Files
MyFastHtml/src/myfasthtml/core/matcher.py
2025-10-25 22:34:00 +02:00

346 lines
12 KiB
Python

import re
from dataclasses import dataclass
from fastcore.basics import NotStr
from myfasthtml.core.testclient import MyFT
class Predicate:
def __init__(self, value):
self.value = value
def validate(self, actual):
raise NotImplementedError
def __str__(self):
return f"{self.__class__.__name__}({self.value})"
def __eq__(self, other):
if not isinstance(other, Predicate):
return False
return self.value == other.value
def __hash__(self):
return hash(self.value)
class StartsWith(Predicate):
def __init__(self, value):
super().__init__(value)
def validate(self, actual):
return actual.startswith(self.value)
class Contains(Predicate):
def __init__(self, value):
super().__init__(value)
def validate(self, actual):
return self.value in actual
class DoesNotContain(Predicate):
def __init__(self, value):
super().__init__(value)
def validate(self, actual):
return self.value not in actual
@dataclass
class DoNotCheck:
desc: str = None
@dataclass
class Empty:
desc: str = None
class ErrorOutput:
def __init__(self, path, element, expected):
self.path = path
self.element = element
self.expected = expected
self.output = []
self.indent = ""
@staticmethod
def _unconstruct_path_item(item):
if "#" in item:
elt_name, elt_id = item.split("#")
return elt_name, "id", elt_id
elif match := re.match(r'(\w+)\[(class|name)=([^]]+)]', item):
return match.groups()
return item, None, None
def __str__(self):
self.compute()
def compute(self):
# first render the path hierarchy
for p in self.path.split(".")[:-1]:
elt_name, attr_name, attr_value = self._unconstruct_path_item(p)
path_str = self._str_element(MyFT(elt_name, {attr_name: attr_value}), keep_open=True)
self._add_to_output(f"{path_str}")
self.indent += " "
# then render the element
if hasattr(self.expected, "tag") and hasattr(self.element, "tag"):
# display the tag and its attributes
tag_str = self._str_element(self.element, self.expected)
self._add_to_output(tag_str)
# Try to show where the differences are
error_str = self._detect_error(self.element, self.expected)
if error_str:
self._add_to_output(error_str)
# render the children
if len(self.expected.children) > 0:
self.indent += " "
element_index = 0
for expected_child in self.expected.children:
if hasattr(expected_child, "tag"):
if element_index < len(self.element.children):
# display the child
element_child = self.element.children[element_index]
child_str = self._str_element(element_child, expected_child, keep_open=False)
self._add_to_output(child_str)
# manage errors in children
child_error_str = self._detect_error(element_child, expected_child)
if child_error_str:
self._add_to_output(child_error_str)
element_index += 1
else:
# When there are fewer children than expected, we display a placeholder
child_str = "! ** MISSING ** !"
self._add_to_output(child_str)
else:
self._add_to_output(expected_child)
self.indent = self.indent[:-2]
self._add_to_output(")")
else:
self._add_to_output(str(self.element))
# Try to show where the differences are
error_str = self._detect_error(self.element, self.expected)
if error_str:
self._add_to_output(error_str)
def _add_to_output(self, msg):
self.output.append(f"{self.indent}{msg}")
@staticmethod
def _str_element(element, expected=None, keep_open=None):
# compare to itself if no expected element is provided
if expected is None:
expected = element
# the attributes are compared to the expected element
elt_attrs = {attr_name: element.attrs.get(attr_name, "** MISSING **") for attr_name in
[attr_name for attr_name in expected.attrs if attr_name is not None]}
elt_attrs_str = " ".join(f'"{attr_name}"="{attr_value}"' for attr_name, attr_value in elt_attrs.items())
#
tag_str = f"({element.tag} {elt_attrs_str}"
# manage the closing tag
if keep_open is False:
tag_str += " ...)" if len(element.children) > 0 else ")"
elif keep_open is True:
tag_str += "..." if elt_attrs_str == "" else " ..."
else:
# close the tag if there are no children
if len(element.children) == 0: tag_str += ")"
return tag_str
def _detect_error(self, element, expected):
if hasattr(expected, "tag") and hasattr(element, "tag"):
tag_str = len(element.tag) * (" " if element.tag == expected.tag else "^")
elt_attrs = {attr_name: element.attrs.get(attr_name, "** MISSING **") for attr_name in expected.attrs}
attrs_in_error = [attr_name for attr_name, attr_value in elt_attrs.items() if
not self._matches(attr_value, expected.attrs[attr_name])]
if attrs_in_error:
elt_attrs_error = " ".join(len(f'"{name}"="{value}"') * ("^" if name in attrs_in_error else " ")
for name, value in elt_attrs.items())
error_str = f" {tag_str} {elt_attrs_error}"
return error_str
else:
if not self._matches(element, expected):
return len(str(element)) * "^"
return None
@staticmethod
def _matches(element, expected):
if element == expected:
return True
elif isinstance(expected, Predicate):
return expected.validate(element)
else:
return element == expected
class ErrorComparisonOutput:
def __init__(self, actual_error_output, expected_error_output):
self.actual_error_output = actual_error_output
self.expected_error_output = expected_error_output
@staticmethod
def adjust(to_adjust, reference):
for index, ref_line in enumerate(reference):
if "^^" in ref_line:
# insert an empty line in to_adjust
to_adjust.insert(index, "")
return to_adjust
def render(self):
# init if needed
if not self.actual_error_output.output:
self.actual_error_output.compute()
if not self.expected_error_output.output:
self.expected_error_output.compute()
actual = self.actual_error_output.output
expected = self.expected_error_output.output
# actual = self.adjust(actual, expected) # does not seem to be needed
expected = self.adjust(expected, actual)
actual_max_length = len(max(actual, key=len))
# expected_max_length = len(max(expected, key=len))
output = []
for a, e in zip(actual, expected):
line = f"{a:<{actual_max_length}} | {e}".rstrip()
output.append(line)
return "\n".join(output)
def matches(actual, expected, path=""):
def print_path(p):
return f"Path : '{p}'\n" if p else ""
def _type(x):
return type(x)
def _debug(elt):
return str(elt) if elt else "None"
def _debug_compare(a, b):
actual_out = ErrorOutput(path, a, b)
expected_out = ErrorOutput(path, b, b)
comparison_out = ErrorComparisonOutput(actual_out, expected_out)
return comparison_out.render()
def _error_msg(msg, _actual=None, _expected=None):
if _actual is None and _expected is None:
debug_info = ""
elif _actual is None:
debug_info = _debug(_expected)
elif _expected is None:
debug_info = _debug(_actual)
else:
debug_info = _debug_compare(_actual, _expected)
return f"{print_path(path)}Error : {msg}\n{debug_info}"
def _assert_error(msg, _actual=None, _expected=None):
assert False, _error_msg(msg, _actual=_actual, _expected=_expected)
def _get_current_path(elt):
if hasattr(elt, "tag"):
res = f"{elt.tag}"
if "id" in elt.attrs:
res += f"#{elt.attrs['id']}"
elif "name" in elt.attrs:
res += f"[name={elt.attrs['name']}]"
elif "class" in elt.attrs:
res += f"[class={elt.attrs['class']}]"
return res
else:
return elt.__class__.__name__
if actual is not None and expected is None:
_assert_error("Actual is not None while expected is None", _actual=actual)
if isinstance(expected, DoNotCheck):
return True
if actual is None and expected is not None:
_assert_error("Actual is None while expected is ", _expected=expected)
# set the path
path += "." + _get_current_path(actual) if path else _get_current_path(actual)
assert _type(actual) == _type(expected) or (hasattr(actual, "tag") and hasattr(expected, "tag")), \
_error_msg("The types are different: ", _actual=actual, _expected=expected)
if isinstance(expected, (list, tuple)):
if len(actual) < len(expected):
_assert_error("Actual is smaller than expected: ", _actual=actual, _expected=expected)
if len(actual) > len(expected):
_assert_error("Actual is bigger than expected: ", _actual=actual, _expected=expected)
for actual_child, expected_child in zip(actual, expected):
assert matches(actual_child, expected_child, path=path)
elif isinstance(expected, NotStr):
to_compare = actual.s.lstrip('\n').lstrip()
assert to_compare.startswith(expected.s), _error_msg("Notstr values are different: ",
_actual=to_compare,
_expected=expected.s)
elif hasattr(expected, "tag"):
# validate the tags names
assert actual.tag == expected.tag, _error_msg("The elements are different.",
_actual=actual.tag,
_expected=expected.tag)
# special case when the expected element is empty
if len(expected.children) > 0 and isinstance(expected.children[0], Empty):
assert len(actual.children) == 0, _error_msg("Actual is not empty:", _actual=actual)
assert len(actual.attrs) == 0, _error_msg("Actual is not empty:", _actual=actual)
return True
# compare the attributes
for expected_attr, expected_value in expected.attrs.items():
assert expected_attr in actual.attrs, _error_msg(f"'{expected_attr}' is not found in Actual.",
_actual=actual.attrs)
if isinstance(expected_value, Predicate):
assert expected_value.validate(actual.attrs[expected_attr]), \
_error_msg(f"The condition '{expected_value}' is not satisfied.",
_actual=actual,
_expected=expected)
else:
assert actual.attrs[expected_attr] == expected.attrs[expected_attr], \
_error_msg(f"The values are different for '{expected_attr}': ",
_actual=actual.attrs[expected_attr],
_expected=expected.attrs[expected_attr])
# compare the children
if len(actual.children) < len(expected.children):
_assert_error("Actual is lesser than expected: ", _actual=actual, _expected=expected)
for actual_child, expected_child in zip(actual.children, expected.children):
assert matches(actual_child, expected_child, path=path)
else:
assert actual == expected, _error_msg("The values are different: ",
_actual=actual,
_expected=expected)
return True