added a whole bunch of functions to Labels.jl for working with raven and avians labels

[?]

AEj8dahVWy718uSSFPe9VSRJ5qX5G8pC2zvFzJJ8yzBd

Nov 9, 2024, 1:30 AM

YODTMMPTZOUTK2OQWHDJB6D3QXZ2FJFYH2XPICCOITCK54EARC2AC

Dependencies

[2] YXAKJSDT added check_change_avianz_species! function to Labels.jl
[3] 2UBDFCJH new files tracked
[4] 2O6SHIVY added avianz_of_raven function to Labels.jl

Change contents

replacement in src/Labels.jl at line 3

[3.15]→[3.15:53](∅→∅)

export aggregate_labels, audiodata_db

[3.15]

[3.53]

export aggregate_labels, audiodata_db, raven_of_avianz, avianz_of_raven, label_summary
replacement in src/Labels.jl at line 5

[3.54]→[2.9:73](∅→∅)

using CSV, DataFrames, Glob, Random, DBInterface, DuckDB, JSON3

[3.54]

[3.111]

using CSV, DataFrames, Glob, Random, DBInterface, DuckDB, JSON3, Dates
edit in src/Labels.jl at line 159

[3.4517]

[3.9]

end

function avianz_file_of_dict(f, payload)
isfile(f) && cp(f, "$f.backup")
open(f, "w") do io
JSON3.write(io, payload)
end
replacement in src/Labels.jl at line 172

[3.199]→[3.199:234](∅→∅)

# map(x -> avianz_of_raven!(x), a)

[3.199]

[3.234]

# map(x -> avianz_of_raven(x), a)
replacement in src/Labels.jl at line 174

[3.235]→[3.235:272](∅→∅)

function avianz_of_raven!(f::String)

[3.235]

[2.74]

function avianz_of_raven(f::String)
replacement in src/Labels.jl at line 189

[3.797]→[2.467:493](∅→∅),[2.493]→[3.821:900](∅→∅),[3.821]→[3.821:900](∅→∅)

avianzf = f * ".data"
open(avianzf, "w") do io
JSON3.write(io, data)
println(io)

[3.797]

[3.900]

basef = f = replace(f, "Table.1.selections.txt" => "")
isfile((basef * ".WAV")) ? avianzf = basef * "WAV.data" : avianzf = basef * "wav.data"
avianz_file_of_dict(avianzf, data)
end

function dict_of_avianz_file(f)
json_string = read(f, String)
x = JSON3.read(json_string)
return x
end

function df_of_avianz_dict(data)
df = DataFrame(
start_time = Float64[],
end_time = Float64[],
low_f = Float64[],
high_f = Float64[],
Species = String[],
)
for (index1, item1) in enumerate(data[2:end])
st, en, lf, hf = item1[1:4]
sp = map(x -> x[:species], item1[5])
lsp = length(sp)
for (index2, item2) in enumerate(sp)
push!(
df,
(
start_time = st,
end_time = en,
low_f = lf,
high_f = hf,
Species = sp[index2],
),
)
end
replacement in src/Labels.jl at line 225

[3.908]→[3.908:938](∅→∅)

#isfile(avianzf) && rm(f)

[3.908]

[2.494]

return df
end

function prepare_df_for_raven(data)
l = length(data.start_time)
df = DataFrame(
"Selection" => collect(1:l),
"View" => ["Spectrogram 1" for x = 1:l],
"Channel" => [1 for x = 1:l],
"Begin Time (s)" => data.start_time,
"End Time (s)" => data.end_time,
"Low Freq (Hz)" => data.low_f,
"High Freq (Hz)" => data.high_f,
"Species" => data.Species,
)
return df
end

# using CSV, DataFrames, JSON3
function raven_of_avianz(file::String)
data = dict_of_avianz_file(file)
if length(data) >= 2 #ignores empty .data files
df = df_of_avianz_dict(data) |> prepare_df_for_raven
outfile = replace(
file,
".WAV.data" => ".Table.1.selections.txt",
".wav.data" => ".Table.1.selections.txt",
)
CSV.write(outfile, df, delim = '\t')
end
edit in src/Labels.jl at line 275

[2.1199]→[2.1199:1483](∅→∅)

end
function read_avianz_data(f)
json_string = read(f, String)
x = JSON3.read(json_string)
return x
end
function write_new_file!(f, payload)
cp(f, "$f.backup")
open(f, "w") do io
JSON3.write(io, payload)
end
replacement in src/Labels.jl at line 277

[2.1512]→[2.1512:1547](∅→∅)

x = read_avianz_data(file)

[2.1512]

[2.1547]

x = dict_of_avianz_file(file)
replacement in src/Labels.jl at line 282

[2.1723]→[2.1723:1774](∅→∅)

write_new_file!(file, payload)

[2.1723]

[2.1774]

avianz_file_of_dict(file, payload)
edit in src/Labels.jl at line 290

[2.1922]

function label_summary(folder::String, avianz = true)
if avianz == true
files = glob("$folder/*.['W','w']['A','a']['V','v'].data")
df = DataFrame(
start_time = Float64[],
end_time = Float64[],
low_f = Float64[],
high_f = Float64[],
Species = String[],
File = String[],
)
for file in files
df1 = dict_of_avianz_file(file) |> x -> df_of_avianz_dict(x)
f = split(file, "/")[end] |> x -> replace(x, ".data" => "")
df1.File = ["$f" for x = 1:length(df1.start_time)]
df = vcat(df, df1)
end
select!(df, [:File, :start_time, :end_time, :low_f, :high_f, :Species])
CSV.write("$folder/label_summary_avianz-$(Dates.today()).csv", df, delim = '\t')
return df
else
files = glob("$folder/*.Table.1.selections.txt")
df = DataFrame(
File = String[],
start_time = Float64[],
end_time = Float64[],
low_f = Float64[],
high_f = Float64[],
Species = String[],
)
for file in files
df1 = DataFrame(CSV.File(file))
for d in eachrow(df1)
f =
split(file, "/")[end] |>
x -> replace(x, ".Table.1.selections.txt" => "")
push!(
df,
(
File = f,
start_time = d."Begin Time (s)",
end_time = d."End Time (s)",
low_f = d."Low Freq (Hz)",
high_f = d."High Freq (Hz)",
Species = d."Species",
),
)
end
end
CSV.write("$folder/label_summary_raven-$(Dates.today()).csv", df, delim = '\t')
return df
end
end