Quick search

Table Of Contents

Source code for kivy.lang.parser


Class used for the parsing of .kv files into rules.
import os

import re
import sys
import traceback
import ast
import importlib
from re import sub, findall
from types import CodeType
from functools import partial
from collections import OrderedDict, defaultdict

import kivy.lang.builder  # imported as absolute to avoid circular import
from kivy.logger import Logger
from kivy.cache import Cache
from kivy import require
from kivy.resources import resource_find
from kivy.utils import rgba
import kivy.metrics as Metrics

__all__ = ('Parser', 'ParserException')

trace = Logger.trace
global_idmap = {}

# register cache for creating new classtype (template)

# all previously included files
__KV_INCLUDES__ = []

# precompile regexp expression
str_re = (

lang_str = re.compile(f"({str_re})", re.DOTALL)
lang_fstr = re.compile(f"([fF](?:{str_re}))", re.DOTALL)

lang_key = re.compile('([a-zA-Z_]+)')
lang_keyvalue = re.compile(r'([a-zA-Z_][a-zA-Z0-9_.]*\.[a-zA-Z0-9_.]+)')
lang_tr = re.compile(r'(_\()')
lang_cls_split_pat = re.compile(', *')

# all the widget handlers, used to correctly unbind all the callbacks then the
# widget is deleted
_handlers = defaultdict(partial(defaultdict, list))

class ProxyApp(object):
    # proxy app object
    # taken from

    __slots__ = ['_obj']

    def __init__(self):
        object.__setattr__(self, '_obj', None)

    def _ensure_app(self):
        app = object.__getattribute__(self, '_obj')
        if app is None:
            from import App
            app = App.get_running_app()
            object.__setattr__(self, '_obj', app)
            # Clear cached application instance, when it stops
            app.bind(on_stop=lambda instance:
                     object.__setattr__(self, '_obj', None))
        return app

    def __getattribute__(self, name):
        object.__getattribute__(self, '_ensure_app')()
        return getattr(object.__getattribute__(self, '_obj'), name)

    def __delattr__(self, name):
        object.__getattribute__(self, '_ensure_app')()
        delattr(object.__getattribute__(self, '_obj'), name)

    def __setattr__(self, name, value):
        object.__getattribute__(self, '_ensure_app')()
        setattr(object.__getattribute__(self, '_obj'), name, value)

    def __bool__(self):
        object.__getattribute__(self, '_ensure_app')()
        return bool(object.__getattribute__(self, '_obj'))

    def __str__(self):
        object.__getattribute__(self, '_ensure_app')()
        return str(object.__getattribute__(self, '_obj'))

    def __repr__(self):
        object.__getattribute__(self, '_ensure_app')()
        return repr(object.__getattribute__(self, '_obj'))

global_idmap['app'] = ProxyApp()
global_idmap['pt'] =
global_idmap['inch'] = Metrics.inch
global_idmap['cm'] =
global_idmap['mm'] =
global_idmap['dp'] = Metrics.dp
global_idmap['sp'] = Metrics.sp
global_idmap['rgba'] = rgba

[docs]class ParserException(Exception): '''Exception raised when something wrong happened in a kv file. ''' def __init__(self, context, line, message, cause=None): self.filename = context.filename or '<inline>' self.line = line sourcecode = context.sourcecode sc_start = max(0, line - 2) sc_stop = min(len(sourcecode), line + 3) sc = ['...'] for x in range(sc_start, sc_stop): if x == line: sc += ['>> %4d:%s' % (line + 1, sourcecode[line][1])] else: sc += [' %4d:%s' % (x + 1, sourcecode[x][1])] sc += ['...'] sc = '\n'.join(sc) message = 'Parser: File "%s", line %d:\n%s\n%s' % ( self.filename, self.line + 1, sc, message) if cause: message += '\n' + ''.join(traceback.format_tb(cause)) super(ParserException, self).__init__(message)
class ParserRuleProperty(object): '''Represent a property inside a rule. ''' __slots__ = ('ctx', 'line', 'name', 'value', 'co_value', 'watched_keys', 'mode', 'count', 'ignore_prev') def __init__(self, ctx, line, name, value, ignore_prev=False): super(ParserRuleProperty, self).__init__() #: Associated parser self.ctx = ctx #: Line of the rule self.line = line #: Name of the property = name #: Value of the property self.value = value #: Compiled value self.co_value = None #: Compilation mode self.mode = None #: Watched keys self.watched_keys = None #: Stats self.count = 0 #: whether previous rules targeting name should be cleared self.ignore_prev = ignore_prev def precompile(self): name = value = self.value # first, remove all the string from the value tmp = sub(lang_str, '', self.value) # detecting how to handle the value according to the key name mode = self.mode if self.mode is None: self.mode = mode = 'exec' if name[:3] == 'on_' else 'eval' if mode == 'eval': # if we don't detect any string/key in it, we can eval and give the # result if, tmp) is None: value = '\n' * self.line + value self.co_value = eval( compile(value, self.ctx.filename or '<string>', 'eval') ) return # ok, we can compile. value = '\n' * self.line + value self.co_value = compile(value, self.ctx.filename or '<string>', mode) # for exec mode, we don't need to watch any keys. if mode == 'exec': return # now, detect obj.prop # find all the fstrings in the value fstrings = lang_fstr.findall(value) expressions = [ast.parse(s) for s in fstrings] wk = set() for s in fstrings: expression = ast.parse(s) wk |= set(self.get_names_from_expression(expression.body[0].value)) # first, remove all the string from the value tmp = sub(lang_str, '', value) idx = tmp.find('#') if idx != -1: tmp = tmp[:idx] # detect key.value inside value, and split them wk |= set(findall(lang_keyvalue, tmp)) if wk: self.watched_keys = [x.split('.') for x in wk] if findall(lang_tr, tmp): if self.watched_keys: self.watched_keys += [['_']] else: self.watched_keys = [['_']] @classmethod def get_names_from_expression(cls, node): """ Look for all the symbols used in an ast node. """ if isinstance(node, ast.Name): yield if isinstance(node, (ast.JoinedStr, ast.BoolOp)): for n in node.values: if isinstance(n, ast.Str): # NOTE: required for python3.6 yield from cls.get_names_from_expression(n.s) else: yield from cls.get_names_from_expression(n.value) if isinstance(node, ast.BinOp): yield from cls.get_names_from_expression(node.right) yield from cls.get_names_from_expression(node.left) if isinstance(node, ast.IfExp): yield from cls.get_names_from_expression(node.test) yield from cls.get_names_from_expression(node.body) yield from cls.get_names_from_expression(node.orelse) if isinstance(node, ast.Subscript): yield from cls.get_names_from_expression(node.value) yield from cls.get_names_from_expression(node.slice) if isinstance(node, ast.Slice): yield from cls.get_names_from_expression(node.lower) yield from cls.get_names_from_expression(node.upper) yield from cls.get_names_from_expression(node.step) if isinstance( node, (ast.ListComp, ast.DictComp, ast.SetComp, ast.GeneratorExp) ): for g in node.generators: yield from cls.get_names_from_expression(g.iter) if isinstance(node, (ast.List, ast.Tuple, ast.Set)): for elt in node.elts: yield from cls.get_names_from_expression(elt) if isinstance(node, ast.Dict): for val in node.values: yield from cls.get_names_from_expression(val) if isinstance(node, ast.UnaryOp): yield from cls.get_names_from_expression(node.operand) if isinstance(node, ast.comprehension): yield from cls.get_names_from_expression(node.iter.value) if isinstance(node, ast.Attribute): if isinstance(node.value, ast.Name): yield f'{}.{node.attr}' if isinstance(node, ast.Call): yield from cls.get_names_from_expression(node.func) for arg in node.args: yield from cls.get_names_from_expression(arg) for keyword in node.keywords: yield from cls.get_names_from_expression(keyword.value) def __repr__(self): return '<ParserRuleProperty name=%r filename=%s:%d ' \ 'value=%r watched_keys=%r>' % (, self.ctx.filename, self.line + 1, self.value, self.watched_keys) class ParserRule(object): '''Represents a rule, in terms of the Kivy internal language. ''' __slots__ = ('ctx', 'line', 'name', 'children', 'id', 'properties', 'canvas_before', 'canvas_root', 'canvas_after', 'handlers', 'level', 'cache_marked', 'avoid_previous_rules') def __init__(self, ctx, line, name, level): super(ParserRule, self).__init__() #: Level of the rule in the kv self.level = level #: Associated parser self.ctx = ctx #: Line of the rule self.line = line #: Name of the rule = name #: List of children to create self.children = [] #: Id given to the rule = None #: Properties associated to the rule = OrderedDict() #: Canvas normal self.canvas_root = None #: Canvas before self.canvas_before = None #: Canvas after self.canvas_after = None #: Handlers associated to the rule self.handlers = [] #: Properties cache list: mark which class have already been checked self.cache_marked = [] #: Indicate if any previous rules should be avoided. self.avoid_previous_rules = False if level == 0: self._detect_selectors() else: self._forbid_selectors() def precompile(self): for x in x.precompile() for x in self.handlers: x.precompile() for x in self.children: x.precompile() if self.canvas_before: self.canvas_before.precompile() if self.canvas_root: self.canvas_root.precompile() if self.canvas_after: self.canvas_after.precompile() def create_missing(self, widget): # check first if the widget class already been processed by this rule cls = widget.__class__ if cls in self.cache_marked: return self.cache_marked.append(cls) for name in if hasattr(widget, name): continue value =[name].co_value if type(value) is CodeType: value = None widget.create_property(name, value, default_value=False) def _forbid_selectors(self): c =[0] if c == '<' or c == '[': raise ParserException( self.ctx, self.line, 'Selectors rules are allowed only at the first level') def _detect_selectors(self): c =[0] if c == '<': self._build_rule() elif c == '[': self._build_template() else: if self.ctx.root is not None: raise ParserException( self.ctx, self.line, 'Only one root object is allowed by .kv') self.ctx.root = self def _build_rule(self): name = if __debug__: trace('Builder: build rule for %s' % name) if name[0] != '<' or name[-1] != '>': raise ParserException(self.ctx, self.line, 'Invalid rule (must be inside <>)') # if the very first name start with a -, avoid previous rules name = name[1:-1] if name[:1] == '-': self.avoid_previous_rules = True name = name[1:] for rule in re.split(lang_cls_split_pat, name): crule = None if not len(rule): raise ParserException(self.ctx, self.line, 'Empty rule detected') if '@' in rule: # new class creation ? # ensure the name is correctly written rule, baseclasses = rule.split('@', 1) if not re.match(lang_key, rule): raise ParserException(self.ctx, self.line, 'Invalid dynamic class name') # save the name in the dynamic classes dict. self.ctx.dynamic_classes[rule] = baseclasses crule = ParserSelectorName(rule) else: # classical selectors. if rule[0] == '.': crule = ParserSelectorClass(rule[1:]) else: crule = ParserSelectorName(rule) self.ctx.rules.append((crule, self)) def _build_template(self): name = exception = ParserException( self.ctx, self.line, 'Deprecated Kivy lang template syntax used "{}". Templates will ' 'be removed in a future version'.format(name)) if name not in ('[FileListEntry@FloatLayout+TreeViewNode]', '[FileIconEntry@Widget]', '[AccordionItemTitle@Label]'): Logger.warning(exception) if __debug__: trace('Builder: build template for %s' % name) if name[0] != '[' or name[-1] != ']': raise ParserException(self.ctx, self.line, 'Invalid template (must be inside [])') item_content = name[1:-1] if '@' not in item_content: raise ParserException(self.ctx, self.line, 'Invalid template name (missing @)') template_name, template_root_cls = item_content.split('@') self.ctx.templates.append((template_name, template_root_cls, self)) def __repr__(self): return '<ParserRule name=%r>' % (, )
[docs]class Parser(object): '''Create a Parser object to parse a Kivy language file or Kivy content. ''' PROP_ALLOWED = ('canvas.before', 'canvas.after') CLASS_RANGE = list(range(ord('A'), ord('Z') + 1)) PROP_RANGE = ( list(range(ord('A'), ord('Z') + 1)) + list(range(ord('a'), ord('z') + 1)) + list(range(ord('0'), ord('9') + 1)) + [ord('_')]) __slots__ = ('rules', 'templates', 'root', 'sourcecode', 'directives', 'filename', 'dynamic_classes') def __init__(self, **kwargs): super(Parser, self).__init__() self.rules = [] self.templates = [] self.root = None self.sourcecode = [] self.directives = [] self.dynamic_classes = {} self.filename = kwargs.get('filename', None) content = kwargs.get('content', None) if content is None: raise ValueError('No content passed') self.parse(content) def execute_directives(self): global __KV_INCLUDES__ for ln, cmd in self.directives: cmd = cmd.strip() if __debug__: trace('Parser: got directive <%s>' % cmd) if cmd[:5] == 'kivy ': version = cmd[5:].strip() if len(version.split('.')) == 2: version += '.0' require(version) elif cmd[:4] == 'set ': try: name, value = cmd[4:].strip().split(' ', 1) except: Logger.exception('') raise ParserException(self, ln, 'Invalid directive syntax') try: value = eval(value, global_idmap) except: Logger.exception('') raise ParserException(self, ln, 'Invalid value') global_idmap[name] = value elif cmd[:8] == 'include ': ref = cmd[8:].strip() force_load = False if ref[:6] == 'force ': ref = ref[6:].strip() force_load = True # if #:include [force] "path with quotes around" if ref[0] == ref[-1] and ref[0] in ('"', "'"): c = ref[:3].count(ref[0]) ref = ref[c:-c] if c != 2 else ref if ref[-3:] != '.kv': Logger.warn('Lang: {0} does not have a valid Kivy' 'Language extension (.kv)'.format(ref)) break if ref in __KV_INCLUDES__: if not os.path.isfile(resource_find(ref) or ref): raise ParserException(self, ln, 'Invalid or unknown file: {0}' .format(ref)) if not force_load: Logger.warn('Lang: {0} has already been included!' .format(ref)) continue else: Logger.debug('Lang: Reloading {0} ' 'because include was forced.' .format(ref)) kivy.lang.builder.Builder.unload_file(ref) kivy.lang.builder.Builder.load_file(ref) continue Logger.debug('Lang: Including file: {0}'.format(0)) __KV_INCLUDES__.append(ref) kivy.lang.builder.Builder.load_file(ref) elif cmd[:7] == 'import ': package = cmd[7:].strip() z = package.split() if len(z) != 2: raise ParserException(self, ln, 'Invalid import syntax') alias, package = z try: if package not in sys.modules: try: mod = importlib.__import__(package) except ImportError: module_name = '.'.join(package.split('.')[:-1]) mod = importlib.__import__(module_name) # resolve the whole thing for part in package.split('.')[1:]: mod = getattr(mod, part) else: mod = sys.modules[package] global_idmap[alias] = mod except ImportError: Logger.exception('') raise ParserException(self, ln, 'Unable to import package %r' % package) else: raise ParserException(self, ln, 'Unknown directive')
[docs] def parse(self, content): '''Parse the contents of a Parser file and return a list of root objects. ''' # Read and parse the lines of the file lines = content.splitlines() if not lines: return num_lines = len(lines) lines = list(zip(list(range(num_lines)), lines)) self.sourcecode = lines[:] if __debug__: trace('Parser: parsing %d lines' % num_lines) # Strip all comments self.strip_comments(lines) # Execute directives self.execute_directives() # Get object from the first level objects, remaining_lines = self.parse_level(0, lines) # Precompile rules tree for rule in objects: rule.precompile() # After parsing, there should be no remaining lines # or there's an error we did not catch earlier. if remaining_lines: ln, content = remaining_lines[0] raise ParserException(self, ln, 'Invalid data (not parsed)')
[docs] def strip_comments(self, lines): '''Remove all comments from all lines in-place. Comments need to be on a single line and not at the end of a line. i.e. a comment line's first non-whitespace character must be a #. ''' # extract directives for ln, line in lines[:]: stripped = line.strip() if stripped[:2] == '#:': self.directives.append((ln, stripped[2:])) if stripped[:1] == '#': lines.remove((ln, line)) if not stripped: lines.remove((ln, line))
[docs] def parse_level(self, level, lines, spaces=0): '''Parse the current level (level * spaces) indentation. ''' indent = spaces * level if spaces > 0 else 0 objects = [] current_object = None current_property = None current_propobject = None i = 0 while i < len(lines): line = lines[i] ln, content = line # Get the number of space tmp = content.lstrip(' \t') # Replace any tab with 4 spaces tmp = content[:len(content) - len(tmp)] tmp = tmp.replace('\t', ' ') # first indent designates the indentation if spaces == 0: spaces = len(tmp) count = len(tmp) if spaces > 0 and count % spaces != 0: raise ParserException(self, ln, 'Invalid indentation, ' 'must be a multiple of ' '%s spaces' % spaces) content = content.strip() rlevel = count // spaces if spaces > 0 else 0 # Level finished if count < indent: return objects, lines[i - 1:] # Current level, create an object elif count == indent: x = content.split(':', 1) if not len(x[0]): raise ParserException(self, ln, 'Identifier missing') if (len(x) == 2 and len(x[1]) and not x[1].lstrip().startswith('#')): raise ParserException(self, ln, 'Invalid data after declaration') name = x[0].rstrip() # if it's not a root rule, then we got some restriction # aka, a valid name, without point or everything else if count != 0: if False in [ord(z) in Parser.PROP_RANGE for z in name]: raise ParserException(self, ln, 'Invalid class name') current_object = ParserRule(self, ln, name, rlevel) current_property = None objects.append(current_object) # Next level, is it a property or an object ? elif count == indent + spaces: x = content.split(':', 1) if not len(x[0]): raise ParserException(self, ln, 'Identifier missing') # It's a class, add to the current object as a children current_property = None name = x[0].rstrip() ignore_prev = name[0] == '-' if ignore_prev: name = name[1:] if ord(name[0]) in Parser.CLASS_RANGE: if ignore_prev: raise ParserException( self, ln, 'clear previous, `-`, not allowed here') _objects, _lines = self.parse_level( level + 1, lines[i:], spaces) current_object.children = _objects lines = _lines i = 0 # It's a property else: if name not in Parser.PROP_ALLOWED: if not all(ord(z) in Parser.PROP_RANGE for z in name): raise ParserException(self, ln, 'Invalid property name') if len(x) == 1: raise ParserException(self, ln, 'Syntax error') value = x[1].strip() if name == 'id': if len(value) <= 0: raise ParserException(self, ln, 'Empty id') if value in ('self', 'root'): raise ParserException( self, ln, 'Invalid id, cannot be "self" or "root"') = value elif len(value): rule = ParserRuleProperty( self, ln, name, value, ignore_prev) if name[:3] == 'on_': current_object.handlers.append(rule) else: ignore_prev = False[name] = rule else: current_property = name current_propobject = None if ignore_prev: # it wasn't consumed raise ParserException( self, ln, 'clear previous, `-`, not allowed here') # Two more levels? elif count == indent + 2 * spaces: if current_property in ( 'canvas', 'canvas.after', 'canvas.before'): _objects, _lines = self.parse_level( level + 2, lines[i:], spaces) rl = ParserRule(self, ln, current_property, rlevel) rl.children = _objects if current_property == 'canvas': current_object.canvas_root = rl elif current_property == 'canvas.before': current_object.canvas_before = rl else: current_object.canvas_after = rl current_property = None lines = _lines i = 0 else: if current_propobject is None: current_propobject = ParserRuleProperty( self, ln, current_property, content) if current_property[:3] == 'on_': current_object.handlers.append(current_propobject) else:[current_property] = \ current_propobject else: current_propobject.value += '\n' + content # Too much indentation, invalid else: raise ParserException(self, ln, 'Invalid indentation (too many levels)') # Check the next line i += 1 return objects, []
class ParserSelector(object): def __init__(self, key): self.key = key.lower() def match(self, widget): raise NotImplementedError def __repr__(self): return '<%s key=%s>' % (self.__class__.__name__, self.key) class ParserSelectorClass(ParserSelector): def match(self, widget): return self.key in widget.cls class ParserSelectorName(ParserSelector): parents = {} def get_bases(self, cls): for base in cls.__bases__: if base.__name__ == 'object': break yield base if base.__name__ == 'Widget': break for cbase in self.get_bases(base): yield cbase def match(self, widget): parents = ParserSelectorName.parents cls = widget.__class__ if cls not in parents: classes = [x.__name__.lower() for x in [cls] + list(self.get_bases(cls))] parents[cls] = classes return self.key in parents[cls] def match_rule_name(self, rule_name): return self.key == rule_name.lower()