Link matching improvements
This commit is contained in:
parent
1da620875a
commit
14d8539298
|
@ -23,7 +23,7 @@ trans2[en_dash] = trans2[" "]
|
||||||
|
|
||||||
patterns = [
|
patterns = [
|
||||||
lambda q: re.compile(
|
lambda q: re.compile(
|
||||||
r"(?<!-)(?:\[\[(?:[^]]+\|)?)?(%s)%s(?:\]\])?"
|
r"(?<!-)\[\[(%s)%s\|(?=.*\]\])"
|
||||||
% (
|
% (
|
||||||
re.escape(q[0]),
|
re.escape(q[0]),
|
||||||
"".join("-?" + (trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
"".join("-?" + (trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
||||||
|
@ -31,10 +31,19 @@ patterns = [
|
||||||
re.I,
|
re.I,
|
||||||
),
|
),
|
||||||
lambda q: re.compile(
|
lambda q: re.compile(
|
||||||
r"(?<!-)\[\[[^|]+\|(%s)%s\]\]" % (re.escape(q[0]), re.escape(q[1:])), re.I
|
r"(?<!-)\[\[(?:(?!File:)(?:[^]]+\|)?)(%s)%s\]\]"
|
||||||
|
% (
|
||||||
|
re.escape(q[0]),
|
||||||
|
"".join("-?" + (trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
||||||
|
),
|
||||||
|
re.I,
|
||||||
),
|
),
|
||||||
lambda q: re.compile(
|
lambda q: re.compile(
|
||||||
r"(?<!-)\[\[[^|]+\|(%s)%s(?:\]\])?"
|
r"(?<!-)\[\[(?!File:)[^|]+\|(%s)%s\]\]" % (re.escape(q[0]), re.escape(q[1:])),
|
||||||
|
re.I,
|
||||||
|
),
|
||||||
|
lambda q: re.compile(
|
||||||
|
r"(?<!-)\[\[(?!File:)[^|]+\|(%s)%s(?:\]\])?"
|
||||||
% (
|
% (
|
||||||
re.escape(q[0]),
|
re.escape(q[0]),
|
||||||
"".join("-?" + (trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
"".join("-?" + (trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
||||||
|
@ -46,7 +55,7 @@ patterns = [
|
||||||
r"(?<!-)(%s)%s"
|
r"(?<!-)(%s)%s"
|
||||||
% (
|
% (
|
||||||
re.escape(q[0]),
|
re.escape(q[0]),
|
||||||
"".join((trans[c] if c in trans else re.escape(c)) for c in q[1:]),
|
"".join((trans2[c] if c in trans2 else re.escape(c)) for c in q[1:]),
|
||||||
),
|
),
|
||||||
re.I,
|
re.I,
|
||||||
),
|
),
|
||||||
|
@ -58,7 +67,9 @@ class NoMatch(Exception):
|
||||||
|
|
||||||
|
|
||||||
re_cite = re.compile(
|
re_cite = re.compile(
|
||||||
r"<ref( [^>]*?)?>\s*({{cite.*?}}|\[https?://[^]]*?\])\s*</ref>", re.I | re.S
|
# r"<ref( [^>]*?)?>\s*({{cite.*?}}|\[https?://[^]]*?\])\s*</ref>", re.I | re.S
|
||||||
|
r"<ref( [^>]*?)?>.*</ref>",
|
||||||
|
re.I | re.S,
|
||||||
)
|
)
|
||||||
|
|
||||||
|
|
||||||
|
@ -173,8 +184,14 @@ def mk_link_matcher(q: str) -> typing.Callable[[str], re.Match[str] | None]:
|
||||||
return search_for_link
|
return search_for_link
|
||||||
|
|
||||||
|
|
||||||
def add_link(m, replacement, text):
|
def add_link(m: re.Match[str], replacement: str, text: str) -> str:
|
||||||
return m.re.sub(lambda m: "[[%s]]" % replacement, text, count=1)
|
"""Add link to text."""
|
||||||
|
|
||||||
|
matched_text = m.group(0)
|
||||||
|
if matched_text.startswith("[[") and matched_text.endswith("|"):
|
||||||
|
return m.re.sub(lambda m: f"[[{replacement}|", text, count=1)
|
||||||
|
else:
|
||||||
|
return m.re.sub(lambda m: f"[[{replacement}]]", text, count=1)
|
||||||
|
|
||||||
|
|
||||||
def find_link_in_chunk(
|
def find_link_in_chunk(
|
||||||
|
@ -303,7 +320,9 @@ def find_link_and_section(q: str, content: str, linkto: str | None = None):
|
||||||
if header:
|
if header:
|
||||||
new_content += header
|
new_content += header
|
||||||
for token_type, text in parse_cite(section_text):
|
for token_type, text in parse_cite(section_text):
|
||||||
if token_type == "text" and not replacement:
|
if token_type != "text" or replacement:
|
||||||
|
new_content += text
|
||||||
|
continue
|
||||||
new_text = ""
|
new_text = ""
|
||||||
for token_type2, text2 in parse_links(text):
|
for token_type2, text2 in parse_links(text):
|
||||||
if token_type2 == "link" and not replacement:
|
if token_type2 == "link" and not replacement:
|
||||||
|
|
Loading…
Reference in a new issue