tests/lapi: add utf8 tests

tests/lapi/utf8_char_test.lua

@@ -0,0 +1,43 @@
+--[[
+SPDX-License-Identifier: ISC
+Copyright (c) 2023-2025, Sergey Bronnikov.
+
+6.5 – UTF-8 Support
+https://www.lua.org/manual/5.3/manual.html#6.5
+
+Synopsis: utf8.char(...)
+]]
+
+local luzer = require("luzer")
+local test_lib = require("lib")
+local MAX_INT = test_lib.MAX_INT
+local MIN_INT = test_lib.MIN_INT
+
+-- The function is introduced in Lua 5.3.
+if test_lib.lua_current_version_lt_than(5, 3) then
+    print("Unsupported version.")
+    os.exit()
+end
+
+local unpack = unpack or table.unpack
+
+local ignored_msgs = {
+    "value out of range",
+}
+
+local function TestOneInput(buf)
+    local fdp = luzer.FuzzedDataProvider(buf)
+    -- Limit count to prevent error "too many results to unpack".
+    local MAX_N = 1000
+    local count = fdp:consume_integer(1, MAX_N)
+    local chars = fdp:consume_integers(MIN_INT, MAX_INT, count)
+    os.setlocale(test_lib.random_locale(fdp), "all")
+    local err_handler = test_lib.err_handler(ignored_msgs)
+    local ok, _ = xpcall(utf8.char, err_handler, unpack(chars))
+    if not ok then return end
+end
+
+local args = {
+    artifact_prefix = "utf8_char_",
+}
+luzer.Fuzz(TestOneInput, nil, args)

tests/lapi/utf8_codepoint_test.lua

@@ -0,0 +1,35 @@
+--[=[[
+SPDX-License-Identifier: ISC
+Copyright (c) 2023-2025, Sergey Bronnikov.
+
+6.5 – UTF-8 Support
+https://www.lua.org/manual/5.3/manual.html#6.5
+
+Synopsis: utf8.codepoint(s [, i [, j [, lax]]])
+]]=]
+
+local luzer = require("luzer")
+local test_lib = require("lib")
+local MAX_INT = test_lib.MAX_INT
+
+-- The function is introduced in Lua 5.3.
+if test_lib.lua_current_version_lt_than(5, 3) then
+    print("Unsupported version.")
+    os.exit()
+end
+
+local function TestOneInput(buf)
+    local fdp = luzer.FuzzedDataProvider(buf)
+    local max_len = fdp:consume_integer(1, MAX_INT)
+    local s = fdp:consume_string(max_len)
+    local i = fdp:consume_integer(0, MAX_INT)
+    local j = fdp:consume_integer(0, MAX_INT)
+    local lax = fdp:consume_boolean()
+    os.setlocale(test_lib.random_locale(fdp), "all")
+    pcall(utf8.codepoint, s, i, j, lax)
+end
+
+local args = {
+    artifact_prefix = "utf8_codepoint_",
+}
+luzer.Fuzz(TestOneInput, nil, args)

tests/lapi/utf8_codes_test.lua

@@ -0,0 +1,42 @@
+--[[
+SPDX-License-Identifier: ISC
+Copyright (c) 2023-2025, Sergey Bronnikov.
+
+6.5 – UTF-8 Support
+https://www.lua.org/manual/5.3/manual.html#6.5
+
+'utf8.codes' does not raise an error on spurious continuation bytes,
+https://github.com/lua/lua/commit/a1089b415a3f5c753aa1b40758ffdaf28d5701b0
+
+Synopsis: utf8.codes(s [, lax])
+]]
+
+local luzer = require("luzer")
+local test_lib = require("lib")
+local MAX_INT = test_lib.MAX_INT
+
+-- The function is introduced in Lua 5.3.
+if test_lib.lua_current_version_lt_than(5, 3) then
+    print("Unsupported version.")
+    os.exit()
+end
+
+local ignored_msgs = {
+    "invalid UTF-8 code",
+}
+
+local function TestOneInput(buf)
+    local fdp = luzer.FuzzedDataProvider(buf)
+    local max_len = fdp:consume_integer(1, MAX_INT)
+    local s = fdp:consume_string(max_len)
+    local lax = fdp:consume_boolean()
+    os.setlocale(test_lib.random_locale(fdp), "all")
+    local err_handler = test_lib.err_handler(ignored_msgs)
+    local ok, _ = xpcall(utf8.codes, err_handler, s, lax)
+    if not ok then return end
+end
+
+local args = {
+    artifact_prefix = "utf8_codes_",
+}
+luzer.Fuzz(TestOneInput, nil, args)

tests/lapi/utf8_len_test.lua

@@ -0,0 +1,35 @@
+--[=[[
+SPDX-License-Identifier: ISC
+Copyright (c) 2023-2025, Sergey Bronnikov.
+
+6.5 – UTF-8 Support
+https://www.lua.org/manual/5.3/manual.html#6.5
+
+Synopsis: utf8.len(s [, i [, j [, lax]]])
+]]=]
+
+local luzer = require("luzer")
+local test_lib = require("lib")
+local MAX_INT = test_lib.MAX_INT
+
+-- The function is introduced in Lua 5.3.
+if test_lib.lua_current_version_lt_than(5, 3) then
+    print("Unsupported version.")
+    os.exit()
+end
+
+local function TestOneInput(buf)
+    local fdp = luzer.FuzzedDataProvider(buf)
+    local max_len = fdp:consume_integer(1, MAX_INT)
+    local s = fdp:consume_string(max_len)
+    local i = fdp:consume_integer(0, MAX_INT)
+    local j = fdp:consume_integer(0, MAX_INT)
+    local lax = fdp:consume_boolean()
+    os.setlocale(test_lib.random_locale(fdp), "all")
+    pcall(utf8.len, s, i, j, lax)
+end
+
+local args = {
+    artifact_prefix = "utf8_len_",
+}
+luzer.Fuzz(TestOneInput, nil, args)

tests/lapi/utf8_offset_test.lua

@@ -0,0 +1,35 @@
+--[[
+SPDX-License-Identifier: ISC
+Copyright (c) 2023-2025, Sergey Bronnikov.
+
+6.5 – UTF-8 Support
+https://www.lua.org/manual/5.3/manual.html#6.5
+
+Synopsis: utf8.offset(s, n [, i])
+]]
+
+local luzer = require("luzer")
+local test_lib = require("lib")
+local MAX_INT = test_lib.MAX_INT
+local MIN_INT = test_lib.MIN_INT
+
+-- The function is introduced in Lua 5.3.
+if test_lib.lua_current_version_lt_than(5, 3) then
+    print("Unsupported version.")
+    os.exit()
+end
+
+local function TestOneInput(buf)
+    local fdp = luzer.FuzzedDataProvider(buf)
+    local max_len = fdp:consume_integer(0, MAX_INT)
+    local s = fdp:consume_string(max_len)
+    local n = fdp:consume_integer(MIN_INT, MAX_INT)
+    local i = fdp:consume_integer(1, MAX_INT)
+    os.setlocale(test_lib.random_locale(fdp), "all")
+    utf8.offset(s, n, i)
+end
+
+local args = {
+    artifact_prefix = "utf8_offset_",
+}
+luzer.Fuzz(TestOneInput, nil, args)