path: root/doc/user/dia_tree

@Section
    @Tag { dia_tree }
    @Title { Trees }
@Begin
@PP
@@Diag offers some symbols for producing tree diagrams, using the
diagrams. @RawIndex { diagrams }
diagrams.tree @SubIndex { @Code "@Tree" symbol }
tree.diagrams @Index { @Code "@Tree" symbol (diagrams) }
@Code "@Tree" symbol, which may appear anywhere within the nodes part:
@ID @OneRow @Code @Verbatim {
@Diag {
    ...
    @Tree { ... }
    ...
}
}
Within this symbol, new symbols {@Code "@LeftSub"}, {@Code "@RightSub"},
diagrams. @RawIndex { diagrams }
diagrams.leftsub @SubIndex { @Code "@LeftSub" symbol }
leftsub.diagrams @Index { @Code "@LeftSub" symbol (diagrams) }
diagrams. @RawIndex { diagrams }
diagrams.rightsub @SubIndex { @Code "@RightSub" symbol }
rightsub.diagrams @Index { @Code "@RightSub" symbol (diagrams) }
{@Code "@FirstSub"}, {@Code "@NextSub"}, and {@Code "@StubSub"} become
available.  The first two are used to get a (non-empty) binary tree:
@ID @OneRow {
@Code @Verbatim {
@Tree {
    @Circle A
    @LeftSub {
        @Circle B
        @LeftSub @Square C
        @RightSub @Square D
    }
    @RightSub @Circle E
}
}
||7ct
@Diag {
@Tree {
    @Circle A
    @LeftSub {
        @Circle B
        @LeftSub @Square C
        @RightSub @Square D
    }
    @RightSub @Circle E
}
}
}
The root of the tree, which must be a single node but may have any
outline, comes first.  After that comes the @Code "@LeftSub" symbol
followed by the left subtree, which must be enclosed in braces unless
it consists of a single node.  After that comes the @Code "@RightSub"
symbol followed by the right subtree, again enclosed in braces unless it
consists of a single node.  These rules apply recursively and will
produce a binary tree of arbitrary size and depth.  If a node has no
left or right subtree, leave out the corresponding @Code "@LeftSub" or
@Code "@RightSub" symbol.
@PP
A similar system using @Code "@FirstSub" and @Code "@NextSub" produces
diagrams. @RawIndex { diagrams }
diagrams.firstsub @SubIndex { @Code "@FirstSub" symbol }
firstsub.diagrams @Index { @Code "@FirstSub" symbol (diagrams) }
diagrams. @RawIndex { diagrams }
diagrams.nextsub @SubIndex { @Code "@NextSub" symbol }
nextsub.diagrams @Index { @Code "@NextSub" symbol (diagrams) }
trees in which each node may have arbitrarily many children:
@ID @OneRow {
@Code @Verbatim {
@Tree {
    @Circle A
    @FirstSub {
        @Circle B
        @FirstSub @Square C
        @NextSub @Square D
    }
    @NextSub @Circle E
    @NextSub @Circle F
}
}
||7ct
@Diag {
@Tree {
    @Circle A
    @FirstSub {
        @Circle B
        @FirstSub @Square C
        @NextSub @Square D
    }
    @NextSub @Circle E
    @NextSub @Circle F
}
}
}
The first subtree is preceded by {@Code "@FirstSub"}, and subsequent
trees are preceded by {@Code "@NextSub"}.  The subtrees are spaced
at equal separations from each other, with the root centred over
them, in contrast to the binary tree arrangement in which the two
subtrees are positioned to the left and right of the root, never
intruding into the space beneath it.
@PP
Although each subtree must contain a node for its root, it is not hard
to get around this:
@ID @OneRow {
@Code @Verbatim {
@Tree
{
@Circle
@FirstSub @Circle
@NextSub pathstyle { noline }
    @Circle outlinestyle { noline }
       ...
@NextSub @Circle
}
}
||7ct
@Diag {
@Tree
{
@Circle
@FirstSub @Circle
@NextSub pathstyle { noline }
    @Circle outlinestyle { noline }
       ...
@NextSub @Circle
}
}
}
Clumsy as this is, it often assists in placing the unenclosed object
in a way consistent with the surrounding nodes, and offers margins
and so forth which help with fine-tuning its position.
@PP
The fifth subtree symbol, {@Code "@StubSub"}, produces a stub subtree:
diagrams. @RawIndex { diagrams }
diagrams.stubsub @SubIndex { @Code "@StubSub" symbol }
stubsub.diagrams @Index { @Code "@StubSub" symbol (diagrams) }
@ID @OneRow {
@Code @Verbatim {
@Tree {
@Circle @Eq { a }
@StubSub @Eq { T tsub a }
}
}
||7ct
@Diag {
@Tree {
@Circle @Eq { a }
@StubSub @Eq { T tsub a }
}
}
}
Unlike the other subtree symbols, {@Code "@StubSub"} is not followed
by a subtree with a node for its root; rather, it is followed by an
arbitrary object, and the path is drawn around this stub object, which
is placed directly underneath the parent node with zero vertical
separation.  In practice, it is usually necessary to attach margins to
the following object; the easiest way to do that is to enclose it in
{@Code "@Box outlinestyle { noline }"}.  An example appears below.
@PP
It is possible to mix the three subtree types, by having binary tree
symbols following some nodes, non-binary tree symbols following
others, and a single {@Code "@StubSub"} following others.  However,
at any one node the subtrees must be all either binary, non-binary,
or stub.
@PP
The subtree symbols have all of the options of {@Code "@Link"}, and
these apply to the link drawn from the parent of the root of the subtree
to the root of the subtree (or anticlockwise around the stub object):
@ID @OneRow {
@Code @Verbatim {
@Tree {
    @Circle A
    @LeftSub
        arrow { yes }
        xlabel { 1 }
    @Circle B
    @RightSub
        arrow { yes }
        xlabel { 2 }
    @Circle C
}
}
||7ct
@Diag {
@Tree {
    @Circle A
    @LeftSub
        arrow { yes }
        xlabel { 1 }
    @Circle B
    @RightSub
        arrow { yes }
        xlabel { 2 }
    @Circle C
}
}
}
To get reverse arrows use @Code "arrow { back }" as usual.
@PP
The subtree symbols do not need @Code from and @Code to options,
because they already know which nodes they are linking together.  However,
you may use @Code from or @Code to to give a tag specifying a particular
diagrams. @RawIndex { diagrams }
diagrams.from @SubIndex { @Code "from" option }
from.diagrams @Index { @Code "from" option (diagrams) }
diagrams. @RawIndex { diagrams }
diagrams.to @SubIndex { @Code "to" option }
to.diagrams @Index { @Code "to" option (diagrams) }
point within the node:
@ID @OneRow {
@Code @Verbatim {
@Tree {
@Circle
@LeftSub from { S } to { N }
    @Isosceles vsize { 2f }
@RightSub from { S } to { N }
    @Isosceles vsize { 2f }
}
}
||7ct
@Diag
{
@Tree {
@Circle
@LeftSub from { S } to { N }
    @Isosceles vsize { 2f }
@RightSub from { S } to { N }
    @Isosceles vsize { 2f }
}
}
}
In this example both links go from the @Code S tag of the parent node to the
@Code N tag of the child node (at the apex of the iscosceles triangle).  These
options also work for {@Code "@StubSub"}, where they refer to the start and
end of the stub path:
@ID @OneRow {
@Code @Verbatim {
@Tree {
@Circle @Eq { a }
@StubSub
    from { SW }
    to { SE }
@Box outlinestyle { noline }
    @Eq { T tsub a }
}
}
||7ct
@Diag {
@Tree {
@Circle @Eq { a }
@StubSub
    from { SW }
    to { SE }
@Box outlinestyle { noline }
    @Eq { T tsub a }
}
}
}
and so the tags both refer to points in the parent node in this case.
@PP
The @Code "@LeftSub" and @Code "@RightSub" symbols have variants called
@Code "@ZeroWidthLeftSub" and @Code "@ZeroWidthRightSub" which are the
diagrams. @RawIndex { diagrams }
diagrams.zerowidthleftsub @SubIndex { @Code "@ZeroWidthLeftSub" symbol }
zerowidthleftsub.diagrams @Index { @Code "@ZeroWidthLeftSub" (diagrams) }
diagrams. @RawIndex { diagrams }
diagrams.zerowidthrightsub @SubIndex { @Code "@ZeroWidthRightSub" symbol }
zerowidthrightsub.diagrams @Index { @Code "@ZeroWidthRightSub" (diagrams) }
same except that the resulting subtrees consume no width:
@ID @OneRow {
@Code @Verbatim {
@Tree {
@Circle
@LeftSub {
    @Circle
    @LeftSub @Square
    @RightSub @Square
}
@RightSub {
    @Circle
    @LeftSub {
        @Circle
        @ZeroWidthLeftSub @Square
        @ZeroWidthRightSub @Square
    }
    @RightSub @Square
} }
}
||7ct
@Diag {
@Tree
{
@Circle
@LeftSub {
    @Circle
    @LeftSub @Square
    @RightSub @Square
}
@RightSub {
    @Circle
    @LeftSub {
        @Circle
        @ZeroWidthLeftSub @Square
        @ZeroWidthRightSub @Square
    }
    @RightSub @Square
}
}
}
}
There is nothing analogous for the other subtree symbols.
@PP
The @Code "@Diag" symbol has a few options for adjusting the appearance
of the tree.  The @Code "treehsep" option determines the horizontal space left
diagrams. @RawIndex { diagrams }
diagrams.treehsep @SubIndex { @Code "treehsep" option }
treehsep.diagrams @Index { @Code "treehsep" option (diagrams) }
between a root and its left subtree, between a root and its right subtree,
and between one subtree and the next when @Code "@NextSub" is used.  The
@Code "treevsep" option determines the vertical space left between a root
diagrams. @RawIndex { diagrams }
diagrams.treevsep @SubIndex { @Code "treevsep" option }
treevsep.diagrams @Index { @Code "treevsep" option (diagrams) }
and its subtrees:
@ID @OneRow {
@Code @Verbatim {
@Diag
    treehsep { 0c }
    treevsep { 0c }
{
@Tree
{
    @Circle A
    @LeftSub @Square B
    @RightSub @Square C
}
}
}
||7ct
@Diag
    treehsep { 0c }
    treevsep { 0c }
{
@Tree
{
    @Circle A
    @LeftSub @Square B
    @RightSub @Square C
}
}
}
These options may also be given to individual subtree symbols, although
@Code "treevsep" works as expected only with @Code "@LeftSub" and
{@Code "@FirstSub"}, since these determine the vertical separation of
all children of their parent.
@PP
The @Code "treehindent" option determines where the root of a non-binary
diagrams. @RawIndex { diagrams }
diagrams.treehindent @SubIndex { @Code "treehindent" option }
treehindent.diagrams @Index { @Code "treehindent" option (diagrams) }
tree is positioned over its subtrees; the value may be @Code "left"
for at left, @Code "ctr" for centred over them (the default),
@Code "right" for at the right, or any length, meaning that far from
the left.  Owing to problems behind the scenes, this option may not be
given to individual subtree symbols; so as a consolation, it is permitted
as an option to the @Code "@Tree" symbol.
@PP
It is not possible to attach tags to nodes within a tree, because
tags are attached automatically by the tree symbols and any extra
tags would disrupt the linking.  However, you can use @Code "@ShowTags"
to find out what these automatic tags are, and use them in a subsequent
links part.  For example, the tag attached to the right child of the left
child of the root of a binary tree is {@Code "L@R@T"}, and in general the
tag records the path from the root to the node, with @Code "T" added to
the end.  The root always has tag {@Code "T"}.  The tree as a whole may
be retagged in the usual way.
@PP
There is an @Code "@HTree" symbol which is the same as
diagrams. @RawIndex { diagrams }
diagrams.htree @SubIndex { @Code "@HTree" symbol }
htree.diagrams @Index { @Code "@HTree" symbol (diagrams) }
@Code "@Tree" except that the tree grows horizontally (from left to
right) instead of vertically.  The same symbols are available within
@Code "@HTree" as within {@Code "@Tree"}; @Code "@LeftSub" and
@Code "@FirstSub" produce what might be called the top subtree, and
@Code "@RightSub" and @Code "@NextSub" produce lower trees.  @Code "@HTree"
has no @Code "treehindent" option; instead, it has an exactly analogous
@Code "treevindent" option.
diagrams. @RawIndex { diagrams }
diagrams.treevindent @SubIndex { @Code "treevindent" option }
treevindent.diagrams @Index { @Code "treevindent" option (diagrams) }
@PP
@Code "@HTree" may be used to get horizontal lists:
@ID @OneRow {
@Code @Verbatim {
@I @Diag
    arrow { yes } treehsep { 1c } {
@HTree {
    @Node A
    @FirstSub {
        @Node B
        @FirstSub @Node C
    }
}
}
}
||7ct
@I @Diag arrow { yes } treehsep { 1c } {
@HTree {
@Node A
@FirstSub {
    @Node B
    @FirstSub @Node C 
}
}
}
}
The braces are clumsy but necessary.  The first node has tag {@Code "T"}, the
second has tag {@Code "S@T"}, the third has tag {@Code "S@S@T"}, and so on.
@End @Section