feat: expand the dataset name regex to allow periods

Co-authored-by: Sebastian Pfitzner <pfitzseb@gmail.com>

Author: mortenpi

src/DataSets.jl

@@ -87,27 +87,49 @@ end
 Check whether a dataset name is valid.
 
 Valid names must start with a letter or a number, the rest of the name can also contain `-`
-and `_` characters. The names can also be hieracicial, with segments separated by forward
-slashes (`/`). Each segment must also start with either a letter or a number. For example:
+and `_` characters. The names can also be hierarchical, with segments separated by forward
+slashes (`/`) or (`.`). Each segment must also start with either a letter or a number.
+
+For example, the following dataset names are valid:
 
     my_data
     my_data_1
     username/data
     organization_name/project-name/data
     123user/456dataset--name
+    username/my_table.csv
+    dataset/v0.1.2
+
+whereas names like this are invalid:
+
+    __mydata__
+    username/.git
+    my...dataset
+
+!!! note "Segment separators"
+
+    In dataset names, both `/` and `.` are considered segment separators from a syntax
+    perspective. While DataSets.jl does not impose any specific interpretation on the
+    dataset name, it is recommended to use `/` to separate segments from a semantic
+    perspective, and to interpret each forward-slash-separated segment as a path separator.
+    Periods would conventionally be used to separate file extensions within a segment.
+
+    E.g. use `username/my-project-data/population.csv`, rather than
+    `username.my-project-data.population.csv` or something like that.
 """
 function check_dataset_name(name::AbstractString)
     if !occursin(DATASET_NAME_REGEX, name)
-        error("DataSet name \"$name\" is invalid. DataSet names must start with a letter and can contain only letters, numbers, `-`, `_` or `/`.")
+        error("DataSet name \"$name\" is invalid. DataSet names must start with a letter or a number, and can contain only letters, numbers, `-` and `_`, or `/` and `.` as segment separators.")
     end
 end
 # DataSet names disallow most punctuation for now, as it may be needed as
 # delimiters in data-related syntax (eg, for the data REPL).
 const DATASET_NAME_REGEX_STRING = raw"""
 [[:alnum:]]
 (?:
-    [-[:alnum:]_]     |
-    / (?=[[:alnum:]])
+    [-[:alnum:]_]      |
+    \.(?=[[:alnum:]])  |
+    \/ (?=[[:alnum:]])
 )*
 """
 const DATASET_NAME_REGEX = Regex("^\n$(DATASET_NAME_REGEX_STRING)\n\$", "x")

test/dataset-names-invalid.txt

@@ -104,20 +104,48 @@ function load_list(filename)
 end
 @testset "Data set name parsing" begin
     @testset "Valid names" begin
-        valid_names = load_list("dataset-names-valid.txt")
-        @test length(valid_names) == 16
+        valid_names = load_list("testnames-valid.txt")
+        @test !isempty(valid_names)
         @testset "Valid name: $name" for name in valid_names
             @test DataSets.check_dataset_name(name) === nothing
             @test DataSets._split_dataspec(name) == (name, nothing, nothing)
+            # Also test that the name is still valid when it appears as part of
+            # a path elements.
+            let path_name = "foo/$(name)"
+                @test DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (path_name, nothing, nothing)
+            end
+            let path_name = "$(name)/foo"
+                @test DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (path_name, nothing, nothing)
+            end
+            let path_name = "foo/$(name)/bar"
+                @test DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (path_name, nothing, nothing)
+            end
         end
     end
 
     @testset "Invalid names" begin
-        invalid_names = load_list("dataset-names-invalid.txt")
-        @test length(invalid_names) == 9
+        invalid_names = load_list("testnames-invalid.txt")
+        @test !isempty(invalid_names)
         @testset "Invalid name: $name" for name in invalid_names
             @test_throws ErrorException DataSets.check_dataset_name(name)
             @test DataSets._split_dataspec(name) == (nothing, nothing, nothing)
+            # Also test that the name is still invalid when it appears as part of
+            # a path elements.
+            let path_name = "foo/$(name)"
+                @test_throws ErrorException DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (nothing, nothing, nothing)
+            end
+            let path_name = "$(name)/foo"
+                @test_throws ErrorException DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (nothing, nothing, nothing)
+            end
+            let path_name = "foo/$(name)/bar"
+                @test_throws ErrorException DataSets.check_dataset_name(path_name) === nothing
+                @test DataSets._split_dataspec(path_name) == (nothing, nothing, nothing)
+            end
         end
     end
 end

test/runtests.jl

@@ -0,0 +1,25 @@
+a b
+a/b/
+a//b
+/a/b
+a/-
+a/ _/b
+a/-a
+a/-1
+.a
+..a
+a.
+a..
+.a.
+a..b
+.abc
+abc.
+abc/.def
+abc/def.
+a./b
+a.-
+_._
+a._b
+a.-b
+./a
+b/../a

test/testnames-invalid.txt

@@ -14,3 +14,11 @@ a/1
 1-2-3
 x_-__
 a---
+a.b
+a.b
+abc.def
+abc/def.ghi
+abc-def.ghi_jkl
+a.b.c
+a_.c
+foo__-.csv