initial commits

saferune.go

@@ -0,0 +1,180 @@
+package htmlescape
+
+var escaped00 []byte = []byte("�")
+var escaped01 []byte = []byte("")
+var escaped02 []byte = []byte("")
+var escaped03 []byte = []byte("")
+var escaped04 []byte = []byte("")
+var escaped05 []byte = []byte("")
+var escaped06 []byte = []byte("")
+var escaped07 []byte = []byte("")
+var escaped08 []byte = []byte("")
+var escaped09 []byte = []byte("	")
+var escaped0A []byte = []byte("
")
+var escaped0B []byte = []byte("")
+var escaped0C []byte = []byte("")
+var escaped0D []byte = []byte("
")
+var escaped0E []byte = []byte("")
+var escaped0F []byte = []byte("")
+
+var escaped10 []byte = []byte("")
+var escaped11 []byte = []byte("")
+var escaped12 []byte = []byte("")
+var escaped13 []byte = []byte("")
+var escaped14 []byte = []byte("")
+var escaped15 []byte = []byte("")
+var escaped16 []byte = []byte("")
+var escaped17 []byte = []byte("")
+var escaped18 []byte = []byte("")
+var escaped19 []byte = []byte("")
+var escaped1A []byte = []byte("")
+var escaped1B []byte = []byte("")
+var escaped1C []byte = []byte("")
+var escaped1D []byte = []byte("")
+var escaped1E []byte = []byte("")
+var escaped1F []byte = []byte("")
+
+const minCtrlBlock1 = 0x7F
+
+var escaped7F []byte = []byte("")
+var escaped80 []byte = []byte("€")
+var escaped81 []byte = []byte("")
+var escaped82 []byte = []byte("‚")
+var escaped83 []byte = []byte("ƒ")
+var escaped84 []byte = []byte("„")
+var escaped85 []byte = []byte("…")
+var escaped86 []byte = []byte("†")
+var escaped87 []byte = []byte("‡")
+var escaped88 []byte = []byte("ˆ")
+var escaped89 []byte = []byte("‰")
+var escaped8A []byte = []byte("Š")
+var escaped8B []byte = []byte("‹")
+var escaped8C []byte = []byte("Œ")
+var escaped8D []byte = []byte("")
+var escaped8E []byte = []byte("Ž")
+var escaped8F []byte = []byte("")
+var escaped90 []byte = []byte("")
+var escaped91 []byte = []byte("‘")
+var escaped92 []byte = []byte("’")
+var escaped93 []byte = []byte("“")
+var escaped94 []byte = []byte("”")
+var escaped95 []byte = []byte("•")
+var escaped96 []byte = []byte("–")
+var escaped97 []byte = []byte("—")
+var escaped98 []byte = []byte("˜")
+var escaped99 []byte = []byte("™")
+var escaped9A []byte = []byte("š")
+var escaped9B []byte = []byte("›")
+var escaped9C []byte = []byte("œ")
+var escaped9D []byte = []byte("")
+var escaped9E []byte = []byte("ž")
+var escaped9F []byte = []byte("Ÿ")
+
+var ctrlBlock0 [][]byte = [][]byte{
+	escaped00,
+	escaped01,
+	escaped02,
+	escaped03,
+	escaped04,
+	escaped05,
+	escaped06,
+	escaped07,
+	escaped08,
+	escaped09,
+	escaped0A,
+	escaped0B,
+	escaped0C,
+	escaped0D,
+	escaped0E,
+	escaped0F,
+
+	escaped10,
+	escaped11,
+	escaped12,
+	escaped13,
+	escaped14,
+	escaped15,
+	escaped16,
+	escaped17,
+	escaped18,
+	escaped19,
+	escaped1A,
+	escaped1B,
+	escaped1C,
+	escaped1D,
+	escaped1E,
+	escaped1F,
+}
+
+var ctrlBlock1 [][]byte = [][]byte{
+	escaped7F,
+
+	escaped80,
+	escaped81,
+	escaped82,
+	escaped83,
+	escaped84,
+	escaped85,
+	escaped86,
+	escaped87,
+	escaped88,
+	escaped89,
+	escaped8A,
+	escaped8B,
+	escaped8C,
+	escaped8D,
+	escaped8E,
+	escaped8F,
+
+	escaped90,
+	escaped91,
+	escaped92,
+	escaped93,
+	escaped94,
+	escaped95,
+	escaped96,
+	escaped97,
+	escaped98,
+	escaped99,
+	escaped9A,
+	escaped9B,
+	escaped9C,
+	escaped9D,
+	escaped9E,
+	escaped9F,
+}
+
+var escapedQuotationMark   []byte = []byte(""")  //          """ """
+var escapedApostrophe      []byte = []byte("'")  //          "'" "'"
+var escapedAmpersand       []byte = []byte("&")  // "&"  "&"
+var escapedLessThanSign    []byte = []byte("<")   // "<"  "&#x3C"
+var escapedGreaterThanSign []byte = []byte(">")   // ">"  "&#x3E"
+
+// safeRune will only escape the rune 'r' if it is necessary or wise to do so.
+// The returned []byte is in UTF-8 — although all the values returned are in the sub-set of Unicode that is also ASCII.
+// So it is both valid UTF-8 and ASCII.
+func safeRune(r rune) (escapedRune []byte, wasEscaped bool) {
+	switch r {
+	case '"':
+		return escapedQuotationMark, true
+	case '\'':
+		return escapedApostrophe, true
+	case '&':
+		return escapedAmpersand, true
+	case '<':
+		return escapedLessThanSign, true
+	case '>':
+		return escapedGreaterThanSign, true
+	default:
+	}
+
+	if int64(0) <= int64(r) && int64(r) < int64(len(ctrlBlock0)) {
+		return ctrlBlock0[int(r)], true
+	}
+
+	if minCtrlBlock1 <= int64(r) && int64(r) < (minCtrlBlock1 + int64(len(ctrlBlock1))) {
+		return ctrlBlock1[int(r) - minCtrlBlock1], true
+	}
+
+	return nil, false
+}