From 5ecb7699394b1c75b13769118d92fd847665d126 Mon Sep 17 00:00:00 2001
From: Stefan Karpinski <stefan.karpinski@post.harvard.edu>
Date: Mon, 10 May 2021 18:18:10 -0400
Subject: [PATCH] show(::String): elide long strings (close #40724) (#40736)

---
 NEWS.md            |  1 +
 base/strings/io.jl | 48 +++++++++++++++++++++++++++++++++++++++++
 test/show.jl       | 54 ++++++++++++++++++++++++++++++++++++++++++----
 3 files changed, 99 insertions(+), 4 deletions(-)

diff --git a/NEWS.md b/NEWS.md
index 2e63729ee09cb4..07db69fecf7dbf 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -55,6 +55,7 @@ New library features
 Standard library changes
 ------------------------
 
+* Long strings are now elided using the syntax `"head" ⋯ 12345 bytes ⋯ "tail"` when displayed in the REPL ([#40736]).
 * `count` and `findall` now accept an `AbstractChar` argument to search for a character in a string ([#38675]).
 * `range` now supports the `range(start, stop)` and `range(start, stop, length)` methods ([#39228]).
 * `range` now supports `start` as an optional keyword argument ([#38041]).
diff --git a/base/strings/io.jl b/base/strings/io.jl
index 3977f137510e34..0f877275ed7595 100644
--- a/base/strings/io.jl
+++ b/base/strings/io.jl
@@ -190,6 +190,54 @@ print(io::IO, s::AbstractString) = for c in s; print(io, c); end
 write(io::IO, s::AbstractString) = (len = 0; for c in s; len += Int(write(io, c))::Int; end; len)
 show(io::IO, s::AbstractString) = print_quoted(io, s)
 
+# show elided string if more than `limit` characters
+function show(
+    io    :: IO,
+    mime  :: MIME"text/plain",
+    str   :: AbstractString;
+    limit :: Union{Int, Nothing} = nothing,
+)
+    # compute limit in default case
+    if limit === nothing
+        get(io, :limit, false) || return show(io, str)
+        limit = max(20, displaysize(io)[2])
+        # one line in collection, seven otherwise
+        get(io, :typeinfo, nothing) === nothing && (limit *= 7)
+    end
+
+    # early out for short strings
+    len = ncodeunits(str)
+    len ≤ limit - 2 && # quote chars
+        return show(io, str)
+
+    # these don't depend on string data
+    units = codeunit(str) == UInt8 ? "bytes" : "code units"
+    skip_text(skip) = " ⋯ $skip $units ⋯ "
+    short = length(skip_text("")) + 4 # quote chars
+    chars = max(limit, short + 1) - short # at least 1 digit
+
+    # figure out how many characters to print in elided case
+    chars -= d = ndigits(len - chars) # first adjustment
+    chars += d - ndigits(len - chars) # second if needed
+    chars = max(0, chars)
+
+    # find head & tail, avoiding O(length(str)) computation
+    head = nextind(str, 0, 1 + (chars + 1) ÷ 2)
+    tail = prevind(str, len + 1, chars ÷ 2)
+
+    # threshold: min chars skipped to make elision worthwhile
+    t = short + ndigits(len - chars) - 1
+    n = tail - head # skipped code units
+    if 4t ≤ n || t ≤ n && t ≤ length(str, head, tail-1)
+        skip = skip_text(n)
+        show(io, SubString(str, 1:prevind(str, head)))
+        print(io, skip) # TODO: bold styled
+        show(io, SubString(str, tail))
+    else
+        show(io, str)
+    end
+end
+
 # optimized methods to avoid iterating over chars
 write(io::IO, s::Union{String,SubString{String}}) =
     GC.@preserve s Int(unsafe_write(io, pointer(s), reinterpret(UInt, sizeof(s))))::Int
diff --git a/test/show.jl b/test/show.jl
index 2c8a6fd67bd4a2..2dde201b06dfb7 100644
--- a/test/show.jl
+++ b/test/show.jl
@@ -808,13 +808,20 @@ Base.methodloc_callback[] = nothing
     # test that no spurious visual lines are added when one element spans multiple lines
     v = fill!(Array{Any}(undef, 9), 0)
     v[1] = "look I'm wide! --- " ^ 9
-    @test replstr(v) == "9-element Vector{Any}:\n  \"look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- \"\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0"
-    @test replstr([fill(0, 9) v]) == "9×2 Matrix{Any}:\n 0  …   \"look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- \"\n 0     0\n 0     0\n 0     0\n 0     0\n 0  …  0\n 0     0\n 0     0\n 0     0"
+    r = replstr(v)
+    @test startswith(r, "9-element Vector{Any}:\n  \"look I'm wide! ---")
+    @test endswith(r, "look I'm wide! --- \"\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0")
+
     # test vertical/diagonal ellipsis
     v = fill!(Array{Any}(undef, 50), 0)
     v[1] = "look I'm wide! --- " ^ 9
-    @test replstr(v) == "50-element Vector{Any}:\n  \"look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- \"\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n ⋮\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0"
-    @test replstr([fill(0, 50) v]) == "50×2 Matrix{Any}:\n 0  …   \"look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- look I'm wide! --- \"\n 0     0\n 0     0\n 0     0\n 0     0\n 0  …  0\n 0     0\n 0     0\n 0     0\n 0     0\n ⋮  ⋱  \n 0     0\n 0     0\n 0     0\n 0     0\n 0  …  0\n 0     0\n 0     0\n 0     0\n 0     0"
+    r = replstr(v)
+    @test startswith(r, "50-element Vector{Any}:\n  \"look I'm wide! ---")
+    @test endswith(r, "look I'm wide! --- \"\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n ⋮\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0\n 0")
+
+    r = replstr([fill(0, 50) v])
+    @test startswith(r, "50×2 Matrix{Any}:\n 0  …   \"look I'm wide! ---")
+    @test endswith(r, "look I'm wide! --- \"\n 0     0\n 0     0\n 0     0\n 0     0\n 0  …  0\n 0     0\n 0     0\n 0     0\n 0     0\n ⋮  ⋱  \n 0     0\n 0     0\n 0     0\n 0     0\n 0  …  0\n 0     0\n 0     0\n 0     0\n 0     0")
 
     # issue #34659
     @test replstr(Int32[]) == "Int32[]"
@@ -825,6 +832,45 @@ Base.methodloc_callback[] = nothing
     @test replstr([zeros(3,0),zeros(2,0)]) == "2-element Vector{Matrix{Float64}}:\n 3×0 Matrix{Float64}\n 2×0 Matrix{Float64}"
 end
 
+# string show with elision
+@testset "string show with elision" begin
+    @testset "elision logic" begin
+        strs = ["A", "∀", "∀A", "A∀", "😃"]
+        for limit = 0:100, len = 0:100, str in strs
+            str = str^len
+            str = str[1:nextind(str, 0, len)]
+            out = sprint() do io
+                show(io, MIME"text/plain"(), str; limit)
+            end
+            lower = length("\"\" ⋯ $(ncodeunits(str)) bytes ⋯ \"\"")
+            limit = max(limit, lower)
+            if length(str) + 2 ≤ limit
+                @test eval(Meta.parse(out)) == str
+            else
+                @test limit-!isascii(str) <= length(out) <= limit
+                re = r"(\"[^\"]*\") ⋯ (\d+) bytes ⋯ (\"[^\"]*\")"
+                m = match(re, out)
+                head = eval(Meta.parse(m.captures[1]))
+                tail = eval(Meta.parse(m.captures[3]))
+                skip = parse(Int, m.captures[2])
+                @test startswith(str, head)
+                @test endswith(str, tail)
+                @test ncodeunits(str) ==
+                    ncodeunits(head) + skip + ncodeunits(tail)
+            end
+        end
+    end
+
+    @testset "default elision limit" begin
+        r = replstr("x"^1000)
+        @test length(r) == 7*80
+        @test r == repr("x"^271) * " ⋯ 459 bytes ⋯ " * repr("x"^270)
+        r = replstr(["x"^1000])
+        @test length(r) < 120
+        @test r == "1-element Vector{String}:\n " * repr("x"^31) * " ⋯ 939 bytes ⋯ " * repr("x"^30)
+    end
+end
+
 # Issue 14121
 @test_repr "(A'x)'"