Changes in the validation of UTF-8

All UTF-8 encoding functionality (including the escape
sequence '\u') accepts all values from the original UTF-8
specification (with sequences of up to six bytes).

By default, the decoding functions in the UTF-8 library do not
accept invalid Unicode code points, such as surrogates. A new
parameter 'nonstrict' makes them accept all code points up to
(2^31)-1, as in the original UTF-8 specification.

2 files changed, 72 insertions, 37 deletions

diff --git a/testes/literals.lua b/testes/literals.lua
index 76c08f12..fc45d4ad 100644
--- a/testes/literals.lua
+++ b/testes/literals.lua
@@ -56,16 +56,23 @@ assert("abc\z
 assert("\u{0}\u{00000000}\x00\0" == string.char(0, 0, 0, 0))
 
 -- limits for 1-byte sequences
-assert("\u{0}\u{7F}" == "\x00\z\x7F")
+assert("\u{0}\u{7F}" == "\x00\x7F")
 
 -- limits for 2-byte sequences
-assert("\u{80}\u{7FF}" == "\xC2\x80\z\xDF\xBF")
+assert("\u{80}\u{7FF}" == "\xC2\x80\xDF\xBF")
 
 -- limits for 3-byte sequences
-assert("\u{800}\u{FFFF}" ==   "\xE0\xA0\x80\z\xEF\xBF\xBF")
+assert("\u{800}\u{FFFF}" ==   "\xE0\xA0\x80\xEF\xBF\xBF")
 
 -- limits for 4-byte sequences
-assert("\u{10000}\u{10FFFF}" == "\xF0\x90\x80\x80\z\xF4\x8F\xBF\xBF")
+assert("\u{10000}\u{1FFFFF}" == "\xF0\x90\x80\x80\xF7\xBF\xBF\xBF")
+
+-- limits for 5-byte sequences
+assert("\u{200000}\u{3FFFFFF}" == "\xF8\x88\x80\x80\x80\xFB\xBF\xBF\xBF\xBF")
+
+-- limits for 6-byte sequences
+assert("\u{4000000}\u{7FFFFFFF}" ==
+       "\xFC\x84\x80\x80\x80\x80\xFD\xBF\xBF\xBF\xBF\xBF")
 
 
 -- Error in escape sequences
@@ -94,7 +101,7 @@ lexerror([["xyz\300"]], [[\300"]])
 lexerror([["   \256"]], [[\256"]])
 
 -- errors in UTF-8 sequences
-lexerror([["abc\u{110000}"]], [[abc\u{110000]])   -- too large
+lexerror([["abc\u{100000000}"]], [[abc\u{100000000]])   -- too large
 lexerror([["abc\u11r"]], [[abc\u1]])    -- missing '{'
 lexerror([["abc\u"]], [[abc\u"]])    -- missing '{'
 lexerror([["abc\u{11r"]], [[abc\u{11r]])    -- missing '}'
diff --git a/testes/utf8.lua b/testes/utf8.lua
index 4b6a57fd..86ec1b00 100644
--- a/testes/utf8.lua
+++ b/testes/utf8.lua
@@ -21,62 +21,59 @@ local justone = "^" .. utf8.charpattern .. "$"
 
 -- 't' is the list of codepoints of 's'
 local function checksyntax (s, t)
+  -- creates a string "return '\u{t[1]}...\u{t[n]}'"
   local ts = {"return '"}
   for i = 1, #t do ts[i + 1] = string.format("\\u{%x}", t[i]) end
   ts[#t + 2] = "'"
   ts = table.concat(ts)
+  -- its execution should result in 's'
   assert(assert(load(ts))() == s)
 end
 
 assert(utf8.offset("alo", 5) == nil)
 assert(utf8.offset("alo", -4) == nil)
 
--- 't' is the list of codepoints of 's'
-local function check (s, t)
-  local l = utf8.len(s) 
+-- 'check' makes several tests over the validity of string 's'.
+-- 't' is the list of codepoints of 's'.
+local function check (s, t, nonstrict)
+  local l = utf8.len(s, 1, -1, nonstrict)
   assert(#t == l and len(s) == l)
-  assert(utf8.char(table.unpack(t)) == s)
+  assert(utf8.char(table.unpack(t)) == s)   -- 't' and 's' are equivalent
 
   assert(utf8.offset(s, 0) == 1)
 
   checksyntax(s, t)
 
-  local t1 = {utf8.codepoint(s, 1, -1)}
+  -- creates new table with all codepoints of 's'
+  local t1 = {utf8.codepoint(s, 1, -1, nonstrict)}
   assert(#t == #t1)
-  for i = 1, #t do assert(t[i] == t1[i]) end
+  for i = 1, #t do assert(t[i] == t1[i]) end   -- 't' is equal to 't1'
 
-  for i = 1, l do
+  for i = 1, l do   -- for all codepoints
     local pi = utf8.offset(s, i)        -- position of i-th char
     local pi1 = utf8.offset(s, 2, pi)   -- position of next char
     assert(string.find(string.sub(s, pi, pi1 - 1), justone))
     assert(utf8.offset(s, -1, pi1) == pi)
     assert(utf8.offset(s, i - l - 1) == pi)
-    assert(pi1 - pi == #utf8.char(utf8.codepoint(s, pi)))
-    for j = pi, pi1 - 1 do 
+    assert(pi1 - pi == #utf8.char(utf8.codepoint(s, pi, pi, nonstrict)))
+    for j = pi, pi1 - 1 do
       assert(utf8.offset(s, 0, j) == pi)
     end
     for j = pi + 1, pi1 - 1 do
       assert(not utf8.len(s, j))
     end
-   assert(utf8.len(s, pi, pi) == 1)
-   assert(utf8.len(s, pi, pi1 - 1) == 1)
-   assert(utf8.len(s, pi) == l - i + 1)
-   assert(utf8.len(s, pi1) == l - i)
-   assert(utf8.len(s, 1, pi) == i)
+   assert(utf8.len(s, pi, pi, nonstrict) == 1)
+   assert(utf8.len(s, pi, pi1 - 1, nonstrict) == 1)
+   assert(utf8.len(s, pi, -1, nonstrict) == l - i + 1)
+   assert(utf8.len(s, pi1, -1, nonstrict) == l - i)
+   assert(utf8.len(s, 1, pi, -1, nonstrict) == i)
   end
 
   local i = 0
-  for p, c in utf8.codes(s) do
+  for p, c in utf8.codes(s, nonstrict) do
     i = i + 1
     assert(c == t[i] and p == utf8.offset(s, i))
-    assert(utf8.codepoint(s, p) == c)
-  end
-  assert(i == #t)
-
-  i = 0
-  for p, c in utf8.codes(s) do
-    i = i + 1
-    assert(c == t[i] and p == utf8.offset(s, i)) 
+    assert(utf8.codepoint(s, p, p, nonstrict) == c)
   end
   assert(i == #t)
 
@@ -105,13 +102,17 @@ do    -- error indication in utf8.len
   check("\xF4\x9F\xBF\xBF", 1)
 end
 
--- error in utf8.codes
-checkerror("invalid UTF%-8 code",
-  function ()
-    local s = "ab\xff"
-    for c in utf8.codes(s) do assert(c) end
-  end)
-
+-- errors in utf8.codes
+do
+  local function errorcodes (s)
+    checkerror("invalid UTF%-8 code",
+      function ()
+        for c in utf8.codes(s) do assert(c) end
+      end)
+  end
+  errorcodes("ab\xff")
+  errorcodes("\u{110000}")
+end
 
 -- error in initial position for offset
 checkerror("position out of range", utf8.offset, "abc", 1, 5)
@@ -141,14 +142,22 @@ do
   assert(#t == 0)
   checkerror("out of range", utf8.codepoint, s, -(#s + 1), 1)
   checkerror("out of range", utf8.codepoint, s, 1, #s + 1)
+  -- surrogates
+  assert(utf8.codepoint("\u{D7FF}") == 0xD800 - 1)
+  assert(utf8.codepoint("\u{E000}") == 0xDFFF + 1)
+  assert(utf8.codepoint("\u{D800}", 1, 1, true) == 0xD800)
+  assert(utf8.codepoint("\u{DFFF}", 1, 1, true) == 0xDFFF)
+  assert(utf8.codepoint("\u{7FFFFFFF}", 1, 1, true) == 0x7FFFFFFF)
 end
 
 assert(utf8.char() == "")
-assert(utf8.char(97, 98, 99) == "abc")
+assert(utf8.char(0, 97, 98, 99, 1) == "\0abc\1")
 
 assert(utf8.codepoint(utf8.char(0x10FFFF)) == 0x10FFFF)
+assert(utf8.codepoint(utf8.char(0x7FFFFFFF), 1, 1, true) == (1<<31) - 1)
 
-checkerror("value out of range", utf8.char, 0x10FFFF + 1)
+checkerror("value out of range", utf8.char, 0x7FFFFFFF + 1)
+checkerror("value out of range", utf8.char, -1)
 
 local function invalid (s)
   checkerror("invalid UTF%-8 code", utf8.codepoint, s)
@@ -158,6 +167,10 @@ end
 -- UTF-8 representation for 0x11ffff (value out of valid range)
 invalid("\xF4\x9F\xBF\xBF")
 
+-- surrogates
+invalid("\u{D800}")
+invalid("\u{DFFF}")
+
 -- overlong sequences
 invalid("\xC0\x80")          -- zero
 invalid("\xC1\xBF")          -- 0x7F (should be coded in 1 byte)
@@ -183,6 +196,21 @@ s = "\0 \x7F\z
 s = string.gsub(s, " ", "")
 check(s, {0,0x7F, 0x80,0x7FF, 0x800,0xFFFF, 0x10000,0x10FFFF})
 
+do
+  -- original UTF-8 values
+  local s = "\u{4000000}\u{7FFFFFFF}"
+  assert(#s == 12)
+  check(s, {0x4000000, 0x7FFFFFFF}, true)
+
+  s = "\u{200000}\u{3FFFFFF}"
+  assert(#s == 10)
+  check(s, {0x200000, 0x3FFFFFF}, true)
+
+  s = "\u{10000}\u{1fffff}"
+  assert(#s == 8)
+  check(s, {0x10000, 0x1FFFFF}, true)
+end
+
 x = "日本語a-4\0éó"
 check(x, {26085, 26412, 35486, 97, 45, 52, 0, 233, 243})