编写一个程序，将二叉树作为输入，并输出最深的节点及其深度。如果存在平局，请打印所有涉及的节点及其深度。每个节点表示为：

T(x,x)

T(x)

T

其中，T是一个或多个字母数字字符的标识符，每个x是另一个节点。

这是二叉树的简单定义：

• 二叉树的头是一个节点。
• 二叉树中的一个节点最多有两个子节点。

例如，输入A(B(C,D(E)))（下面）将输出3:E。

下面的树是5、11和4之间的三向关系，其深度也是3（从0开始）：

输入2(7(2,6(5,11)),5(9(4)))（下面）将输出3:5,11,4。

这是代码高尔夫，因此以字节为单位的最短代码获胜。

果酱（CJam）49 47

0q')/{'(/):U;,+:TW>{T:W];}*TW={U]',*}*T(}/;W':@

0                 " Push 0 ";
q                 " Read the whole input ";
')/               " Split the input by ')' ";
{                 " For each item ";
  '(/             " Split by '(' ";
  )               " Extract the last item of the array ";
  :U;             " Assign the result to U, and discard it ";
  ,               " Get the array length ";
  +               " Add the top two items of the stack, which are the array length and the number initialized to 0 ";
  :T              " Assign the result to T ";
  W>{             " If T>W, while W is always initialized to -1 ";
    T:W];         " Set T to W, and empty the stack ";
  }*
  TW={            " If T==W ";
    U]',*         " Push U and add a ',' between everything in the stack, if there were more than one ";
  }*
  T(              " Push T and decrease by one ";
}/
;                 " Discard the top item, which should be now -1 ";
W                 " Push W ";
':                " Push ':' ";
@                 " Rotate the 3rd item to the top ";

Haskell，186个字节

p@(n,s)%(c:z)=maybe((n,s++[c])%z)(\i->p:(n+i,"")%z)$lookup c$zip"),("[-1..1];p%_=[p]
(n,w)&(i,s)|i>n=(i,show i++':':s)|i==n=(n,w++',':s);p&_=p
main=interact$snd.foldl(&)(0,"").((0,"")%)

完整的程序，在上生成树，在上stdin生成指定的输出格式stdout：

& echo '2(7(2,6(5,11)),5(9(4)))' | runhaskell 32557-Deepest.hs 
3:5,11,4

& echo 'A(B(C,D(E)))' | runhaskell 32557-Deepest.hs 
3:E

高尔夫球代码的指南（添加了更好的名称，类型签名，注释以及一些子表达式，并拔出并命名了-否则使用相同的代码；非高尔夫球版本将不会混淆使用编号打断节点，也不会找到最深的节点带有输出格式。）：

type Label = String         -- the label on a node
type Node = (Int, Label)    -- the depth of a node, and its label

-- | Break a string into nodes, counting the depth as we go
number :: Node -> String -> [Node]
number node@(n, label) (c:cs) =
    maybe addCharToNode startNewNode $ lookup c adjustTable
  where
    addCharToNode = number (n, label ++ [c]) cs
        -- ^ append current character onto label, and keep numbering rest

    startNewNode adjust = node : number (n + adjust, "") cs
        -- ^ return current node, and the number the rest, adjusting the depth

    adjustTable = zip "),(" [-1..1]
        -- ^ map characters that end node labels onto depth adjustments
        -- Equivalent to [ (')',-1), (',',0), ('(',1) ]

number node _ = [node]      -- default case when there is no more input

-- | Accumulate into the set of deepest nodes, building the formatted output
deepest :: (Int, String) -> Node -> (Int, String)
deepest (m, output) (n, label)
    | n > m     = (n, show n ++ ':' : label)    -- n is deeper tham what we have
    | n == m    = (m, output ++ ',' : label)    -- n is as deep, so add on label
deepest best _ = best                           -- otherwise, not as deep

main' :: IO ()
main' = interact $ getOutput . findDeepest . numberNodes
  where
    numberNodes :: String -> [Node]
    numberNodes = number (0, "")

    findDeepest :: [Node] -> (Int, String)
    findDeepest = foldl deepest (0, "")

    getOutput :: (Int, String) -> String
    getOutput = snd

GolfScript（75个字符）

竞争不是特别激烈，但是已经足够引起人们的兴趣了：

{.48<{"'^"\39}*}%','-)](+0.{;.@.@>-\}:^;@:Z~{2$2$={@@.}*;}:^;Z~\-])':'@','*

该代码分为三个阶段。首先，我们预处理输入字符串：

# In regex terms, this is s/([ -\/])/'^\1'/g
{.48<{"'^"\39}*}%
# Remove all commas
','-
# Rotate the ' which was added after the closing ) to the start
)](+

我们已经将转换A(B(C,D(E)))为'A'^('B'^('C'^'D'^('E'^)''^)''^)。如果我们给它分配一个合适的块，^可以通过~评估字符串来进行有用的处理。

其次，我们找到最大深度：

0.
# The block we assign to ^ assumes that the stack is
#   max-depth current-depth string
# It discards the string and updates max-depth
{;.@.@>-\}:^;
@:Z~

最后，我们选择最深的节点并构建输出：

# The block we assign to ^ assumes that the stack is
#   max-depth current-depth string
# If max-depth == current-depth it pushes the string under them on the stack
# Otherwise it discards the string
{2$2$={@@.}*;}:^;
# Eval
Z~
# The stack now contains
#   value1 ... valuen max-depth 0
# Get a positive value for the depth, collect everything into an array, and pop the depth
\-])
# Final rearranging for the desired output
':'@','*

Perl 5-85

^{请随时编辑此帖子以更正字符数。我使用sayfeature，但是我不知道这些标志如何使其在没有声明的情况下正确运行use 5.010;。}

$_=$t=<>,$i=0;$t=$_,$i++while s/\w+(\((?R)(,(?R))?\))?/$1/g,/\w/;@x=$t=~/\w+/gs;say"$i:@x"

ideone上的演示

输出以空格分隔，而不是逗号分隔。

该代码仅使用递归正则表达式删除林中树的根，直到无法删除为止。然后，最后一个字符串之前的字符串将包含最深层的所有叶节点。

样品运行

2
0:2

2(3(4(5)),6(7))
3:5

2(7(2,6(5,11)),5(9(4)))
3:5 11 4

1(2(3(4,5),6(7,8)),9(10(11,12),13(14,15)))
3:4 5 7 8 11 12 14 15

VB.net

Function FindDeepest(t$) As String
  Dim f As New List(Of String)
  Dim m = 0
  Dim d = 0
  Dim x = ""
  For Each c In t
    Select Case c
      Case ","
        If d = m Then f.Add(x)
        x = ""
      Case "("
        d += 1
        If d > m Then f.Clear() :
        m = d
        x = ""
      Case ")"
        If d = m Then f.Add(x) : x = ""
        d -= 1
      Case Else
        x += c
    End Select
  Next
  Return m & ":" & String.Join(",", f)
End Function

假设：节点值不能包含,，(，)

Javascript（E6）120

迭代版

m=d=0,n=[''];
prompt().split(/,|(\(|\))/).map(e=>e&&(e=='('?m<++d&&(n[m=d]=''):e==')'?d--:n[d]+=' '+e));
alert(m+':'+n[m])

松散且可测试

F= a=> (
    m=d=0,n=[''],
    a.split(/,|(\(|\))/)
    .map(e=>e && (e=='(' ? m < ++d && (n[m=d]='') : e==')' ? d-- : n[d]+=' '+e)),
    m+':'+n[m]
)

在Firefox控制台中进行测试：

['A', '2(7(2,6(5,11)),5(9(4)))', 'A(B(C,D(E)))']
.map(x => x + ' --> ' + F(x)).join('\n')

输出量

“ A-> 0：A

2（7（2,6（5,11）），5（9（4）））-> 3：5 11 4

A（B（C，D（E）））-> 3：E“