DataURI parsing and creation

Author: s-celles

Project.toml

@@ -3,6 +3,9 @@ uuid = "5c2747f8-b7ea-4ff2-ba2e-563bfd36b1d4"
 authors = ["Jacob Quinn", "Sam O'Connor", "contributors: https://github.com/JuliaWeb/URIs.jl/graphs/contributors"]
 version = "1.5.1"
 
+[deps]
+Base64 = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"
+
 [compat]
 julia = "1.6"
 

README.md

@@ -3,7 +3,7 @@
 [![Documentation (stable)][docs-stable-img]][docs-stable-url] [![Documentation (dev)][docs-dev-img]][docs-dev-url] [![Build Status](https://github.com/JuliaWeb/URIs.jl/workflows/CI/badge.svg)](https://github.com/JuliaWeb/URIs.jl/actions)
 
 `URIs` is a Julia package for parsing and working with Uniform Resource
-Identifiers, as defined in [RFC 3986](https://www.ietf.org/rfc/rfc3986.txt).
+Identifiers, as defined in [RFC 3986](https://www.ietf.org/rfc/rfc3986.txt) and Data URI, as defined in [RFC 2397](https://datatracker.ietf.org/doc/html/rfc2397).
 
 Read the [**stable documentation** on
 JuliaHub][docs-stable-url], or see the [development

docs/src/index.md

@@ -2,9 +2,11 @@
 
 `URIs` is a Julia package for parsing and working with Uniform Resource
 Identifiers, as defined in [RFC 3986](https://www.ietf.org/rfc/rfc3986.txt).
+It also supports data URI as defined in [RFC 2397](https://datatracker.ietf.org/doc/html/rfc2397).
 
 ## Tutorial
 
+### URI
 ```@meta
 DocTestSetup = quote
     using URIs
@@ -48,6 +50,23 @@ Dict{String,String} with 2 entries:
   "y" => "hi"
 ```
 
+### Data URI
+
+```@meta
+DocTestSetup = quote
+    using URIs
+end
+```
+
+Parsing Data URIs from a string can be done with the [`DataURI`](@ref) constructor:
+
+```jldoctest
+julia> u = DataURI("data:,A%20brief%20note")
+DataURI("data:,A%20brief%20note")
+```
+
+See reference and unit tests for more usage.
+
 ## Reference
 
 ```@docs
@@ -61,5 +80,7 @@ escapepath
 resolvereference
 URIs.splitpath
 Base.isvalid(::URI)
+DataURI
+getdata
 ```
 

src/URIs.jl

@@ -4,6 +4,7 @@ export URI,
        queryparams, queryparampairs, absuri, splitfilepath,
        escapeuri, unescapeuri, escapepath,
        resolvereference
+export DataURI, getdata
 
 import Base.==
 
@@ -703,6 +704,7 @@ function __init__()
     return
 end
 
+include("data_uri.jl")
 include("deprecate.jl")
 
 end # module

src/data_uri.jl

@@ -0,0 +1,60 @@
+using Base64
+
+const DATA_URI_PATTERNS = [
+    r"data:(?<mediatype>[^;]+)(?:;(?<parameters>.+))?,(?<data>.+)",
+    r"data:,(?<data>.+)"
+]
+
+"""
+    DataURI(; mediatype="", encode=false, data="")
+    DataURI(str) = parse(DataURI, str::String)
+
+A type representing a Data URI (e.g. a data URL scheme as defined in RFC 2397). Can be constructed from distinct parts using the various
+supported keyword arguments, or from a string.
+"""
+struct DataURI
+    uri::String
+    mediatype::DataType
+    isbase64::Bool
+    data::SubString{String}
+    parameters::Vector{Pair{String, String}}
+    
+end
+
+function DataURI(; mediatype=MIME"text/plain", encoded=false, data="", parameters=Pair{String, String}[])
+    base64 = encoded ? ";base64" : ""
+    s_parameters = ""
+    for (key, value) in parameters
+        s_parameters *= ";$(key)=$(value)"
+    end
+    s_mediatype = string(mediatype.parameters[1])
+    uri = "data:$(s_mediatype)$(s_parameters)$(base64),$(data)"
+    DataURI(uri, mediatype, encoded, data, parameters)
+end
+
+function DataURI(str)
+    m = match(DATA_URI_PATTERNS[1], str)
+    if !isnothing(m)
+        groups = m.captures
+        mediatype, parameters, data = groups
+        mime = MIME{Symbol(mediatype)}
+        isbase64 = endswith(parameters, "base64")
+        _parameters = filter(p -> p != "base64", split(parameters, ";"))
+        parameters = Pair{String, String}[]
+        for p in _parameters
+            key, value = split(p, "=")
+            push!(parameters, key => value)
+        end
+    else
+        m = match(DATA_URI_PATTERNS[2], str)
+        data, = m.captures
+        mime = MIME""
+        isbase64 = false
+        parameters = Pair{String, String}[]
+    end
+    DataURI(str, mime, isbase64, data, parameters)
+end
+
+function getdata(data_uri::DataURI)
+    data_uri.isbase64 ? base64decode(data_uri.data) : data_uri.data
+end

test/data_uri.jl

@@ -0,0 +1,78 @@
+using Test
+
+const larry = "data:image/gif;base64,R0lGODdhMAAwAPAAAAAAAP///ywAAAAAMAAwAAAC8IyPqcvt3wCcDkiLc7C0qwyGHhSWpjQu5yqmCYsapyuvUUlvONmOZtfzgFzByTB10QgxOR0TqBQejhRNzOfkVJ+5YiUqrXF5Y5lKh/DeuNcP5yLWGsEbtLiOSpa/TPg7JpJHxyendzWTBfX0cxOnKPjgBzi4diinWGdkF8kjdfnycQZXZeYGejmJlZeGl9i2icVqaNVailT6F5iJ90m6mvuTS4OK05M0vDk0Q4XUtwvKOzrcd3iq9uisF81M1OIcR7lEewwcLp7tuNNkM3uNna3F2JQFo97Vriy/Xl4/f1cf5VWzXyym7PHhhx4dbgYKAAA7"
+const greek_characters = "data:text/plain;charset=iso-8859-7,%be%fg%be"
+const brief_note = "data:,A%20brief%20note"
+
+
+@testset "Data URI" begin
+    @testset "parsing" begin
+        @testset "larry" begin
+            data_uri = DataURI(larry)
+            @test data_uri.mediatype == MIME"image/gif"
+            @test data_uri.isbase64
+            @test length(data_uri.parameters) == 0
+        end
+
+        @testset "greek_characters" begin
+            data_uri = DataURI(greek_characters)
+            @test data_uri.mediatype == MIME"text/plain"
+            @test !data_uri.isbase64
+            @test getdata(data_uri) == "%be%fg%be"
+            @test length(data_uri.parameters) == 1
+            @test data_uri.parameters[1][1] == "charset"
+            @test data_uri.parameters[1][2] == "iso-8859-7"
+        end
+
+        @testset "brief_note" begin
+            data_uri = DataURI(brief_note)
+            @test data_uri.mediatype == MIME""
+            @test !data_uri.isbase64
+            @test getdata(data_uri) == "A%20brief%20note"
+            @test length(data_uri.parameters) == 0
+        end
+
+        @testset "img_pluto" begin
+            open("resources/sample_data_uri/img_pluto.txt") do f
+                content = read(f, String)
+                data_uri = DataURI(content)
+                @test data_uri.mediatype == MIME"image/jpeg"
+                @test data_uri.isbase64
+                @test length(data_uri.parameters) == 0    
+            end
+        end
+
+        @testset "audio_meow" begin
+            open("resources/sample_data_uri/audio_meow.txt") do f
+                content = read(f, String)
+                data_uri = DataURI(content)
+                @test data_uri.mediatype == MIME"audio/mpeg"
+                @test data_uri.isbase64
+                @test length(data_uri.parameters) == 0    
+            end
+        end
+    end
+
+    @testset "creating" begin
+        @testset "larry" begin
+            data_uri = DataURI(; mediatype=MIME"image/gif", encoded=true, data="R0lGODdhMAAwAPAAAAAAAP///ywAAAAAMAAwAAAC8IyPqcvt3wCcDkiLc7C0qwyGHhSWpjQu5yqmCYsapyuvUUlvONmOZtfzgFzByTB10QgxOR0TqBQejhRNzOfkVJ+5YiUqrXF5Y5lKh/DeuNcP5yLWGsEbtLiOSpa/TPg7JpJHxyendzWTBfX0cxOnKPjgBzi4diinWGdkF8kjdfnycQZXZeYGejmJlZeGl9i2icVqaNVailT6F5iJ90m6mvuTS4OK05M0vDk0Q4XUtwvKOzrcd3iq9uisF81M1OIcR7lEewwcLp7tuNNkM3uNna3F2JQFo97Vriy/Xl4/f1cf5VWzXyym7PHhhx4dbgYKAAA7")
+            @test data_uri.uri == larry
+        end
+
+        @testset "greek_characters" begin
+            data_uri = DataURI(; mediatype=MIME"text/plain", data="%be%fg%be", parameters=["charset" => "iso-8859-7"])
+            @test data_uri.uri == greek_characters
+        end
+
+        @testset "greek_characters using default mediatype" begin
+            data_uri = DataURI(; data="%be%fg%be", parameters=["charset" => "iso-8859-7"])
+            @test data_uri.uri == greek_characters
+        end
+
+        @testset "brief_note" begin
+            data_uri = DataURI(; mediatype=MIME"", data="A%20brief%20note")
+            @test data_uri.uri == brief_note
+        end
+    end
+
+end

test/resources/sample_data_uri/audio_meow.txt

@@ -3,6 +3,7 @@ using Test
 
 include("uri.jl")
 include("url.jl")
+include("data_uri.jl")
 
 # https://github.com/JuliaWeb/URIs.jl/issues/42
 struct CustomString <: AbstractString