, etc. which can do pretty much Anything with strings or character lists. Also, they are very easy to use and the resources to learn how to use them extremely plentiful.

Good luck.

@svanya [shell] works on OS X too.

Hi There,

I've made significant progress with the patch, using my initial concept of connecting [route] serially. I have a few problems still to address.

I'm currently connecting sub patches for each parameter serially, which works but is less than ideal.

• Using this approach means that the URL sub patch will need to be manually redone for each video, on each computer.
• Data stops flowing past the pitch parameter whenever it's at a -0 value. -1, -2, -3, and positive values are all okay, but not -0.
• The position parameter is counting in ms., which cause parameters down the line to stutter.

The fix for all of this would be to access each parameter directly from [udpreceive]. I've tried a few of the list-abs objects, but have not gotten anything to work yet.

Attached is an overview of the patch and the fov sub patch.

@mattm Hello Matt...... I probably don't understand exactly the message string that is being received, but if I do then in extended I would parse it like this (as I think the curly braces might be "dropped" or you should be able to remove them with bytes2any/any2bytes and [list-filter]).......
There is a way to replace [cxc_split] in extended and vanilla, but my neurons that held that info are sleeping at the moment........ I will edit this post when they wake up..... parse_custom.pd
Can you post a complete message so that we can see?
David.

@whale-av Hi David,

I see where you're going, and I'm trying something similar parsing with [list split], and using [list length] of the entire message, subtracted by the [list length] after each parameter. It sorta works, but I'm running into issues.

Anyway, as requested, here is a complete message. I've gone ahead and parsed it, to make it easier on the eyes.

print:
123
34 102 111 118 34 58 56 52 44
34 105 100 34 58 34 107 101 100 34 44
34 112 105 116 99 104 34 58 48 44
34 112 111 115 105 116 105 111 110 34 58 57 51 52 51 44
34 114 111 108 108 34 58 48 44
34 115 116 97 116 101 34 58 50 44
34 117 114 108 34 58 34 102 105 108 101 58 47 47 47 85 115 101 114 115 47 108 97 112 116 111 112 111 108 105 115 116 47 68 101 115 107 116 111 112 47 85 108 116 114 97 70 108 105 103 104 116 46 109 112 52 34 44
34 121 97 119 34 58 48
125

Thanks, Matt

@mattm Hello Matt.......
Does this help.......... parse_custom_maybe.pd
David.

@whale-av Hi David,
Wow, so clean and efficient! It works wonderfully except for one problem, the patch doesn't seem to like negative values for "pitch" and 'yaw". I notice a forward slash in the console, at the end of the "pitch" and yaw' values. I'm assuming that it's truncating the decimal places at a certain point.

Here's a complete message, with negative values:
34 102 111 118 34 58 57 48 44 34 105 100 34 58 34 107 101 100 34 44 34 112 105 116 99 104 34 58 45 49 46 49 51 49 56 53 53 54 48 55 48 51 50 55 55 53 57 44 34 112 111 115 105 116 105 111 110 34 58 49 50 53 50 56 57 44 34 114 111 108 108 34 58 48 44 34 115 116 97 116 101 34 58 50 44 34 117 114 108 34 58 34 102 105 108 101 58 47 47 47 85 115 101 114 115 47 108 97 112 116 111 112 111 108 105 115 116 47 68 101 115 107 116 111 112 47 86 101 103 97 115 46 109 112 52 34 44 34 121 97 119 34 58 45 50 46 51 49 51 57 50 55 49 55 51 54 49 52 53 48 50

Thank you for your time with this!

Regards, Matt

@mattm Hello Matt.....
Check this for positive numbers as well please........
It is a rather poor solution, and it throws errors to the terminal, but should do the job. I will post again if I can clean it up..... but I am not sure that it is possible...
I might be able to filter the messages to the terminal though.
parse_custom_neg.pd
David.

@whale-av This is really terrific. I did not know about cxc_split. I also see the value of placing an enitre string in a message for programming and debug purposes. I'll let you know if I figure out why that error is bring logged in the console.

Thanks again,

Matt

For those of you who like pain, or know way more about this object than i do: the following works GREAT in the real shell yet in the [shell] object, due to its limitations (or mine :---) (as I know them) for instance not being able to take ";" it will not work in pd.

(I had thought i could just pass the string to the script and it would take it as a variable to parse)

So if you choose to pursue this, do please, post back. I must move on to other things.

str="34 102 111 118 34 58 57 48 44 34 105 100 34 58 34 107 101 100 34 44 34 112 105 116 99 104 34 58 45 49 46 49 51 49 56 53 53 54 48 55 48 51 50 55 55 53 57 44 34 112 111 115 105 116 105 111 110 34 58 49 50 53 50 56 57 44 34 114 111 108 108 34 58 48 44 34 115 116 97 116 101 34 58 50 44 34 117 114 108 34 58 34 102 105 108 101 58 47 47 47 85 115 101 114 115 47 108 97 112 116 111 112 111 108 105 115 116 47 68 101 115 107 116 111 112 47 86 101 103 97 115 46 109 112 52 34 44 34 121 97 119 34 58 45 50 46 51 49 51 57 50 55 49 55 51 54 49 52 53 48 50"

echo "$str" | awk '{ for(i=1;i<=NF;i++) printf("%c",$i); print "";  }' ```

fyi: I tried creating a script and just pass it to the script but that failed due to how it wants to pass/absorb its $-variables.```
Insert Code Here

Couldn't keep away. Figure I am the glutton for punishment I was talking about.

But did get it worked out:

Steps:

Put the code below as script named "asc2txt" in /root with executable permissions (of course).

The attached patched (borrowing from @whale-av's) includes how you put it to use.

parse_custom_neg_sv.pd

image just in case below

#! /bin/bash
var=""
while read line
do
    name=$(printf "\x$(printf %x $line)") 
    var=$var$name
done < <(cat /tmp/test.txt | tr ' ' '\n')
echo "$var" | sed 's/:/ /g' 

also as a footnote I think there is a way (I think I've even used it before) to dump a list directly into a textfile instead of having to write the line, but can't remember it right now).

Ciao

@mattm Hello Matt,
The errors are because the string 0.1234 is seen as a float, and I have connected that directly to the number boxes for pitch and yaw..... no error..... but it arrives at the [* -1] as a symbol which throws the error. The string -0.1234 has its "-" stripped and arrives as a float at the [* -1] so no error...... but it arrives at the same number boxes as a symbol at the same time....... and they complain.
So the end result is that the correct number arrives, but always arrives at the same time by another route as a symbol.... and throws an error to the terminal. Fortunately for the patch the erroneous symbols are simply dumped...... so the patch works...... but it is a horrible hack!
David.