fix: inline html

This commit is contained in:
2026-03-30 09:54:23 +02:00
parent 3075719963
commit 3970c6eb47
2 changed files with 49 additions and 4 deletions

View File

@@ -22,7 +22,7 @@ function mask_plain(s, t) {
gsub(/\$/, "\034P8\034", t)
return t
}
BEGIN { in_plain = 0 }
BEGIN { in_plain = 0; in_script_style = 0 }
{
line = $0
out = ""
@@ -48,5 +48,41 @@ BEGIN { in_plain = 0 }
in_plain = 0
}
}
print out
tmp_line = out
out2 = ""
while (1) {
if (!in_script_style) {
pos_script = match(tolower(tmp_line), /<script([ >]|$)/)
script_start = RSTART; script_len = RLENGTH
pos_style = match(tolower(tmp_line), /<style([ >]|$)/)
style_start = RSTART; style_len = RLENGTH
if (pos_script == 0 && pos_style == 0) {
out2 = out2 tmp_line
break
}
if (pos_script > 0 && (pos_style == 0 || pos_script < pos_style)) {
out2 = out2 substr(tmp_line, 1, script_start + script_len - 1)
tmp_line = substr(tmp_line, script_start + script_len)
in_script_style = 1
end_tag = "</script>"
} else {
out2 = out2 substr(tmp_line, 1, style_start + style_len - 1)
tmp_line = substr(tmp_line, style_start + style_len)
in_script_style = 1
end_tag = "</style>"
}
} else {
pos_end = match(tolower(tmp_line), end_tag)
if (pos_end == 0) {
out2 = out2 mask_plain(tmp_line)
tmp_line = ""
break
}
out2 = out2 mask_plain(substr(tmp_line, 1, RSTART - 1)) substr(tmp_line, RSTART, RLENGTH)
tmp_line = substr(tmp_line, RSTART + RLENGTH)
in_script_style = 0
}
}
print out2
}

View File

@@ -4,7 +4,7 @@ BEGIN {
}
{
if ($0 ~ /^<pre>/) in_pre = 1
if ($0 ~ /<pre>/) in_pre = 1
if (in_pre) {
if (in_p) { print "</p>"; in_p = 0 }
@@ -13,7 +13,16 @@ BEGIN {
next
}
if ($0 ~ /^<\/?(div|table|p|[ou]l|h[1-6]|[bh]r|blockquote|li|hr|section|article|nav|aside|header|footer|dl|dt|dd)/) {
if ($0 ~ /^<\/?(div|table|p|[ou]l|h[1-6]|[bh]r|blockquote|li|hr|section|article|nav|aside|header|footer|dl|dt|dd|script|style|iframe|details|summary|figure|figcaption|audio|video|picture)/) {
if (in_p) {
print "</p>"
in_p = 0
}
print
next
}
if ($0 ~ /^[\t ]*!([a-zA-Z])?\[[^\]]*\](\([^)]*\))?[\t ]*$/ || $0 ~ /^[\t ]*!!?\[[^\]]*\](\([^)]*\))?[\t ]*$/) {
if (in_p) {
print "</p>"
in_p = 0