added parse

[?]

AEj8dahVWy718uSSFPe9VSRJ5qX5G8pC2zvFzJJ8yzBd

Jan 13, 2025, 7:24 PM

6XJX6ED3CDARYVCX5XCZMRE4UB3PHD2JAWB67DZQFEXV336VERZQC

Dependencies

[2] 5M7JW5OV new functiont in Labels.jl
[3] QA2TJZRA ripped ML out into the SkraakML repo, compiles quicker now
[4] 2UBDFCJH new files tracked
[5] E3Y55MPR added perceptual colour maps and changed get_image_from_sample
[6] 4BTZNCRM catch up commit before starting work
[*] NV7FXZ5Q first commit
[*] YODTMMPT added a whole bunch of functions to Labels.jl for working with raven and avians labels

Change contents

edit in src/Skraak.jl at line 11

[8.345651]

[8.356019]

include("Parse.jl")
file addition: Parse.jl (----------)

[8.6598]

#=
a = ["201012_123456.wav", "201014_123456.WAV", "201217_123456.wav", "211122_123456.WAV"]

b = ["121020_123456.WAV", "141020_123456.wav", "171220_123456.WAV", "221121_123456.wav"]

c = ["20230609_103000.WAV", "20241109_201504.wav"]

d = [
"120119_003002.wav",
"180120_231502.wav",
"170122_010005.wav",
"010419_234502.WAV",
"310320_231502.wav",
"220824_231502.WAV",
"240123_231502.wav",
]
e = ["XYZ123_7689_20230609_103000.WAV", "string 20241109_201504.wav"]
f = [
"abcdefg__1234_180120_231502.wav",
"string 120119_003002.wav",
"ABCD EFG___170122_010005.wav",
"BHD_1234 010419_234502.WAV",
"cill xyz 310320_231502.wav",
"220824_231502.WAV",
"240123_231502.wav",
]
=#

using Dates
#Assumes it's getting a folder of 1 recording session, not random files.
#For 6 digit dates the year digits need to have less variation than the day digits.
#Does not handle miss shapen dates filenamesBad = ["20230609_1030qa.WAV", "20241109_20wp04.wav"]
function date_time_of_fname(filenames::Vector{String})
pattern = r"(\d{6}|\d{8})_\d{6}"
f2 = [match(pattern, str).match for str in filenames]
#date_time format must be the same for whole vector
g = map(x -> length(x), f2) |> x -> unique(x)
@assert length(g) == 1 "Different date formats in vector" #all same length
@assert length(f2[1]) == 15 || length(f2[1]) == 13 "Wrong length to be a date_time"
raw_dt = map(x -> (split(x, "_")[1], split(x, "_")[2]), f2)
d = parse_date_strings(first.(raw_dt))
t = parse_time_strings(last.(raw_dt))
dt = map((x, y) -> (x..., y...), d, t)
return map(x -> DateTime(x...), dt)
end

function parse_time_strings(times)
h = map(x -> tryparse(Int64, x[1:2]), times)
m = map(x -> tryparse(Int64, x[3:4]), times)
s = map(x -> tryparse(Int64, x[5:6]), times)
return zip(h, m, s) |> collect
end

function parse_date_strings(dates)
if length(dates[1]) == 8
y = map(x -> tryparse(Int64, x[1:4]), dates)
m = map(x -> tryparse(Int64, x[5:6]), dates)
d = map(x -> tryparse(Int64, x[7:8]), dates)
elseif length(dates[1]) == 6
y, m, d = parse_short_date_strings(dates)
else
error("Date is not 8 or 6 digits long")
end
return zip(y, m, d) |> collect
end

# 6 digit dates
function parse_short_date_strings(dates)
@assert length(dates) > 1 "Not enough files to work out YYMMDD v DDMMYY"
x1 = map(x -> tryparse(Int64, x[1:2]), dates)
m = map(x -> tryparse(Int64, x[3:4]), dates)
x2 = map(x -> tryparse(Int64, x[5:6]), dates)
if length(dates) > 1
length(unique(x2)) >= length(unique(x1)) ? (y, d) = ((x1 .+ 2000), x2) :
(y, d) = ((x2 .+ 2000), x1) #assumes year was in 2000's, if 1 file or variance in y and y is equal, it assumes DDMMYY
else
@info "Not enough files to work out YYMMDD v DDMMYY, assuming DDMMYY" #check this is sensible I think DDMMYY is most common
y, d = (x2 .+ 2000), x1
end
return y, m, d
end
edit in src/Labels.jl at line 361

[9.4280]

[2.626]

end

function folder_summary()
a=glob("*/*.data")
replacement in src/Clips.jl at line 229

[3.5040]→[3.837:863](∅→∅),[3.2054]→[3.837:863](∅→∅)

#x -> RGB.(x) |>

[3.5040]

[3.5041]

#x -> RGB.(x) |>
file addition: Parse.jl (----------)

[8.6598]

#=
a = ["201012_123456.wav", "201014_123456.WAV", "201217_123456.wav", "211122_123456.WAV"]

b = ["121020_123456.WAV", "141020_123456.wav", "171220_123456.WAV", "221121_123456.wav"]

c = ["20230609_103000.WAV", "20241109_201504.wav"]

d = [
"120119_003002.wav",
"180120_231502.wav",
"170122_010005.wav",
"010419_234502.WAV",
"310320_231502.wav",
"220824_231502.WAV",
"240123_231502.wav",
]
e = ["XYZ123_7689_20230609_103000.WAV", "string 20241109_201504.wav"]
f = [
"abcdefg__1234_180120_231502.wav",
"string 120119_003002.wav",
"ABCD EFG___170122_010005.wav",
"BHD_1234 010419_234502.WAV",
"cill xyz 310320_231502.wav",
"220824_231502.WAV",
"240123_231502.wav",
]
=#

using Dates
#Assumes it's getting a folder of 1 recording session, not random files.
#For 6 digit dates the year digits need to have less variation than the day digits.
#Does not handle miss shapen dates filenamesBad = ["20230609_1030qa.WAV", "20241109_20wp04.wav"]
function date_time_of_fname(filenames::Vector{String})
pattern = r"(\d{6}|\d{8})_\d{6}"
f2 = [match(pattern, str).match for str in filenames]
#date_time format must be the same for whole vector
g = map(x -> length(x), f2) |> x -> unique(x)
@assert length(g) == 1 "Different date formats in vector" #all same length
@assert length(f2[1]) == 15 || length(f2[1]) == 13 "Wrong length to be a date_time"
raw_dt = map(x -> (split(x, "_")[1], split(x, "_")[2]), f2)
d = parse_date_strings(first.(raw_dt))
t = parse_time_strings(last.(raw_dt))
dt = map((x, y) -> (x..., y...), d, t)
return map(x -> DateTime(x...), dt)
end

function parse_time_strings(times)
h = map(x -> tryparse(Int64, x[1:2]), times)
m = map(x -> tryparse(Int64, x[3:4]), times)
s = map(x -> tryparse(Int64, x[5:6]), times)
return zip(h, m, s) |> collect
end

function parse_date_strings(dates)
if length(dates[1]) == 8
y = map(x -> tryparse(Int64, x[1:4]), dates)
m = map(x -> tryparse(Int64, x[5:6]), dates)
d = map(x -> tryparse(Int64, x[7:8]), dates)
elseif length(dates[1]) == 6
y, m, d = parse_short_date_strings(dates)
else
error("Date is not 8 or 6 digits long")
end
return zip(y, m, d) |> collect
end

# 6 digit dates
function parse_short_date_strings(dates)
@assert length(dates) > 1 "Not enough files to work out YYMMDD v DDMMYY"
x1 = map(x -> tryparse(Int64, x[1:2]), dates)
m = map(x -> tryparse(Int64, x[3:4]), dates)
x2 = map(x -> tryparse(Int64, x[5:6]), dates)
if length(dates) > 1
length(unique(x2)) >= length(unique(x1)) ? (y, d) = ((x1 .+ 2000), x2) :
(y, d) = ((x2 .+ 2000), x1) #assumes year was in 2000's, if 1 file or variance in y and y is equal, it assumes DDMMYY
else
@info "Not enough files to work out YYMMDD v DDMMYY, assuming DDMMYY" #check this is sensible I think DDMMYY is most common
y, d = (x2 .+ 2000), x1
end
return y, m, d
end