Artifact Content

Artifact 428131a1792dffe049d41886353b1bea777b1fb3:

#!/usr/bin/env python
# -*- coding: utf-8 -*-

This file is part of the web2py Web Framework (Copyrighted, 2007-2011).
License: LGPLv3 (

Author: Thadeus Burgess


- Thank you to Massimo Di Pierro for creating the original gluon/
- Thank you to Jonathan Lundell for extensively testing the regex on Jython.
- Thank you to Limodou (creater of uliweb) who inspired the block-element support for web2py.

import os
import re
import cgi
import cStringIO
import logging
    from restricted import RestrictedError
    def RestrictedError(a,b,c):
        return RuntimeError

class Node(object):
    Basic Container Object
    def __init__(self, value = None, pre_extend = False):
        self.value = value
        self.pre_extend = pre_extend

    def __str__(self):
        return str(self.value)

class SuperNode(Node):
    def __init__(self, name = '', pre_extend = False): = name
        self.value = None
        self.pre_extend = pre_extend

    def __str__(self):
        if self.value:
            return str(self.value)
            raise SyntaxError("Undefined parent block ``%s``. \n" % + \
"You must define a block before referencing it.\nMake sure you have not left out an ``{{end}}`` tag." )

    def __repr__(self):
        return "%s->%s" % (, self.value)

class BlockNode(Node):
    Block Container.

    This Node can contain other Nodes and will render in a hierarchical order
    of when nodes were added.


        {{ block test }}
            This is default block test
        {{ end }}
    def __init__(self, name = '', pre_extend = False, delimiters = ('{{','}}')):
        name - Name of this Node.
        self.nodes = [] = name
        self.pre_extend = pre_extend
        self.left, self.right = delimiters

    def __repr__(self):
        lines = ['%sblock %s%s' % (self.left,,self.right)]
        for node in self.nodes:
        lines.append('%send%s' % (self.left, self.right))
        return ''.join(lines)

    def __str__(self):
        Get this BlockNodes content, not including child Nodes
        lines = []
        for node in self.nodes:
            if not isinstance(node, BlockNode):
        return ''.join(lines)

    def append(self, node):
        Add an element to the nodes.

        Keyword Arguments

        - node -- Node object or string to append.
        if isinstance(node, str) or isinstance(node, Node):
            raise TypeError("Invalid type; must be instance of ``str`` or ``BlockNode``. %s" % node)

    def extend(self, other):
        Extend the list of nodes with another BlockNode class.

        Keyword Arguments

        - other -- BlockNode or Content object to extend from.
        if isinstance(other, BlockNode):
            raise TypeError("Invalid type; must be instance of ``BlockNode``. %s" % other)

    def output(self, blocks):
        Merges all nodes into a single string.

        blocks -- Dictionary of blocks that are extending
        from this template.
        lines = []
        # Get each of our nodes
        for node in self.nodes:
            # If we have a block level node.
            if isinstance(node, BlockNode):
                # If we can override this block.
                if in blocks:
                    # Override block from vars.
                # Else we take the default
            # Else its just a string
        # Now combine all of our lines together.
        return ''.join(lines)

class Content(BlockNode):
    Parent Container -- Used as the root level BlockNode.

    Contains functions that operate as such.
    def __init__(self, name = "ContentBlock", pre_extend = False):
        Keyword Arguments

        name -- Unique name for this BlockNode
        """ = name
        self.nodes = []
        self.blocks = {}
        self.pre_extend = pre_extend

    def __str__(self):
        lines = []
        # For each of our nodes
        for node in self.nodes:
            # If it is a block node.
            if isinstance(node, BlockNode):
                # And the node has a name that corresponds with a block in us
                if in self.blocks:
                    # Use the overriding output.
                    # Otherwise we just use the nodes output.
                # It is just a string, so include it.
        # Merge our list together.
        return ''.join(lines)

    def _insert(self, other, index = 0):
        Inserts object at index.
        if isinstance(other, str) or isinstance(other, Node):
            self.nodes.insert(index, other)
            raise TypeError("Invalid type, must be instance of ``str`` or ``Node``.")

    def insert(self, other, index = 0):
        Inserts object at index.

        You may pass a list of objects and have them inserted.
        if isinstance(other, (list, tuple)):
            # Must reverse so the order stays the same.
            for item in other:
                self._insert(item, index)
            self._insert(other, index)

    def append(self, node):
        Adds a node to list. If it is a BlockNode then we assign a block for it.
        if isinstance(node, str) or isinstance(node, Node):
            if isinstance(node, BlockNode):
                self.blocks[] = node
            raise TypeError("Invalid type, must be instance of ``str`` or ``BlockNode``. %s" % node)

    def extend(self, other):
        Extends the objects list of nodes with another objects nodes
        if isinstance(other, BlockNode):
            raise TypeError("Invalid type; must be instance of ``BlockNode``. %s" % other)

    def clear_content(self):
        self.nodes = []

class TemplateParser(object):

    r_tag = re.compile(r'(\{\{.*?\}\})', re.DOTALL)

    r_multiline = re.compile(r'(""".*?""")|(\'\'\'.*?\'\'\')', re.DOTALL)

    # These are used for re-indentation.
    # Indent + 1
    re_block = re.compile('^(elif |else:|except:|except |finally:).*$',
    # Indent - 1
    re_unblock = re.compile('^(return|continue|break|raise)( .*)?$', re.DOTALL)
    # Indent - 1
    re_pass = re.compile('^pass( .*)?$', re.DOTALL)

    def __init__(self, text,
                 name    = "ParserContainer",
                 context = dict(),
                 path    = 'views/',
                 writer  = 'response.write',
                 lexers  = {},
                 delimiters = ('{{','}}'),
                 _super_nodes = [],
        text -- text to parse
        context -- context to parse in
        path -- folder path to templates
        writer -- string of writer class to use
        lexers -- dict of custom lexers to use.
        delimiters -- for example ('{{','}}')
        _super_nodes -- a list of nodes to check for inclusion
                        this should only be set by "self.extend"
                        It contains a list of SuperNodes from a child
                        template that need to be handled.

        # Keep a root level name. = name
        # Raw text to start parsing.
        self.text = text
        # Writer to use (refer to the default for an example).
        # This will end up as
        # "%s(%s, escape=False)" % (self.writer, value)
        self.writer = writer

        # Dictionary of custom name lexers to use.
        if isinstance(lexers, dict):
            self.lexers = lexers
            self.lexers = {}

        # Path of templates
        self.path = path
        # Context for templates.
        self.context = context

        # allow optional alternative delimiters
        self.delimiters = delimiters
        if delimiters!=('{{','}}'):
            escaped_delimiters = (re.escape(delimiters[0]),re.escape(delimiters[1]))
            self.r_tag = re.compile(r'(%s.*?%s)' % escaped_delimiters, re.DOTALL)

        # Create a root level Content that everything will go into.
        self.content = Content(name=name)

        # Stack will hold our current stack of nodes.
        # As we descend into a node, it will be added to the stack
        # And when we leave, it will be removed from the stack.
        # self.content should stay on the stack at all times.
        self.stack = [self.content]

        # This variable will hold a reference to every super block
        # that we come across in this template.
        self.super_nodes = []

        # This variable will hold a reference to the child
        # super nodes that need handling.
        self.child_super_nodes = _super_nodes

        # This variable will hold a reference to every block
        # that we come across in this template
        self.blocks = {}

        # Begin parsing.

    def to_string(self):
        Return the parsed template with correct indentation.

        Used to make it easier to port to python3.
        return self.reindent(str(self.content))

    def __str__(self):
        "Make sure str works exactly the same as python 3"
        return self.to_string()

    def __unicode__(self):
        "Make sure str works exactly the same as python 3"
        return self.to_string()

    def reindent(self, text):
        Reindents a string of unindented python code.

        # Get each of our lines into an array.
        lines = text.split('\n')

        # Our new lines
        new_lines = []

        # Keeps track of how many indents we have.
        # Used for when we need to drop a level of indentation
        # only to reindent on the next line.
        credit = 0

        # Current indentation
        k = 0

        # THINGS TO KNOW

        # k += 1 means indent
        # k -= 1 means unindent
        # credit = 1 means unindent on the next line.

        for raw_line in lines:
            line = raw_line.strip()

            # ignore empty lines
            if not line:

            # If we have a line that contains python code that
            # should be unindented for this line of code.
            # and then reindented for the next line.
            if TemplateParser.re_block.match(line):
                k = k + credit - 1

            # We obviously can't have a negative indentation
            k = max(k,0)

            # Add the indentation!
            new_lines.append(' '*(4*k)+line)

            # Bank account back to 0 again :(
            credit = 0

            # If we are a pass block, we obviously de-dent.
            if TemplateParser.re_pass.match(line):
                k -= 1

            # If we are any of the following, de-dent.
            # However, we should stay on the same level
            # But the line right after us will be de-dented.
            # So we add one credit to keep us at the level
            # while moving back one indentation level.
            if TemplateParser.re_unblock.match(line):
                credit = 1
                k -= 1

            # If we are an if statement, a try, or a semi-colon we
            # probably need to indent the next line.
            if line.endswith(':') and not line.startswith('#'):
                k += 1

        # This must come before so that we can raise an error with the
        # right content.
        new_text = '\n'.join(new_lines)

        if k > 0:
            self._raise_error('missing "pass" in view', new_text)
        elif k < 0:
            self._raise_error('too many "pass" in view', new_text)

        return new_text

    def _raise_error(self, message='', text=None):
        Raise an error using itself as the filename and textual content.
        raise RestrictedError(, text or self.text, message)

    def _get_file_text(self, filename):
        Attempt to open ``filename`` and retrieve its text.

        This will use self.path to search for the file.

        # If they didn't specify a filename, how can we find one!
        if not filename.strip():
            self._raise_error('Invalid template filename')

        # Get the filename; filename looks like ``"template.html"``.
        # We need to eval to remove the quotes and get the string type.
        filename = eval(filename, self.context)

        # Get the path of the file on the system.
        filepath = os.path.join(self.path, filename)

        # try to read the text.
            fileobj = open(filepath, 'rb')
            text =
        except IOError:
            self._raise_error('Unable to open included view file: ' + filepath)

        return text

    def include(self, content, filename):
        Include ``filename`` here.
        text = self._get_file_text(filename)

        t = TemplateParser(text,
                           name    = filename,
                           context = self.context,
                           path    = self.path,
                           writer  = self.writer,
                           delimiters = self.delimiters)


    def extend(self, filename):
        Extend ``filename``. Anything not declared in a block defined by the
        parent will be placed in the parent templates ``{{include}}`` block.
        text = self._get_file_text(filename)

        # Create out nodes list to send to the parent
        super_nodes = []
        # We want to include any non-handled nodes.
        # And our nodes as well.

        t = TemplateParser(text,
                    name         = filename,
                    context      = self.context,
                    path         = self.path,
                    writer       = self.writer,
                    delimiters   = self.delimiters,
                    _super_nodes = super_nodes)

        # Make a temporary buffer that is unique for parent
        # template.
        buf = BlockNode(name='__include__' + filename, delimiters=self.delimiters)
        pre = []

        # Iterate through each of our nodes
        for node in self.content.nodes:
            # If a node is a block
            if isinstance(node, BlockNode):
                # That happens to be in the parent template
                if in t.content.blocks:
                    # Do not include it

            if isinstance(node, Node):
                # Or if the node was before the extension
                # we should not include it
                if node.pre_extend:

            # Otherwise, it should go int the
            # Parent templates {{include}} section.

        # Clear our current nodes. We will be replacing this with
        # the parent nodes.
        self.content.nodes = []

        # Set our include, unique by filename
        t.content.blocks['__include__' + filename] = buf

        # Make sure our pre_extended nodes go first

        # Then we extend our blocks

        # Work off the parent node.
        self.content = t.content

    def parse(self, text):

        # Basically, r_tag.split will split the text into
        # an array containing, 'non-tag', 'tag', 'non-tag', 'tag'
        # so if we alternate this variable, we know
        # what to look for. This is alternate to
        # line.startswith("{{")
        in_tag = False
        extend = None
        pre_extend = True

        # Use a list to store everything in
        # This is because later the code will "look ahead"
        # for missing strings or brackets.
        ij = self.r_tag.split(text)
        # j = current index
        # i = current item
        for j in range(len(ij)):
            i = ij[j]

            if i:
                if len(self.stack) == 0:
                    self._raise_error('The "end" tag is unmatched, please check if you have a starting "block" tag')

                # Our current element in the stack.
                top = self.stack[-1]

                if in_tag:
                    line = i

                    # If we are missing any strings!!!!
                    # This usually happens with the following example
                    # template code
                    # {{a = '}}'}}
                    # or
                    # {{a = '}}blahblah{{'}}
                    # This will fix these
                    # This is commented out because the current template
                    # system has this same limitation. Since this has a
                    # performance hit on larger templates, I do not recommend
                    # using this code on production systems. This is still here
                    # for "i told you it *can* be fixed" purposes.
#                    if line.count("'") % 2 != 0 or line.count('"') % 2 != 0:
#                        # Look ahead
#                        la = 1
#                        nextline = ij[j+la]
#                        # As long as we have not found our ending
#                        # brackets keep going
#                        while '}}' not in nextline:
#                            la += 1
#                            nextline += ij[j+la]
#                            # clear this line, so we
#                            # don't attempt to parse it
#                            # this is why there is an "if i"
#                            # around line 530
#                            ij[j+la] = ''
#                        # retrieve our index.
#                        index = nextline.index('}}')
#                        # Everything before the new brackets
#                        before = nextline[:index+2]
#                        # Everything after
#                        after = nextline[index+2:]
#                        # Make the next line everything after
#                        # so it parses correctly, this *should* be
#                        # all html
#                        ij[j+1] = after
#                        # Add everything before to the current line
#                        line += before

                    # Get rid of '{{' and '}}'
                    line = line[2:-2].strip()

                    # This is bad juju, but let's do it anyway
                    if not line:

                    # We do not want to replace the newlines in code,
                    # only in block comments.
                    def remove_newline(re_val):
                        # Take the entire match and replace newlines with
                        # escaped newlines.
                        return'\n', '\\n')

                    # Perform block comment escaping.
                    # This performs escaping ON anything
                    # in between """ and """
                    line = re.sub(TemplateParser.r_multiline,

                    if line.startswith('='):
                        # IE: {{=response.title}}
                        name, value = '=', line[1:].strip()
                        v = line.split(' ', 1)
                        if len(v) == 1:
                            # Example
                            # {{ include }}
                            # {{ end }}
                            name = v[0]
                            value = ''
                            # Example
                            # {{ block pie }}
                            # {{ include "layout.html" }}
                            # {{ for i in range(10): }}
                            name = v[0]
                            value = v[1]

                    # This will replace newlines in block comments
                    # with the newline character. This is so that they
                    # retain their formatting, but squish down to one
                    # line in the rendered template.

                    # First check if we have any custom lexers
                    if name in self.lexers:
                        # Pass the information to the lexer
                        # and allow it to inject in the environment

                        # You can define custom names such as
                        # '{{<<variable}}' which could potentially
                        # write unescaped version of the variable.
                        self.lexers[name](parser    = self,
                                          value     = value,
                                          top       = top,
                                          stack     = self.stack,)

                    elif name == '=':
                        # So we have a variable to insert into
                        # the template
                        buf = "\n%s(%s)" % (self.writer, value)
                        top.append(Node(buf, pre_extend = pre_extend))

                    elif name == 'block' and not value.startswith('='):
                        # Make a new node with name.
                        node = BlockNode(name = value.strip(),
                                         pre_extend = pre_extend,
                                         delimiters = self.delimiters)

                        # Append this node to our active node

                        # Make sure to add the node to the stack.
                        # so anything after this gets added
                        # to this node. This allows us to
                        # "nest" nodes.

                    elif name == 'end' and not value.startswith('='):
                        # We are done with this node.

                        # Save an instance of it
                        self.blocks[] = top

                        # Pop it.

                    elif name == 'super' and not value.startswith('='):
                        # Get our correct target name
                        # If they just called {{super}} without a name
                        # attempt to assume the top blocks name.
                        if value:
                            target_node = value
                            target_node =

                        # Create a SuperNode instance
                        node = SuperNode(name = target_node,
                                            pre_extend = pre_extend)

                        # Add this to our list to be taken care of

                        # And put in in the tree

                    elif name == 'include' and not value.startswith('='):
                        # If we know the target file to include
                        if value:
                            self.include(top, value)

                        # Otherwise, make a temporary include node
                        # That the child node will know to hook into.
                            include_node = BlockNode(name = '__include__' +,
                                                     pre_extend = pre_extend,
                                                     delimiters = self.delimiters)

                    elif name == 'extend' and not value.startswith('='):
                        # We need to extend the following
                        # template.
                        extend = value
                        pre_extend = False

                        # If we don't know where it belongs
                        # we just add it anyways without formatting.
                        if line and in_tag:

                            # Split on the newlines >.<
                            tokens = line.split('\n')

                            # We need to look for any instances of
                            # for i in range(10):
                            #   = i
                            # pass
                            # So we can properly put a response.write() in place.
                            continuation = False
                            len_parsed = 0
                            for k in range(len(tokens)):

                                tokens[k] = tokens[k].strip()
                                len_parsed += len(tokens[k])

                                if tokens[k].startswith('='):
                                    if tokens[k].endswith('\\'):
                                        continuation = True
                                        tokens[k] = "\n%s(%s" % (self.writer, tokens[k][1:].strip())
                                        tokens[k] = "\n%s(%s)" % (self.writer, tokens[k][1:].strip())
                                elif continuation:
                                    tokens[k] += ')'
                                    continuation = False

                            buf = "\n%s" % '\n'.join(tokens)
                            top.append(Node(buf, pre_extend = pre_extend))

                    # It is HTML so just include it.
                    buf = "\n%s(%r, escape=False)" % (self.writer, i)
                    top.append(Node(buf, pre_extend = pre_extend))

            # Remember: tag, not tag, tag, not tag
            in_tag = not in_tag

        # Make a list of items to remove from child
        to_rm = []

        # Go through each of the children nodes
        for node in self.child_super_nodes:
            # If we declared a block that this node wants to include
            if in self.blocks:
                # Go ahead and include it!
                node.value = self.blocks[]
                # Since we processed this child, we don't need to
                # pass it along to the parent

        # Remove some of the processed nodes
        for node in to_rm:
            # Since this is a pointer, it works beautifully.
            # Sometimes I miss C-Style pointers... I want my asterisk...

        # If we need to extend a template.
        if extend:

# We need this for integration with gluon
def parse_template(filename,
                   path    = 'views/',
                   context = dict(),
                   lexers  = {},
                   delimiters = ('{{','}}')
    filename can be a view filename in the views folder or an input stream
    path is the path of a views folder
    context is a dictionary of symbols used to render the template

    # First, if we have a str try to open the file
    if isinstance(filename, str):
            fp = open(os.path.join(path, filename), 'rb')
            text =
        except IOError:
            raise RestrictedError(filename, '', 'Unable to find the file')
        text =

    # Use the file contents to get a parsed template and return it.
    return str(TemplateParser(text, context=context, path=path, lexers=lexers, delimiters=delimiters))

def get_parsed(text):
    Returns the indented python code of text. Useful for unit testing.

    return str(TemplateParser(text))

# And this is a generic render function.
# Here for integration with gluon.
def render(content = "hello world",
           stream = None,
           filename = None,
           path = None,
           context = {},
           lexers  = {},
           delimiters = ('{{','}}')
    >>> render()
    'hello world'
    >>> render(content='abc')
    >>> render(content='abc\\'')
    >>> render(content='a"\\'bc')
    >>> render(content='a\\nbc')
    >>> render(content='a"bcd"e')
    >>> render(content="'''a\\nc'''")
    >>> render(content="'''a\\'c'''")
    >>> render(content='{{for i in range(a):}}{{=i}}<br />{{pass}}', context=dict(a=5))
    '0<br />1<br />2<br />3<br />4<br />'
    >>> render(content='{%for i in range(a):%}{%=i%}<br />{%pass%}', context=dict(a=5),delimiters=('{%','%}'))
    '0<br />1<br />2<br />3<br />4<br />'
    >>> render(content="{{='''hello\\nworld'''}}")
    >>> render(content='{{for i in range(3):\\n=i\\npass}}')
    # Here to avoid circular Imports
        from globals import Response
        # Working standalone. Build a mock Response object.
        class Response():
            def __init__(self):
                self.body = cStringIO.StringIO()
            def write(self, data, escape=True):
                if not escape:
                elif hasattr(data,'xml') and callable(data.xml):
                    # make it a string
                    if not isinstance(data, (str, unicode)):
                        data = str(data)
                    elif isinstance(data, unicode):
                        data = data.encode('utf8', 'xmlcharrefreplace')
                    data = cgi.escape(data, True).replace("'","&#x27;")

        # A little helper to avoid escaping.
        class NOESCAPE():
            def __init__(self, text):
                self.text = text
            def xml(self):
                return self.text
        # Add it to the context so we can use it.
        context['NOESCAPE'] = NOESCAPE

    # If we don't have anything to render, why bother?
    if not content and not stream and not filename:
        raise SyntaxError, "Must specify a stream or filename or content"

    # Here for legacy purposes, probably can be reduced to something more simple.
    close_stream = False
    if not stream:
        if filename:
            stream = open(filename, 'rb')
            close_stream = True
        elif content:
            stream = cStringIO.StringIO(content)

    # Get a response class.
    context['response'] = Response()

    # Execute the template.
    code = str(TemplateParser(, context=context, path=path, lexers=lexers, delimiters=delimiters))
        exec(code) in context
    except Exception:
        # for i,line in enumerate(code.split('\n')): print i,line

    if close_stream:

    # Returned the rendered content.
    return context['response'].body.getvalue()

if __name__ == '__main__':
    import doctest