Latex sanitizing

1 files changed, 105 insertions, 0 deletions

diff --git a/utils.go b/utils.go
index e91ae44..3a1dd6a 100644
--- a/utils.go
+++ b/utils.go
@@ -4,8 +4,48 @@ import (
 	"bytes"
 	"strings"
 	"unicode"
+
+	"golang.org/x/text/unicode/norm"
 )
 
+var accents = map[string]string{
+	"'":  "\u0301",
+	"`":  "\u0300",
+	"^":  "\u0302",
+	"\"": "\u0308",
+	"~":  "\u0303",
+	"=":  "\u0304",
+	".":  "\u0307",
+	"u":  "\u0306",
+	"v":  "\u030C",
+	"H":  "\u030B",
+	"t":  "\u0361",
+	"c":  "\u0327",
+	"d":  "\u0323",
+	"b":  "\u0331",
+}
+
+var commands = map[string]string{
+	"aa": "å",
+	"AA": "Å",
+	"oe": "œ",
+	"ae": "æ",
+	"OE": "Œ",
+	"AE": "Æ",
+	"ss": "ß",
+	"o":  "ø",
+	"O":  "Ø",
+	"L":  "Ł",
+	"l":  "ł",
+	"$":  "$",
+	"{":  "{",
+	"}":  "}",
+	"_":  "_",
+	"#":  "#",
+	"%":  "%",
+	"&":  "&",
+}
+
 // reader for LaTeX strings with methods to easily handle special chars, etc
 type reader struct {
 	*strings.Reader
@@ -61,6 +101,71 @@ func (r *reader) peek() rune {
 	return ch
 }
 
+func (r *reader) eatSpace() {
+	for c := r.readRune(); c != eof; c = r.readRune() {
+		if !unicode.IsSpace(c) {
+			r.UnreadRune()
+			break
+		}
+	}
+}
+
+func (r *reader) readLetter() (d rune) {
+	c := r.readRune()
+	switch c {
+	case '{':
+		blevel := 1
+		for c != eof {
+			c = r.readRune()
+			if c == '{' {
+				blevel += 1
+			} else if c == '}' {
+				blevel -= 1
+				if blevel == 0 {
+					break
+				}
+			} else {
+				if d == rune(0) {
+					d = c
+				}
+			}
+		}
+	default:
+		d = c
+		return
+	}
+	if d == rune(0) {
+		d = ' '
+	}
+	return
+}
+
+func removeLatexCommands(s string) string {
+	r := newReader(s)
+	var buf bytes.Buffer
+	for c := r.readRune(); c != eof; c = r.readRune() {
+		switch c {
+		case '\\':
+			command := r.readCommand()
+			if command == "" {
+				command = string(r.readRune())
+			}
+			if v, in := commands[command]; in {
+				buf.WriteString(v)
+			} else if v, in := accents[command]; in {
+				r.eatSpace()
+				l := r.readLetter()
+				buf.WriteRune(l)
+				buf.WriteString(v)
+			}
+		case '{', '}':
+		default:
+			buf.WriteRune(c)
+		}
+	}
+	return string(norm.NFC.Bytes(buf.Bytes()))
+}
+
 // 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