"""
Handling of tree structures given in a special 'dotted' syntax.
This represents trees of nodes with strings as tags,
in a readable and writable and easy to parse syntax.

There are two main functions, unparse_sexpr and parse_string.
When parsing, the result is by default given in 'sexpr' format:
each node is a tuple of the form

    (tag, ) or (tag, node) or (tag, node, node) ...

The following invariant is intended to hold for every node x,

    parse_string(unparse_sexpr(x)) == x

Currently the following invariant has been tested for some strings:

    unparse_sexpr(parse_string(s)).strip() == s.strip()

[It only holds on stripped results but may be fixed sometime.]

"""


class Node(object):
    __slots__ = 'tag', 'children', 'index',

    def __init__(self, tag, children, index):
        self.tag = tag
        self.children = children
        self.index = index

    def as_sexpr(self):
        return (self.tag,) + tuple([c.as_sexpr() for c in self.children])

    def __repr__(self):
        return '%s(%r, %r, %r)' % (
            self.__class__.__name__,
            self.tag,
            self.children,
            self.index)


class _GLUECLAMP_:

    _imports_ = (
        '_parent.FileIO:IO',
    )

    ##
    # The name of attributes that are configurable in instances.
    #
    _chgable_ = 'node', 'dotchar'

    ##
    # The character that begins the 'dotted' indentation.
    dotchar = '.'

    ##
    # The character that quotes lines beginning with dots and itself.
    quotechar = '\\'

    ##
    # Construct a new node.
    # @return
    # This variant returns a tuple in s-expression form.
    # @param tag a string
    # @param children a sequence of nodes
    # @param lineindex line index of tag, not used in s-expressions

    def node_sexpr(self, tag, children, lineindex):
        return (tag,) + tuple(children)

    ##
    # Construct a new node.
    # @return
    # This variant returns a Node object.
    # @param tag a string
    # @param children a sequence of nodes
    # @param lineindex line index of beginning tag, first line = 0

    def node_node(self, tag, children, lineindex):
        return Node(tag, tuple(children), lineindex)

    node = node_node

    def parse_file(self, file, src=None):
        return self.parse_string(self.IO.read_file(file), src)

    ##
    # Parse a dotted tree text given as a sequence of lines.
    # @param pos
    # @param tag     list with first line of tag, if any
    # @param lineindex line index of tag
    # @param it      iterator yielding remaining lines
    # @return a triple (index, nextvar, node) where
    # index is the index of line 'nextvar',
    # nextvar is the first line of next node to parse, and
    # node is the resulting node of this parse.

    def parse_iter(self, pos, tag, lineindex, it, src=None):
        dotchar = self.dotchar
        quotechar = self.quotechar
        children = []
        firstline = lineindex

        while 1:
            try:
                lineindex, nextvar = next(it)
            except StopIteration:
                nextvar = None
                break
            if not nextvar.startswith(dotchar):
                tag.append(nextvar)
            else:
                break
        for (i, t) in enumerate(tag):
            if (t.startswith(quotechar+dotchar) or
                    t.startswith(quotechar+quotechar+dotchar)):
                tag[i] = t[len(quotechar):]
        if tag == ['']:
            tag = '\n'
        else:
            tag = '\n'.join(tag)
        while 1:
            if (nextvar is None or len(nextvar) <= pos
                or nextvar[pos] != dotchar or
                    not nextvar.startswith(dotchar*(pos+1))):
                return lineindex, nextvar, self.node(tag, children, firstline)
            if len(nextvar) > pos+1 and nextvar[pos+1] == dotchar:
                if src is None:
                    raise SyntaxError('Level must increase with 1 max')
                else:
                    src.error('Level must increase with 1 max', lineindex)
            lineindex, nextvar, child = self.parse_iter(pos+1, [nextvar[pos+1:]],
                                                        lineindex, it, src)
            children.append(child)

    def parse_lines(self, lines, src=None):
        it = enumerate(lines)
        lineindex, nextvar, node = self.parse_iter(0, [], 0, it, src)
        assert nextvar is None
        return node

    def parse_string(self, string, src=None):
        if string:
            lines = string.split('\n')
        else:
            lines = []
        return self.parse_lines(lines, src)

    ##
    # Unparse a tree given on Node form

    def unparse_node(self, node):
        return self.unparse_sexpr(node.as_sexpr())

    ##
    # Unparse a tree given on sexpr form
    # @return a string in dotted-tree form
    def unparse_sexpr(self, sexpr):
        li = []

        def unparse(depth, sexpr):
            li.append(self.unparse_tag(depth, sexpr[0]))
            for x in sexpr[1:]:
                unparse(depth+1, x)

        unparse(0, sexpr)
        return '\n'.join(li)

    def unparse_tag(self, depth, tag):
        dotchar, quotechar = self.dotchar, self.quotechar
        tag = tag.split('\n')
        for (i, t) in enumerate(tag):
            if (t.startswith(dotchar) or
                    t.startswith(quotechar + dotchar)):
                tag[i] = quotechar + t
        tag = '\n'.join(tag)
        tag = dotchar*depth+tag
        return tag


def test_1():
    # Test parsing to sexpr's and back
    # for a variety of cases
    from guppy import Root
    dt = Root().guppy.gsl.DottedTree
    dt.node = dt.node_sexpr
    parse = dt.parse_string
    unparse = dt.unparse_sexpr

    for x, y in [
        ['', ('',)],
        ['a', ('a',)],
        ['.a', ('', ('a',))],
        ['a\n.b', ('a', ('b',))],
        ['a\nb\n.c', ('a\nb', ('c',))],
        ["""\n.a\n..a""", ('\n', ('a', ('a',)))],
        ["""hello\n.a\n.b\n..ba\nx\n..bb""",
            ('hello', ('a',), ('b', ('ba\nx',), ('bb',)))],
        # Quoting dots
        [r'\.', ('.',)],
        [r'.\.', ('', ('.',))],
        # Preserving quote
        ['\\', ('\\',)],
        ['.\n\\', ('', ('\n\\',))],
        # Quoting quote-dots
        [r'\\.', (r'\.',)],
        # Preserving whitespace starting a tag
        # Or should it be stripped? I think better not, it would complicate transparency.
        [r'. tag', ('', (' tag', ))],

        # Preserving initial whitespace
        [' ', (' ',)],
        # Preserving initial newline
        ['\n', ('\n',)],
        ['\na', ('\na',)],

        # A n intended usage example
        ['''
initial
text
.aspect for guppy.hsp
..returns
...type A
...latex
~\\
\..~|begincolorbox|~raw::~LaTeX~\\
~\\
~~~{\textbackslash}{\textbackslash}begin{\{}center{\}}~\\
.aspect for guppy.gsl
..contains DottedTree
''',

         ('\ninitial\ntext',
          ('aspect for guppy.hsp',
           ('returns',
            ('type A',),
            ('latex\n~\\\n..~|begincolorbox|~raw::~LaTeX~\\\n~\\\n~~~{\textbackslash}{\textbackslash}begin{\\{}center{\\}}~\\',))),
             ('aspect for guppy.gsl',
              ('contains DottedTree\n',)))]

    ]:
        z = parse(x)
        if y is not None:
            assert z == y
        assert unparse(z).strip() == x.strip()

    # Unparsing x and then parsing should give back x transparently
    # for any tree x, involving any combination of dots, quotes and other characters.

    # List of special chars and one normal
    chars = [dt.quotechar, dt.dotchar, '\n', ' ', 'a']

    import random

    ##
    # Generate a random node with random number of children.
    # Shuffles the chars list to make the tag string.
    # @param maxchild maximum number of children
    def randnode(maxchild):
        numchild = random.randint(0, maxchild)
        random.shuffle(chars)
        tag = ''.join(chars)
        children = [randnode(maxchild-1) for i in range(numchild)]
        return dt.node(tag, children, 0)

    for i in range(10):
        y = randnode(3)
        x = unparse(y)
        z = parse(x)
        assert z == y


def test_2():
    # Test parsing to Node
    # that the line lineindex are correct
    # They start from 0, since enumerate() generates them,
    # It seemed inconsistent to change them to start from 1.
    # Which will be made in error prints.

    from guppy import Root
    dt = Root().guppy.gsl.DottedTree
    parse = dt.parse_string
    unparse = dt.unparse_node

    node = parse("""\
line 0
.line 1
..line 2
line 3
.line 4
""")

    exp = Node('line 0', (
        Node('line 1',
             (Node('line 2\nline 3', (), 2),), 1),
        Node('line 4\n', (), 4)), 0)
    assert repr(node) == repr(exp)


def test_main():
    test_1()
    test_2()