use textwidth for string display truncation (#55442)

It makes a big difference when displaying strings that have width-2 or
width-0 characters.

Author: stevengj

base/strings/io.jl

@@ -214,35 +214,29 @@ function show(
         # one line in collection, seven otherwise
         get(io, :typeinfo, nothing) === nothing && (limit *= 7)
     end
+    limit = max(0, limit-2) # quote chars
 
     # early out for short strings
-    len = ncodeunits(str)
-    len ≤ limit - 2 && # quote chars
-        return show(io, str)
+    check_textwidth(str, limit) && 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)
+    # longest possible replacement string for omitted chars
+    max_replacement = skip_text(ncodeunits(str) * 100) # *100 for 2 inner quote chars
 
-    # find head & tail, avoiding O(length(str)) computation
-    head = nextind(str, 0, 1 + (chars + 1) ÷ 2)
-    tail = prevind(str, len + 1, chars ÷ 2)
+    head, tail = string_truncate_boundaries(str, limit, max_replacement, Val(:center))
 
     # 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)))
-        printstyled(io, skip; color=:light_yellow, bold=true)
-        show(io, SubString(str, tail))
+    afterhead = nextind(str, head)
+    n = tail - afterhead # skipped code units
+    replacement = skip_text(n)
+    t = ncodeunits(replacement) # length of replacement (textwidth == ncodeunits here)
+    @views if 4t ≤ n || t ≤ n && t ≤ textwidth(str[afterhead:prevind(str,tail)])
+        show(io, str[begin:head])
+        printstyled(io, replacement; color=:light_yellow, bold=true)
+        show(io, str[tail:end])
     else
         show(io, str)
     end

base/strings/util.jl

@@ -613,22 +613,25 @@ function ctruncate(str::AbstractString, maxwidth::Integer, replacement::Union{Ab
     end
 end
 
+# return whether textwidth(str) <= maxwidth
+function check_textwidth(str::AbstractString, maxwidth::Integer)
+    # check efficiently for early return if str is wider than maxwidth
+    total_width = 0
+    for c in str
+        total_width += textwidth(c)
+        total_width > maxwidth && return false
+    end
+    return true
+end
+
 function string_truncate_boundaries(
             str::AbstractString,
             maxwidth::Integer,
             replacement::Union{AbstractString,AbstractChar},
             ::Val{mode},
             prefer_left::Bool = true) where {mode}
-
     maxwidth >= 0 || throw(ArgumentError("maxwidth $maxwidth should be non-negative"))
-
-    # check efficiently for early return if str is less wide than maxwidth
-    total_width = 0
-    for c in str
-        total_width += textwidth(c)
-        total_width > maxwidth && break
-    end
-    total_width <= maxwidth && return nothing
+    check_textwidth(str, maxwidth) && return nothing
 
     l0, _ = left, right = firstindex(str), lastindex(str)
     width = textwidth(replacement)

test/show.jl

@@ -928,19 +928,19 @@ end
 # string show with elision
 @testset "string show with elision" begin
     @testset "elision logic" begin
-        strs = ["A", "∀", "∀A", "A∀", "😃"]
+        strs = ["A", "∀", "∀A", "A∀", "😃", "x̂"]
         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 ⋯ \"\"")
+            lower = textwidth("\"\" ⋯ $(ncodeunits(str)) bytes ⋯ \"\"")
             limit = max(limit, lower)
-            if length(str) + 2 ≤ limit
+            if textwidth(str) + 2 ≤ limit+1 && !contains(out, '⋯')
                 @test eval(Meta.parse(out)) == str
             else
-                @test limit-!isascii(str) <= length(out) <= limit
+                @test limit-2 <= textwidth(out) <= limit
                 re = r"(\"[^\"]*\") ⋯ (\d+) bytes ⋯ (\"[^\"]*\")"
                 m = match(re, out)
                 head = eval(Meta.parse(m.captures[1]))
@@ -956,11 +956,11 @@ 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)
+        @test length(r) == 7*80-1
+        @test r == repr("x"^270) * " ⋯ 460 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)
+        @test r == "1-element Vector{String}:\n " * repr("x"^30) * " ⋯ 940 bytes ⋯ " * repr("x"^30)
     end
 end