Chat Messages: components 201-220

internal/markdown_inline.go

@@ -0,0 +1,128 @@
+package internal
+
+import "strings"
+
+// mdStyle is the inline style in effect.
+type mdStyle struct{ strong, em, strike bool }
+
+// ParseInline splits text into styled spans: emphasis, code, strikethrough and links.
+// An opener without its closer stays literal, as while a message streams in.
+func ParseInline(s string) []MdSpan {
+	var out []MdSpan
+	parseInline(s, mdStyle{}, &out)
+	return out
+}
+
+func mdEmit(out *[]MdSpan, sp MdSpan) {
+	if sp.Text == "" {
+		return
+	}
+	if n := len(*out); n > 0 {
+		last := &(*out)[n-1]
+		if sp.URL == "" && last.URL == "" && last.Strong == sp.Strong && last.Em == sp.Em && last.Code == sp.Code && last.Strike == sp.Strike {
+			last.Text += sp.Text
+			return
+		}
+	}
+	*out = append(*out, sp)
+}
+
+func parseInline(s string, st mdStyle, out *[]MdSpan) {
+	plain := func(t string) {
+		mdEmit(out, MdSpan{Text: t, Strong: st.strong, Em: st.em, Strike: st.strike})
+	}
+	for i := 0; i < len(s); {
+		switch c := s[i]; {
+		case c == '\\' && i+1 < len(s) && strings.IndexByte("\\`*_~[]()#!>-|", s[i+1]) >= 0:
+			plain(s[i+1 : i+2])
+			i += 2
+			continue
+		case c == '`':
+			if end := strings.IndexByte(s[i+1:], '`'); end >= 0 {
+				mdEmit(out, MdSpan{Text: s[i+1 : i+1+end], Code: true, Strong: st.strong, Em: st.em, Strike: st.strike})
+				i += end + 2
+				continue
+			}
+		case c == '[':
+			if text, url, n := mdLink(s[i:]); n > 0 {
+				mdEmit(out, MdSpan{Text: text, URL: url, Strong: st.strong, Em: st.em, Strike: st.strike})
+				i += n
+				continue
+			}
+		case c == '*' || c == '_' || c == '~':
+			if n := mdDelimited(s, i, st, out); n > 0 {
+				i += n
+				continue
+			}
+		}
+		j := i + 1
+		for j < len(s) && strings.IndexByte("\\`[*_~", s[j]) < 0 {
+			j++
+		}
+		plain(s[i:j])
+		i = j
+	}
+}
+
+// mdDelimited parses emphasis or strikethrough opening at i and returns its length, or 0.
+func mdDelimited(s string, i int, st mdStyle, out *[]MdSpan) int {
+	c := s[i]
+	n := 1
+	if i+1 < len(s) && s[i+1] == c {
+		n = 2
+	}
+	if c == '~' && n != 2 {
+		return 0
+	}
+	// Intraword underscores stay literal, as in snake_case.
+	if c == '_' && i > 0 && mdWord(s[i-1]) {
+		return 0
+	}
+	marker := s[i : i+n]
+	body := i + n
+	if body >= len(s) || s[body] == ' ' {
+		return 0
+	}
+	end := -1
+	for j := body + 1; j+n <= len(s); j++ {
+		if s[j:j+n] == marker && s[j-1] != ' ' && (n == 2 || j+1 >= len(s) || s[j+1] != c) {
+			end = j
+			break
+		}
+	}
+	if end < 0 {
+		return 0
+	}
+	inner := st
+	switch {
+	case c == '~':
+		inner.strike = true
+	case n == 2:
+		inner.strong = true
+	default:
+		inner.em = true
+	}
+	parseInline(s[body:end], inner, out)
+	return end + n - i
+}
+
+func mdWord(b byte) bool {
+	return b >= 'a' && b <= 'z' || b >= 'A' && b <= 'Z' || b >= '0' && b <= '9'
+}
+
+// mdLink parses [text](url) at the start of s and returns its length, or 0.
+func mdLink(s string) (text, url string, n int) {
+	close := strings.Index(s, "](")
+	if close < 1 || strings.IndexByte(s[1:close], '[') >= 0 {
+		return "", "", 0
+	}
+	end := strings.IndexByte(s[close+2:], ')')
+	if end < 0 {
+		return "", "", 0
+	}
+	url = strings.TrimSpace(s[close+2 : close+2+end])
+	if url == "" || strings.ContainsAny(url, " \n") {
+		return "", "", 0
+	}
+	return s[1:close], url, close + 3 + end
+}