Merge pull request #978 from julia-vscode/newuri

Rewrite uri handling

Author: pfitzseb

Project.toml

@@ -13,18 +13,18 @@ REPL = "3fa0cd96-eef1-5676-8a61-b3b8758bbffb"
 StaticLint = "b3cc710f-9c33-5bdb-a03d-a94903873e97"
 SymbolServer = "cf896787-08d5-524d-9de7-132aaa0cb996"
 Tokenize = "0796e94c-ce3b-5d07-9a54-7f471281c624"
-URIParser = "30578b45-9adc-5946-b283-645ec420af67"
+URIs = "5c2747f8-b7ea-4ff2-ba2e-563bfd36b1d4"
 UUIDs = "cf7118a7-6976-5b1a-9a39-7adc72f591a4"
 
 [compat]
 CSTParser = "3.3"
 JSON = "0.20, 0.21"
 JSONRPC = "1.1"
+URIs = "1.3"
 JuliaFormatter = "0.20.0, 0.21, 0.22"
 StaticLint = "8.0"
 SymbolServer = "7.1"
 Tokenize = "0.5.10"
-URIParser = "0.4.1"
 julia = "1"
 
 [extras]

src/LanguageServer.jl

@@ -1,5 +1,4 @@
 module LanguageServer
-import URIParser
 using JSON, REPL, CSTParser, JuliaFormatter, SymbolServer, StaticLint
 using CSTParser: EXPR, Tokenize.Tokens, Tokenize.Tokens.kind, headof, parentof, valof, to_codeobject
 using StaticLint: refof, scopeof, bindingof
@@ -10,8 +9,12 @@ using JSONRPC: Outbound, @dict_readable
 
 export LanguageServerInstance, runserver
 
+include("URIs2/URIs2.jl")
+using .URIs2
+
+JSON.lower(uri::URI) = string(uri)
+
 include("exception_types.jl")
-include("uri2.jl")
 include("protocol/protocol.jl")
 include("extensions/extensions.jl")
 include("document.jl")

src/URIs2/URIs2.jl

@@ -0,0 +1,203 @@
+module URIs2
+
+import URIs
+
+export URI, uri2filepath, filepath2uri, @uri_str
+
+struct URI
+    scheme::Union{String,Nothing}
+    authority::Union{String,Nothing}
+    path::String
+    query::Union{String,Nothing}
+    fragment::Union{String,Nothing}
+end
+
+function percent_decode(str::AbstractString)
+    return URIs.unescapeuri(str)
+end
+
+function URI(value::AbstractString)
+    m = match(r"^(([^:/?#]+?):)?(\/\/([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?", value)
+
+    m===nothing && error("Invalid argument.")
+
+    return URI(
+        m.captures[2],
+        m.captures[4]===nothing ? nothing : percent_decode(m.captures[4]),
+        m.captures[5]===nothing ? nothing : percent_decode(m.captures[5]),
+        m.captures[7]===nothing ? nothing : percent_decode(m.captures[7]),
+        m.captures[9]===nothing ? nothing : percent_decode(m.captures[9])
+    )
+end
+
+function URI(;
+    scheme::Union{AbstractString,Nothing}=nothing,
+    authority::Union{AbstractString,Nothing}=nothing,
+    path::AbstractString="",
+    query::Union{AbstractString,Nothing}=nothing,
+    fragment::Union{AbstractString,Nothing}=nothing
+    )
+    return URI(scheme, authority, path, query, fragment)
+end
+
+@inline function is_rfc3986_unreserved(c::Char)
+    return 'A' <= c <= 'Z' ||
+        'a' <= c <= 'z' ||
+        '0' <= c <= '9' ||
+        c == '-' ||
+        c == '.' ||
+        c == '_' ||
+        c == '~'
+end
+
+@inline function is_rfc3986_sub_delim(c::Char)
+    return c == '!' ||
+        c == '$' ||
+        c == '&' ||
+        c == '\'' ||
+        c == '(' ||
+        c == ')' ||
+        c == '*' ||
+        c == '+' ||
+        c == ',' ||
+        c == ';' ||
+        c == '='
+end
+
+@inline function is_rfc3986_pchar(c::Char)
+    return is_rfc3986_unreserved(c) ||
+        is_rfc3986_sub_delim(c) ||
+        c == ':' ||
+        c == '@'
+end
+
+@inline function is_rfc3986_query(c::Char)
+    return is_rfc3986_pchar(c) || c=='/' || c=='?'
+end
+
+@inline function is_rfc3986_fragment(c::Char)
+    return is_rfc3986_pchar(c) || c=='/' || c=='?'
+end
+
+@inline function is_rfc3986_userinfo(c::Char)
+    return is_rfc3986_unreserved(c) ||
+        is_rfc3986_sub_delim(c) ||
+        c == ':'
+end
+
+@inline function is_rfc3986_reg_name(c::Char)
+    return is_rfc3986_unreserved(c) ||
+        is_rfc3986_sub_delim(c)
+end
+
+function encode(io::IO, s::AbstractString, issafe::Function)
+    for c in s
+        if issafe(c)
+            print(io, c)
+        else
+            print(io, '%')
+            print(io, uppercase(string(Int(c), base=16, pad=2)))
+        end
+    end
+end
+
+@inline function is_ipv4address(s::AbstractString)
+    if length(s)==1
+        return '0' <= s[1] <= '9'
+    elseif length(s)==2
+        return '1' <= s[1] <= '9' && '0' <= s[2] <= '9'
+    elseif length(s)==3
+        return (s[1]=='1' && '0' <= s[2] <= '9' && '0' <= s[3] <= '9') ||
+            (s[1]=='2' && '0' <= s[2] <= '4' && '0' <= s[3] <= '9') ||
+            (s[1]=='2' && s[2] == '5' && '0' <= s[3] <= '5')
+    else
+        return false
+    end
+end
+
+@inline function is_ipliteral(s::AbstractString)
+    # TODO Implement this
+    return false
+end
+
+function encode_host(io::IO, s::AbstractString)
+    if is_ipv4address(s) || is_ipliteral(s)
+        print(io, s)
+    else
+        # The host must be a reg-name
+        encode(io, s, is_rfc3986_reg_name)
+    end
+end
+
+function encode_path(io::IO, s::AbstractString)
+    # TODO Write our own version
+    print(io, URIs.escapepath(s))
+end
+
+function Base.print(io::IO, uri::URI)
+    scheme = uri.scheme
+    authority = uri.authority
+    path = uri.path
+    query = uri.query
+    fragment = uri.fragment
+
+ 	if scheme!==nothing
+        print(io, scheme)
+        print(io, ':')
+ 	end
+
+ 	if authority!==nothing
+        print(io, "//")
+
+		idx = findfirst("@", authority)
+		if idx !== nothing
+			# <user>@<auth>
+			userinfo = SubString(authority, 1:idx.start-1)
+			host_and_port = SubString(authority, idx.start + 1)
+			encode(io, userinfo, is_rfc3986_userinfo)
+            print(io, '@')
+        else
+            host_and_port = SubString(authority, 1)
+		end
+
+		idx3 = findfirst(":", host_and_port)
+		if idx3 === nothing
+            encode_host(io, host_and_port)
+		else
+			# <auth>:<port>
+            encode_host(io, SubString(host_and_port, 1:idx3.start-1))
+			print(io, SubString(host_and_port, idx3.start))
+        end
+     end
+
+     # Append path
+     encode_path(io, path)
+
+    if query!==nothing
+        print(io, '?')
+        encode(io, query, is_rfc3986_query)
+    end
+
+ 	if fragment!==nothing
+        print(io, '#')
+        encode(io, fragment, is_rfc3986_fragment)
+    end
+
+    return nothing
+end
+
+function Base.string(uri::URI)
+    io = IOBuffer()
+
+    print(io, uri)
+
+    return String(take!(io))
+end
+
+macro uri_str(ex)
+    return URI(ex)
+end
+
+include("uri_helpers.jl")
+
+end

src/URIs2/uri_helpers.jl

@@ -0,0 +1,57 @@
+
+function uri2filepath(uri::URI)
+    if uri.scheme != "file"
+        return nothing
+    end
+
+    path = uri.path
+    host = uri.authority
+
+    if host!==nothing && host != "" && length(path) > 1
+        # unc path: file://shares/c$/far/boo
+        value = "//$host$path"
+    elseif length(path) >= 3 &&
+            path[1] == '/' &&
+            isascii(path[2]) && isletter(path[2]) &&
+            path[3] == ':'
+        # windows drive letter: file:///c:/far/boo
+        value = lowercase(path[2]) * path[3:end]
+    else
+        # other path
+        value = path
+    end
+
+    if Sys.iswindows()
+        value = replace(value, '/' => '\\')
+    end
+
+    return value
+end
+
+function filepath2uri(path::String)
+    isabspath(path) || error("Relative path `$path` is not valid.")
+
+    path = normpath(path)
+
+    if Sys.iswindows()
+        path = replace(path, "\\" => "/")
+    end
+
+    authority = ""
+
+    if startswith(path, "//")
+        # UNC path //foo/bar/foobar
+        idx = findnext("/", path, 3)
+        if idx===nothing
+            authority = path[3:end]
+            path = "/"
+        else
+            authority = path[3:idx.start-1]
+            path = path[idx.start:end]
+        end
+    elseif length(path)>=2 && isascii(path[1]) && isletter(path[1]) && path[2]==':'
+        path = string('/', lowercase(path[1]), SubString(path, 2))
+    end
+
+    return URI(scheme="file", authority=authority, path=path)
+end

src/document.jl

@@ -1,5 +1,5 @@
 mutable struct Document
-    _uri::String
+    _uri::URI
     _path::String
     _content::String
     _line_offsets::Union{Nothing,Vector{Int}}
@@ -11,7 +11,7 @@ mutable struct Document
     _version::Int
     server
     root::Document
-    function Document(uri::AbstractString, text::AbstractString, workspace_file::Bool, server=nothing)
+    function Document(uri::URI, text::AbstractString, workspace_file::Bool, server=nothing)
         path = something(uri2filepath(uri), "")
         path == "" || isabspath(path) || throw(LSRelativePath("Relative path `$path` is not valid."))
         cst = CSTParser.parse(text, true)

src/languageserverinstance.jl

@@ -28,7 +28,7 @@ For normal usage, the language server can be instantiated with
 mutable struct LanguageServerInstance
     jr_endpoint::Union{JSONRPC.JSONRPCEndpoint,Nothing}
     workspaceFolders::Set{String}
-    _documents::Dict{URI2,Document}
+    _documents::Dict{URI,Document}
 
     env_path::String
     depot_path::String
@@ -67,7 +67,7 @@ mutable struct LanguageServerInstance
         new(
             JSONRPC.JSONRPCEndpoint(pipe_in, pipe_out, err_handler),
             Set{String}(),
-            Dict{URI2,Document}(),
+            Dict{URI,Document}(),
             env_path,
             depot_path,
             SymbolServer.SymbolServerInstance(depot_path, symserver_store_path; symbolcache_upstream = symbolcache_upstream),
@@ -101,11 +101,11 @@ function Base.display(server::LanguageServerInstance)
     end
 end
 
-function hasdocument(server::LanguageServerInstance, uri::URI2)
+function hasdocument(server::LanguageServerInstance, uri::URI)
     return haskey(server._documents, uri)
 end
 
-function getdocument(server::LanguageServerInstance, uri::URI2)
+function getdocument(server::LanguageServerInstance, uri::URI)
     return server._documents[uri]
 end
 
@@ -121,11 +121,11 @@ function getdocuments_value(server::LanguageServerInstance)
     return values(server._documents)
 end
 
-function setdocument!(server::LanguageServerInstance, uri::URI2, doc::Document)
+function setdocument!(server::LanguageServerInstance, uri::URI, doc::Document)
     server._documents[uri] = doc
 end
 
-function deletedocument!(server::LanguageServerInstance, uri::URI2)
+function deletedocument!(server::LanguageServerInstance, uri::URI)
     doc = getdocument(server, uri)
     StaticLint.clear_meta(getcst(doc))
     delete!(server._documents, uri)

src/protocol/basic.jl

@@ -21,7 +21,7 @@ mutable struct DocumentFilter <: Outbound
     pattern::Union{String,Missing}
 end
 const DocumentSelector = Vector{DocumentFilter}
-const DocumentUri = String
+const DocumentUri = URI
 
 @dict_readable struct Position
     line::Int
@@ -56,7 +56,7 @@ Location(f::String, line::Integer) = Location(f, Range(line))
 end
 
 @dict_readable struct WorkspaceFolder
-    uri::String
+    uri::DocumentUri
     name::String
 end
 

src/protocol/features.jl

@@ -120,7 +120,7 @@ end
 
 struct DocumentLink <: Outbound
     range::Range
-    target::Union{String,Missing}
+    target::Union{DocumentUri,Missing}
     tooltip::Union{String,Missing}
     data::Union{Any,Missing}
 end