diff options
author | 最萌小汐 <sumneko@hotmail.com> | 2019-11-22 23:26:32 +0800 |
---|---|---|
committer | 最萌小汐 <sumneko@hotmail.com> | 2019-11-22 23:26:32 +0800 |
commit | d0ff66c9abe9d6abbca12fd811e0c3cb69c1033a (patch) | |
tree | bb34518d70b85de7656dbdbe958dfa221a3ff3b3 /script/src/parser | |
parent | 0a2c2ad15e1ec359171fb0dd4c72e57c5b66e9ba (diff) | |
download | lua-language-server-d0ff66c9abe9d6abbca12fd811e0c3cb69c1033a.zip |
整理一下目录结构
Diffstat (limited to 'script/src/parser')
-rw-r--r-- | script/src/parser/ast.lua | 1912 | ||||
-rw-r--r-- | script/src/parser/calcline.lua | 93 | ||||
-rw-r--r-- | script/src/parser/grammar.lua | 630 | ||||
-rw-r--r-- | script/src/parser/init.lua | 9 | ||||
-rw-r--r-- | script/src/parser/lines.lua | 187 | ||||
-rw-r--r-- | script/src/parser/parse.lua | 41 | ||||
-rw-r--r-- | script/src/parser/relabel.lua | 361 | ||||
-rw-r--r-- | script/src/parser/split.lua | 9 |
8 files changed, 3242 insertions, 0 deletions
diff --git a/script/src/parser/ast.lua b/script/src/parser/ast.lua new file mode 100644 index 00000000..38cd8576 --- /dev/null +++ b/script/src/parser/ast.lua @@ -0,0 +1,1912 @@ +local tonumber = tonumber +local string_char = string.char +local utf8_char = utf8.char +local type = type +local table = table + +local Errs +local State +local function pushError(err) + if err.finish < err.start then + err.finish = err.start + end + local last = Errs[#Errs] + if last then + if last.start <= err.start and last.finish >= err.finish then + return + end + end + err.level = err.level or 'error' + Errs[#Errs+1] = err + return err +end + +-- goto 单独处理 +local RESERVED = { + ['and'] = true, + ['break'] = true, + ['do'] = true, + ['else'] = true, + ['elseif'] = true, + ['end'] = true, + ['false'] = true, + ['for'] = true, + ['function'] = true, + ['if'] = true, + ['in'] = true, + ['local'] = true, + ['nil'] = true, + ['not'] = true, + ['or'] = true, + ['repeat'] = true, + ['return'] = true, + ['then'] = true, + ['true'] = true, + ['until'] = true, + ['while'] = true, +} + +local VersionOp = { + ['&'] = {'Lua 5.3', 'Lua 5.4'}, + ['~'] = {'Lua 5.3', 'Lua 5.4'}, + ['|'] = {'Lua 5.3', 'Lua 5.4'}, + ['<<'] = {'Lua 5.3', 'Lua 5.4'}, + ['>>'] = {'Lua 5.3', 'Lua 5.4'}, + ['//'] = {'Lua 5.3', 'Lua 5.4'}, +} + +local function checkOpVersion(op, start) + local versions = VersionOp[op] + if not versions then + return + end + for i = 1, #versions do + if versions[i] == State.Version then + return + end + end + pushError { + type = 'UNSUPPORT_SYMBOL', + start = start, + finish = start + #op - 1, + version = versions, + info = { + version = State.Version, + } + } +end + +local Exp + +local function expSplit(list, start, finish, level) + if start == finish then + return list[start] + end + local info = Exp[level] + if not info then + return + end + local func = info[1] + return func(list, start, finish, level) +end + +local function binaryForward(list, start, finish, level) + local info = Exp[level] + for i = finish-1, start+2, -1 do + local op = list[i] + if info[op] then + local e1 = expSplit(list, start, i-2, level) + if not e1 then + goto CONTINUE + end + local e2 = expSplit(list, i+1, finish, level+1) + if not e2 then + goto CONTINUE + end + checkOpVersion(op, list[i-1]) + return { + type = 'binary', + op = op, + start = e1.start, + finish = e2.finish, + [1] = e1, + [2] = e2, + } + end + ::CONTINUE:: + end + return expSplit(list, start, finish, level+1) +end + +local function binaryBackward(list, start, finish, level) + local info = Exp[level] + for i = start+2, finish-1 do + local op = list[i] + if info[op] then + local e1 = expSplit(list, start, i-2, level+1) + if not e1 then + goto CONTINUE + end + local e2 = expSplit(list, i+1, finish, level) + if not e2 then + goto CONTINUE + end + checkOpVersion(op, list[i-1]) + return { + type = 'binary', + op = op, + start = e1.start, + finish = e2.finish, + [1] = e1, + [2] = e2, + } + end + ::CONTINUE:: + end + return expSplit(list, start, finish, level+1) +end + +local function unary(list, start, finish, level) + local info = Exp[level] + local op = list[start+1] + if info[op] then + local e1 = expSplit(list, start+2, finish, level) + if e1 then + checkOpVersion(op, list[start]) + return { + type = 'unary', + op = op, + start = list[start], + finish = e1.finish, + [1] = e1, + } + end + end + return expSplit(list, start, finish, level+1) +end + +local function checkMissEnd(start) + if not State.MissEndErr then + return + end + local err = State.MissEndErr + State.MissEndErr = nil + local _, finish = State.Lua:find('[%w_]+', start) + if not finish then + return + end + err.info.related = { start, finish } + pushError { + type = 'MISS_END', + start = start, + finish = finish, + } +end + +Exp = { + { + ['or'] = true, + binaryForward, + }, + { + ['and'] = true, + binaryForward, + }, + { + ['<='] = true, + ['>='] = true, + ['<'] = true, + ['>'] = true, + ['~='] = true, + ['=='] = true, + binaryForward, + }, + { + ['|'] = true, + binaryForward, + }, + { + ['~'] = true, + binaryForward, + }, + { + ['&'] = true, + binaryForward, + }, + { + ['<<'] = true, + ['>>'] = true, + binaryForward, + }, + { + ['..'] = true, + binaryBackward, + }, + { + ['+'] = true, + ['-'] = true, + binaryForward, + }, + { + ['*'] = true, + ['//'] = true, + ['/'] = true, + ['%'] = true, + binaryForward, + }, + { + ['^'] = true, + binaryBackward, + }, + { + ['not'] = true, + ['#'] = true, + ['~'] = true, + ['-'] = true, + unary, + }, +} + +local Defs = { + Nil = function (pos) + return { + type = 'nil', + start = pos, + finish = pos + 2, + } + end, + True = function (pos) + return { + type = 'boolean', + start = pos, + finish = pos + 3, + [1] = true, + } + end, + False = function (pos) + return { + type = 'boolean', + start = pos, + finish = pos + 4, + [1] = false, + } + end, + LongComment = function (beforeEq, afterEq, str, finish, missPos) + State.Comments[#State.Comments+1] = { + start = beforeEq, + finish = finish, + } + if missPos then + local endSymbol = ']' .. ('='):rep(afterEq-beforeEq) .. ']' + local s, _, w = str:find('(%][%=]*%])[%c%s]*$') + if s then + pushError { + type = 'ERR_LCOMMENT_END', + start = missPos - #str + s - 1, + finish = missPos - #str + s + #w - 2, + info = { + symbol = endSymbol, + }, + fix = { + title = 'FIX_LCOMMENT_END', + { + start = missPos - #str + s - 1, + finish = missPos - #str + s + #w - 2, + text = endSymbol, + } + }, + } + end + pushError { + type = 'MISS_SYMBOL', + start = missPos, + finish = missPos, + info = { + symbol = endSymbol, + }, + fix = { + title = 'ADD_LCOMMENT_END', + { + start = missPos, + finish = missPos, + text = endSymbol, + } + }, + } + end + end, + CLongComment = function (start1, finish1, start2, finish2) + pushError { + type = 'ERR_C_LONG_COMMENT', + start = start1, + finish = finish2 - 1, + fix = { + title = 'FIX_C_LONG_COMMENT', + { + start = start1, + finish = finish1 - 1, + text = '--[[', + }, + { + start = start2, + finish = finish2 - 1, + text = '--]]' + }, + } + } + end, + CCommentPrefix = function (start, finish) + pushError { + type = 'ERR_COMMENT_PREFIX', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_COMMENT_PREFIX', + { + start = start, + finish = finish - 1, + text = '--', + }, + } + } + return false + end, + String = function (start, quote, str, finish) + return { + type = 'string', + start = start, + finish = finish - 1, + [1] = str, + [2] = quote, + } + end, + LongString = function (beforeEq, afterEq, str, missPos) + if missPos then + local endSymbol = ']' .. ('='):rep(afterEq-beforeEq) .. ']' + local s, _, w = str:find('(%][%=]*%])[%c%s]*$') + if s then + pushError { + type = 'ERR_LSTRING_END', + start = missPos - #str + s - 1, + finish = missPos - #str + s + #w - 2, + info = { + symbol = endSymbol, + }, + fix = { + title = 'FIX_LSTRING_END', + { + start = missPos - #str + s - 1, + finish = missPos - #str + s + #w - 2, + text = endSymbol, + } + }, + } + end + pushError { + type = 'MISS_SYMBOL', + start = missPos, + finish = missPos, + info = { + symbol = endSymbol, + }, + fix = { + title = 'ADD_LSTRING_END', + { + start = missPos, + finish = missPos, + text = endSymbol, + } + }, + } + end + return '[' .. ('='):rep(afterEq-beforeEq) .. '[', str + end, + Char10 = function (char) + char = tonumber(char) + if not char or char < 0 or char > 255 then + return '' + end + return string_char(char) + end, + Char16 = function (pos, char) + if State.Version == 'Lua 5.1' then + pushError { + type = 'ERR_ESC', + start = pos-1, + finish = pos, + version = {'Lua 5.2', 'Lua 5.3', 'Lua 5.4', 'LuaJIT'}, + info = { + version = State.Version, + } + } + return char + end + return string_char(tonumber(char, 16)) + end, + CharUtf8 = function (pos, char) + if State.Version ~= 'Lua 5.3' + and State.Version ~= 'Lua 5.4' + and State.Version ~= 'LuaJIT' + then + pushError { + type = 'ERR_ESC', + start = pos-3, + finish = pos-2, + version = {'Lua 5.3', 'Lua 5.4', 'LuaJIT'}, + info = { + version = State.Version, + } + } + return char + end + if #char == 0 then + pushError { + type = 'UTF8_SMALL', + start = pos-3, + finish = pos, + } + return '' + end + local v = tonumber(char, 16) + if not v then + for i = 1, #char do + if not tonumber(char:sub(i, i), 16) then + pushError { + type = 'MUST_X16', + start = pos + i - 1, + finish = pos + i - 1, + } + end + end + return '' + end + if State.Version == 'Lua 5.4' then + if v < 0 or v > 0x7FFFFFFF then + pushError { + type = 'UTF8_MAX', + start = pos-3, + finish = pos+#char, + info = { + min = '00000000', + max = '7FFFFFFF', + } + } + end + else + if v < 0 or v > 0x10FFFF then + pushError { + type = 'UTF8_MAX', + start = pos-3, + finish = pos+#char, + version = v <= 0x7FFFFFFF and 'Lua 5.4' or nil, + info = { + min = '000000', + max = '10FFFF', + } + } + end + end + if v >= 0 and v <= 0x10FFFF then + return utf8_char(v) + end + return '' + end, + Number = function (start, number, finish) + local n = tonumber(number) + if n then + State.LastNumber = { + type = 'number', + start = start, + finish = finish - 1, + [1] = n, + } + return State.LastNumber + else + pushError { + type = 'MALFORMED_NUMBER', + start = start, + finish = finish - 1, + } + State.LastNumber = { + type = 'number', + start = start, + finish = finish - 1, + [1] = 0, + } + return State.LastNumber + end + end, + FFINumber = function (start, symbol) + if math.type(State.LastNumber[1]) == 'float' then + pushError { + type = 'UNKNOWN_SYMBOL', + start = start, + finish = start + #symbol - 1, + info = { + symbol = symbol, + } + } + State.LastNumber[1] = 0 + return + end + if State.Version ~= 'LuaJIT' then + pushError { + type = 'UNSUPPORT_SYMBOL', + start = start, + finish = start + #symbol - 1, + version = 'LuaJIT', + info = { + version = State.Version, + } + } + State.LastNumber[1] = 0 + end + end, + ImaginaryNumber = function (start, symbol) + if State.Version ~= 'LuaJIT' then + pushError { + type = 'UNSUPPORT_SYMBOL', + start = start, + finish = start + #symbol - 1, + version = 'LuaJIT', + info = { + version = State.Version, + } + } + end + State.LastNumber[1] = 0 + end, + Name = function (start, str, finish) + local isKeyWord + if RESERVED[str] then + isKeyWord = true + elseif str == 'goto' then + if State.Version ~= 'Lua 5.1' and State.Version ~= 'LuaJIT' then + isKeyWord = true + end + end + if isKeyWord then + pushError { + type = 'KEYWORD', + start = start, + finish = finish - 1, + } + end + return { + type = 'name', + start = start, + finish = finish - 1, + [1] = str, + } + end, + Simple = function (first, ...) + if ... then + local obj = { + type = 'simple', + start = first.start, + first, ..., + } + local last = obj[#obj] + obj.finish = last.finish + return obj + elseif first == '' then + return nil + else + return first + end + end, + SimpleCall = function (simple) + if not simple then + return nil + end + if simple.type ~= 'simple' then + pushError { + type = 'EXP_IN_ACTION', + start = simple.start, + finish = simple.finish, + } + return simple + end + local last = simple[#simple] + if last.type == 'call' then + return simple + end + local colon = simple[#simple-1] + if colon and colon.type == ':' then + -- 型如 `obj:method`,将错误让给MISS_SYMBOL + return simple + end + pushError { + type = 'EXP_IN_ACTION', + start = simple[1].start, + finish = last.finish, + } + return simple + end, + Exp = function (first, ...) + if not ... then + return first + end + local list = {first, ...} + return expSplit(list, 1, #list, 1) + end, + Prefix = function (start, exp, finish) + exp.brackets = true + return exp + end, + Index = function (start, exp, finish) + return { + type = 'index', + start = start, + finish = finish - 1, + [1] = exp, + } + end, + Call = function (start, arg, finish) + if arg == nil then + return { + type = 'call', + start = start, + finish = finish - 1, + } + end + if arg.type == 'list' then + arg.type = 'call' + arg.start = start + arg.finish = finish - 1 + return arg + end + local obj = { + type = 'call', + start = start, + finish = finish - 1, + [1] = arg, + } + return obj + end, + DOTS = function (start) + return { + type = '...', + start = start, + finish = start + 2, + } + end, + DotsAsArg = function (obj) + State.Dots[#State.Dots] = true + return obj + end, + DotsAsExp = function (obj) + if not State.Dots[#State.Dots] then + pushError { + type = 'UNEXPECT_DOTS', + start = obj.start, + finish = obj.finish, + } + end + return obj + end, + COLON = function (start) + return { + type = ':', + start = start, + finish = start, + } + end, + DOT = function (start) + return { + type = '.', + start = start, + finish = start, + } + end, + Function = function (start, argStart, arg, argFinish, ...) + local obj = { + type = 'function', + start = start, + arg = arg, + argStart = argStart - 1, + argFinish = argFinish, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + if obj.argFinish > obj.finish then + obj.argFinish = obj.finish + end + checkMissEnd(start) + return obj + end, + NamedFunction = function (start, name, argStart, arg, argFinish, ...) + local obj = { + type = 'function', + start = start, + name = name, + arg = arg, + argStart = argStart - 1, + argFinish = argFinish, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + if obj.argFinish > obj.finish then + obj.argFinish = obj.finish + end + checkMissEnd(start) + return obj + end, + LocalFunction = function (start, name, argStart, arg, argFinish, ...) + local obj = { + type = 'localfunction', + start = start, + name = name, + arg = arg, + argStart = argStart - 1, + argFinish = argFinish, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + if obj.argFinish > obj.finish then + obj.argFinish = obj.finish + end + + if name.type ~= 'name' then + pushError { + type = 'UNEXPECT_LFUNC_NAME', + start = name.start, + finish = name.finish, + } + end + + checkMissEnd(start) + return obj + end, + Table = function (start, ...) + local args = {...} + local max = #args + local finish = args[max] - 1 + local table = { + type = 'table', + start = start, + finish = finish + } + start = start + 1 + local wantField = true + for i = 1, max-1 do + local arg = args[i] + local isField = type(arg) == 'table' + local isEmmy = isField and arg.type:sub(1, 4) == 'emmy' + if wantField and not isField then + pushError { + type = 'MISS_EXP', + start = start, + finish = arg - 1, + } + elseif not wantField and isField and not isEmmy then + pushError { + type = 'MISS_SEP_IN_TABLE', + start = start, + finish = arg.start-1, + } + end + if isField then + table[#table+1] = arg + if not isEmmy then + wantField = false + start = arg.finish + 1 + end + else + wantField = true + start = arg + end + end + return table + end, + NewField = function (key, value) + return { + type = 'pair', + start = key.start, + finish = value.finish, + key, value, + } + end, + NewIndex = function (key, value) + return { + type = 'pair', + start = key.start, + finish = value.finish, + key, value, + } + end, + List = function (first, second, ...) + if second then + local list = { + type = 'list', + start = first.start, + first, second, ... + } + local last = list[#list] + list.finish = last.finish + return list + elseif type(first) == 'table' then + return first + else + return nil + end + end, + ArgList = function (...) + if ... == '' then + return nil + end + local args = table.pack(...) + local list = {} + local max = args.n + args.n = nil + local wantName = true + for i = 1, max do + local obj = args[i] + if type(obj) == 'number' then + if wantName then + pushError { + type = 'MISS_NAME', + start = obj, + finish = obj, + } + end + wantName = true + else + if not wantName then + pushError { + type = 'MISS_SYMBOL', + start = obj.start-1, + finish = obj.start-1, + info = { + symbol = ',', + } + } + end + wantName = false + list[#list+1] = obj + if obj.type == '...' then + if i < max then + local a = args[i+1] + local b = args[max] + pushError { + type = 'ARGS_AFTER_DOTS', + start = type(a) == 'number' and a or a.start, + finish = type(b) == 'number' and b or b.finish, + } + end + break + end + end + end + if wantName then + local last = args[max] + pushError { + type = 'MISS_NAME', + start = last+1, + finish = last+1, + } + end + if #list == 0 then + return nil + elseif #list == 1 then + return list[1] + else + list.type = 'list' + list.start = list[1].start + list.finish = list[#list].finish + return list + end + end, + CallArgList = function (start, ...) + local args = {...} + local max = #args + local finish = args[max] - 1 + local exps = { + type = 'list', + start = start, + finish = finish, + } + local wantExp = true + for i = 1, max-1 do + local arg = args[i] + local isExp = type(arg) == 'table' + if wantExp and not isExp then + pushError { + type = 'MISS_EXP', + start = start, + finish = arg - 1, + } + elseif not wantExp and isExp then + pushError { + type = 'MISS_SYMBOL', + start = start, + finish = arg.start-1, + info = { + symbol = ',', + } + } + end + if isExp then + exps[#exps+1] = arg + wantExp = false + start = arg.finish + 1 + else + wantExp = true + start = arg + end + end + if wantExp then + pushError { + type = 'MISS_EXP', + start = start, + finish = finish, + } + end + if #exps == 0 then + return nil + elseif #exps == 1 then + return exps[1] + else + return exps + end + end, + Nothing = function () + return nil + end, + None = function() + return + end, + Skip = function () + return false + end, + Set = function (keys, values) + return { + type = 'set', + keys, values, + } + end, + LocalTag = function (...) + if not ... or ... == '' then + return nil + end + local tags = {...} + for i, tag in ipairs(tags) do + if State.Version ~= 'Lua 5.4' then + pushError { + type = 'UNSUPPORT_SYMBOL', + start = tag.start, + finish = tag.finish, + version = 'Lua 5.4', + info = { + version = State.Version, + } + } + elseif tag[1] ~= 'const' and tag[1] ~= 'close' then + pushError { + type = 'UNKNOWN_TAG', + start = tag.start, + finish = tag.finish, + info = { + tag = tag[1], + } + } + elseif i > 1 then + pushError { + type = 'MULTI_TAG', + start = tag.start, + finish = tag.finish, + info = { + tag = tag[1], + } + } + end + end + return tags + end, + LocalName = function (name, tags) + name.tags = tags + return name + end, + Local = function (keys, values) + return { + type = 'local', + keys, values, + } + end, + DoBody = function (...) + if ... == '' then + return { + type = 'do', + } + else + return { + type = 'do', + ... + } + end + end, + Do = function (start, action, finish) + action.start = start + action.finish = finish - 1 + checkMissEnd(start) + return action + end, + Break = function (finish, ...) + if State.Break > 0 then + local breakChunk = { + type = 'break', + } + if not ... then + return breakChunk + end + local action = select(-1, ...) + if not action then + return breakChunk + end + if State.Version == 'Lua 5.1' or State.Version == 'LuaJIT' then + pushError { + type = 'ACTION_AFTER_BREAK', + start = finish - #'break', + finish = finish - 1, + } + end + return breakChunk, action + else + pushError { + type = 'BREAK_OUTSIDE', + start = finish - #'break', + finish = finish - 1, + } + if not ... then + return false + end + local action = select(-1, ...) + if not action then + return false + end + return action + end + end, + BreakStart = function () + State.Break = State.Break + 1 + end, + BreakEnd = function () + State.Break = State.Break - 1 + end, + Return = function (start, exp, finish) + if not finish then + finish = exp + exp = { + type = 'return', + start = start, + finish = finish - 1, + } + else + if exp.type == 'list' then + exp.type = 'return' + exp.start = start + exp.finish = finish - 1 + else + exp = { + type = 'return', + start = start, + finish = finish - 1, + [1] = exp, + } + end + end + return exp + end, + Label = function (start, name, finish) + if State.Version == 'Lua 5.1' then + pushError { + type = 'UNSUPPORT_SYMBOL', + start = start, + finish = finish - 1, + version = {'Lua 5.2', 'Lua 5.3', 'Lua 5.4', 'LuaJIT'}, + info = { + version = State.Version, + } + } + return false + end + name.type = 'label' + local labels = State.Label[#State.Label] + local str = name[1] + if labels[str] then + --pushError { + -- type = 'REDEFINE_LABEL', + -- start = name.start, + -- finish = name.finish, + -- info = { + -- label = str, + -- related = {labels[str].start, labels[str].finish}, + -- } + --} + else + labels[str] = name + end + return name + end, + GoTo = function (start, name, finish) + if State.Version == 'Lua 5.1' then + pushError { + type = 'UNSUPPORT_SYMBOL', + start = start, + finish = finish - 1, + version = {'Lua 5.2', 'Lua 5.3', 'Lua 5.4', 'LuaJIT'}, + info = { + version = State.Version, + } + } + return false + end + name.type = 'goto' + local labels = State.Label[#State.Label] + labels[#labels+1] = name + return name + end, + -- TODO 这里的检查不完整,但是完整的检查比较复杂,开销比较高 + -- 不能jump到另一个局部变量的作用域 + -- 函数会切断goto与label + -- 不能从block外jump到block内,但是可以从block内jump到block外 + BlockStart = function () + State.Label[#State.Label+1] = {} + State.Dots[#State.Dots+1] = false + end, + BlockEnd = function () + local labels = State.Label[#State.Label] + State.Label[#State.Label] = nil + State.Dots[#State.Dots] = nil + for i = 1, #labels do + local name = labels[i] + local str = name[1] + if not labels[str] then + pushError { + type = 'NO_VISIBLE_LABEL', + start = name.start, + finish = name.finish, + info = { + label = str, + } + } + end + end + end, + IfBlock = function (exp, start, ...) + local obj = { + filter = exp, + start = start, + ... + } + local max = #obj + obj.finish = obj[max] + obj[max] = nil + return obj + end, + ElseIfBlock = function (exp, start, ...) + local obj = { + filter = exp, + start = start, + ... + } + local max = #obj + obj.finish = obj[max] + obj[max] = nil + return obj + end, + ElseBlock = function (start, ...) + local obj = { + start = start, + ... + } + local max = #obj + obj.finish = obj[max] + obj[max] = nil + return obj + end, + If = function (start, ...) + local obj = { + type = 'if', + start = start, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + checkMissEnd(start) + return obj + end, + Loop = function (start, arg, min, max, step, ...) + local obj = { + type = 'loop', + start = start, + arg = arg, + min = min, + max = max, + step = step, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + checkMissEnd(start) + return obj + end, + In = function (start, arg, exp, ...) + local obj = { + type = 'in', + start = start, + arg = arg, + exp = exp, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + checkMissEnd(start) + return obj + end, + While = function (start, filter, ...) + local obj = { + type = 'while', + start = start, + filter = filter, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj[max] = nil + checkMissEnd(start) + return obj + end, + Repeat = function (start, ...) + local obj = { + type = 'repeat', + start = start, + ... + } + local max = #obj + obj.finish = obj[max] - 1 + obj.filter = obj[max-1] + obj[max] = nil + obj[max-1] = nil + return obj + end, + Lua = function (...) + if ... == '' then + return {} + end + return {...} + end, + + -- EmmyLua 支持 + EmmyName = function (start, str) + return { + type = 'emmyName', + start = start, + finish = start + #str - 1, + [1] = str, + } + end, + DirtyEmmyName = function (pos) + pushError { + type = 'MISS_NAME', + level = 'warning', + start = pos, + finish = pos, + } + return { + type = 'emmyName', + start = pos-1, + finish = pos-1, + [1] = '' + } + end, + EmmyClass = function (class, startPos, extends) + if extends and extends[1] == '' then + extends.start = startPos + end + return { + type = 'emmyClass', + start = class.start, + finish = (extends or class).finish, + [1] = class, + [2] = extends, + } + end, + EmmyType = function (typeDef) + return typeDef + end, + EmmyCommonType = function (...) + local result = { + type = 'emmyType', + ... + } + for i = 1, #result // 2 do + local startPos = result[i * 2] + local emmyName = result[i * 2 + 1] + if emmyName[1] == '' then + emmyName.start = startPos + end + result[i + 1] = emmyName + end + for i = #result // 2 + 2, #result do + result[i] = nil + end + result.start = result[1].start + result.finish = result[#result].finish + return result + end, + EmmyArrayType = function (start, emmy, _, finish) + emmy.type = 'emmyArrayType' + emmy.start = start + emmy.finish = finish - 1 + return emmy + end, + EmmyTableType = function (start, keyType, valueType, finish) + return { + type = 'emmyTableType', + start = start, + finish = finish - 1, + [1] = keyType, + [2] = valueType, + } + end, + EmmyFunctionType = function (start, args, returns, finish) + local result = { + start = start, + finish = finish - 1, + type = 'emmyFunctionType', + args = args, + returns = returns, + } + return result + end, + EmmyFunctionRtns = function (...) + return {...} + end, + EmmyFunctionArgs = function (...) + local args = {...} + args[#args] = nil + return args + end, + EmmyAlias = function (name, emmyName, ...) + return { + type = 'emmyAlias', + start = name.start, + finish = emmyName.finish, + name, + emmyName, + ... + } + end, + EmmyParam = function (argName, emmyName, option, ...) + local emmy = { + type = 'emmyParam', + option = option, + argName, + emmyName, + ... + } + emmy.start = emmy[1].start + emmy.finish = emmy[#emmy].finish + return emmy + end, + EmmyReturn = function (start, type, name, finish, option) + local emmy = { + type = 'emmyReturn', + option = option, + start = start, + finish = finish - 1, + [1] = type, + [2] = name, + } + return emmy + end, + EmmyField = function (access, fieldName, ...) + local obj = { + type = 'emmyField', + access, fieldName, + ... + } + obj.start = obj[2].start + obj.finish = obj[3].finish + return obj + end, + EmmyGenericBlock = function (genericName, parentName) + return { + start = genericName.start, + finish = parentName and parentName.finish or genericName.finish, + genericName, + parentName, + } + end, + EmmyGeneric = function (...) + local emmy = { + type = 'emmyGeneric', + ... + } + emmy.start = emmy[1].start + emmy.finish = emmy[#emmy].finish + return emmy + end, + EmmyVararg = function (typeName) + return { + type = 'emmyVararg', + start = typeName.start, + finish = typeName.finish, + typeName, + } + end, + EmmyLanguage = function (language) + return { + type = 'emmyLanguage', + start = language.start, + finish = language.finish, + language, + } + end, + EmmySee = function (start, className, methodName, finish) + return { + type = 'emmySee', + start = start, + finish = finish - 1, + className, methodName + } + end, + EmmyOverLoad = function (EmmyFunctionType) + EmmyFunctionType.type = 'emmyOverLoad' + return EmmyFunctionType + end, + EmmyIncomplete = function (emmyName) + emmyName.type = 'emmyIncomplete' + return emmyName + end, + EmmyComment = function (...) + local lines = {...} + for i = 2, #lines do + local line = lines[i] + if line:sub(1, 1) == '|' then + lines[i] = '\n' .. line:sub(2) + end + end + return { + type = 'emmyComment', + [1] = table.concat(lines, '\n'), + } + end, + EmmyOption = function (options) + if not options or options == '' then + return nil + end + local option = {} + for _, pair in ipairs(options) do + if pair.type == 'pair' then + local key = pair[1] + local value = pair[2] + if key.type == 'name' then + option[key[1]] = value[1] + end + end + end + return option + end, + EmmyTypeEnum = function (default, enum, comment) + enum.type = 'emmyEnum' + if default ~= '' then + enum.default = true + end + enum.comment = comment + return enum + end, + + -- 捕获错误 + UnknownSymbol = function (start, symbol) + pushError { + type = 'UNKNOWN_SYMBOL', + start = start, + finish = start + #symbol - 1, + info = { + symbol = symbol, + } + } + return + end, + UnknownAction = function (start, symbol) + pushError { + type = 'UNKNOWN_SYMBOL', + start = start, + finish = start + #symbol - 1, + info = { + symbol = symbol, + } + } + return false + end, + DirtyName = function (pos) + pushError { + type = 'MISS_NAME', + start = pos, + finish = pos, + } + return { + type = 'name', + start = pos-1, + finish = pos-1, + [1] = '' + } + end, + DirtyExp = function (pos) + pushError { + type = 'MISS_EXP', + start = pos, + finish = pos, + } + return { + type = 'name', + start = pos, + finish = pos, + [1] = '' + } + end, + MissExp = function (pos) + pushError { + type = 'MISS_EXP', + start = pos, + finish = pos, + } + end, + MissExponent = function (start, finish) + pushError { + type = 'MISS_EXPONENT', + start = start, + finish = finish - 1, + } + end, + MissQuote1 = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '"' + } + } + end, + MissQuote2 = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = "'" + } + } + end, + MissEscX = function (pos) + pushError { + type = 'MISS_ESC_X', + start = pos-2, + finish = pos+1, + } + end, + MissTL = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '{', + } + } + end, + MissTR = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '}', + } + } + return pos + 1 + end, + MissBR = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = ']', + } + } + return pos + 1 + end, + MissPL = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '(', + } + } + end, + DirtyPR = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = ')', + } + } + return pos + 1 + end, + MissPR = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = ')', + } + } + end, + ErrEsc = function (pos) + pushError { + type = 'ERR_ESC', + start = pos-1, + finish = pos, + } + end, + MustX16 = function (pos, str) + pushError { + type = 'MUST_X16', + start = pos, + finish = pos + #str - 1, + } + end, + MissAssign = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '=', + } + } + end, + MissTableSep = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = ',' + } + } + end, + MissField = function (pos) + pushError { + type = 'MISS_FIELD', + start = pos, + finish = pos, + } + end, + MissMethod = function (pos) + pushError { + type = 'MISS_METHOD', + start = pos, + finish = pos, + } + end, + MissLabel = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = '::', + } + } + end, + MissEnd = function (pos) + State.MissEndErr = pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = 'end', + } + } + end, + MissDo = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = 'do', + } + } + end, + MissComma = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = ',', + } + } + end, + MissIn = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = 'in', + } + } + end, + MissUntil = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = 'until', + } + } + end, + MissThen = function (pos) + pushError { + type = 'MISS_SYMBOL', + start = pos, + finish = pos, + info = { + symbol = 'then', + } + } + end, + ExpInAction = function (start, exp, finish) + pushError { + type = 'EXP_IN_ACTION', + start = start, + finish = finish - 1, + } + return exp + end, + AfterReturn = function (rtn, ...) + if not ... then + return rtn + end + local action = select(-1, ...) + if not action then + return rtn + end + pushError { + type = 'ACTION_AFTER_RETURN', + start = rtn.start, + finish = rtn.finish, + } + return rtn, action + end, + MissIf = function (start, block) + pushError { + type = 'MISS_SYMBOL', + start = start, + finish = start, + info = { + symbol = 'if', + } + } + return block + end, + MissGT = function (start) + pushError { + type = 'MISS_SYMBOL', + start = start, + finish = start, + info = { + symbol = '>' + } + } + end, + ErrAssign = function (start, finish) + pushError { + type = 'ERR_ASSIGN_AS_EQ', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_ASSIGN_AS_EQ', + { + start = start, + finish = finish - 1, + text = '=', + } + } + } + end, + ErrEQ = function (start, finish) + pushError { + type = 'ERR_EQ_AS_ASSIGN', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_EQ_AS_ASSIGN', + { + start = start, + finish = finish - 1, + text = '==', + } + } + } + return '==' + end, + ErrUEQ = function (start, finish) + pushError { + type = 'ERR_UEQ', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_UEQ', + { + start = start, + finish = finish - 1, + text = '~=', + } + } + } + return '==' + end, + ErrThen = function (start, finish) + pushError { + type = 'ERR_THEN_AS_DO', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_THEN_AS_DO', + { + start = start, + finish = finish - 1, + text = 'then', + } + } + } + end, + ErrDo = function (start, finish) + pushError { + type = 'ERR_DO_AS_THEN', + start = start, + finish = finish - 1, + fix = { + title = 'FIX_DO_AS_THEN', + { + start = start, + finish = finish - 1, + text = 'do', + } + } + } + end, +} + +local function init(state, errs) + State = state + Errs = errs +end + +return { + defs = Defs, + init = init, +} diff --git a/script/src/parser/calcline.lua b/script/src/parser/calcline.lua new file mode 100644 index 00000000..26f475d9 --- /dev/null +++ b/script/src/parser/calcline.lua @@ -0,0 +1,93 @@ +local m = require 'lpeglabel' + +local row +local fl +local NL = (m.P'\r\n' + m.S'\r\n') * m.Cp() / function (pos) + row = row + 1 + fl = pos +end +local ROWCOL = (NL + m.P(1))^0 +local function rowcol(str, n) + row = 1 + fl = 1 + ROWCOL:match(str:sub(1, n)) + local col = n - fl + 1 + return row, col +end + +local function rowcol_utf8(str, n) + row = 1 + fl = 1 + ROWCOL:match(str:sub(1, n)) + return row, utf8.len(str, fl, n) +end + +local function position(str, _row, _col) + local cur = 1 + local row = 1 + while true do + if row == _row then + return cur + _col - 1 + elseif row > _row then + return cur - 1 + end + local pos = str:find('[\r\n]', cur) + if not pos then + return #str + end + row = row + 1 + if str:sub(pos, pos+1) == '\r\n' then + cur = pos + 2 + else + cur = pos + 1 + end + end +end + +local function position_utf8(str, _row, _col) + local cur = 1 + local row = 1 + while true do + if row == _row then + return utf8.offset(str, _col, cur) + elseif row > _row then + return cur - 1 + end + local pos = str:find('[\r\n]', cur) + if not pos then + return #str + end + row = row + 1 + if str:sub(pos, pos+1) == '\r\n' then + cur = pos + 2 + else + cur = pos + 1 + end + end +end + +local NL = m.P'\r\n' + m.S'\r\n' + +local function line(str, row) + local count = 0 + local res + local LINE = m.Cmt((1 - NL)^0, function (_, _, c) + count = count + 1 + if count == row then + res = c + return false + end + return true + end) + local MATCH = (LINE * NL)^0 * LINE + MATCH:match(str) + return res +end + +return { + rowcol = rowcol, + rowcol_utf8 = rowcol_utf8, + position = position, + position_utf8 = position_utf8, + line = line, +} diff --git a/script/src/parser/grammar.lua b/script/src/parser/grammar.lua new file mode 100644 index 00000000..568b8544 --- /dev/null +++ b/script/src/parser/grammar.lua @@ -0,0 +1,630 @@ +local re = require 'parser.relabel' +local m = require 'lpeglabel' +local ast = require 'parser.ast' + +local scriptBuf = '' +local compiled = {} +local parser +local defs = ast.defs + +-- goto 可以作为名字,合法性之后处理 +local RESERVED = { + ['and'] = true, + ['break'] = true, + ['do'] = true, + ['else'] = true, + ['elseif'] = true, + ['end'] = true, + ['false'] = true, + ['for'] = true, + ['function'] = true, + ['if'] = true, + ['in'] = true, + ['local'] = true, + ['nil'] = true, + ['not'] = true, + ['or'] = true, + ['repeat'] = true, + ['return'] = true, + ['then'] = true, + ['true'] = true, + ['until'] = true, + ['while'] = true, +} + +defs.nl = (m.P'\r\n' + m.S'\r\n') +defs.s = m.S' \t' +defs.S = - defs.s +defs.ea = '\a' +defs.eb = '\b' +defs.ef = '\f' +defs.en = '\n' +defs.er = '\r' +defs.et = '\t' +defs.ev = '\v' +defs['nil'] = m.Cp() / function () return nil end +defs['false'] = m.Cp() / function () return false end +defs.NotReserved = function (_, _, str) + if RESERVED[str] then + return false + end + return true +end +defs.Reserved = function (_, _, str) + if RESERVED[str] then + return true + end + return false +end +defs.None = function () end +defs.np = m.Cp() / function (n) return n+1 end + +m.setmaxstack(1000) + +local eof = re.compile '!. / %{SYNTAX_ERROR}' + +local function grammar(tag) + return function (script) + scriptBuf = script .. '\r\n' .. scriptBuf + compiled[tag] = re.compile(scriptBuf, defs) * eof + end +end + +local function errorpos(pos, err) + return { + type = 'UNKNOWN', + start = pos or 0, + finish = pos or 0, + err = err, + } +end + +grammar 'Comment' [[ +Comment <- LongComment / '--' ShortComment +LongComment <- ('--[' {} {:eq: '='* :} {} '[' + {(!CommentClose .)*} + (CommentClose {} / {} {})) + -> LongComment + / ( + {} '/*' {} + (!'*/' .)* + {} '*/' {} + ) + -> CLongComment +CommentClose <- ']' =eq ']' +ShortComment <- (!%nl .)* +]] + +grammar 'Sp' [[ +Sp <- (Comment / %nl / %s)* +Sps <- (Comment / %nl / %s)+ +]] + +grammar 'Common' [[ +Word <- [a-zA-Z0-9_] +Cut <- !Word +X16 <- [a-fA-F0-9] +Rest <- (!%nl .)* + +AND <- Sp {'and'} Cut +BREAK <- Sp 'break' Cut +DO <- Sp 'do' Cut + / Sp ({} 'then' Cut {}) -> ErrDo +ELSE <- Sp 'else' Cut +ELSEIF <- Sp 'elseif' Cut +END <- Sp 'end' Cut +FALSE <- Sp 'false' Cut +FOR <- Sp 'for' Cut +FUNCTION <- Sp 'function' Cut +GOTO <- Sp 'goto' Cut +IF <- Sp 'if' Cut +IN <- Sp 'in' Cut +LOCAL <- Sp 'local' Cut +NIL <- Sp 'nil' Cut +NOT <- Sp 'not' Cut +OR <- Sp {'or'} Cut +REPEAT <- Sp 'repeat' Cut +RETURN <- Sp 'return' Cut +THEN <- Sp 'then' Cut + / Sp ({} 'do' Cut {}) -> ErrThen +TRUE <- Sp 'true' Cut +UNTIL <- Sp 'until' Cut +WHILE <- Sp 'while' Cut + +Esc <- '\' -> '' + EChar +EChar <- 'a' -> ea + / 'b' -> eb + / 'f' -> ef + / 'n' -> en + / 'r' -> er + / 't' -> et + / 'v' -> ev + / '\' + / '"' + / "'" + / %nl + / ('z' (%nl / %s)*) -> '' + / ({} 'x' {X16 X16}) -> Char16 + / ([0-9] [0-9]? [0-9]?) -> Char10 + / ('u{' {} {Word*} '}') -> CharUtf8 + -- 错误处理 + / 'x' {} -> MissEscX + / 'u' !'{' {} -> MissTL + / 'u{' Word* !'}' {} -> MissTR + / {} -> ErrEsc + +BOR <- Sp {'|'} +BXOR <- Sp {'~'} !'=' +BAND <- Sp {'&'} +Bshift <- Sp {BshiftList} +BshiftList <- '<<' + / '>>' +Concat <- Sp {'..'} +Adds <- Sp {AddsList} +AddsList <- '+' + / '-' +Muls <- Sp {MulsList} +MulsList <- '*' + / '//' + / '/' + / '%' +Unary <- Sp {} {UnaryList} +UnaryList <- NOT + / '#' + / '-' + / '~' !'=' +POWER <- Sp {'^'} + +BinaryOp <- Sp {} {'or'} Cut + / Sp {} {'and'} Cut + / Sp {} {'<=' / '>=' / '<'!'<' / '>'!'>' / '~=' / '=='} + / Sp {} ({} '=' {}) -> ErrEQ + / Sp {} ({} '!=' {}) -> ErrUEQ + / Sp {} {'|'} + / Sp {} {'~'} + / Sp {} {'&'} + / Sp {} {'<<' / '>>'} + / Sp {} {'..'} !'.' + / Sp {} {'+' / '-'} + / Sp {} {'*' / '//' / '/' / '%'} + / Sp {} {'^'} +UnaryOp <- Sp {} {'not' Cut / '#' / '~' !'=' / '-' !'-'} + +PL <- Sp '(' +PR <- Sp ')' +BL <- Sp '[' !'[' !'=' +BR <- Sp ']' +TL <- Sp '{' +TR <- Sp '}' +COMMA <- Sp ',' +SEMICOLON <- Sp ';' +DOTS <- Sp ({} '...') -> DOTS +DOT <- Sp ({} '.' !'.') -> DOT +COLON <- Sp ({} ':' !':') -> COLON +LABEL <- Sp '::' +ASSIGN <- Sp '=' !'=' +AssignOrEQ <- Sp ({} '==' {}) -> ErrAssign + / Sp '=' + +Nothing <- {} -> Nothing + +DirtyBR <- BR {} / {} -> MissBR +DirtyTR <- TR {} / {} -> MissTR +DirtyPR <- PR {} / {} -> DirtyPR +DirtyLabel <- LABEL / {} -> MissLabel +NeedPR <- PR / {} -> MissPR +NeedEnd <- END / {} -> MissEnd +NeedDo <- DO / {} -> MissDo +NeedAssign <- ASSIGN / {} -> MissAssign +NeedComma <- COMMA / {} -> MissComma +NeedIn <- IN / {} -> MissIn +NeedUntil <- UNTIL / {} -> MissUntil +]] + +grammar 'Nil' [[ +Nil <- Sp ({} -> Nil) NIL +]] + +grammar 'Boolean' [[ +Boolean <- Sp ({} -> True) TRUE + / Sp ({} -> False) FALSE +]] + +grammar 'String' [[ +String <- Sp ({} StringDef {}) + -> String +StringDef <- {'"'} + {~(Esc / !%nl !'"' .)*~} -> 1 + ('"' / {} -> MissQuote1) + / {"'"} + {~(Esc / !%nl !"'" .)*~} -> 1 + ("'" / {} -> MissQuote2) + / ('[' {} {:eq: '='* :} {} '[' %nl? + {(!StringClose .)*} -> 1 + (StringClose / {})) + -> LongString +StringClose <- ']' =eq ']' +]] + +grammar 'Number' [[ +Number <- Sp ({} {NumberDef} {}) -> Number + NumberSuffix? + ErrNumber? +NumberDef <- Number16 / Number10 +NumberSuffix<- ({} {[uU]? [lL] [lL]}) -> FFINumber + / ({} {[iI]}) -> ImaginaryNumber +ErrNumber <- ({} {([0-9a-zA-Z] / '.')+}) -> UnknownSymbol + +Number10 <- Float10 Float10Exp? + / Integer10 Float10? Float10Exp? +Integer10 <- [0-9]+ ('.' [0-9]*)? +Float10 <- '.' [0-9]+ +Float10Exp <- [eE] [+-]? [0-9]+ + / ({} [eE] [+-]? {}) -> MissExponent + +Number16 <- '0' [xX] Float16 Float16Exp? + / '0' [xX] Integer16 Float16? Float16Exp? +Integer16 <- X16+ ('.' X16*)? + / ({} {Word*}) -> MustX16 +Float16 <- '.' X16+ + / '.' ({} {Word*}) -> MustX16 +Float16Exp <- [pP] [+-]? [0-9]+ + / ({} [pP] [+-]? {}) -> MissExponent +]] + +grammar 'Name' [[ +Name <- Sp ({} NameBody {}) + -> Name +NameBody <- {[a-zA-Z_] [a-zA-Z0-9_]*} +FreeName <- Sp ({} {NameBody=>NotReserved} {}) + -> Name +MustName <- Name / DirtyName +DirtyName <- {} -> DirtyName +]] + +grammar 'Exp' [[ +Exp <- (UnUnit (BinaryOp (UnUnit / {} -> DirtyExp))*) + -> Exp +UnUnit <- ExpUnit + / UnaryOp+ (ExpUnit / {} -> DirtyExp) +ExpUnit <- Nil + / Boolean + / String + / Number + / DOTS -> DotsAsExp + / Table + / Function + / Simple + +Simple <- (Prefix (Sp Suffix)*) + -> Simple +Prefix <- Sp ({} PL DirtyExp DirtyPR) + -> Prefix + / FreeName +Index <- ({} BL DirtyExp DirtyBR) -> Index +Suffix <- DOT Name / DOT {} -> MissField + / Method (!(Sp CallStart) {} -> MissPL)? + / ({} Table {}) -> Call + / ({} String {}) -> Call + / Index + / ({} PL CallArgList DirtyPR) -> Call +Method <- COLON Name / COLON {} -> MissMethod +CallStart <- PL + / TL + / '"' + / "'" + / '[' '='* '[' + +DirtyExp <- Exp + / {} -> DirtyExp +MaybeExp <- Exp / MissExp +MissExp <- {} -> MissExp +ExpList <- Sp (MaybeExp (COMMA (MaybeExp))*) + -> List +MustExpList <- Sp (Exp (COMMA (MaybeExp))*) + -> List +CallArgList <- Sp ({} (COMMA {} / Exp)+ {}) + -> CallArgList + / %nil +NameList <- (MustName (COMMA MustName)*) + -> List + +ArgList <- (DOTS -> DotsAsArg / Name / Sp {} COMMA)* + -> ArgList + +Table <- Sp ({} TL TableFields? DirtyTR) + -> Table +TableFields <- (Emmy / TableSep {} / TableField)+ +TableSep <- COMMA / SEMICOLON +TableField <- NewIndex / NewField / Exp +NewIndex <- Sp (Index NeedAssign DirtyExp) + -> NewIndex +NewField <- (MustName ASSIGN DirtyExp) + -> NewField + +Function <- Sp ({} FunctionBody {}) + -> Function +FuncArg <- PL {} ArgList {} NeedPR + / {} {} -> MissPL Nothing {} +FunctionBody<- FUNCTION BlockStart FuncArg + (Emmy / !END Action)* + BlockEnd + NeedEnd + +BlockStart <- {} -> BlockStart +BlockEnd <- {} -> BlockEnd + +-- 纯占位,修改了 `relabel.lua` 使重复定义不抛错 +Action <- !END . +Set <- END +Emmy <- '---@' +]] + +grammar 'Action' [[ +Action <- Sp (CrtAction / UnkAction) +CrtAction <- Semicolon + / Do + / Break + / Return + / Label + / GoTo + / If + / For + / While + / Repeat + / NamedFunction + / LocalFunction + / Local + / Set + / Call + / ExpInAction +UnkAction <- ({} {Word+}) + -> UnknownAction + / ({} '//' {} (LongComment / ShortComment)) + -> CCommentPrefix + / ({} {. (!Sps !CrtAction .)*}) + -> UnknownAction +ExpInAction <- Sp ({} Exp {}) + -> ExpInAction + +Semicolon <- SEMICOLON + -> Skip +SimpleList <- (Simple (COMMA Simple)*) + -> List + +Do <- Sp ({} 'do' Cut DoBody NeedEnd {}) + -> Do +DoBody <- (Emmy / !END Action)* + -> DoBody + +Break <- BREAK ({} Semicolon* AfterBreak?) + -> Break +AfterBreak <- Sp !END !UNTIL !ELSEIF !ELSE Action +BreakStart <- {} -> BreakStart +BreakEnd <- {} -> BreakEnd + +Return <- (ReturnBody Semicolon* AfterReturn?) + -> AfterReturn +ReturnBody <- Sp ({} RETURN MustExpList? {}) + -> Return +AfterReturn <- Sp !END !UNTIL !ELSEIF !ELSE Action + +Label <- Sp ({} LABEL MustName DirtyLabel {}) -> Label + +GoTo <- Sp ({} GOTO MustName {}) -> GoTo + +If <- Sp ({} IfBody {}) + -> If +IfHead <- (IfPart -> IfBlock) + / ({} ElseIfPart -> ElseIfBlock) + -> MissIf + / ({} ElsePart -> ElseBlock) + -> MissIf +IfBody <- IfHead + (ElseIfPart -> ElseIfBlock)* + (ElsePart -> ElseBlock)? + NeedEnd +IfPart <- IF DirtyExp THEN + {} (Emmy / !ELSEIF !ELSE !END Action)* {} + / IF DirtyExp {}->MissThen + {} {} +ElseIfPart <- ELSEIF DirtyExp THEN + {} (Emmy / !ELSE !ELSEIF !END Action)* {} + / ELSEIF DirtyExp {}->MissThen + {} {} +ElsePart <- ELSE + {} (Emmy / !END Action)* {} + +For <- Loop / In + / FOR + +Loop <- Sp ({} LoopBody {}) + -> Loop +LoopBody <- FOR LoopStart LoopFinish LoopStep NeedDo + BreakStart + (Emmy / !END Action)* + BreakEnd + NeedEnd +LoopStart <- MustName AssignOrEQ DirtyExp +LoopFinish <- NeedComma DirtyExp +LoopStep <- COMMA DirtyExp + / NeedComma Exp + / Nothing + +In <- Sp ({} InBody {}) + -> In +InBody <- FOR InNameList NeedIn ExpList NeedDo + BreakStart + (Emmy / !END Action)* + BreakEnd + NeedEnd +InNameList <- &IN DirtyName + / NameList + +While <- Sp ({} WhileBody {}) + -> While +WhileBody <- WHILE DirtyExp NeedDo + BreakStart + (Emmy / !END Action)* + BreakEnd + NeedEnd + +Repeat <- Sp ({} RepeatBody {}) + -> Repeat +RepeatBody <- REPEAT + BreakStart + (Emmy / !UNTIL Action)* + BreakEnd + NeedUntil DirtyExp + +LocalTag <- (Sp '<' Sp MustName Sp LocalTagEnd)* + -> LocalTag +LocalTagEnd <- '>' / {} -> MissGT +Local <- (LOCAL LocalNameList (AssignOrEQ ExpList)?) + -> Local +Set <- (SimpleList AssignOrEQ ExpList?) + -> Set +LocalNameList + <- (LocalName (COMMA LocalName)*) + -> List +LocalName <- (MustName LocalTag) + -> LocalName + +Call <- Simple + -> SimpleCall + +LocalFunction + <- Sp ({} LOCAL FunctionNamedBody {}) + -> LocalFunction + +NamedFunction + <- Sp ({} FunctionNamedBody {}) + -> NamedFunction +FunctionNamedBody + <- FUNCTION FuncName BlockStart FuncArg + (Emmy / !END Action)* + BlockEnd + NeedEnd +FuncName <- (MustName (DOT MustName)* FuncMethod?) + -> Simple +FuncMethod <- COLON Name / COLON {} -> MissMethod + +-- 占位 +Emmy <- '---@' +]] + +grammar 'Emmy' [[ +Emmy <- EmmyAction + / EmmyComments +EmmyAction <- EmmySp '---' %s* '@' EmmyBody ShortComment +EmmySp <- (!'---' Comment / %s / %nl)* +EmmyComments <- EmmyComment+ + -> EmmyComment +EmmyComment <- EmmySp '---' %s* !'@' {(!%nl .)*} +EmmyBody <- 'class' %s+ EmmyClass -> EmmyClass + / 'type' %s+ EmmyType -> EmmyType + / 'alias' %s+ EmmyAlias -> EmmyAlias + / 'param' %s+ EmmyParam -> EmmyParam + / 'return' %s+ EmmyReturn -> EmmyReturn + / 'field' %s+ EmmyField -> EmmyField + / 'generic' %s+ EmmyGeneric -> EmmyGeneric + / 'vararg' %s+ EmmyVararg -> EmmyVararg + / 'language' %s+ EmmyLanguage -> EmmyLanguage + / 'see' %s+ EmmySee -> EmmySee + / 'overload' %s+ EmmyOverLoad -> EmmyOverLoad + / EmmyIncomplete + +EmmyName <- ({} {[a-zA-Z_] [a-zA-Z0-9_]*}) + -> EmmyName +MustEmmyName <- EmmyName / DirtyEmmyName +DirtyEmmyName <- {} -> DirtyEmmyName +EmmyLongName <- ({} {(!%nl .)+}) + -> EmmyName +EmmyIncomplete <- MustEmmyName + -> EmmyIncomplete + +EmmyClass <- (MustEmmyName EmmyParentClass?) +EmmyParentClass <- %s* {} ':' %s* MustEmmyName + +EmmyType <- EmmyFunctionType + / EmmyTableType + / EmmyArrayType + / EmmyCommonType +EmmyCommonType <- EmmyTypeNames + -> EmmyCommonType +EmmyTypeNames <- EmmyTypeName (%s* {} '|' %s* !String EmmyTypeName)* +EmmyTypeName <- EmmyFunctionType + / EmmyTableType + / EmmyArrayType + / MustEmmyName +EmmyTypeEnum <- %s* (%nl %s* '---')? '|' EmmyEnum + -> EmmyTypeEnum +EmmyEnum <- %s* {'>'?} %s* String (EmmyEnumComment / (!%nl !'|' .)*) +EmmyEnumComment <- %s* '#' %s* {(!%nl .)*} + +EmmyAlias <- MustEmmyName %s* EmmyType EmmyTypeEnum* + +EmmyParam <- MustEmmyName %s* EmmyType %s* EmmyOption %s* EmmyTypeEnum* +EmmyOption <- Table? + -> EmmyOption + +EmmyReturn <- {} %nil %nil {} Table -> EmmyOption + / {} EmmyType (%s* EmmyName/%nil) {} EmmyOption + +EmmyField <- (EmmyFieldAccess MustEmmyName %s* EmmyType) +EmmyFieldAccess <- ({'public'} Cut %s*) + / ({'protected'} Cut %s*) + / ({'private'} Cut %s*) + / {} -> 'public' + +EmmyGeneric <- EmmyGenericBlock + (%s* ',' %s* EmmyGenericBlock)* +EmmyGenericBlock<- (MustEmmyName %s* (':' %s* EmmyType)?) + -> EmmyGenericBlock + +EmmyVararg <- EmmyType + +EmmyLanguage <- MustEmmyName + +EmmyArrayType <- ({} MustEmmyName -> EmmyCommonType {} '[' DirtyBR) + -> EmmyArrayType + / ({} PL EmmyCommonType DirtyPR '[' DirtyBR) + -> EmmyArrayType + +EmmyTableType <- ({} 'table' Cut '<' %s* EmmyType %s* ',' %s* EmmyType %s* '>' {}) + -> EmmyTableType + +EmmyFunctionType<- ({} 'fun' Cut %s* EmmyFunctionArgs %s* EmmyFunctionRtns {}) + -> EmmyFunctionType +EmmyFunctionArgs<- ('(' %s* EmmyFunctionArg %s* (',' %s* EmmyFunctionArg %s*)* DirtyPR) + -> EmmyFunctionArgs + / '(' %nil DirtyPR -> None + / %nil +EmmyFunctionRtns<- (':' %s* EmmyType (%s* ',' %s* EmmyType)*) + -> EmmyFunctionRtns + / %nil +EmmyFunctionArg <- MustEmmyName %s* ':' %s* EmmyType + +EmmySee <- {} MustEmmyName %s* '#' %s* MustEmmyName {} +EmmyOverLoad <- EmmyFunctionType +]] + +grammar 'Lua' [[ +Lua <- Head? + (Emmy / Action)* -> Lua + BlockEnd + Sp +Head <- '#' (!%nl .)* +]] + +return function (self, lua, mode) + local gram = compiled[mode] or compiled['Lua'] + local r, _, pos = gram:match(lua) + if not r then + local err = errorpos(pos) + return nil, err + end + + return r +end diff --git a/script/src/parser/init.lua b/script/src/parser/init.lua new file mode 100644 index 00000000..30596dbe --- /dev/null +++ b/script/src/parser/init.lua @@ -0,0 +1,9 @@ +local api = { + grammar = require 'parser.grammar', + parse = require 'parser.parse', + split = require 'parser.split', + calcline = require 'parser.calcline', + lines = require 'parser.lines', +} + +return api diff --git a/script/src/parser/lines.lua b/script/src/parser/lines.lua new file mode 100644 index 00000000..f2f076e1 --- /dev/null +++ b/script/src/parser/lines.lua @@ -0,0 +1,187 @@ +local m = require 'lpeglabel' + +local function utf8_len(buf, start, finish) + local len, pos = utf8.len(buf, start, finish) + if len then + return len + end + return 1 + utf8_len(buf, start, pos-1) + utf8_len(buf, pos+1, finish) +end + +local function Line(start, line, finish) + line.start = start + line.finish = finish - 1 + return line +end + +local function Space(...) + local line = {...} + local sp = 0 + local tab = 0 + for i = 1, #line do + if line[i] == ' ' then + sp = sp + 1 + elseif line[i] == '\t' then + tab = tab + 1 + end + line[i] = nil + end + line.sp = sp + line.tab = tab + return line +end + +local parser = m.P{ +'Lines', +Lines = m.Ct(m.V'Line'^0 * m.V'LastLine'), +Line = m.Cp() * m.V'Indent' * (1 - m.V'Nl')^0 * m.Cp() * m.V'Nl' / Line, +LastLine= m.Cp() * m.V'Indent' * (1 - m.V'Nl')^0 * m.Cp() / Line, +Nl = m.P'\r\n' + m.S'\r\n', +Indent = m.C(m.S' \t')^0 / Space, +} + +local mt = {} +mt.__index = mt + +function mt:position(row, col, code) + if row < 1 then + return 1 + end + code = code or self.code + if row > #self then + if code == 'utf8' then + return utf8_len(self.buf) + 1 + else + return #self.buf + 1 + end + end + local line = self[row] + local next_line = self[row+1] + local start = line.start + local finish + if next_line then + finish = next_line.start - 1 + else + finish = #self.buf + 1 + end + local pos + if code == 'utf8' then + pos = utf8.offset(self.buf, col, start) or finish + else + pos = start + col - 1 + end + if pos < start then + pos = start + elseif pos > finish then + pos = finish + end + return pos +end + +local function isCharByte(byte) + if not byte then + return false + end + -- [0-9] + if byte >= 48 and byte <= 57 then + return true + end + -- [A-Z] + if byte >= 65 and byte <= 90 then + return true + end + -- [a-z] + if byte >= 97 and byte <= 122 then + return true + end + -- <utf8> + if byte >= 128 then + return true + end + return false +end + +function mt:positionAsChar(row, col, code) + local pos = self:position(row, col, code) + if isCharByte(self.buf:byte(pos, pos)) then + return pos + elseif isCharByte(self.buf:byte(pos+1, pos+1)) then + return pos + 1 + end + return pos +end + +function mt:rowcol(pos, code) + if pos < 1 then + return 1, 1 + end + code = code or self.code + if pos >= #self.buf + 1 then + local start = self[#self].start + if code == 'utf8' then + return #self, utf8_len(self.buf, start) + 1 + else + return #self, #self.buf - start + 2 + end + end + local min = 1 + local max = #self + for _ = 1, 100 do + if max == min then + local start = self[min].start + if code == 'utf8' then + return min, utf8_len(self.buf, start, pos) + else + return min, pos - start + 1 + end + end + local row = (max - min) // 2 + min + local start = self[row].start + if pos < start then + max = row + elseif pos > start then + local next_start = self[row + 1].start + if pos < next_start then + if code == 'utf8' then + return row, utf8_len(self.buf, start, pos) + else + return row, pos - start + 1 + end + elseif pos > next_start then + min = row + 1 + else + return row + 1, 1 + end + else + return row, 1 + end + end + error('rowcol failed!') +end + +function mt:line(i) + local start, finish = self:range(i) + return self.buf:sub(start, finish) +end + +function mt:range(i) + if i < 1 or i > #self then + return 0, 0 + end + return self[i].start, self[i].finish +end + +function mt:set_code(code) + self.code = code +end + +return function (self, buf, code) + local lines, err = parser:match(buf) + if not lines then + return nil, err + end + lines.buf = buf + lines.code = code + + return setmetatable(lines, mt) +end diff --git a/script/src/parser/parse.lua b/script/src/parser/parse.lua new file mode 100644 index 00000000..6ad79d9b --- /dev/null +++ b/script/src/parser/parse.lua @@ -0,0 +1,41 @@ +local ast = require 'parser.ast' + +local Errs +local State + +local function pushError(err) + if err.finish < err.start then + err.finish = err.start + end + local last = Errs[#Errs] + if last then + if last.start <= err.start and last.finish >= err.finish then + return + end + end + err.level = err.level or 'error' + Errs[#Errs+1] = err + return err +end + +return function (self, lua, mode, version) + Errs = {} + State= { + Break = 0, + Label = {{}}, + Dots = {true}, + Version = version, + Comments = {}, + Lua = lua, + } + ast.init(State, Errs) + local suc, res, err = xpcall(self.grammar, debug.traceback, self, lua, mode) + if not suc then + return nil, res + end + if not res then + pushError(err) + return nil, Errs + end + return res, Errs, State.Comments +end diff --git a/script/src/parser/relabel.lua b/script/src/parser/relabel.lua new file mode 100644 index 00000000..ac902403 --- /dev/null +++ b/script/src/parser/relabel.lua @@ -0,0 +1,361 @@ +-- $Id: re.lua,v 1.44 2013/03/26 20:11:40 roberto Exp $ + +-- imported functions and modules +local tonumber, type, print, error = tonumber, type, print, error +local pcall = pcall +local setmetatable = setmetatable +local tinsert, concat = table.insert, table.concat +local rep = string.rep +local m = require"lpeglabel" + +-- 'm' will be used to parse expressions, and 'mm' will be used to +-- create expressions; that is, 're' runs on 'm', creating patterns +-- on 'mm' +local mm = m + +-- pattern's metatable +local mt = getmetatable(mm.P(0)) + + + +-- No more global accesses after this point +_ENV = nil + + +local any = m.P(1) +local dummy = mm.P(false) + + +local errinfo = { + NoPatt = "no pattern found", + ExtraChars = "unexpected characters after the pattern", + + ExpPatt1 = "expected a pattern after '/'", + + ExpPatt2 = "expected a pattern after '&'", + ExpPatt3 = "expected a pattern after '!'", + + ExpPatt4 = "expected a pattern after '('", + ExpPatt5 = "expected a pattern after ':'", + ExpPatt6 = "expected a pattern after '{~'", + ExpPatt7 = "expected a pattern after '{|'", + + ExpPatt8 = "expected a pattern after '<-'", + + ExpPattOrClose = "expected a pattern or closing '}' after '{'", + + ExpNumName = "expected a number, '+', '-' or a name (no space) after '^'", + ExpCap = "expected a string, number, '{}' or name after '->'", + + ExpName1 = "expected the name of a rule after '=>'", + ExpName2 = "expected the name of a rule after '=' (no space)", + ExpName3 = "expected the name of a rule after '<' (no space)", + + ExpLab1 = "expected a label after '{'", + + ExpNameOrLab = "expected a name or label after '%' (no space)", + + ExpItem = "expected at least one item after '[' or '^'", + + MisClose1 = "missing closing ')'", + MisClose2 = "missing closing ':}'", + MisClose3 = "missing closing '~}'", + MisClose4 = "missing closing '|}'", + MisClose5 = "missing closing '}'", -- for the captures + + MisClose6 = "missing closing '>'", + MisClose7 = "missing closing '}'", -- for the labels + + MisClose8 = "missing closing ']'", + + MisTerm1 = "missing terminating single quote", + MisTerm2 = "missing terminating double quote", +} + +local function expect (pattern, label) + return pattern + m.T(label) +end + + +-- Pre-defined names +local Predef = { nl = m.P"\n" } + + +local mem +local fmem +local gmem + + +local function updatelocale () + mm.locale(Predef) + Predef.a = Predef.alpha + Predef.c = Predef.cntrl + Predef.d = Predef.digit + Predef.g = Predef.graph + Predef.l = Predef.lower + Predef.p = Predef.punct + Predef.s = Predef.space + Predef.u = Predef.upper + Predef.w = Predef.alnum + Predef.x = Predef.xdigit + Predef.A = any - Predef.a + Predef.C = any - Predef.c + Predef.D = any - Predef.d + Predef.G = any - Predef.g + Predef.L = any - Predef.l + Predef.P = any - Predef.p + Predef.S = any - Predef.s + Predef.U = any - Predef.u + Predef.W = any - Predef.w + Predef.X = any - Predef.x + mem = {} -- restart memoization + fmem = {} + gmem = {} + local mt = {__mode = "v"} + setmetatable(mem, mt) + setmetatable(fmem, mt) + setmetatable(gmem, mt) +end + + +updatelocale() + + + +local I = m.P(function (s,i) print(i, s:sub(1, i-1)); return i end) + + +local function getdef (id, defs) + local c = defs and defs[id] + if not c then + error("undefined name: " .. id) + end + return c +end + + +local function mult (p, n) + local np = mm.P(true) + while n >= 1 do + if n%2 >= 1 then np = np * p end + p = p * p + n = n/2 + end + return np +end + +local function equalcap (s, i, c) + if type(c) ~= "string" then return nil end + local e = #c + i + if s:sub(i, e - 1) == c then return e else return nil end +end + + +local S = (Predef.space + "--" * (any - Predef.nl)^0)^0 + +local name = m.C(m.R("AZ", "az", "__") * m.R("AZ", "az", "__", "09")^0) + +local arrow = S * "<-" + +-- a defined name only have meaning in a given environment +local Def = name * m.Carg(1) + +local num = m.C(m.R"09"^1) * S / tonumber + +local String = "'" * m.C((any - "'" - m.P"\n")^0) * expect("'", "MisTerm1") + + '"' * m.C((any - '"' - m.P"\n")^0) * expect('"', "MisTerm2") + + +local defined = "%" * Def / function (c,Defs) + local cat = Defs and Defs[c] or Predef[c] + if not cat then + error("name '" .. c .. "' undefined") + end + return cat +end + +local Range = m.Cs(any * (m.P"-"/"") * (any - "]")) / mm.R + +local item = defined + Range + m.C(any - m.P"\n") + +local Class = + "[" + * (m.C(m.P"^"^-1)) -- optional complement symbol + * m.Cf(expect(item, "ExpItem") * (item - "]")^0, mt.__add) + / function (c, p) return c == "^" and any - p or p end + * expect("]", "MisClose8") + +local function adddef (t, k, exp) + if t[k] then + -- TODO 改了一下这里的代码,重复定义不会抛错 + --error("'"..k.."' already defined as a rule") + else + t[k] = exp + end + return t +end + +local function firstdef (n, r) return adddef({n}, n, r) end + + +local function NT (n, b) + if not b then + error("rule '"..n.."' used outside a grammar") + else return mm.V(n) + end +end + + +local exp = m.P{ "Exp", + Exp = S * ( m.V"Grammar" + + m.Cf(m.V"Seq" * (S * "/" * expect(S * m.V"Seq", "ExpPatt1"))^0, mt.__add) ); + Seq = m.Cf(m.Cc(m.P"") * m.V"Prefix" * (S * m.V"Prefix")^0, mt.__mul); + Prefix = "&" * expect(S * m.V"Prefix", "ExpPatt2") / mt.__len + + "!" * expect(S * m.V"Prefix", "ExpPatt3") / mt.__unm + + m.V"Suffix"; + Suffix = m.Cf(m.V"Primary" * + ( S * ( m.P"+" * m.Cc(1, mt.__pow) + + m.P"*" * m.Cc(0, mt.__pow) + + m.P"?" * m.Cc(-1, mt.__pow) + + "^" * expect( m.Cg(num * m.Cc(mult)) + + m.Cg(m.C(m.S"+-" * m.R"09"^1) * m.Cc(mt.__pow) + + name * m.Cc"lab" + ), + "ExpNumName") + + "->" * expect(S * ( m.Cg((String + num) * m.Cc(mt.__div)) + + m.P"{}" * m.Cc(nil, m.Ct) + + m.Cg(Def / getdef * m.Cc(mt.__div)) + ), + "ExpCap") + + "=>" * expect(S * m.Cg(Def / getdef * m.Cc(m.Cmt)), + "ExpName1") + ) + )^0, function (a,b,f) if f == "lab" then return a + mm.T(b) else return f(a,b) end end ); + Primary = "(" * expect(m.V"Exp", "ExpPatt4") * expect(S * ")", "MisClose1") + + String / mm.P + + Class + + defined + + "%" * expect(m.P"{", "ExpNameOrLab") + * expect(S * m.V"Label", "ExpLab1") + * expect(S * "}", "MisClose7") / mm.T + + "{:" * (name * ":" + m.Cc(nil)) * expect(m.V"Exp", "ExpPatt5") + * expect(S * ":}", "MisClose2") + / function (n, p) return mm.Cg(p, n) end + + "=" * expect(name, "ExpName2") + / function (n) return mm.Cmt(mm.Cb(n), equalcap) end + + m.P"{}" / mm.Cp + + "{~" * expect(m.V"Exp", "ExpPatt6") + * expect(S * "~}", "MisClose3") / mm.Cs + + "{|" * expect(m.V"Exp", "ExpPatt7") + * expect(S * "|}", "MisClose4") / mm.Ct + + "{" * expect(m.V"Exp", "ExpPattOrClose") + * expect(S * "}", "MisClose5") / mm.C + + m.P"." * m.Cc(any) + + (name * -arrow + "<" * expect(name, "ExpName3") + * expect(">", "MisClose6")) * m.Cb("G") / NT; + Label = num + name; + Definition = name * arrow * expect(m.V"Exp", "ExpPatt8"); + Grammar = m.Cg(m.Cc(true), "G") + * m.Cf(m.V"Definition" / firstdef * (S * m.Cg(m.V"Definition"))^0, + adddef) / mm.P; +} + +local pattern = S * m.Cg(m.Cc(false), "G") * expect(exp, "NoPatt") / mm.P + * S * expect(-any, "ExtraChars") + +local function lineno (s, i) + if i == 1 then return 1, 1 end + local adjustment = 0 + -- report the current line if at end of line, not the next + if s:sub(i,i) == '\n' then + i = i-1 + adjustment = 1 + end + local rest, num = s:sub(1,i):gsub("[^\n]*\n", "") + local r = #rest + return 1 + num, (r ~= 0 and r or 1) + adjustment +end + +local function calcline (s, i) + if i == 1 then return 1, 1 end + local rest, line = s:sub(1,i):gsub("[^\n]*\n", "") + local col = #rest + return 1 + line, col ~= 0 and col or 1 +end + + +local function splitlines(str) + local t = {} + local function helper(line) tinsert(t, line) return "" end + helper((str:gsub("(.-)\r?\n", helper))) + return t +end + +local function compile (p, defs) + if mm.type(p) == "pattern" then return p end -- already compiled + p = p .. " " -- for better reporting of column numbers in errors when at EOF + local ok, cp, label, poserr = pcall(function() return pattern:match(p, 1, defs) end) + if not ok and cp then + if type(cp) == "string" then + cp = cp:gsub("^[^:]+:[^:]+: ", "") + end + error(cp, 3) + end + if not cp then + local lines = splitlines(p) + local line, col = lineno(p, poserr) + local err = {} + tinsert(err, "L" .. line .. ":C" .. col .. ": " .. errinfo[label]) + tinsert(err, lines[line]) + tinsert(err, rep(" ", col-1) .. "^") + error("syntax error(s) in pattern\n" .. concat(err, "\n"), 3) + end + return cp +end + +local function match (s, p, i) + local cp = mem[p] + if not cp then + cp = compile(p) + mem[p] = cp + end + return cp:match(s, i or 1) +end + +local function find (s, p, i) + local cp = fmem[p] + if not cp then + cp = compile(p) / 0 + cp = mm.P{ mm.Cp() * cp * mm.Cp() + 1 * mm.V(1) } + fmem[p] = cp + end + local i, e = cp:match(s, i or 1) + if i then return i, e - 1 + else return i + end +end + +local function gsub (s, p, rep) + local g = gmem[p] or {} -- ensure gmem[p] is not collected while here + gmem[p] = g + local cp = g[rep] + if not cp then + cp = compile(p) + cp = mm.Cs((cp / rep + 1)^0) + g[rep] = cp + end + return cp:match(s) +end + + +-- exported names +local re = { + compile = compile, + match = match, + find = find, + gsub = gsub, + updatelocale = updatelocale, + calcline = calcline +} + +return re diff --git a/script/src/parser/split.lua b/script/src/parser/split.lua new file mode 100644 index 00000000..6ce4a4e7 --- /dev/null +++ b/script/src/parser/split.lua @@ -0,0 +1,9 @@ +local m = require 'lpeglabel' + +local NL = m.P'\r\n' + m.S'\r\n' +local LINE = m.C(1 - NL) + +return function (str) + local MATCH = m.Ct((LINE * NL)^0 * LINE) + return MATCH:match(str) +end |