排版依存语法树

Question

我尝试排版依赖树：

有没有可以做到这一点的软件包？它应该与兼容xelatex，即不基于pstricks。我在谷歌或 CTAN 上找不到任何东西。如果没有包，我会尝试用来实现这一点forest，但这似乎并不简单。如何才能直接做到这一点？

编辑：我找到了（一半）解决方案forest：

\documentclass{article}

\usepackage{forest}



\forestset{
dg edges/.style={for tree={parent anchor=south, child anchor=north,align=center,base=bottom,where n children=0{tier=word,edge=dotted}{}}},
}


\begin{document}

\begin{forest}
dg edges
[V
  [N
    [D [the] ]
     [child] ]
  [reads]
  [N
    [D [a] ]
    [book] ] ]
\end{forest}


\end{document}

这样虚线就正确了，单词也与基线对齐了。但是，节点的对齐方式不正确。如果我可以让 N 恰好位于书籍上方，V 位于阅读上方，并在样式定义中执行此操作，这将是首选解决方案。有没有办法在 中执行此操作forest？

Answer 1

这是使用 的一次尝试tikz tree。为了将文本对齐到底部，dfont定义了一种样式。

代码

\documentclass{article}
\usepackage{tikz}
\usetikzlibrary{shapes,trees,calc}
\begin{document}
\tikzset{ 
treenode/.style = {inner sep=0pt, outer sep=2pt, font=\sffamily},
edge from parent/.style={draw, edge from parent path=
    {(\tikzparentnode.south) -- (\tikzchildnode.north)}},
dfont/.style={dashed,font=\vphantom{Ag},anchor=north}  % to align the text at the bottom
}

\begin{tikzpicture}
 [
% Children and edges style
    level distance=1cm,
    level 1/.style={sibling distance=2cm},
    level 2/.style={sibling distance=2cm},
    level 3/.style={sibling distance=3.5cm},
    level 4/.style={sibling distance=6cm},
    level 5/.style={sibling distance=3cm}
    ]
%% Draw events and edges
 \node (g) [treenode] {are}
            child {node[treenode] (a) {We}}    % left
            child {node[treenode] (b) {trying} % right  
                     child[missing]
            child {node[treenode] (c) {to}
                     child[missing]
            child {node[treenode](d) {understand}
                     child[missing]
            child {node[treenode] (e) {difference}    
                   child {node[treenode](f){the}}
                   child[missing]}
}
}
};  
\node (a1) at ($(a)+(0,-6)$){};
\draw[dfont] (a) -- (a|-a1)node[]{We};
\draw[dfont] (g) -- (g|-a1)node[]{are};
\draw[dfont] (b) -- (b|-a1)node[]{trying};
\draw[dfont] (c) -- (c|-a1)node[]{to};
\draw[dfont] (d) -- (d|-a1)node[]{understand};
\draw[dfont] (f) -- (f|-a1)node[]{the};
\draw[dfont] (e) -- (e|-a1)node[]{difference.};
\end{tikzpicture}
\end{document}

Answer 2

有点笨拙，但展示了如何指定自定义增长函数以及在底部复制单词的巧妙方法。我从 Jesse 的回答中借鉴了几种风格：

\documentclass[tikz, border=5]{standalone}
% An (incomplete) growth function
\makeatletter
\def\tikz@grow@tree{%
  \tikzset{shift=(270:\tikzleveldistance)}%
  \ifcase\tikznumberofchildren%
  \or     
  \or
     \tikzset{shift=(0:{(\tikznumberofcurrentchild-1)*\tikzsiblingdistance})}
  \else%
     \tikzset{shift=(0:{(\tikznumberofcurrentchild-1-\tikznumberofchildren+2)*\tikzsiblingdistance})}
  \fi
}

\tikzset{%
  dependency tree/.code={\let\tikz@grow=\tikz@grow@tree},
  leaf level/.store in=\tikzleaflevel,
  tree node/.style={inner sep=0pt, outer sep=2pt, font=\sffamily},
     edge from parent/.style={draw, edge from parent path={(\tikzparentnode.south) -- (\tikzchildnode.north)}},
  leaf/.style={level distance=(\tikzleaflevel-\the\tikztreelevel)*1cm,
    edge from parent/.append style={dashed}, execute at begin node=\strut}
}
\def\wordnode#1{ node [tree node] {#1} child [leaf] { node {#1} } }

\begin{document}
\begin{tikzpicture}[level distance=1cm,dependency tree, leaf level=8]

 \node [tree node] 
   {are}
   child { \wordnode{We} }    
   child [leaf] { node {are} }
   child { \wordnode {trying} 
     child { \wordnode {to} 
       child  { \wordnode{understand} 
         child [sibling distance=3cm] { \wordnode{difference}           
           child [xscale=-1, sibling distance=1.5cm] { \wordnode{the}
    }}}}};  

\end{tikzpicture}
\end{document}

Answer 3

这个calign with current edge选项起了作用：

\documentclass{article}

\usepackage{forest}


\forestset{
dg edges/.style={for tree={parent anchor=south, child anchor=north,align=center,base=bottom,where n children=0{tier=word,edge=dotted,calign with current edge}{}}},
}


\begin{document}

\begin{forest}
dg edges
[V
  [N
    [D [the] ]
     [child] ]
  [reads]
  [N
    [D [a] ]
    [book] ] ]
\end{forest}


\end{document}