From ee8db6b1eb7df6fa0da3fb5c09e1497232ab335d Mon Sep 17 00:00:00 2001
From: Alex Wietek <awietek@pks.mpg.de>
Date: Thu, 10 Aug 2023 14:14:48 +0200
Subject: [PATCH] Added FileCollection

---
 src/Dumper.jl          |  3 ++-
 src/file_collection.jl | 38 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)
 create mode 100644 src/file_collection.jl

diff --git a/src/Dumper.jl b/src/Dumper.jl
index 474bef4..114de52 100644
--- a/src/Dumper.jl
+++ b/src/Dumper.jl
@@ -2,10 +2,11 @@ module Dumper
 using HDF5
 using Printf
 
-export DumpFile, write_data!, dump!, read_data
+export DumpFile, write_data!, dump!, read_data, FileCollection, filenames, read_collection_h5
 
 include("create_extensible.jl")
 include("append_extensible.jl")
+include("file_collection.jl")
 
 struct DumpFile
     filename::AbstractString
diff --git a/src/file_collection.jl b/src/file_collection.jl
new file mode 100644
index 0000000..f133d70
--- /dev/null
+++ b/src/file_collection.jl
@@ -0,0 +1,38 @@
+struct FileCollection
+    directory::AbstractString
+    regex::Regex
+end
+
+function filenames(collection::FileCollection)
+    files = readdir(collection.directory)
+    matched_files = String[]
+    for file in files
+        m = match(collection.regex, file)
+        if !isnothing(m)
+            push!(matched_files, file)
+        end
+    end
+    return map(fl -> joinpath(collection.directory, fl), matched_files)
+end
+
+function read_collection_h5(collection::FileCollection, tag::AbstractString)
+    files = readdir(collection.directory)
+    data = Dict()
+    for file in files
+        m = match(collection.regex, file)
+        if !isnothing(m)
+            if length(m) == 1
+                param = m[keys(m)[1]]
+            else
+                param = ntuple(i -> m[keys(m)[i]], length(m))
+            end
+
+            # read hdf5 data
+            file = joinpath(collection.directory, file)
+            dset = h5read(file, tag)
+            ndims = length(size(dset))
+            data[param] = permutedims(dset, ndims:-1:1)
+        end
+    end
+    return sort(collect(data), by = x->x[1])
+end