import macros;macro ImportExpand(s:untyped):untyped = parseStmt($s[2])
# source: src/cplib/tmpl/sheep.nim
ImportExpand "cplib/tmpl/sheep" <=== "when not declared CPLIB_TMPL_SHEEP:\n    const CPLIB_TMPL_SHEEP* = 1\n    {.warning[UnusedImport]: off.}\n    {.hint[XDeclaredButNotUsed]: off.}\n    import algorithm\n    import sequtils\n    import tables\n    import macros\n    import math\n    import sets\n    import strutils\n    import strformat\n    import sugar\n    import heapqueue\n    import streams\n    import deques\n    import bitops\n    import std/lenientops\n    import options\n    #入力系\n    {.emit: \"\"\"\n    #include <cstdio>\n    #include <cstdint>\n    #include <cstring>\n    #include <sys/mman.h>\n    #include <sys/stat.h>\n\n    namespace cplib_sheep_input {\n    constexpr std::size_t buffer_size = 1U << 20;\n    char buffer[buffer_size];\n    std::size_t cursor = 0;\n    std::size_t length = 0;\n    const char* mapped = nullptr;\n    bool initialized = false;\n\n    inline void initialize() {\n      if (initialized) return;\n      initialized = true;\n\n      struct stat st;\n      const int fd = fileno(stdin);\n      if (fstat(fd, &st) == 0 && S_ISREG(st.st_mode) && st.st_size > 0) {\n        void* p = mmap(nullptr, static_cast<std::size_t>(st.st_size),\n                       PROT_READ, MAP_PRIVATE, fd, 0);\n        if (p != MAP_FAILED) {\n          mapped = static_cast<const char*>(p);\n          length = static_cast<std::size_t>(st.st_size);\n          madvise(const_cast<char*>(mapped), length, MADV_SEQUENTIAL);\n        }\n      }\n    }\n\n    inline int get_char() {\n      initialize();\n      if (mapped != nullptr) {\n        if (cursor == length) return -1;\n        return static_cast<unsigned char>(mapped[cursor++]);\n      }\n\n      if (cursor == length) {\n        length = fread_unlocked(buffer, 1, buffer_size, stdin);\n        cursor = 0;\n        if (length == 0) return -1;\n      }\n      return static_cast<unsigned char>(buffer[cursor++]);\n    }\n\n    inline bool refill() {\n      length = fread_unlocked(buffer, 1, buffer_size, stdin);\n      cursor = 0;\n      return length != 0;\n    }\n\n    inline bool has_eight_digits(const char* source) {\n      std::uint64_t bytes;\n      std::memcpy(&bytes, source, sizeof(bytes));\n      constexpr std::uint64_t high_nibbles = 0xf0f0f0f0f0f0f0f0ULL;\n      return (bytes & high_nibbles) == 0x3030303030303030ULL &&\n             ((bytes + 0x0606060606060606ULL) & high_nibbles) ==\n                 0x3030303030303030ULL;\n    }\n\n    inline unsigned parse_eight_digits(const char* source) {\n#if defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__\n      std::uint64_t digits;\n      std::memcpy(&digits, source, sizeof(digits));\n      digits -= 0x3030303030303030ULL;\n      digits = (digits * 10 + (digits >> 8)) & 0x00ff00ff00ff00ffULL;\n      digits = (digits * 100 + (digits >> 16)) & 0x0000ffff0000ffffULL;\n      return static_cast<unsigned>(\n          (digits * 10000 + (digits >> 32)) & 0xffffffffULL);\n#else\n      unsigned result = 0;\n      for (int i = 0; i < 8; ++i) {\n        result = result * 10U + static_cast<unsigned>(source[i] - '0');\n      }\n      return result;\n#endif\n    }\n\n    inline long long read_int() {\n      initialize();\n\n      if (mapped != nullptr) {\n        while (cursor < length && mapped[cursor] <= ' ') ++cursor;\n        if (cursor == length) return 0;\n\n        const bool negative = mapped[cursor] == '-';\n        if (negative) {\n          ++cursor;\n          if (cursor == length) return 0;\n        }\n\n        if (!negative && length - cursor >= 9 &&\n            has_eight_digits(mapped + cursor)) {\n          const unsigned value = parse_eight_digits(mapped + cursor);\n          const unsigned ninth = static_cast<unsigned>(mapped[cursor + 8] - '0');\n          if (ninth >= 10U) {\n            cursor += 8;\n            return static_cast<long long>(value);\n          }\n          if (length - cursor >= 10 &&\n              static_cast<unsigned>(mapped[cursor + 9] - '0') >= 10U) {\n            cursor += 9;\n            return static_cast<long long>(value * 10U + ninth);\n          }\n        }\n\n        long long value = 0;\n        if (negative) {\n          while (length - cursor >= 2) {\n            const unsigned first = static_cast<unsigned>(mapped[cursor] - '0');\n            const unsigned second = static_cast<unsigned>(mapped[cursor + 1] - '0');\n            if (first >= 10U || second >= 10U) break;\n            value = value * 100 - static_cast<long long>(first * 10U + second);\n            cursor += 2;\n          }\n          if (cursor < length) {\n            const unsigned digit = static_cast<unsigned>(mapped[cursor] - '0');\n            if (digit < 10U) {\n              value = value * 10 - static_cast<long long>(digit);\n              ++cursor;\n            }\n          }\n        } else {\n          while (length - cursor >= 2) {\n            const unsigned first = static_cast<unsigned>(mapped[cursor] - '0');\n            const unsigned second = static_cast<unsigned>(mapped[cursor + 1] - '0');\n            if (first >= 10U || second >= 10U) break;\n            value = value * 100 + static_cast<long long>(first * 10U + second);\n            cursor += 2;\n          }\n          if (cursor < length) {\n            const unsigned digit = static_cast<unsigned>(mapped[cursor] - '0');\n            if (digit < 10U) {\n              value = value * 10 + static_cast<long long>(digit);\n              ++cursor;\n            }\n          }\n        }\n        return value;\n      }\n\n      for (;;) {\n        if (cursor == length && !refill()) return 0;\n        while (cursor < length && buffer[cursor] <= ' ') ++cursor;\n        if (cursor < length) break;\n      }\n\n      const bool negative = buffer[cursor] == '-';\n      if (negative) ++cursor;\n      long long value = 0;\n\n      for (;;) {\n        if (!negative && length - cursor >= 9 &&\n            has_eight_digits(buffer + cursor)) {\n          const unsigned first_eight = parse_eight_digits(buffer + cursor);\n          const unsigned ninth = static_cast<unsigned>(buffer[cursor + 8] - '0');\n          if (ninth >= 10U) {\n            cursor += 8;\n            return static_cast<long long>(first_eight);\n          }\n          if (length - cursor >= 10 &&\n              static_cast<unsigned>(buffer[cursor + 9] - '0') >= 10U) {\n            cursor += 9;\n            return static_cast<long long>(first_eight * 10U + ninth);\n          }\n        }\n\n        while (length - cursor >= 2) {\n          const unsigned first = static_cast<unsigned>(buffer[cursor] - '0');\n          const unsigned second = static_cast<unsigned>(buffer[cursor + 1] - '0');\n          if (first >= 10U) return value;\n          if (second >= 10U) {\n            value = negative\n                ? value * 10 - static_cast<long long>(first)\n                : value * 10 + static_cast<long long>(first);\n            ++cursor;\n            return value;\n          }\n          value = negative\n              ? value * 100 - static_cast<long long>(first * 10U + second)\n              : value * 100 + static_cast<long long>(first * 10U + second);\n          cursor += 2;\n        }\n\n        if (cursor < length) {\n          const unsigned digit = static_cast<unsigned>(buffer[cursor] - '0');\n          if (digit >= 10U) return value;\n          value = negative\n              ? value * 10 - static_cast<long long>(digit)\n              : value * 10 + static_cast<long long>(digit);\n          ++cursor;\n        }\n        if (!refill()) return value;\n      }\n    }\n\n    template <class T>\n    inline void read_int_array(T* output, std::size_t count) {\n      for (std::size_t i = 0; i < count; ++i) {\n        output[i] = static_cast<T>(read_int());\n      }\n    }\n    } // namespace cplib_sheep_input\n    \"\"\".}\n\n    proc sheepGetChar(): cint {.importcpp: \"cplib_sheep_input::get_char()\", nodecl, inline.}\n    proc sheepReadInt(): clonglong {.importcpp: \"cplib_sheep_input::read_int()\", nodecl, inline.}\n    proc sheepReadIntArray(values: ptr int, count: csize_t) {.importcpp: \"cplib_sheep_input::read_int_array(@)\", nodecl, inline.}\n\n    proc ii(): int {.inline.} = sheepReadInt().int\n    proc lii(N: int): seq[int] {.inline.} =\n        result = newSeq[int](N)\n        if N > 0:\n            sheepReadIntArray(addr result[0], N.csize_t)\n\n    proc si(): string {.inline.} =\n        var c = sheepGetChar()\n        while c >= 0 and c <= ord(' '):\n            c = sheepGetChar()\n        while c > ord(' '):\n            result.add(char(c))\n            c = sheepGetChar()\n    \n    # 出力系\n    {.emit: \"\"\"\n    #include <cstddef>\n    #include <cstdint>\n    #include <cstring>\n    #include <type_traits>\n\n    namespace cplib_sheep_output {\n    struct FourDigits {\n      char data[10000][4];\n      FourDigits() {\n        for (unsigned i = 0; i < 10000; ++i) {\n          data[i][0] = static_cast<char>('0' + i / 1000);\n          data[i][1] = static_cast<char>('0' + i / 100 % 10);\n          data[i][2] = static_cast<char>('0' + i / 10 % 10);\n          data[i][3] = static_cast<char>('0' + i % 10);\n        }\n      }\n    };\n\n    inline const FourDigits& four_digits() {\n      static const FourDigits table;\n      return table;\n    }\n\n    inline char* write_small(char* output, unsigned value,\n                             const FourDigits& table) {\n      if (value >= 1000) {\n        std::memcpy(output, table.data[value], 4);\n        return output + 4;\n      }\n      if (value >= 100) {\n        std::memcpy(output, table.data[value] + 1, 3);\n        return output + 3;\n      }\n      if (value >= 10) {\n        std::memcpy(output, table.data[value] + 2, 2);\n        return output + 2;\n      }\n      *output++ = static_cast<char>('0' + value);\n      return output;\n    }\n\n    template <class Unsigned>\n    inline char* write_unsigned(char* output, Unsigned value,\n                                const FourDigits& table) {\n      unsigned chunks[5];\n      unsigned count = 0;\n      while (value >= 10000) {\n        const Unsigned quotient = value / 10000;\n        chunks[count++] = static_cast<unsigned>(value - quotient * 10000);\n        value = quotient;\n      }\n      output = write_small(output, static_cast<unsigned>(value), table);\n      while (count != 0) {\n        std::memcpy(output, table.data[chunks[--count]], 4);\n        output += 4;\n      }\n      return output;\n    }\n\n    template <class Integer>\n    inline std::size_t join_signed(\n            const Integer* values, std::size_t count, char* output,\n            const char* separator, std::size_t separator_length) {\n      const FourDigits& table = four_digits();\n      char* cursor = output;\n      using Unsigned = typename std::make_unsigned<Integer>::type;\n      for (std::size_t i = 0; i < count; ++i) {\n        const Integer value = values[i];\n        Unsigned magnitude = static_cast<Unsigned>(value);\n        if (value < 0) {\n          *cursor++ = '-';\n          magnitude = Unsigned(0) - magnitude;\n        }\n        cursor = write_unsigned(cursor, magnitude, table);\n        if (i + 1 != count) {\n          std::memcpy(cursor, separator, separator_length);\n          cursor += separator_length;\n        }\n      }\n      return static_cast<std::size_t>(cursor - output);\n    }\n\n    template <class Integer>\n    inline std::size_t join_unsigned(\n            const Integer* values, std::size_t count, char* output,\n            const char* separator, std::size_t separator_length) {\n      const FourDigits& table = four_digits();\n      char* cursor = output;\n      for (std::size_t i = 0; i < count; ++i) {\n        cursor = write_unsigned(cursor, values[i], table);\n        if (i + 1 != count) {\n          std::memcpy(cursor, separator, separator_length);\n          cursor += separator_length;\n        }\n      }\n      return static_cast<std::size_t>(cursor - output);\n    }\n    }  // namespace cplib_sheep_output\n    \"\"\".}\n\n    proc sheepJoinI32(values: ptr int32, count: csize_t, output: ptr char,\n                      separator: cstring, separatorLen: csize_t): csize_t\n        {.importcpp: \"cplib_sheep_output::join_signed(@)\", nodecl.}\n    proc sheepJoinI64(values: ptr int64, count: csize_t, output: ptr char,\n                      separator: cstring, separatorLen: csize_t): csize_t\n        {.importcpp: \"cplib_sheep_output::join_signed(@)\", nodecl.}\n    proc sheepJoinU32(values: ptr uint32, count: csize_t, output: ptr char,\n                      separator: cstring, separatorLen: csize_t): csize_t\n        {.importcpp: \"cplib_sheep_output::join_unsigned(@)\", nodecl.}\n    proc sheepJoinU64(values: ptr uint64, count: csize_t, output: ptr char,\n                      separator: cstring, separatorLen: csize_t): csize_t\n        {.importcpp: \"cplib_sheep_output::join_unsigned(@)\", nodecl.}\n\n    # 1. 実際の処理を行う proc (openArray を受け取る)\n    proc print_internal(prop: tuple[f: File, sepc: string, endc: string, flush: bool], args: openArray[string]) =\n        for i in 0 ..< args.len:\n            prop.f.write(args[i])\n            if i != args.len - 1:\n                prop.f.write(prop.sepc)\n            else:\n                prop.f.write(prop.endc)\n        if prop.flush:\n            prop.f.flushFile()\n\n    # 2. ユーザーが呼び出すためのインターフェース (varargs を受け取る)\n    proc print*(prop: tuple[f: File, sepc: string, endc: string, flush: bool], args: varargs[string, `$`]) =\n        # varargs は内部では openArray として扱えるので、そのまま渡せる\n        print_internal(prop, args)\n\n    proc print*(args: varargs[string, `$`]) =\n        # こちらも内部用の proc を呼ぶ\n        print_internal((f: stdout, sepc: \" \", endc: \"\\n\", flush: false), args)\n\n    proc sheepPrintWithSeparator(sep: string,\n            args: varargs[string, `$`]) =\n        print_internal((f: stdout, sepc: sep, endc: \"\\n\", flush: false), args)\n\n    macro getSymbolName(x: typed): string = x.toStrLit\n    macro debug*(args: varargs[untyped]): untyped =\n        when defined(debug):\n            result = newNimNode(nnkStmtList, args)\n            template prop(e: string = \"\"): untyped = (f: stderr, sepc: \"\", endc: e, flush: true)\n            for i, arg in args:\n                if arg.kind == nnkStrLit:\n                    result.add(quote do: print(prop(), \"\\\"\", `arg`, \"\\\"\"))\n                else:\n                    result.add(quote do: print(prop(\": \"), getSymbolName(`arg`)))\n                    result.add(quote do: print(prop(), `arg`))\n                if i != args.len - 1: result.add(quote do: print(prop(), \", \"))\n                else: result.add(quote do: print(prop(), \"\\n\"))\n        else:\n            return (quote do: discard)\n    #chmin,chmax\n    template `max=`(x, y) =\n        let yVal = y # yが計算式の場合に評価を1回にするため\n        if x < yVal:\n            x = yVal\n\n    template `min=`(x, y) =\n        let yVal = y\n        if x > yVal:\n            x = yVal\n    proc chmin[T](x: var T, y: T):bool=\n        if x > y:\n            x = y\n            return true\n        return false\n    proc chmax[T](x: var T, y: T):bool=\n        if x < y:\n            x = y\n            return true\n        return false\n    #bit演算\n    proc `%`*(x: int, y: int): int =\n        result = x mod y\n        if y > 0 and result < 0: result += y\n        if y < 0 and result > 0: result += y\n    proc `//`*(x: int, y: int): int{.inline.} =\n        result = x div y\n        if y > 0 and result * y > x: result -= 1\n        if y < 0 and result * y < x: result -= 1\n    proc `%=`(x: var int, y: int): void = x = x%y\n    proc `//=`(x: var int, y: int): void = x = x//y\n    proc `**`(x: int, y: int): int = x^y\n    proc `**=`(x: var int, y: int): void = x = x^y\n    proc `^`(x: int, y: int): int = x xor y\n    proc `|`(x: int, y: int): int = x or y\n    proc `&`(x: int, y: int): int = x and y\n    proc `>>`(x: int, y: int): int = x shr y\n    proc `<<`(x: int, y: int): int = x shl y\n    proc `~`(x: int): int = not x\n    proc `^=`(x: var int, y: int): void = x = x ^ y\n    proc `&=`(x: var int, y: int): void = x = x & y\n    proc `|=`(x: var int, y: int): void = x = x | y\n    proc `>>=`(x: var int, y: int): void = x = x >> y\n    proc `<<=`(x: var int, y: int): void = x = x << y\n    proc `[]`(x: int, n: int): bool = (x and (1 shl n)) != 0\n    #便利な変換\n    proc `!`(x: char, a = '0'): int = int(x)-int(a)\n    #定数\n    when not declared CPLIB_UTILS_CONSTANTS:\n        const CPLIB_UTILS_CONSTANTS* = 1\n        const INF32*: int32 = 1001000027.int32\n        const INF64*: int = int(3300300300300300491)\n    \n    const INF = INF64\n    #converter\n\n    #range\n    iterator range(start: int, ends: int, step: int): int =\n        var i = start\n        if step < 0:\n            while i > ends:\n                yield i\n                i += step\n        elif step > 0:\n            while i < ends:\n                yield i\n                i += step\n    iterator range(ends: int): int = (for i in 0..<ends: yield i)\n    iterator range(start: int, ends: int): int = (for i in\n            start..<ends: yield i)\n\n    # 整数は4桁テーブルを使うC++フォーマッタへまとめて渡す。\n    proc sheepJoinImpl[T](a: openArray[T], sep: string): string =\n        if a.len == 0:\n            return \"\"\n        when T is SomeSignedInt and sizeof(T) == 8:\n            result = newString(a.len * 20 + (a.len - 1) * sep.len)\n            let written = sheepJoinI64(cast[ptr int64](unsafeAddr a[0]),\n                a.len.csize_t, addr result[0], sep.cstring, sep.len.csize_t)\n            result.setLen(written.int)\n        elif T is SomeSignedInt and sizeof(T) == 4:\n            result = newString(a.len * 11 + (a.len - 1) * sep.len)\n            let written = sheepJoinI32(cast[ptr int32](unsafeAddr a[0]),\n                a.len.csize_t, addr result[0], sep.cstring, sep.len.csize_t)\n            result.setLen(written.int)\n        elif T is SomeUnsignedInt and sizeof(T) == 8:\n            result = newString(a.len * 20 + (a.len - 1) * sep.len)\n            let written = sheepJoinU64(cast[ptr uint64](unsafeAddr a[0]),\n                a.len.csize_t, addr result[0], sep.cstring, sep.len.csize_t)\n            result.setLen(written.int)\n        elif T is SomeUnsignedInt and sizeof(T) == 4:\n            result = newString(a.len * 10 + (a.len - 1) * sep.len)\n            let written = sheepJoinU32(cast[ptr uint32](unsafeAddr a[0]),\n                a.len.csize_t, addr result[0], sep.cstring, sep.len.csize_t)\n            result.setLen(written.int)\n        elif compiles(a[0].umod) and compiles(a[0].val):\n            # Montgomery表現を含め、公開値へ正規化してから一括変換する。\n            var canonical = newSeq[uint32](a.len)\n            for i, value in a:\n                canonical[i] = value.val.uint32\n            result = sheepJoinImpl(canonical, sep)\n        else:\n            result = newStringOfCap(a.len * 4)\n            for i, value in a:\n                if i != 0:\n                    result.add(sep)\n                result.add($value)\n\n    proc join*[T: not string](a: openArray[T], sep: string = \"\"): string {.inline.} =\n        sheepJoinImpl(a, sep)\n\n    # Python風に print(*X) と書くと、Xを空白区切りで1行に出力する。\n    template `*`*[T](values: openArray[T]): string =\n        sheepJoinImpl(values, \" \")\n\n    # 最後の文字列引数をsepと誤認しないよう、名前付きsepはマクロで処理する。\n    macro print*(args: varargs[untyped]): untyped =\n        var sep = newLit(\" \")\n        var hasSep = false\n        var values: seq[NimNode]\n        for arg in args:\n            if arg.kind == nnkExprEqExpr and arg[0].eqIdent(\"sep\"):\n                if hasSep:\n                    error(\"sep can only be specified once\", arg)\n                sep = arg[1]\n                hasSep = true\n            else:\n                values.add(arg)\n        var splatValues: NimNode\n        if values.len == 1:\n            if values[0].kind == nnkPrefix and values[0][0].eqIdent(\"*\"):\n                splatValues = values[0][1]\n            elif values[0].kind in nnkCallKinds and values[0].len == 3 and\n                    values[0][0].eqIdent(\"sheepJoinImpl\"):\n                # オーバーロード解決時に *values が先に展開された場合。\n                splatValues = values[0][1]\n        if not splatValues.isNil:\n            let joined = newCall(bindSym\"sheepJoinImpl\", splatValues, sep)\n            result = newCall(bindSym\"sheepPrintWithSeparator\", newLit(\" \"), joined)\n        else:\n            result = newCall(bindSym\"sheepPrintWithSeparator\", sep)\n            for value in values:\n                result.add(value)\n\n    proc dump[T](arr:seq[seq[T]])=\n        for i in 0..<len(arr):\n            echo arr[i]\n\n    proc sum(slice:HSlice[int,int]):int=\n        return (slice.a+slice.b)*len(slice)//2\n    \n    proc `<`[T](l,r:seq[T]):bool=\n        for i in 0..<min(len(l),len(r)):\n            if l[i] > r[i]:\n                return false\n            elif l[i] < r[i]:\n                return true\n        return len(l) < len(r)\n    \n    # Yes/No\n    proc yes*(b: bool = true): void = print(if b: \"Yes\" else: \"No\")\n\n    template dblock(body: untyped) =\n        when defined(debug):\n            block:\n                body\n"


var H,W = ii()
var A = newseqwith(H,lii(W))


var S = newseqwith(H,uint(0))
for i in range(H):
    for j in range(W):
        S[i] += A[i][j].uint()

var T = S.sum()
S.applyit(it + T)

print(S.join("\n"))