update hashing to Julia 1.13 and use column names in data frame hashing (#3507)

Author: bkamins

src/abstractdataframe/abstractdataframe.jl

@@ -1902,8 +1902,9 @@ end
 const hashdf_seed = UInt == UInt32 ? 0xfd8bb02e : 0x6215bada8c8c46de
 
 function Base.hash(df::AbstractDataFrame, h::UInt)
-    h += hashdf_seed
-    h += hash(size(df))
+    h ⊻= hashdf_seed
+    h = hash(size(df), h)
+    h = hash(_names(df), h)
     for i in 1:size(df, 2)
         h = hash(df[!, i], h)
     end

src/groupeddataframe/utils.jl

@@ -58,7 +58,8 @@ end
 # Calculate the vector of `df` rows hash values.
 function hashrows(cols::Tuple{Vararg{AbstractVector}}, skipmissing::Bool)
     len = length(cols[1])
-    rhashes = zeros(UInt, len)
+    ref_val = @static Base.VERSION >= v"1.13.0-DEV" ? Base.HASH_SEED : UInt(0)
+    rhashes = fill(ref_val, len)
     missings = fill(false, skipmissing ? len : 0)
     for (i, col) in enumerate(cols)
         rp = DataAPI.refpool(col)

src/join/core.jl

@@ -40,7 +40,9 @@ Base.hash(ocr1::OnColRow, h::UInt) = throw(MethodError(hash, (ocr1, h)))
 function _prehash(oc::OnCol)
     h = oc.h
     resize!(h, oc.len)
-    fill!(h, Base.tuplehash_seed)
+    h0 = @static Base.VERSION >= v"1.13.0-DEV" ? Base.HASH_SEED : UInt(0)
+    h0 ⊻= Base.tuplehash_seed
+    fill!(h, h0)
     for col in reverse(oc.cols)
         h .= hash.(col, h)
     end

test/dataframe.jl

@@ -425,7 +425,12 @@ end
 
     @test hash(DataFrame([1 2; 3 4], :auto)) == hash(DataFrame([1 2; 3 4], :auto))
     @test hash(DataFrame([1 2; 3 4], :auto)) != hash(DataFrame([1 3; 2 4], :auto))
-    @test hash(DataFrame([1 2; 3 4], :auto)) == hash(DataFrame([1 2; 3 4], :auto), zero(UInt))
+    @test hash(DataFrame([1 2; 3 4], :auto)) != hash(DataFrame([1 2; 3 4], [:x2, :x1]))
+
+    @test hash(DataFrame([1 2; 3 4], :auto), UInt(10)) == hash(DataFrame([1 2; 3 4], :auto), UInt(10))
+    @test hash(DataFrame([1 2; 3 4], :auto), UInt(10)) != hash(DataFrame([1 2; 3 4], :auto), UInt(11))
+    @test hash(DataFrame([1 2; 3 4], :auto), UInt(10)) != hash(DataFrame([1 3; 2 4], :auto), UInt(10))
+    @test hash(DataFrame([1 2; 3 4], :auto), UInt(10)) != hash(DataFrame([1 2; 3 4], [:x2, :x1]), UInt(10))
 end
 
 @testset "deleteat!" begin