aboutsummaryrefslogtreecommitdiff
path: root/ctec/parse_utils.py
diff options
context:
space:
mode:
Diffstat (limited to 'ctec/parse_utils.py')
-rw-r--r--ctec/parse_utils.py80
1 files changed, 41 insertions, 39 deletions
diff --git a/ctec/parse_utils.py b/ctec/parse_utils.py
index 890d36f..7ba2ccf 100644
--- a/ctec/parse_utils.py
+++ b/ctec/parse_utils.py
@@ -37,16 +37,16 @@ T1 = TypeVar('T1')
T2 = TypeVar('T2')
T3 = TypeVar('T3')
-ParseResult = Optional[Tuple[T, str]]
-Parser = Callable[[str], ParseResult[T]]
+ParseResult = Optional[Tuple[T, bytes]]
+Parser = Callable[[bytes], ParseResult[T]]
-def as_predicate(parser: Parser[T]) -> Callable[[str], bool]:
- def check(text: str) -> bool:
- return parser(text) is not None
+def as_predicate(parser: Parser[T]) -> Callable[[int], bool]:
+ def check(text: int) -> bool:
+ return parser(bytes([text])) is not None
return check
def alt(*parsers: Parser[T]) -> Parser[T]:
- def parse(text: str) -> ParseResult[T]:
+ def parse(text: bytes) -> ParseResult[T]:
for parser in parsers[:-1]:
result = parser(text)
if result is not None:
@@ -54,56 +54,58 @@ def alt(*parsers: Parser[T]) -> Parser[T]:
return parsers[-1](text)
return parse
-def tag(tag_text: str) -> Parser[str]:
- def parse(text: str) -> ParseResult[str]:
+def tag(tag_text: bytes) -> Parser[bytes]:
+ def parse(text: bytes) -> ParseResult[bytes]:
if text.startswith(tag_text):
return tag_text, text[len(tag_text):]
return None
return parse
# case-insensitive tag
-def itag(tag_text: str) -> Parser[str]:
- def parse(text: str) -> ParseResult[str]:
- if text.casefold().startswith(tag_text.casefold()):
+def itag(tag_text: bytes) -> Parser[bytes]:
+ def parse(text: bytes) -> ParseResult[bytes]:
+ tag_str = tag_text.decode()
+ text_str = text.decode()
+ if text_str.casefold().startswith(tag_str.casefold()):
return tag_text, text[len(tag_text):]
return None
return parse
-def take_while0(predicate: Callable[[str], bool]) -> Parser[str]:
- def parse(text: str) -> ParseResult[str]:
+def take_while0(predicate: Callable[[int], bool]) -> Parser[bytes]:
+ def parse(text: bytes) -> ParseResult[bytes]:
for i in range(len(text)):
if not predicate(text[i]):
return text[:i], text[i:]
- return text, ""
+ return text, b""
return parse
-def take_while1(predicate: Callable[[str], bool]) -> Parser[str]:
- def parse(text: str) -> ParseResult[str]:
+def take_while1(predicate: Callable[[int], bool]) -> Parser[bytes]:
+ def parse(text: bytes) -> ParseResult[bytes]:
if len(text) == 0 or not predicate(text[0]):
return None
for i in range(1, len(text)):
if not predicate(text[i]):
return text[:i], text[i:]
- return text, ""
+ return text, b""
return parse
-def take_till1(predicate: Callable[[str], bool]) -> Parser[str]:
+def take_till1(predicate: Callable[[int], bool]) -> Parser[bytes]:
return take_while1(lambda x: not predicate(x))
-def take_n(n: int) -> Parser[str]:
- def parse(text: str) -> ParseResult[str]:
+def take_n(n: int) -> Parser[bytes]:
+ def parse(text: bytes) -> ParseResult[bytes]:
if len(text) < n:
return None
return text[:n], text[n:]
return parse
-def any_char(text: str) -> ParseResult[str]:
+def any_char(text: bytes) -> ParseResult[bytes]:
if len(text) > 0:
- return text[0], text[1:]
+ return text[0:1], text[1:]
return None
def all_consuming(parser: Parser[T], *, debug=False) -> Parser[T]:
- def parse(text: str) -> ParseResult[T]:
+ def parse(text: bytes) -> ParseResult[T]:
parsed_result = parser(text)
if parsed_result is None:
if debug:
@@ -114,11 +116,11 @@ def all_consuming(parser: Parser[T], *, debug=False) -> Parser[T]:
if debug:
print('all_consuming: leftover text {}', repr(extra))
return None
- return result, ''
+ return result, b''
return parse
def map_parser(parser: Parser[T1], mapper: Callable[[T1], T2]) -> Parser[T2]:
- def parse(text: str) -> ParseResult[T2]:
+ def parse(text: bytes) -> ParseResult[T2]:
parsed_result = parser(text)
if parsed_result is None:
return None
@@ -127,7 +129,7 @@ def map_parser(parser: Parser[T1], mapper: Callable[[T1], T2]) -> Parser[T2]:
return parse
def and_then(first_parser: Parser[T1], get_second_parser: Callable[[T1], Parser[T2]]) -> Parser[T2]:
- def parse(text: str) -> ParseResult[T2]:
+ def parse(text: bytes) -> ParseResult[T2]:
parsed_result = first_parser(text)
if parsed_result is None:
return None
@@ -136,7 +138,7 @@ def and_then(first_parser: Parser[T1], get_second_parser: Callable[[T1], Parser[
return parse
def opt(parser: Parser[T]) -> Parser[Optional[T]]:
- def parse(text: str) -> ParseResult[Optional[T]]:
+ def parse(text: bytes) -> ParseResult[Optional[T]]:
result = parser(text)
if result is None:
return None, text
@@ -144,7 +146,7 @@ def opt(parser: Parser[T]) -> Parser[Optional[T]]:
return parse
def verify(parser: Parser[T], predicate: Callable[[T], bool]) -> Parser[T]:
- def parse(text: str) -> ParseResult[T]:
+ def parse(text: bytes) -> ParseResult[T]:
parsed_result = parser(text)
if parsed_result is None:
return None
@@ -155,7 +157,7 @@ def verify(parser: Parser[T], predicate: Callable[[T], bool]) -> Parser[T]:
return parse
def many0(parser: Parser[T]) -> Parser[List[T]]:
- def parse(text: str) -> ParseResult[List[T]]:
+ def parse(text: bytes) -> ParseResult[List[T]]:
result = []
parser_result = parser(text)
while parser_result is not None:
@@ -166,7 +168,7 @@ def many0(parser: Parser[T]) -> Parser[List[T]]:
return parse
def many1(parser: Parser[T]) -> Parser[List[T]]:
- def parse(text: str) -> ParseResult[List[T]]:
+ def parse(text: bytes) -> ParseResult[List[T]]:
parser_result = parser(text)
if parser_result is None:
return None
@@ -182,7 +184,7 @@ def many1(parser: Parser[T]) -> Parser[List[T]]:
return parse
def many_m_n(parser: Parser[T], min_inclusive: int, max_inclusive: int) -> Parser[List[T]]:
- def parse(text: str) -> ParseResult[List[T]]:
+ def parse(text: bytes) -> ParseResult[List[T]]:
result: List[T] = []
while len(result) < min_inclusive:
parser_result = parser(text)
@@ -201,7 +203,7 @@ def many_m_n(parser: Parser[T], min_inclusive: int, max_inclusive: int) -> Parse
return parse
def separated_many0(parser: Parser[T], separator_parser: Parser) -> Parser[List[T]]:
- def parse(text: str) -> ParseResult[List[T]]:
+ def parse(text: bytes) -> ParseResult[List[T]]:
result = []
while True:
parser_result = parser(text)
@@ -221,7 +223,7 @@ def separated_many1(parser: Parser[T], separator_parser: Parser) -> Parser[List[
return verify(separated_many0(parser, separator_parser), lambda result: len(result) > 0)
def delimited(before_parser: Parser[T1], parser: Parser[T], after_parser: Parser[T2]) -> Parser[T]:
- def parse(text: str) -> ParseResult[T]:
+ def parse(text: bytes) -> ParseResult[T]:
before_result = before_parser(text)
if before_result is None:
return None
@@ -241,7 +243,7 @@ def delimited(before_parser: Parser[T1], parser: Parser[T], after_parser: Parser
return parse
def pair(first_parser: Parser[T1], second_parser: Parser[T2]) -> Parser[Tuple[T1, T2]]:
- def parse(text: str) -> ParseResult[Tuple[T1, T2]]:
+ def parse(text: bytes) -> ParseResult[Tuple[T1, T2]]:
first_parsed_result = first_parser(text)
if first_parsed_result is None:
return None
@@ -256,7 +258,7 @@ def pair(first_parser: Parser[T1], second_parser: Parser[T2]) -> Parser[Tuple[T1
return parse
def triple(first_parser: Parser[T1], second_parser: Parser[T2], third_parser: Parser[T3]) -> Parser[Tuple[T1, T2, T3]]:
- def parse(text: str) -> ParseResult[Tuple[T1, T2, T3]]:
+ def parse(text: bytes) -> ParseResult[Tuple[T1, T2, T3]]:
first_parsed_result = first_parser(text)
if first_parsed_result is None:
return None
@@ -286,7 +288,7 @@ def followed(parser: Parser[T], after_parser: Parser[T1]) -> Parser[T]:
return map_parser(pair(parser, after_parser), first)
def separated_pair(first_parser: Parser[T1], between_parser: Parser[T], second_parser: Parser[T2]) -> Parser[Tuple[T1, T2]]:
- def parse(text: str) -> ParseResult[Tuple[T1, T2]]:
+ def parse(text: bytes) -> ParseResult[Tuple[T1, T2]]:
first_parsed_result = first_parser(text)
if first_parsed_result is None:
return None
@@ -306,7 +308,7 @@ def separated_pair(first_parser: Parser[T1], between_parser: Parser[T], second_p
return parse
def separated_triple(first_parser: Parser[T1], between12_parser: Parser, second_parser: Parser[T2], between23_parser: Parser, third_parser: Parser[T3]) -> Parser[Tuple[T1, T2, T3]]:
- def parse(text: str) -> ParseResult[Tuple[T1, T2, T3]]:
+ def parse(text: bytes) -> ParseResult[Tuple[T1, T2, T3]]:
first_parsed_result = first_parser(text)
if first_parsed_result is None:
return None
@@ -335,5 +337,5 @@ def separated_triple(first_parser: Parser[T1], between12_parser: Parser, second_
return (first_result, second_result, third_result), extra
return parse
-def string_concat(parser: Parser[Sequence[str]]) -> Parser[str]:
- return map_parser(parser, ''.join)
+def string_concat(parser: Parser[Sequence[bytes]]) -> Parser[bytes]:
+ return map_parser(parser, b''.join)
generated by cgit v1.2.3 (git 2.25.1) at 2024-04-19 07:26:22 +0000