Merge pull request #6 from JuliaString/spj/makegeneric

Make API more generic

Author: ScottPJones

src/HTML_Entities.jl

@@ -29,62 +29,8 @@ struct HTML_Table{T} <: AbstractEntityTable
 end
 
 function __init__()
-    global _tab =
+    global default =
         HTML_Table(StrTables.load(joinpath(Pkg.dir("HTML_Entities"), "data", "html.dat"))...)
     nothing
 end
-
-const _empty_str = ""
-const _empty_str_vec = Vector{String}()
-
-function _get_str(ind)
-    ind <= _tab.base32 && return string(Char(_tab.val16[ind]))
-    ind <= _tab.base2c && return string(Char(_tab.val32[ind - _tab.base32] + 0x10000))
-    val = _tab.val2c[ind - _tab.base2c]
-    string(Char(val>>>16), Char(val&0xffff))
-end
-    
-function _get_strings(val::T, tab::Vector{T}, ind::Vector{UInt16}) where {T}
-    rng = searchsorted(tab, val)
-    isempty(rng) && return _empty_str_vec
-    _tab.nam[ind[rng]]
-end
-
-function lookupname(str::AbstractString)
-    rng = searchsorted(_tab.nam, str)
-    isempty(rng) ? _empty_str : _get_str(_tab.ind[rng.start])
-end
-
-matchchar(ch::UInt32) =
-    (ch <= 0x0ffff
-     ? _get_strings(ch%UInt16, _tab.val16, _tab.ind16)
-     : (ch <= 0x1ffff ? _get_strings(ch%UInt16, _tab.val32, _tab.ind32) : _empty_str_vec))
-matchchar(ch::Char) = matchchar(UInt32(ch))
-
-matches(str::AbstractString) = matches(convert(Vector{Char}, str))
-function matches(vec::Vector{Char})
-    if length(vec) == 1
-        matchchar(vec[1])
-    elseif length(vec) == 2 && (vec[1] <= '\uffff' && vec[2] <= '\uffff')
-        _get_strings(vec[1]%UInt32<<16 | vec[2]%UInt32, _tab.val2c, _tab.ind2c)
-    else
-        _empty_str_vec
-    end
-end
-
-longestmatches(str::AbstractString) = longestmatches(convert(Vector{Char}, str))
-function longestmatches(vec::Vector{Char})
-    isempty(vec) && return _empty_str_vec
-    if length(vec) >= 2 && (vec[1] <= '\uffff' && vec[2] <= '\uffff')
-        res = _get_strings(vec[1]%UInt32<<16 | vec[2]%UInt32, _tab.val2c, _tab.ind2c)
-        isempty(res) || return res
-        # Fall through and check only the first character
-    end
-    matchchar(vec[1])
-end
-
-completions(str::AbstractString) = completions(convert(String, str))
-completions(str::String) = StrTables.matchfirst(_tab.nam, str)
-
-end # module
-
+end # module HTML_Entities

test/runtests.jl

@@ -7,47 +7,54 @@ using HTML_Entities
 
 HE = HTML_Entities
 
+he_matchchar(ch)       = HE.matchchar(HE.default, ch)
+he_lookupname(nam)     = HE.lookupname(HE.default, nam)
+he_longestmatches(str) = HE.longestmatches(HE.default, str)
+he_matches(str)        = HE.matches(HE.default, str)
+he_completions(str)    = HE.completions(HE.default, str)
+
+
 @testset "HTML_Entities" begin
 @testset "lookupname" begin
-    @test HE.lookupname(SubString("My name is Spock", 12)) == ""
-    @test HE.lookupname("foobar") == ""
-    @test HE.lookupname("nle")    == "\u2270"
-    @test HE.lookupname("Pscr")   == "\U1d4ab"
-    @test HE.lookupname("lvnE")   == "\u2268\ufe00"
+    @test he_lookupname(SubString("My name is Spock", 12)) == ""
+    @test he_lookupname("foobar") == ""
+    @test he_lookupname("nle")    == "\u2270"
+    @test he_lookupname("Pscr")   == "\U1d4ab"
+    @test he_lookupname("lvnE")   == "\u2268\ufe00"
 end
 
 @testset "matches" begin
-    @test isempty(HE.matches(""))
-    @test isempty(HE.matches("\u201f"))
-    @test isempty(HE.matches(SubString("This is \u201f", 9)))
+    @test isempty(he_matches(""))
+    @test isempty(he_matches("\u201f"))
+    @test isempty(he_matches(SubString("This is \u201f", 9)))
     for (chrs, exp) in (("\u2270", ["nle", "nleq"]),
                         ("\U1d4ab", ["Pscr"]),
                         ("\U1d51e", ["afr"]),
                         ("\u2268\ufe00", ["lvertneqq", "lvnE"]))
-        res = HE.matches(chrs)
+        res = he_matches(chrs)
         @test length(res) >= length(exp)
         @test intersect(res, exp) == exp
     end
 end
 
 @testset "longestmatches" begin
-    @test isempty(HE.longestmatches("\u201f abcd"))
-    @test isempty(HE.longestmatches(SubString("This is \U201f abcd", 9)))
+    @test isempty(he_longestmatches("\u201f abcd"))
+    @test isempty(he_longestmatches(SubString("This is \U201f abcd", 9)))
     for (chrs, exp) in (("\u2270 abcd", ["nle", "nleq"]),
                         ("\U1d4ab abcd", ["Pscr"]),
                         ("\u2268\ufe00 silly", ["lvertneqq", "lvnE"]))
-        res = HE.longestmatches(chrs)
+        res = he_longestmatches(chrs)
         @test length(res) >= length(exp)
         @test intersect(res, exp) == exp
     end
 end
 
 @testset "completions" begin
-    @test isempty(HE.completions("ScottPaulJones"))
-    @test isempty(HE.completions(SubString("My name is Scott", 12)))
+    @test isempty(he_completions("ScottPaulJones"))
+    @test isempty(he_completions(SubString("My name is Scott", 12)))
     for (chrs, exp) in (("and", ["and", "andand", "andd", "andslope", "andv"]),
                         ("um", ["umacr", "uml"]))
-        res = HE.completions(chrs)
+        res = he_completions(chrs)
         @test length(res) >= length(exp)
         @test intersect(res, exp) == exp
     end