Fix JuliaData/CSV.jl#306 by ensuring sentinel values get matched before trying to parse type values. Turns out we were already doing this for Dates/Strings, just not other type parsers

Author: quinnj

src/Parsers.jl

@@ -488,17 +488,19 @@ Sentinel(sentinels::Union{String, Vector{String}}) = Sentinel(defaultparser, Tri
 @inline function parse!(s::Sentinel, io::IO, r::Result{T}; kwargs...) where {T}
     # @debug "xparse Sentinel - $T"
     pos = position(io)
+    if !isempty(s.sentinels.leaves)
+        if match!(s.sentinels, io, r)
+            setfield!(r, 3, Int64(pos))
+            r.code &= ~INVALID
+            return r
+        end
+    end
     parse!(s.next, io, r; kwargs...)
     # @debug "Sentinel - $T: result.code=$(result.code), result.result=$(result.result)"
     if !ok(r.code)
         if isempty(s.sentinels.leaves) && position(io) == pos
             r.code &= ~INVALID
             r.code |= (SENTINEL | ifelse(eof(io), EOF, SUCCESS))
-        else
-            fastseek!(io, pos)
-            if match!(s.sentinels, io, r)
-                r.code &= ~INVALID
-            end
         end
     end
     return r

test/bools.jl

@@ -100,12 +100,12 @@ r = Parsers.parse(Parsers.Sentinel(["NA"]), IOBuffer("NA"), Bool)
 @test r.code == SENTINEL | EOF
 @test r.pos == 0
 r = Parsers.parse(Parsers.Sentinel(["false"]), IOBuffer("false"), Bool)
-@test r.result === false
-@test r.code == OK | EOF
+@test r.result === missing
+@test r.code == SENTINEL | EOF
 @test r.pos == 0
 r = Parsers.parse(Parsers.Sentinel(["fals"]), IOBuffer("falsee"), Bool)
-@test r.result === false
-@test r.code == OK
+@test r.result === missing
+@test r.code == SENTINEL
 @test r.pos == 0
 r = Parsers.parse(Parsers.Sentinel(["fals"]), IOBuffer("fals"), Bool)
 @test r.result === missing

test/runtests.jl

@@ -663,6 +663,15 @@ end
 @test Parsers.tryparse(int2str, SubString("101"), Int) === 101
 @test Parsers.tryparse(int2str, IOBuffer(SubString("101")), Int) === 101
 
+# https://github.com/JuliaData/CSV.jl/issues/306
+# ensure sentinels are matched before trying to parse type values
+r = Parsers.parse(Parsers.Delimited(Parsers.Quoted(Parsers.Sentinel(["1"]))), IOBuffer("1"), Int)
+@test r.result === missing
+@test r.code === SENTINEL | EOF
+r = Parsers.parse(Parsers.Delimited(Parsers.Quoted(Parsers.Sentinel(["1"]))), IOBuffer("1"), String)
+@test r.result === missing
+@test r.code === SENTINEL | EOF
+
 end # @testset
 
 end # @testset