neovim/test/unit/testutil.lua

913 lines
24 KiB
Lua
Raw Normal View History

2014-08-30 06:35:42 -07:00
local ffi = require('ffi')
local formatc = require('test.unit.formatc')
local Set = require('test.unit.set')
local Preprocess = require('test.unit.preprocess')
local t_global = require('test.testutil')
local paths = t_global.paths
local assert = require('luassert')
local say = require('say')
local check_cores = t_global.check_cores
local dedent = t_global.dedent
local neq = t_global.neq
local map = vim.tbl_map
local eq = t_global.eq
local trim = vim.trim
2014-08-30 06:35:42 -07:00
-- add some standard header locations
2024-01-15 12:41:22 -07:00
for _, p in ipairs(paths.include_paths) do
2014-08-30 06:35:42 -07:00
Preprocess.add_to_include_path(p)
end
2023-12-13 06:04:24 -07:00
local child_pid = nil --- @type integer?
2023-04-03 04:01:23 -07:00
--- @generic F: function
--- @param func F
--- @return F
local function only_separate(func)
return function(...)
2017-03-11 13:48:16 -07:00
if child_pid ~= 0 then
error('This function must be run in a separate process only')
end
return func(...)
end
end
2023-04-03 04:01:23 -07:00
--- @class ChildCall
--- @field func function
--- @field args any[]
--- @class ChildCallLog
--- @field func string
--- @field args any[]
--- @field ret any?
local child_calls_init = {} --- @type ChildCall[]
local child_calls_mod = nil --- @type ChildCall[]
local child_calls_mod_once = nil --- @type ChildCall[]?
local function child_call(func, ret)
return function(...)
local child_calls = child_calls_mod or child_calls_init
2017-03-11 13:48:16 -07:00
if child_pid ~= 0 then
child_calls[#child_calls + 1] = { func = func, args = { ... } }
return ret
else
return func(...)
end
end
end
-- Run some code at the start of the child process, before running the test
-- itself. Is supposed to be run in `before_each`.
2023-04-03 04:01:23 -07:00
--- @param func function
local function child_call_once(func, ...)
if child_pid ~= 0 then
child_calls_mod_once[#child_calls_mod_once + 1] = { func = func, args = { ... } }
else
func(...)
end
end
2023-04-03 04:01:23 -07:00
local child_cleanups_mod_once = nil --- @type ChildCall[]?
-- Run some code at the end of the child process, before exiting. Is supposed to
-- be run in `before_each` because `after_each` is run after child has exited.
local function child_cleanup_once(func, ...)
local child_cleanups = child_cleanups_mod_once
if child_pid ~= 0 then
child_cleanups[#child_cleanups + 1] = { func = func, args = { ... } }
else
func(...)
end
end
-- Unittests are run from debug nvim binary in lua interpreter mode.
local libnvim = ffi.C
local lib = setmetatable({}, {
2017-03-11 13:48:16 -07:00
__index = only_separate(function(_, idx)
return libnvim[idx]
end),
__newindex = child_call(function(_, idx, val)
libnvim[idx] = val
end),
})
local init = only_separate(function()
for _, c in ipairs(child_calls_init) do
c.func(unpack(c.args))
end
libnvim.event_init()
libnvim.early_init(nil)
if child_calls_mod then
for _, c in ipairs(child_calls_mod) do
c.func(unpack(c.args))
end
end
if child_calls_mod_once then
for _, c in ipairs(child_calls_mod_once) do
c.func(unpack(c.args))
end
child_calls_mod_once = nil
end
end)
local deinit = only_separate(function()
if child_cleanups_mod_once then
for _, c in ipairs(child_cleanups_mod_once) do
c.func(unpack(c.args))
end
child_cleanups_mod_once = nil
end
end)
2014-08-30 06:35:42 -07:00
-- a Set that keeps around the lines we've already seen
local cdefs_init = Set:new()
local cdefs_mod = nil
local imported = Set:new()
local pragma_pack_id = 1
2014-08-30 06:35:42 -07:00
-- some things are just too complex for the LuaJIT C parser to digest. We
-- usually don't need them anyway.
2023-04-03 04:01:23 -07:00
--- @param body string
local function filter_complex_blocks(body)
2023-04-03 04:01:23 -07:00
local result = {} --- @type string[]
2014-08-30 06:35:42 -07:00
for line in body:gmatch('[^\r\n]+') do
if
not (
string.find(line, '(^)', 1, true) ~= nil
or string.find(line, '_ISwupper', 1, true)
or string.find(line, '_Float')
or string.find(line, '__s128')
or string.find(line, '__u128')
or string.find(line, 'msgpack_zone_push_finalizer')
or string.find(line, 'msgpack_unpacker_reserve_buffer')
or string.find(line, 'value_init_')
or string.find(line, 'UUID_NULL') -- static const uuid_t UUID_NULL = {...}
or string.find(line, 'inline _Bool')
-- used by macOS headers
or string.find(line, 'typedef enum : ')
or string.find(line, 'mach_vm_range_recipe')
)
then
2014-08-30 06:35:42 -07:00
result[#result + 1] = line
end
end
return table.concat(result, '\n')
2014-08-30 06:35:42 -07:00
end
local cdef = ffi.cdef
local cimportstr
local previous_defines_init = [[
typedef struct { char bytes[16]; } __attribute__((aligned(16))) __uint128_t;
typedef struct { char bytes[16]; } __attribute__((aligned(16))) __float128;
]]
2023-04-03 04:01:23 -07:00
local preprocess_cache_init = {} --- @type table<string,string>
local previous_defines_mod = ''
2023-04-03 04:01:23 -07:00
local preprocess_cache_mod = nil --- @type table<string,string>
local function is_child_cdefs()
2023-04-03 04:01:23 -07:00
return os.getenv('NVIM_TEST_MAIN_CDEFS') ~= '1'
end
2014-08-30 06:35:42 -07:00
-- use this helper to import C files, you can pass multiple paths at once,
-- this helper will return the C namespace of the nvim library.
2023-04-03 04:01:23 -07:00
local function cimport(...)
local previous_defines --- @type string
local preprocess_cache --- @type table<string,string>
local cdefs
if is_child_cdefs() and preprocess_cache_mod then
preprocess_cache = preprocess_cache_mod
previous_defines = previous_defines_mod
cdefs = cdefs_mod
else
preprocess_cache = preprocess_cache_init
previous_defines = previous_defines_init
cdefs = cdefs_init
2014-08-30 06:35:42 -07:00
end
for _, path in ipairs({ ... }) do
if not (path:sub(1, 1) == '/' or path:sub(1, 1) == '.' or path:sub(2, 2) == ':') then
path = './' .. path
end
2017-03-11 13:48:16 -07:00
if not preprocess_cache[path] then
2023-04-03 04:01:23 -07:00
local body --- @type string
body, previous_defines = Preprocess.preprocess(previous_defines, path)
-- format it (so that the lines are "unique" statements), also filter out
-- Objective-C blocks
if os.getenv('NVIM_TEST_PRINT_I') == '1' then
local lnum = 0
for line in body:gmatch('[^\n]+') do
lnum = lnum + 1
print(lnum, line)
end
end
body = formatc(body)
body = filter_complex_blocks(body)
-- add the formatted lines to a set
local new_cdefs = Set:new()
for line in body:gmatch('[^\r\n]+') do
line = trim(line)
-- give each #pragma pack a unique id, so that they don't get removed
-- if they are inserted into the set
-- (they are needed in the right order with the struct definitions,
-- otherwise luajit has wrong memory layouts for the structs)
if line:match('#pragma%s+pack') then
2023-04-03 04:01:23 -07:00
--- @type string
line = line .. ' // ' .. pragma_pack_id
pragma_pack_id = pragma_pack_id + 1
end
new_cdefs:add(line)
end
-- subtract the lines we've already imported from the new lines, then add
-- the new unique lines to the old lines (so they won't be imported again)
new_cdefs:diff(cdefs)
cdefs:union(new_cdefs)
-- request a sorted version of the new lines (same relative order as the
-- original preprocessed file) and feed that to the LuaJIT ffi
local new_lines = new_cdefs:to_table()
if os.getenv('NVIM_TEST_PRINT_CDEF') == '1' then
for lnum, line in ipairs(new_lines) do
print(lnum, line)
end
end
body = table.concat(new_lines, '\n')
preprocess_cache[path] = body
end
cimportstr(preprocess_cache, path)
2014-08-30 06:35:42 -07:00
end
return lib
end
2014-08-30 06:35:42 -07:00
2023-04-03 04:01:23 -07:00
local function cimport_immediate(...)
2017-03-11 13:48:16 -07:00
local saved_pid = child_pid
child_pid = 0
local err, emsg = pcall(cimport, ...)
2017-03-11 13:48:16 -07:00
child_pid = saved_pid
if not err then
2023-04-03 04:01:23 -07:00
io.stderr:write(tostring(emsg) .. '\n')
assert(false)
else
return lib
2014-08-30 06:35:42 -07:00
end
end
2023-04-03 04:01:23 -07:00
--- @param preprocess_cache table<string,string[]>
--- @param path string
local function _cimportstr(preprocess_cache, path)
if imported:contains(path) then
return lib
end
local body = preprocess_cache[path]
if body == '' then
return lib
2014-08-30 06:35:42 -07:00
end
cdef(body)
imported:add(path)
2014-08-30 06:35:42 -07:00
return lib
end
if is_child_cdefs() then
cimportstr = child_call(_cimportstr, lib)
else
cimportstr = _cimportstr
end
2014-08-30 06:35:42 -07:00
2017-01-07 09:12:18 -07:00
local function alloc_log_new()
local log = {
log = {}, --- @type ChildCallLog[]
lib = cimport('./src/nvim/memory.h'), --- @type table<string,function>
original_functions = {}, --- @type table<string,function>
null = { ['\0:is_null'] = true },
}
2023-04-03 04:01:23 -07:00
local allocator_functions = { 'malloc', 'free', 'calloc', 'realloc' }
2023-04-03 04:01:23 -07:00
function log:save_original_functions()
for _, funcname in ipairs(allocator_functions) do
if not self.original_functions[funcname] then
self.original_functions[funcname] = self.lib['mem_' .. funcname]
end
end
end
2023-04-03 04:01:23 -07:00
log.save_original_functions = child_call(log.save_original_functions)
2023-04-03 04:01:23 -07:00
function log:set_mocks()
for _, k in ipairs(allocator_functions) do
do
local kk = k
self.lib['mem_' .. k] = function(...)
2023-04-03 04:01:23 -07:00
--- @type ChildCallLog
local log_entry = { func = kk, args = { ... } }
self.log[#self.log + 1] = log_entry
if kk == 'free' then
self.original_functions[kk](...)
else
log_entry.ret = self.original_functions[kk](...)
end
for i, v in ipairs(log_entry.args) do
if v == nil then
-- XXX This thing thinks that {NULL} ~= {NULL}.
log_entry.args[i] = self.null
end
end
if self.hook then
self:hook(log_entry)
end
if log_entry.ret then
return log_entry.ret
end
end
end
end
end
2023-04-03 04:01:23 -07:00
log.set_mocks = child_call(log.set_mocks)
2023-04-03 04:01:23 -07:00
function log:clear()
self.log = {}
end
2023-04-03 04:01:23 -07:00
function log:check(exp)
eq(exp, self.log)
self:clear()
end
2023-04-03 04:01:23 -07:00
function log:clear_tmp_allocs(clear_null_frees)
2023-04-03 04:01:23 -07:00
local toremove = {} --- @type integer[]
local allocs = {} --- @type table<string,integer>
for i, v in ipairs(self.log) do
if v.func == 'malloc' or v.func == 'calloc' then
allocs[tostring(v.ret)] = i
elseif v.func == 'realloc' or v.func == 'free' then
if allocs[tostring(v.args[1])] then
toremove[#toremove + 1] = allocs[tostring(v.args[1])]
if v.func == 'free' then
toremove[#toremove + 1] = i
end
elseif clear_null_frees and v.args[1] == self.null then
toremove[#toremove + 1] = i
end
if v.func == 'realloc' then
allocs[tostring(v.ret)] = i
end
end
end
table.sort(toremove)
for i = #toremove, 1, -1 do
table.remove(self.log, toremove[i])
end
end
2023-04-03 04:01:23 -07:00
function log:setup()
log:save_original_functions()
log:set_mocks()
end
2023-04-03 04:01:23 -07:00
function log:before_each() end
2023-04-03 04:01:23 -07:00
function log:after_each() end
2023-04-03 04:01:23 -07:00
log:setup()
2023-04-03 04:01:23 -07:00
return log
end
2014-08-30 06:35:42 -07:00
-- take a pointer to a C-allocated string and return an interned
-- version while also freeing the memory
local function internalize(cdata, len)
2014-08-30 06:35:42 -07:00
ffi.gc(cdata, ffi.C.free)
return ffi.string(cdata, len)
end
local cstr = ffi.typeof('char[?]')
local function to_cstr(string)
return cstr(#string + 1, string)
2014-08-30 06:35:42 -07:00
end
cimport_immediate('./test/unit/fixtures/posix.h')
2023-04-03 04:01:23 -07:00
local sc = {}
function sc.fork()
return tonumber(ffi.C.fork())
end
function sc.pipe()
local ret = ffi.new('int[2]', { -1, -1 })
2023-04-03 04:01:23 -07:00
ffi.errno(0)
local res = ffi.C.pipe(ret)
if res ~= 0 then
2023-04-03 04:01:23 -07:00
local err = ffi.errno(0)
assert(res == 0, ('pipe() error: %u: %s'):format(err, ffi.string(ffi.C.strerror(err))))
2023-04-03 04:01:23 -07:00
end
assert(ret[0] ~= -1 and ret[1] ~= -1)
return ret[0], ret[1]
end
--- @return string
function sc.read(rd, len)
local ret = ffi.new('char[?]', len, { 0 })
2023-04-03 04:01:23 -07:00
local total_bytes_read = 0
ffi.errno(0)
while total_bytes_read < len do
local bytes_read =
tonumber(ffi.C.read(rd, ffi.cast('void*', ret + total_bytes_read), len - total_bytes_read))
2023-04-03 04:01:23 -07:00
if bytes_read == -1 then
local err = ffi.errno(0)
2023-04-03 04:01:23 -07:00
if err ~= ffi.C.kPOSIXErrnoEINTR then
assert(false, ('read() error: %u: %s'):format(err, ffi.string(ffi.C.strerror(err))))
end
2023-04-03 04:01:23 -07:00
elseif bytes_read == 0 then
break
else
total_bytes_read = total_bytes_read + bytes_read
end
2023-04-03 04:01:23 -07:00
end
return ffi.string(ret, total_bytes_read)
end
function sc.write(wr, s)
local wbuf = to_cstr(s)
local total_bytes_written = 0
ffi.errno(0)
while total_bytes_written < #s do
local bytes_written = tonumber(
ffi.C.write(wr, ffi.cast('void*', wbuf + total_bytes_written), #s - total_bytes_written)
)
2023-04-03 04:01:23 -07:00
if bytes_written == -1 then
local err = ffi.errno(0)
if err ~= ffi.C.kPOSIXErrnoEINTR then
assert(
false,
("write() error: %u: %s ('%s')"):format(err, ffi.string(ffi.C.strerror(err)), s)
)
end
2023-04-03 04:01:23 -07:00
elseif bytes_written == 0 then
break
else
total_bytes_written = total_bytes_written + bytes_written
end
2023-04-03 04:01:23 -07:00
end
return total_bytes_written
end
sc.close = ffi.C.close
--- @param pid integer
--- @return integer
function sc.wait(pid)
ffi.errno(0)
local stat_loc = ffi.new('int[1]', { 0 })
2023-04-03 04:01:23 -07:00
while true do
local r = ffi.C.waitpid(pid, stat_loc, ffi.C.kPOSIXWaitWUNTRACED)
if r == -1 then
local err = ffi.errno(0)
if err == ffi.C.kPOSIXErrnoECHILD then
break
elseif err ~= ffi.C.kPOSIXErrnoEINTR then
assert(false, ('waitpid() error: %u: %s'):format(err, ffi.string(ffi.C.strerror(err))))
end
2023-04-03 04:01:23 -07:00
else
assert(r == pid)
end
2023-04-03 04:01:23 -07:00
end
return stat_loc[0]
end
sc.exit = ffi.C._exit
2023-04-03 04:01:23 -07:00
--- @param lst string[]
--- @return string
2016-11-01 02:20:42 -07:00
local function format_list(lst)
2023-04-03 04:01:23 -07:00
local ret = {} --- @type string[]
2016-11-01 02:20:42 -07:00
for _, v in ipairs(lst) do
ret[#ret + 1] = assert:format({ v, n = 1 })[1]
2016-11-01 02:20:42 -07:00
end
2023-04-03 04:01:23 -07:00
return table.concat(ret, ', ')
2016-11-01 02:20:42 -07:00
end
if os.getenv('NVIM_TEST_PRINT_SYSCALLS') == '1' then
for k_, v_ in pairs(sc) do
(function(k, v)
sc[k] = function(...)
local rets = { v(...) }
io.stderr:write(('%s(%s) = %s\n'):format(k, format_list({ ... }), format_list(rets)))
2016-11-01 02:20:42 -07:00
return unpack(rets)
end
end)(k_, v_)
end
end
local function just_fail(_)
return false
end
say:set('assertion.just_fail.positive', '%s')
say:set('assertion.just_fail.negative', '%s')
assert:register(
'assertion',
'just_fail',
just_fail,
'assertion.just_fail.positive',
'assertion.just_fail.negative'
)
local hook_fnamelen = 30
local hook_sfnamelen = 30
local hook_numlen = 5
local hook_msglen = 1 + 1 + 1 + (1 + hook_fnamelen) + (1 + hook_sfnamelen) + (1 + hook_numlen) + 1
local tracehelp = dedent([[
Trace: either in the format described below or custom debug output starting
with `>`. Latter lines still have the same width in byte.
Trace type: _r_eturn from function , function _c_all, _l_ine executed,
_t_ail return, _C_ount (should not actually appear),
_s_aved from previous run for reference, _>_ for custom debug
output.
Function type: _L_ua function, _C_ function, _m_ain part of chunk,
function that did _t_ail call.
Function name type: _g_lobal, _l_ocal, _m_ethod, _f_ield, _u_pvalue,
space for unknown.
Source file name Function name Line
(trunc to 30 bytes, no .lua) (truncated to last 30 bytes) number
CWN SSSSSSSSSSSSSSSSSSSSSSSSSSSSSS:FFFFFFFFFFFFFFFFFFFFFFFFFFFFFF:LLLLL\n
]])
local function child_sethook(wr)
2023-04-03 04:01:23 -07:00
local trace_level_str = os.getenv('NVIM_TEST_TRACE_LEVEL')
local trace_level = 0
if trace_level_str and trace_level_str ~= '' then
--- @type number
trace_level = assert(tonumber(trace_level_str))
end
2023-04-03 04:01:23 -07:00
if trace_level <= 0 then
return
end
2023-04-03 04:01:23 -07:00
local trace_only_c = trace_level <= 1
2023-04-03 04:01:23 -07:00
--- @type debuginfo?, string?, integer
local prev_info, prev_reason, prev_lnum
2023-04-03 04:01:23 -07:00
--- @param reason string
--- @param lnum integer
--- @param use_prev boolean
local function hook(reason, lnum, use_prev)
2023-04-03 04:01:23 -07:00
local info = nil --- @type debuginfo?
if use_prev then
info = prev_info
elseif reason ~= 'tail return' then -- tail return
info = debug.getinfo(2, 'nSl')
end
if trace_only_c and (not info or info.what ~= 'C') and not use_prev then
2023-04-03 04:01:23 -07:00
--- @cast info -nil
if info.source:sub(-9) == '_spec.lua' then
prev_info = info
prev_reason = 'saved'
prev_lnum = lnum
end
return
end
if trace_only_c and not use_prev and prev_reason then
hook(prev_reason, prev_lnum, true)
prev_reason = nil
end
local whatchar = ' '
local namewhatchar = ' '
local funcname = ''
local source = ''
local msgchar = reason:sub(1, 1)
if reason == 'count' then
msgchar = 'C'
end
if info then
funcname = (info.name or ''):sub(1, hook_fnamelen)
whatchar = info.what:sub(1, 1)
namewhatchar = info.namewhat:sub(1, 1)
if namewhatchar == '' then
namewhatchar = ' '
end
source = info.source
if source:sub(1, 1) == '@' then
if source:sub(-4, -1) == '.lua' then
source = source:sub(1, -5)
end
source = source:sub(-hook_sfnamelen, -1)
end
lnum = lnum or info.currentline
end
-- assert(-1 <= lnum and lnum <= 99999)
2023-04-03 04:01:23 -07:00
local lnum_s = lnum == -1 and 'nknwn' or ('%u'):format(lnum)
--- @type string
local msg = ( -- lua does not support %*
''
.. msgchar
.. whatchar
.. namewhatchar
.. ' '
.. source
.. (' '):rep(hook_sfnamelen - #source)
.. ':'
.. funcname
.. (' '):rep(hook_fnamelen - #funcname)
.. ':'
.. ('0'):rep(hook_numlen - #lnum_s)
.. lnum_s
.. '\n'
)
-- eq(hook_msglen, #msg)
sc.write(wr, msg)
end
debug.sethook(hook, 'crl')
end
local trace_end_msg = ('E%s\n'):format((' '):rep(hook_msglen - 2))
2023-04-03 04:01:23 -07:00
--- @type function
local _debug_log
local debug_log = only_separate(function(...)
return _debug_log(...)
end)
local function itp_child(wr, func)
2023-04-03 04:01:23 -07:00
--- @param s string
_debug_log = function(s)
s = s:sub(1, hook_msglen - 2)
sc.write(wr, '>' .. s .. (' '):rep(hook_msglen - 2 - #s) .. '\n')
end
local status, result = pcall(init)
if status then
collectgarbage('stop')
child_sethook(wr)
status, result = pcall(func)
debug.sethook()
end
sc.write(wr, trace_end_msg)
if not status then
local emsg = tostring(result)
if #emsg > 99999 then
emsg = emsg:sub(1, 99999)
end
sc.write(wr, ('-\n%05u\n%s'):format(#emsg, emsg))
deinit()
else
sc.write(wr, '+\n')
deinit()
end
collectgarbage('restart')
collectgarbage()
sc.write(wr, '$\n')
sc.close(wr)
sc.exit(status and 0 or 1)
end
local function check_child_err(rd)
2023-04-03 04:01:23 -07:00
local trace = {} --- @type string[]
local did_traceline = false
local maxtrace = tonumber(os.getenv('NVIM_TEST_MAXTRACE')) or 1024
while true do
local traceline = sc.read(rd, hook_msglen)
if #traceline ~= hook_msglen then
if #traceline == 0 then
break
else
trace[#trace + 1] = 'Partial read: <' .. trace .. '>\n'
end
end
if traceline == trace_end_msg then
did_traceline = true
break
end
trace[#trace + 1] = traceline
if #trace > maxtrace then
table.remove(trace, 1)
end
end
local res = sc.read(rd, 2)
if #res == 2 then
local err = ''
if res ~= '+\n' then
eq('-\n', res)
local len_s = sc.read(rd, 5)
local len = tonumber(len_s)
neq(0, len)
if os.getenv('NVIM_TEST_TRACE_ON_ERROR') == '1' and #trace ~= 0 then
2023-04-03 04:01:23 -07:00
--- @type string
err = '\nTest failed, trace:\n' .. tracehelp
for _, traceline in ipairs(trace) do
2023-04-03 04:01:23 -07:00
--- @type string
err = err .. traceline
end
end
2023-04-03 04:01:23 -07:00
--- @type string
err = err .. sc.read(rd, len + 1)
end
local eres = sc.read(rd, 2)
if eres ~= '$\n' then
if #trace == 0 then
err = '\nTest crashed, no trace available (check NVIM_TEST_TRACE_LEVEL)\n'
else
err = '\nTest crashed, trace:\n' .. tracehelp
for i = 1, #trace do
err = err .. trace[i]
end
end
if not did_traceline then
2023-04-03 04:01:23 -07:00
--- @type string
err = err .. '\nNo end of trace occurred'
end
2024-01-15 12:41:22 -07:00
local cc_err, cc_emsg = pcall(check_cores, paths.test_luajit_prg, true)
if not cc_err then
2023-04-03 04:01:23 -07:00
--- @type string
err = err .. '\ncheck_cores failed: ' .. cc_emsg
end
end
if err ~= '' then
assert.just_fail(err)
end
end
end
local function itp_parent(rd, pid, allow_failure, location)
local ok, emsg = pcall(check_child_err, rd)
local status = sc.wait(pid)
sc.close(rd)
if not ok then
if allow_failure then
io.stderr:write('Errorred out (' .. status .. '):\n' .. tostring(emsg) .. '\n')
os.execute([[
sh -c "source ci/common/test.sh
2024-01-15 12:41:22 -07:00
check_core_dumps --delete \"]] .. paths.test_luajit_prg .. [[\""]])
else
error(tostring(emsg) .. '\nexit code: ' .. status)
end
elseif status ~= 0 then
if not allow_failure then
error('child process errored out with status ' .. status .. '!\n\n' .. location)
end
end
end
local function gen_itp(it)
child_calls_mod = {}
child_calls_mod_once = {}
child_cleanups_mod_once = {}
preprocess_cache_mod = map(function(v)
return v
end, preprocess_cache_init)
previous_defines_mod = previous_defines_init
cdefs_mod = cdefs_init:copy()
2016-11-04 08:20:58 -07:00
local function itp(name, func, allow_failure)
if allow_failure and os.getenv('NVIM_TEST_RUN_FAILING_TESTS') ~= '1' then
-- FIXME Fix tests with this true
return
end
-- Pre-emptively calculating error location, wasteful, ugh!
-- But the way this code messes around with busted implies the real location is strictly
-- not available in the parent when an actual error occurs. so we have to do this here.
local location = debug.traceback()
it(name, function()
local rd, wr = sc.pipe()
2017-03-11 13:48:16 -07:00
child_pid = sc.fork()
if child_pid == 0 then
sc.close(rd)
itp_child(wr, func)
else
sc.close(wr)
2017-04-01 03:17:25 -07:00
local saved_child_pid = child_pid
2017-03-11 13:48:16 -07:00
child_pid = nil
itp_parent(rd, saved_child_pid, allow_failure, location)
end
end)
end
return itp
end
local function cppimport(path)
2024-01-15 12:41:22 -07:00
return cimport(paths.test_source_path .. '/test/includes/pre/' .. path)
end
cimport(
'./src/nvim/types_defs.h',
'./src/nvim/main.h',
'./src/nvim/os/time.h',
'./src/nvim/os/fs.h'
)
local function conv_enum(etab, eval)
local n = tonumber(eval)
return etab[n] or n
end
local function array_size(arr)
return ffi.sizeof(arr) / ffi.sizeof(arr[0])
end
local function kvi_size(kvi)
return array_size(kvi.init_array)
end
local function kvi_init(kvi)
kvi.capacity = kvi_size(kvi)
kvi.items = kvi.init_array
return kvi
end
local function kvi_destroy(kvi)
if kvi.items ~= kvi.init_array then
lib.xfree(kvi.items)
end
end
local function kvi_new(ct)
return kvi_init(ffi.new(ct))
end
2017-10-29 15:48:32 -07:00
local function make_enum_conv_tab(m, values, skip_pref, set_cb)
child_call_once(function()
local ret = {}
for _, v in ipairs(values) do
local str_v = v
if v:sub(1, #skip_pref) == skip_pref then
str_v = v:sub(#skip_pref + 1)
end
2017-10-29 15:48:32 -07:00
ret[tonumber(m[v])] = str_v
end
set_cb(ret)
end)
end
local function ptr2addr(ptr)
return tonumber(ffi.cast('intptr_t', ffi.cast('void *', ptr)))
end
local s = ffi.new('char[64]', { 0 })
local function ptr2key(ptr)
ffi.C.snprintf(s, ffi.sizeof(s), '%p', ffi.cast('void *', ptr))
return ffi.string(s)
end
feat(extmark): support proper multiline ranges The removes the previous restriction that nvim_buf_set_extmark() could not be used to highlight arbitrary multi-line regions The problem can be summarized as follows: let's assume an extmark with a hl_group is placed covering the region (5,0) to (50,0) Now, consider what happens if nvim needs to redraw a window covering the lines 20-30. It needs to be able to ask the marktree what extmarks cover this region, even if they don't begin or end here. Therefore the marktree needs to be augmented with the information covers a point, not just what marks begin or end there. To do this, we augment each node with a field "intersect" which is a set the ids of the marks which overlap this node, but only if it is not part of the set of any parent. This ensures the number of nodes that need to be explicitly marked grows only logarithmically with the total number of explicitly nodes (and thus the number of of overlapping marks). Thus we can quickly iterate all marks which overlaps any query position by looking up what leaf node contains that position. Then we only need to consider all "start" marks within that leaf node, and the "intersect" set of that node and all its parents. Now, and the major source of complexity is that the tree restructuring operations (to ensure that each node has T-1 <= size <= 2*T-1) also need to update these sets. If a full inner node is split in two, one of the new parents might start to completely overlap some ranges and its ids will need to be moved from its children's sets to its own set. Similarly, if two undersized nodes gets joined into one, it might no longer completely overlap some ranges, and now the children which do needs to have the have the ids in its set instead. And then there are the pivots! Yes the pivot operations when a child gets moved from one parent to another.
2020-11-22 02:10:37 -07:00
local function is_asan()
cimport('./src/nvim/version.h')
local status, res = pcall(function()
return lib.version_cflags
end)
feat(extmark): support proper multiline ranges The removes the previous restriction that nvim_buf_set_extmark() could not be used to highlight arbitrary multi-line regions The problem can be summarized as follows: let's assume an extmark with a hl_group is placed covering the region (5,0) to (50,0) Now, consider what happens if nvim needs to redraw a window covering the lines 20-30. It needs to be able to ask the marktree what extmarks cover this region, even if they don't begin or end here. Therefore the marktree needs to be augmented with the information covers a point, not just what marks begin or end there. To do this, we augment each node with a field "intersect" which is a set the ids of the marks which overlap this node, but only if it is not part of the set of any parent. This ensures the number of nodes that need to be explicitly marked grows only logarithmically with the total number of explicitly nodes (and thus the number of of overlapping marks). Thus we can quickly iterate all marks which overlaps any query position by looking up what leaf node contains that position. Then we only need to consider all "start" marks within that leaf node, and the "intersect" set of that node and all its parents. Now, and the major source of complexity is that the tree restructuring operations (to ensure that each node has T-1 <= size <= 2*T-1) also need to update these sets. If a full inner node is split in two, one of the new parents might start to completely overlap some ranges and its ids will need to be moved from its children's sets to its own set. Similarly, if two undersized nodes gets joined into one, it might no longer completely overlap some ranges, and now the children which do needs to have the have the ids in its set instead. And then there are the pivots! Yes the pivot operations when a child gets moved from one parent to another.
2020-11-22 02:10:37 -07:00
if status then
return ffi.string(res):match('-fsanitize=[a-z,]*address')
else
return false
end
end
--- @class test.unit.testutil.module
local M = {
2014-08-30 06:35:42 -07:00
cimport = cimport,
cppimport = cppimport,
internalize = internalize,
ffi = ffi,
lib = lib,
2014-08-30 06:35:42 -07:00
cstr = cstr,
to_cstr = to_cstr,
2023-04-03 04:01:23 -07:00
NULL = ffi.cast('void*', 0),
OK = 1,
FAIL = 0,
alloc_log_new = alloc_log_new,
gen_itp = gen_itp,
only_separate = only_separate,
child_call_once = child_call_once,
child_cleanup_once = child_cleanup_once,
sc = sc,
conv_enum = conv_enum,
2017-10-29 15:48:32 -07:00
array_size = array_size,
kvi_destroy = kvi_destroy,
kvi_size = kvi_size,
kvi_init = kvi_init,
kvi_new = kvi_new,
make_enum_conv_tab = make_enum_conv_tab,
ptr2addr = ptr2addr,
ptr2key = ptr2key,
debug_log = debug_log,
feat(extmark): support proper multiline ranges The removes the previous restriction that nvim_buf_set_extmark() could not be used to highlight arbitrary multi-line regions The problem can be summarized as follows: let's assume an extmark with a hl_group is placed covering the region (5,0) to (50,0) Now, consider what happens if nvim needs to redraw a window covering the lines 20-30. It needs to be able to ask the marktree what extmarks cover this region, even if they don't begin or end here. Therefore the marktree needs to be augmented with the information covers a point, not just what marks begin or end there. To do this, we augment each node with a field "intersect" which is a set the ids of the marks which overlap this node, but only if it is not part of the set of any parent. This ensures the number of nodes that need to be explicitly marked grows only logarithmically with the total number of explicitly nodes (and thus the number of of overlapping marks). Thus we can quickly iterate all marks which overlaps any query position by looking up what leaf node contains that position. Then we only need to consider all "start" marks within that leaf node, and the "intersect" set of that node and all its parents. Now, and the major source of complexity is that the tree restructuring operations (to ensure that each node has T-1 <= size <= 2*T-1) also need to update these sets. If a full inner node is split in two, one of the new parents might start to completely overlap some ranges and its ids will need to be moved from its children's sets to its own set. Similarly, if two undersized nodes gets joined into one, it might no longer completely overlap some ranges, and now the children which do needs to have the have the ids in its set instead. And then there are the pivots! Yes the pivot operations when a child gets moved from one parent to another.
2020-11-22 02:10:37 -07:00
is_asan = is_asan,
2014-08-30 06:35:42 -07:00
}
--- @class test.unit.testutil: test.unit.testutil.module, test.testutil
M = vim.tbl_extend('error', M, t_global)
return M