fix not looking at all conten in filterchunkrev! and also not remove duplicates when not filtering

Author: KristofferC

stdlib/REPL/src/History/resumablefiltering.jl

@@ -253,24 +253,26 @@ end
 
 
 """
-    filterchunkrev!(out, candidates, spec; idx, maxtime, maxresults) -> Int
+    filterchunkrev!(out, candidates, spec; idx, maxtime, maxresults, seen) -> Int
 
 Incrementally filter `candidates[1:idx]` in reverse order.
 
 Pushes matches onto `out` until either `maxtime` is exceeded or `maxresults`
-collected, then returns the new resume index. Only unique entries (by content)
-are added to avoid showing duplicate history items.
+collected, then returns the new resume index. When `seen` is provided (a Set{String}),
+only unique entries (by content) are added to avoid showing duplicate history items.
 """
 function filterchunkrev!(out::Vector{HistEntry}, candidates::DenseVector{HistEntry},
                          spec::FilterSpec, idx::Int = length(candidates);
-                         maxtime::Float64 = Inf, maxresults::Int = length(candidates))
-    seen = Set(e.content for e in out)
+                         maxtime::Float64 = Inf, maxresults::Int = length(candidates),
+                         seen::Union{Nothing,Set{String}} = nothing)
     batchsize = clamp(length(candidates) ÷ 512, 10, 1000)
     for batch in Iterators.partition(idx:-1:1, batchsize)
         time() > maxtime && break
         for outer idx in batch
             entry = candidates[idx]
-            entry.content ∈ seen && continue
+            if !isnothing(seen) && entry.content ∈ seen
+                continue
+            end
             if !isempty(spec.modes)
                 entry.mode ∈ spec.modes || continue
             end
@@ -296,7 +298,7 @@ function filterchunkrev!(out::Vector{HistEntry}, candidates::DenseVector{HistEnt
                 end
             end
             matchfail && continue
-            push!(seen, entry.content)
+            !isnothing(seen) && push!(seen, entry.content)
             pushfirst!(out, entry)
             length(out) == maxresults && break
         end

stdlib/REPL/src/History/search.jl

@@ -62,6 +62,7 @@ function run_display!((; term, pstate), events::Channel{Symbol}, hist::Vector{Hi
     cands_temp = HistEntry[]
     # Filter state
     filter_idx = 0
+    filter_seen = nothing
     # Event loop
     while true
         event = @lock events if !isempty(events) take!(events) end
@@ -153,10 +154,16 @@ function run_display!((; term, pstate), events::Channel{Symbol}, hist::Vector{Hi
                 end
             end
             # Start filtering candidates
+            # Only deduplicate when user has entered a search query. When browsing
+            # with no filter (empty query), show all history including duplicates.
+            isfiltering = !isempty(filter_spec.exacts) || !isempty(filter_spec.negatives) ||
+                          !isempty(filter_spec.regexps) || !isempty(filter_spec.modes)
+            filter_seen = isfiltering ? Set{String}() : nothing
             filter_idx = filterchunkrev!(
                 state, cands_current;
                 maxtime = time() + 0.01,
-                maxresults = outsize[1])
+                maxresults = outsize[1],
+                seen = filter_seen)
             if filter_idx == 0
                 cands_cachestate = addcache!(
                     cands_cache, cands_cachestate, cands_cond => state.candidates)
@@ -187,7 +194,8 @@ function run_display!((; term, pstate), events::Channel{Symbol}, hist::Vector{Hi
                 state.scroll, state.selection, state.hover)
             filter_idx = filterchunkrev!(
                 state, cands_current, filter_idx;
-                maxtime = time() + 0.01)
+                maxtime = time() + 0.01,
+                seen = filter_seen)
             if filter_idx == 0
                 cands_cachestate = addcache!(
                     cands_cache, cands_cachestate, cands_cond => state.candidates)
@@ -204,10 +212,11 @@ function run_display!((; term, pstate), events::Channel{Symbol}, hist::Vector{Hi
 end
 
 function filterchunkrev!(state::SelectorState, candidates::DenseVector{HistEntry}, idx::Int = length(candidates);
-                         maxtime::Float64 = Inf, maxresults::Int = length(candidates))
+                         maxtime::Float64 = Inf, maxresults::Int = length(candidates),
+                         seen::Union{Nothing,Set{String}} = nothing)
     oldlen = length(state.candidates)
     idx = filterchunkrev!(state.candidates, candidates, state.filter, idx;
-                          maxtime = maxtime, maxresults = maxresults)
+                          maxtime = maxtime, maxresults = maxresults, seen = seen)
     newlen = length(state.candidates)
     newcands = view(state.candidates, (oldlen + 1):newlen)
     gfound = Int[]

stdlib/REPL/test/history.jl

@@ -356,16 +356,22 @@ end
                     HistEntry(:julia, now(UTC), "println(\"hello\")", 6),  # duplicate
                     HistEntry(:julia, now(UTC), "tan(π/4)", 7),
                 ]
-                cset = ConditionSet("")  # Match all
+                # When filtering with seen Set, duplicates are removed
+                cset = ConditionSet("cos")
                 spec = FilterSpec(cset)
-                @test filterchunkrev!(results, dup_entries, spec) == 0
-                # Should only get unique entries
+                seen = Set{String}()
+                @test filterchunkrev!(results, dup_entries, spec; seen=seen) == 0
+                # Should only get unique entries matching the filter
                 # Since we iterate in reverse (7->1), we keep the most recent occurrence of each unique content
-                @test length(results) == 4
-                @test results[1] == dup_entries[4]  # sin(π)
-                @test results[2] == dup_entries[5]  # cos(2π) - most recent
-                @test results[3] == dup_entries[6]  # println("hello") - most recent
-                @test results[4] == dup_entries[7]  # tan(π/4)
+                @test length(results) == 1
+                @test results[1] == dup_entries[5]  # cos(2π) - most recent
+                # When browsing without seen Set, duplicates are kept
+                empty!(results)
+                cset2 = ConditionSet("")  # No filter
+                spec2 = FilterSpec(cset2)
+                @test filterchunkrev!(results, dup_entries, spec2) == 0
+                @test length(results) == 7  # All entries, including duplicates
+                @test results == dup_entries
             end
         end
         @testset "Strictness comparison" begin