1# Process html entity - {, ¯, ", ...
2import re
3
4from ..common.entities import entities
5from ..common.utils import fromCodePoint, isValidEntityCode
6from .state_inline import StateInline
7
8DIGITAL_RE = re.compile(r"^&#((?:x[a-f0-9]{1,6}|[0-9]{1,7}));", re.IGNORECASE)
9NAMED_RE = re.compile(r"^&([a-z][a-z0-9]{1,31});", re.IGNORECASE)
10
11
12def entity(state: StateInline, silent: bool) -> bool:
13 pos = state.pos
14 maximum = state.posMax
15
16 if state.src[pos] != "&":
17 return False
18
19 if pos + 1 >= maximum:
20 return False
21
22 if state.src[pos + 1] == "#":
23 if match := DIGITAL_RE.search(state.src[pos:]):
24 if not silent:
25 match1 = match.group(1)
26 code = (
27 int(match1[1:], 16) if match1[0].lower() == "x" else int(match1, 10)
28 )
29
30 token = state.push("text_special", "", 0)
31 token.content = (
32 fromCodePoint(code)
33 if isValidEntityCode(code)
34 else fromCodePoint(0xFFFD)
35 )
36 token.markup = match.group(0)
37 token.info = "entity"
38
39 state.pos += len(match.group(0))
40 return True
41
42 else:
43 if (match := NAMED_RE.search(state.src[pos:])) and match.group(1) in entities:
44 if not silent:
45 token = state.push("text_special", "", 0)
46 token.content = entities[match.group(1)]
47 token.markup = match.group(0)
48 token.info = "entity"
49
50 state.pos += len(match.group(0))
51 return True
52
53 return False