def get_char(self, cp_or_sequence):
"""
Get the char object of a code point.
:param cp_or_sequence: Code point or sequence of the character
to get.
:raises NotInLGR: If the code point does not exist.
>>> cd = Repertoire()
>>> char = cd.add_char([0x002A])
>>> c = cd.get_char([0x002A])
>>> c is char
True
"""
assert len(cp_or_sequence), "there should be at least one char"
origin = CharBase.from_cp_or_sequence(cp_or_sequence)
idx = origin.as_index()
if idx not in self._chardict:
raise NotInLGR(cp_or_sequence)
chars = self._chardict[idx]
try:
list_idx = chars.index(origin)
except ValueError:
logger.error("Code point '%s' does not exist",
format_cp(cp_or_sequence))
raise NotInLGR(cp_or_sequence)
char = chars[list_idx]
return char
def del_char(self, cp_or_sequence):
"""
Delete a character from the LGR.
:param cp_or_sequence: code point or code point sequence to delete.
:raises NotInLGR: If the code point does not exist.
>>> cd = Repertoire()
>>> _ = cd.add_char([0x002A])
>>> 0x002A in cd
True
>>> cd.del_char([0x002A])
>>> 0x002A in cd
False
>>> cd.del_char([0x002B]) # doctest: +IGNORE_EXCEPTION_DETAIL
Traceback (most recent call last):
...
NotInLGR:
"""
assert len(cp_or_sequence), "there should be at least one char"
char = CharBase.from_cp_or_sequence(cp_or_sequence)
if not self._del_char(char):
logger.error("Code point '%s' does not exist",
format_cp(cp_or_sequence))
raise NotInLGR(cp_or_sequence)
def del_range(self, first_cp, last_cp):
"""
Delete a range of characters from the LGR.
Note: This MUST be the exact same range that was added,
meaning you cannot delete partial sub-ranges!
:param first_cp: First code point of the range.
:param last_cp: Last code point of the range.
:raises NotInLGR: If the range does not exist.
>>> cd = Repertoire()
>>> cd.add_range(0x002A, 0x0030)
>>> cd.del_range(0x002A, 0x0030)
>>> 0x002A in cd
False
>>> cd.del_range(0x002A, 0x0030) # doctest: +IGNORE_EXCEPTION_DETAIL
Traceback (most recent call last):
...
NotInLGR:
"""
assert first_cp < last_cp, "range must be defined in order"
if (first_cp, last_cp) not in self.ranges:
logger.error("Range '%s - %s' does not exist", format_cp(first_cp),
format_cp(last_cp))
raise NotInLGR(first_cp)
for cp in range(first_cp, last_cp + 1):
char = RangeChar(cp, first_cp, last_cp)
if not self._del_char(char):
# TODO: clean-up range on error
# This should only happen if range insertion failed
# -> inconsistent state for now
logger.critical("Range '%s - %s' is missing code point %s",
format_cp(first_cp), format_cp(last_cp),
format_cp(cp))
raise NotInLGR(cp)
# Remove and sort by first cp
self.ranges.remove((first_cp, last_cp))
def get_chars_from_prefix(self, cp, only_variants=False):
"""
Return the list of characters starting with cp.
:param cp: The first codepoint of the characters.
:return: List of characters, ordered by decreasing length.
:param only_variants: Only return chars with variants.
:raises NotInLGR: If the code point does not exist.
"""
if cp not in self._chardict:
raise NotInLGR(cp)
if not only_variants:
iterable = self._chardict[cp]
else:
iterable = [v for v in self._chardict[cp] if v.has_variant()]
return sorted(iterable, key=lambda x: len(x), reverse=True)