Skip to content

Commit 27154c7

Browse files
joshdayjoshday
committed
add Node implementation
1 parent cbacf1a commit 27154c7

File tree

4 files changed

+197
-18
lines changed

4 files changed

+197
-18
lines changed

‎.gitignore‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
/Manifest.toml
2-
*test.xml
32
*generated_xsd.jl
3+
*.xml

‎src/XML.jl‎

Lines changed: 9 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -120,7 +120,7 @@ const INDENT = " "
120120
showxml(x; depth=0) = (io=IOBuffer(); showxml(io, x); print(String(take!(io))))
121121

122122
# assumes '\n' occurs in String
123-
showxml(io::IO, x::String; depth=0) = show(io, x)
123+
showxml(io::IO, x::String; depth=0) = print(io, INDENT^depth, x)
124124

125125

126126
#-----------------------------------------------------------------------------# DTD
@@ -219,11 +219,16 @@ attributes(o::Element) = getfield(o, :attributes)
219219
Base.getindex(o::Element, i::Integer) = children(o)[i]
220220
Base.lastindex(o::Element) = lastindex(children(o))
221221
Base.setindex!(o::Element, val::Element, i::Integer) = setindex!(children(o), val, i)
222+
Base.push!(o::Element, val::Element) = push!(children(o), val)
222223

223224
Base.getproperty(o::Element, x::Symbol) = attributes(o)[x]
224225
Base.setproperty!(o::Element, x::Symbol, val) = (attributes(o)[x] = string(val))
225226
Base.propertynames(o::Element) = collect(keys(attributes(o)))
226227

228+
Base.get(o::Element, key::Symbol, val) = hasproperty(o, key) ? getproperty(o, key) : val
229+
Base.get!(o::Element, key::Symbol, val) = hasproperty(o, key) ? getproperty(o, key) : setproperty!(o, key, val)
230+
231+
227232

228233

229234
#-----------------------------------------------------------------------------# Document
@@ -313,4 +318,7 @@ function add_children!(e::Element, o::XMLTokenIterator, until::String)
313318
end
314319
end
315320

321+
#-----------------------------------------------------------------------------# Node
322+
include("node.jl")
323+
316324
end

‎src/node.jl‎

Lines changed: 146 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,146 @@
1+
@enum(NodeType,
2+
DOCUMENT_NODE, # prolog & root ELEMENT_NODE
3+
DTD_NODE, # <!DOCTYPE ...>
4+
DECLARATION_NODE, # <?xml attributes... ?>
5+
COMMENT_NODE, # <!-- ... -->
6+
CDATA_NODE, # <![CDATA[...]]>
7+
ELEMENT_NODE, # <NAME attributes... >
8+
)
9+
10+
#-----------------------------------------------------------------------------# Node
11+
struct Node
12+
type::NodeType
13+
tag::Union{Nothing, String}
14+
attributes::Union{Nothing, Dict{Symbol, String}}
15+
children::Union{Nothing, Vector{Union{String, Node}}, String}
16+
Node(type, tag, attr, children::Nothing) = new(type, tag, attr, children)
17+
function Node(type, tag, attributes, children::Vector)
18+
new(type, tag, attributes, Vector{Union{String,Node}}(children))
19+
end
20+
function Node(type, tag, attributes, children::AbstractString)
21+
new(type, tag, attributes, String(children))
22+
end
23+
end
24+
function Base.:(==)(a::Node, b::Node)
25+
a.type == b.type &&
26+
a.tag == b.tag &&
27+
a.attributes == b.attributes &&
28+
a.children == b.children
29+
end
30+
31+
Base.show(io::IO, o::Node) = showxml(io, o)
32+
33+
Base.write(io::IO, o::Node) = showxml(io, o)
34+
35+
36+
document(children...) = Node(DOCUMENT_NODE, nothing, nothing, collect(children))
37+
dtd(content::AbstractString) = Node(DTD_NODE, nothing, nothing, content)
38+
declaration(; attrs...) = Node(DECLARATION_NODE, nothing, OrderedDict(k=>string(v) for (k,v) in pairs(attrs)), nothing)
39+
comment(content::AbstractString) = Node(COMMENT_NODE, nothing, nothing, content)
40+
cdata(content::AbstractString) = Node(CDATA_NODE, nothing, nothing, content)
41+
42+
m(tag::String, children...; attrs...) = Node(ELEMENT_NODE, tag, OrderedDict(k=>string(v) for (k,v) in pairs(attrs)), collect(children))
43+
# m(tag::String, child::String; attrs...) = Node(ELEMENT_NODE, tag, OrderedDict(k=>string(v) for (k,v) in pairs(attrs)), child)
44+
Base.getproperty(::typeof(m), sym::Symbol) = (f(args...; kw...) = m(string(sym), args...; kw...))
45+
46+
function check(o::Node)
47+
if type == DOCUMENT_NODE
48+
isnothing(o.tag) || error("A DOCUMENT_NODE should not have a tag.")
49+
isnothing(o.attributes) || error("A DOCUMENT_NODE should not have attributes.")
50+
o.children isa Vector{Node} || error("DOCUMENT_NODE children should be Vector{Node}.")
51+
elseif type == DTD_NODE
52+
isnothing(o.tag) || error("A DTD_NODE should not have a tag.")
53+
isnothing(o.children) || error("A DTD_NODE should not have children.")
54+
elseif type == DECLARATION_NODE
55+
isnothing(o.children) || error("A DECLARATION_NODE should not have children.")
56+
elseif type == COMMENT_NODE
57+
isnothing(o.tag) || error("A COMMENT_NODE should not have a tag.")
58+
isnothing(o.attributes) || error("A COMMENT_NODE should not have attributes.")
59+
o.children isa String || error("A COMMENT_NODE's child should be a String.")
60+
elseif type == CDATA_NODE
61+
isnothing(o.tag) || error("A CDATA_NODE should not have a tag.")
62+
isnothing(o.attributes) || error("A CDATA_NODE should not have attributes.")
63+
o.children isa String || error("A CDATA_NODE's child should be a String.")
64+
end
65+
end
66+
67+
attr_string(o::Node) = join(" $k=$(repr(v))" for (k,v) in o.attributes)
68+
69+
#-----------------------------------------------------------------------------# show
70+
function showxml(io::IO, o::Node; depth=0)
71+
p(args...) = printstyled(io, args...; color=depth + 1)
72+
if o.type == DOCUMENT_NODE
73+
for (i,child) in enumerate(o.children)
74+
showxml(io, child; depth=0)
75+
i != length(o.children) && println(io)
76+
end
77+
elseif o.type == DTD_NODE
78+
p(INDENT^depth, "<!doctype ", o.children, '>')
79+
elseif o.type == DECLARATION_NODE
80+
p(INDENT^depth, "<?", o.tag, attr_string(o), "?>")
81+
elseif o.type == COMMENT_NODE
82+
p(INDENT^depth, "<!-- ", o.children, " -->")
83+
elseif o.type == CDATA_NODE
84+
p(INDENT^depth, "<![CDATA[", o.children, "]]>")
85+
elseif o.type == ELEMENT_NODE
86+
p(INDENT^depth, '<', o.tag, attr_string(o))
87+
if isnothing(o.children)
88+
p(" />")
89+
else
90+
if length(o.children) == 1 && o.children[1] isa String
91+
p('>')
92+
showxml(io, o.children[1])
93+
p("</", o.tag, '>')
94+
else
95+
p('>', '\n')
96+
for child in o.children
97+
showxml(io, child, depth=depth+1)
98+
println(io)
99+
end
100+
p(INDENT^depth, "</", o.tag, '>')
101+
end
102+
end
103+
end
104+
end
105+
106+
107+
Base.getindex(o::Node, i::Integer) = o.children[i]
108+
Base.setindex!(o::Node, val, i::Integer) = setindex!(o.children, val, i)
109+
Base.lastindex(o::Node) = lastindex(o.children)
110+
111+
#-----------------------------------------------------------------------------# From XMLTokenIterator
112+
function Node(itr::XMLTokenIterator)
113+
doc = Node(DOCUMENT_NODE, nothing, nothing, [])
114+
add_children!(doc, itr, "END_OF_FILE")
115+
return doc
116+
end
117+
118+
119+
readnode(file::String) = open(io -> Node(XMLTokenIterator(io)), file, "r")
120+
121+
function add_children!(e::Node, o::XMLTokenIterator, until::String)
122+
s = ""
123+
c = e.children
124+
while s != until
125+
next = iterate(o, -1) # if state == 0, io will get reset to original position
126+
isnothing(next) && break
127+
T, s = next[1]
128+
if T == COMMENTTOKEN
129+
push!(c, comment(strip(replace(s, "<!--" => "", "-->" => ""))))
130+
elseif T == CDATATOKEN
131+
push!(c, cdata(replace(s, "<![CDATA[" => "", "]]>" => "")))
132+
elseif T == ELEMENTSELFCLOSEDTOKEN
133+
push!(c, Node(ELEMENT_NODE, get_tag(s), get_attributes(s), nothing))
134+
elseif T == ELEMENTTOKEN
135+
child = Node(ELEMENT_NODE, get_tag(s), get_attributes(s), [])
136+
add_children!(child, o, "</$(child.tag)>")
137+
push!(c, child)
138+
elseif T == TEXTTOKEN
139+
push!(c, unescape(s))
140+
elseif T == DTDTOKEN
141+
push!(c, dtd(replace(s, "<!doctype " => "", "<!DOCTYPE " => "", '>' => "")))
142+
elseif T == DECLARATIONTOKEN
143+
push!(c, Node(DECLARATION_NODE, get_tag(s), get_attributes(s), nothing))
144+
end
145+
end
146+
end

‎test/runtests.jl‎

Lines changed: 41 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -3,25 +3,50 @@ using Downloads: download
33
using Test
44

55
@testset "XML.jl" begin
6-
@testset "example.kml" begin
7-
doc = Document(joinpath(@__DIR__, "example.kml"))
8-
write("test.xml", doc)
9-
doc2 = Document("test.xml")
10-
@test doc == doc2
11-
end
6+
@testset "Separate structs" begin
7+
@testset "example.kml" begin
8+
doc = Document(joinpath(@__DIR__, "example.kml"))
9+
write("test.xml", doc)
10+
doc2 = Document("test.xml")
11+
@test doc == doc2
12+
end
13+
14+
@testset "books.xml" begin
15+
doc = Document(joinpath(@__DIR__, "books.xml"))
16+
write("test.xml", doc)
17+
doc2 = Document("test.xml")
18+
@test doc == doc2
19+
end
1220

13-
@testset "books.xml" begin
14-
doc = Document(joinpath(@__DIR__, "books.xml"))
15-
write("test.xml", doc)
16-
doc2 = Document("test.xml")
17-
@test doc == doc2
21+
@testset "KML spec" begin
22+
doc = Document(download("http://schemas.opengis.net/kml/2.2.0/ogckml22.xsd"))
23+
write("test.xml", doc)
24+
doc2 = Document("test.xml")
25+
@test doc == doc2
26+
end
1827
end
1928

20-
@testset "KML spec" begin
21-
doc = Document(download("http://schemas.opengis.net/kml/2.2.0/ogckml22.xsd"))
22-
write("test.xml", doc)
23-
doc2 = Document("test.xml")
24-
@test doc == doc2
29+
@testset "Node" begin
30+
@testset "example.kml" begin
31+
doc = XML.readnode(joinpath(@__DIR__, "example.kml"))
32+
write("test.xml", doc)
33+
doc2 = XML.readnode("test.xml")
34+
@test doc == doc2
35+
end
36+
37+
@testset "books.xml" begin
38+
doc = XML.readnode(joinpath(@__DIR__, "books.xml"))
39+
write("test.xml", doc)
40+
doc2 = XML.readnode("test.xml")
41+
@test doc == doc2
42+
end
43+
44+
@testset "KML spec" begin
45+
doc = XML.readnode(download("http://schemas.opengis.net/kml/2.2.0/ogckml22.xsd"))
46+
write("test.xml", doc)
47+
doc2 = XML.readnode("test.xml")
48+
@test doc == doc2
49+
end
2550
end
2651

2752
# cleanup

0 commit comments

Comments
 (0)