commonMain.HtmlEntity.kt Maven / Gradle / Ivy
/*
* Copyright 2019-2021 Mamoe Technologies and contributors.
*
* 此源代码的使用受 GNU AFFERO GENERAL PUBLIC LICENSE version 3 许可证的约束, 可以在以下链接找到该许可证.
* Use of this source code is governed by the GNU AGPLv3 license that can be found through the following link.
*
* https://github.com/mamoe/mirai/blob/dev/LICENSE
*/
package net.mamoe.mirai.utils
@Suppress("RegExpRedundantEscape")
private val STR_TO_CHAR_PATTERN = """\&(\#?[A-Za-z0-9]+?)\;""".toRegex()
public fun String.decodeHtmlEscape(): String = replace(STR_TO_CHAR_PATTERN) { match ->
STR_TO_CHAR_MAPPINGS[match.value]?.let { return@replace it }
val match1 = match.groups[1]!!.value
if (match1.length > 1 && match1[0] == '#') {
if (match1.length > 2) {
if (match1[1] == 'x') { // hex
match1.substring(2).toIntOrNull(16)?.let {
return@replace it.toChar().toString()
}
}
}
match1.substring(1).toIntOrNull()?.let {
return@replace it.toChar().toString()
}
}
match.value
}
private val STR_TO_CHAR_MAPPINGS: Map by lazy {
//
val result = HashMap(223)
result["&"] = "\u0026"
result["<"] = "\u003c"
result[">"] = "\u003e"
result[" "] = "\u00a0"
result["¡"] = "\u00a1"
result["¢"] = "\u00a2"
result["£"] = "\u00a3"
result["¤"] = "\u00a4"
result["¥"] = "\u00a5"
result["¦"] = "\u00a6"
result["§"] = "\u00a7"
result["¨"] = "\u00a8"
result["©"] = "\u00a9"
result["ª"] = "\u00aa"
result["«"] = "\u00ab"
result["¬"] = "\u00ac"
result[""] = "\u00ad"
result["®"] = "\u00ae"
result["¯"] = "\u00af"
result["°"] = "\u00b0"
result["±"] = "\u00b1"
result["²"] = "\u00b2"
result["³"] = "\u00b3"
result["´"] = "\u00b4"
result["µ"] = "\u00b5"
result["¶"] = "\u00b6"
result["·"] = "\u00b7"
result["¸"] = "\u00b8"
result["¹"] = "\u00b9"
result["º"] = "\u00ba"
result["»"] = "\u00bb"
result["¼"] = "\u00bc"
result["½"] = "\u00bd"
result["¾"] = "\u00be"
result["¿"] = "\u00bf"
result["À"] = "\u00c0"
result["Á"] = "\u00c1"
result["Â"] = "\u00c2"
result["Ã"] = "\u00c3"
result["Ä"] = "\u00c4"
result["Å"] = "\u00c5"
result["Æ"] = "\u00c6"
result["Ç"] = "\u00c7"
result["È"] = "\u00c8"
result["É"] = "\u00c9"
result["Ê"] = "\u00ca"
result["Ë"] = "\u00cb"
result["Ì"] = "\u00cc"
result["Í"] = "\u00cd"
result["Î"] = "\u00ce"
result["Ï"] = "\u00cf"
result["Ð"] = "\u00d0"
result["Ñ"] = "\u00d1"
result["Ò"] = "\u00d2"
result["Ó"] = "\u00d3"
result["Ô"] = "\u00d4"
result["Õ"] = "\u00d5"
result["Ö"] = "\u00d6"
result["×"] = "\u00d7"
result["Ø"] = "\u00d8"
result["Ù"] = "\u00d9"
result["Ú"] = "\u00da"
result["Û"] = "\u00db"
result["Ü"] = "\u00dc"
result["Ý"] = "\u00dd"
result["Þ"] = "\u00de"
result["ß"] = "\u00df"
result["à"] = "\u00e0"
result["á"] = "\u00e1"
result["â"] = "\u00e2"
result["ã"] = "\u00e3"
result["ä"] = "\u00e4"
result["å"] = "\u00e5"
result["æ"] = "\u00e6"
result["ç"] = "\u00e7"
result["è"] = "\u00e8"
result["é"] = "\u00e9"
result["ê"] = "\u00ea"
result["ë"] = "\u00eb"
result["ì"] = "\u00ec"
result["í"] = "\u00ed"
result["î"] = "\u00ee"
result["ï"] = "\u00ef"
result["ð"] = "\u00f0"
result["ñ"] = "\u00f1"
result["ò"] = "\u00f2"
result["ó"] = "\u00f3"
result["ô"] = "\u00f4"
result["õ"] = "\u00f5"
result["ö"] = "\u00f6"
result["÷"] = "\u00f7"
result["ø"] = "\u00f8"
result["ù"] = "\u00f9"
result["ú"] = "\u00fa"
result["û"] = "\u00fb"
result["ü"] = "\u00fc"
result["ý"] = "\u00fd"
result["þ"] = "\u00fe"
result["ÿ"] = "\u00ff"
result["ƒ"] = "\u0192"
result["Α"] = "\u0391"
result["Β"] = "\u0392"
result["Γ"] = "\u0393"
result["Δ"] = "\u0394"
result["Ε"] = "\u0395"
result["Ζ"] = "\u0396"
result["Η"] = "\u0397"
result["Θ"] = "\u0398"
result["Ι"] = "\u0399"
result["Κ"] = "\u039a"
result["Λ"] = "\u039b"
result["Μ"] = "\u039c"
result["Ν"] = "\u039d"
result["Ξ"] = "\u039e"
result["Ο"] = "\u039f"
result["Π"] = "\u03a0"
result["Ρ"] = "\u03a1"
result["Σ"] = "\u03a3"
result["Τ"] = "\u03a4"
result["Υ"] = "\u03a5"
result["Φ"] = "\u03a6"
result["Χ"] = "\u03a7"
result["Ψ"] = "\u03a8"
result["Ω"] = "\u03a9"
result["α"] = "\u03b1"
result["β"] = "\u03b2"
result["γ"] = "\u03b3"
result["δ"] = "\u03b4"
result["ε"] = "\u03b5"
result["ζ"] = "\u03b6"
result["η"] = "\u03b7"
result["θ"] = "\u03b8"
result["ι"] = "\u03b9"
result["κ"] = "\u03ba"
result["λ"] = "\u03bb"
result["μ"] = "\u03bc"
result["ν"] = "\u03bd"
result["ξ"] = "\u03be"
result["ο"] = "\u03bf"
result["π"] = "\u03c0"
result["ρ"] = "\u03c1"
result["ς"] = "\u03c2"
result["σ"] = "\u03c3"
result["τ"] = "\u03c4"
result["υ"] = "\u03c5"
result["φ"] = "\u03c6"
result["χ"] = "\u03c7"
result["ψ"] = "\u03c8"
result["ω"] = "\u03c9"
result["ϑ"] = "\u03d1"
result["ϒ"] = "\u03d2"
result["ϖ"] = "\u03d6"
result["•"] = "\u2022"
result["…"] = "\u2026"
result["′"] = "\u2032"
result["″"] = "\u2033"
result["‾"] = "\u203e"
result["⁄"] = "\u2044"
result["℘"] = "\u2118"
result["ℑ"] = "\u2111"
result["ℜ"] = "\u211c"
result["™"] = "\u2122"
result["ℵ"] = "\u2135"
result["←"] = "\u2190"
result["↑"] = "\u2191"
result["→"] = "\u2192"
result["↓"] = "\u2193"
result["↔"] = "\u2194"
result["↵"] = "\u21b5"
result["⇐"] = "\u21d0"
result["⇑"] = "\u21d1"
result["⇒"] = "\u21d2"
result["⇓"] = "\u21d3"
result["⇔"] = "\u21d4"
result["∀"] = "\u2200"
result["∂"] = "\u2202"
result["∃"] = "\u2203"
result["∅"] = "\u2205"
result["∇"] = "\u2207"
result["∈"] = "\u2208"
result["∉"] = "\u2209"
result["∋"] = "\u220b"
result["∏"] = "\u220f"
result["∑"] = "\u2211"
result["−"] = "\u2212"
result["∗"] = "\u2217"
result["√"] = "\u221a"
result["∝"] = "\u221d"
result["∞"] = "\u221e"
result["∠"] = "\u2220"
result["∧"] = "\u2227"
result["∨"] = "\u2228"
result["∩"] = "\u2229"
result["∪"] = "\u222a"
result["∫"] = "\u222b"
result["∴"] = "\u2234"
result["∼"] = "\u223c"
result["≅"] = "\u2245"
result["≈"] = "\u2248"
result["≠"] = "\u2260"
result["≡"] = "\u2261"
result["≤"] = "\u2264"
result["≥"] = "\u2265"
result["⊂"] = "\u2282"
result["⊃"] = "\u2283"
result["⊄"] = "\u2284"
result["⊆"] = "\u2286"
result["⊇"] = "\u2287"
result["⊕"] = "\u2295"
result["⊗"] = "\u2297"
result["⊥"] = "\u22a5"
result["⋅"] = "\u22c5"
result["⌈"] = "\u2308"
result["⌉"] = "\u2309"
result["⌊"] = "\u230a"
result["⌋"] = "\u230b"
result["⟨"] = "\u2329"
result["⟩"] = "\u232a"
result["◊"] = "\u25ca"
result["♠"] = "\u2660"
result["♣"] = "\u2663"
result["♥"] = "\u2665"
result["♦"] = "\u2666"
//
result
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy