1 files changed, 149 insertions, 0 deletions
diff --git a/utils.go b/utils.go
new file mode 100644
index 0000000..1ca8298
--- /dev/null
+++ b/utils.go
@@ -0,0 +1,149 @@
+package bibtex
+
+import (
+	"bytes"
+	"strings"
+	"unicode"
+)
+
+// reader for LaTeX strings with methods to easily handle special chars, etc
+type reader struct {
+	*strings.Reader
+}
+
+func newReader(s string) *reader {
+	return &reader{Reader: strings.NewReader(s)}
+}
+
+func (r *reader) readRune() rune {
+	c, _, _ := r.Reader.ReadRune()
+	return c
+}
+
+func (r *reader) readBraces() string {
+	bracelevel := 1
+	var c rune
+	var buf bytes.Buffer
+	for bracelevel > 0 {
+		c = r.readRune()
+		switch {
+		case c == '{':
+			bracelevel += 1
+		case c == '}':
+			bracelevel -= 1
+		case c == eof:
+			return buf.String()
+		}
+		buf.WriteRune(c)
+	}
+	return buf.String()
+}
+
+func (r *reader) skipBraces() {
+	bracelevel := 1
+	var c rune
+	for bracelevel > 0 {
+		c = r.readRune()
+		switch {
+		case c == '{':
+			bracelevel += 1
+		case c == '}':
+			bracelevel -= 1
+		case c == eof:
+			return
+		}
+	}
+}
+
+func (r *reader) peek() rune {
+	ch := r.readRune()
+	r.UnreadRune()
+	return ch
+}
+
+// length of a LaTeX string. A special char counts as one, braces count as one
+// to be consitent with the original BibTeX implementation, but this seems
+// stupid
+func length(s string) (res int) {
+	reader := newReader(s)
+	res = 0
+	blevel := 0
+	for c := reader.readRune(); c != eof; c = reader.readRune() {
+		switch {
+		case c == '{' && blevel == 0:
+			blevel += 1
+			if c := reader.peek(); c == '\\' {
+				// special char, count as one
+				reader.skipBraces()
+				blevel = 0
+			}
+		case c == '{':
+			blevel += 1
+		case c == '}':
+			blevel -= 1
+		}
+		res += 1
+	}
+	return
+}
+
+func (r *reader) searchBraces() (bool, bool) {
+	bracelevel := 1
+	var c rune
+	for bracelevel > 0 {
+		c = r.readRune()
+		switch {
+		case c == '{':
+			bracelevel += 1
+		case c == '}':
+			bracelevel -= 1
+		case unicode.IsLower(c):
+			return true, true
+		case unicode.IsUpper(c):
+			return false, true
+		case c == eof:
+			return false, false
+		}
+	}
+	return false, false
+}
+
+func (r *reader) readCommand() string {
+	var buf bytes.Buffer
+	for c := r.readRune(); c != eof; c = r.readRune() {
+		if !unicode.IsLetter(c) {
+			r.UnreadRune()
+			break
+		} else {
+			buf.WriteRune(c)
+		}
+	}
+	return buf.String()
+}
+
+func (t *Token) isLower() bool {
+	reader := newReader(t.Text)
+	for c := reader.readRune(); c != eof; c = reader.readRune() {
+		switch {
+		case c == '{':
+			if c := reader.readRune(); c == '\\' {
+				command := reader.readCommand()
+				if lcommands[command] {
+					return true
+				} else if ucommands[command] {
+					return false
+				} else if a, b := reader.searchBraces(); b {
+					return a
+				}
+			} else {
+				reader.UnreadRune()
+				reader.skipBraces()
+			}
+		case unicode.IsLower(c):
+			return true
+		case unicode.IsUpper(c):
+			return false
+		}
+	}
+	return false
+}