Hand-made word splitter for BouncerServ

Remove the (direct) dependency on shlex (go-scfg still depends on it).

Co-authored-by: Simon Ser <contact@emersion.fr>

go.mod

@@ -6,7 +6,6 @@ require (
 	git.sr.ht/~emersion/go-scfg v0.0.0-20201019143924-142a8aa629fc
 	git.sr.ht/~sircmpwn/go-bare v0.0.0-20210331145808-46f9b5e5bcf9
 	github.com/emersion/go-sasl v0.0.0-20200509203442-7bfe0ed36a21
-	github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510
 	github.com/klauspost/compress v1.11.13 // indirect
 	github.com/mattn/go-sqlite3 v1.14.6
 	github.com/pires/go-proxyproto v0.5.0

service.go

@@ -21,8 +21,8 @@ import (
 	"strconv"
 	"strings"
 	"time"
+	"unicode"
 
-	"github.com/google/shlex"
 	"golang.org/x/crypto/bcrypt"
 	"gopkg.in/irc.v3"
 )
@@ -63,10 +63,63 @@ func sendServicePRIVMSG(dc *downstreamConn, text string) {
 	})
 }
 
+func splitWords(s string) ([]string, error) {
+	var words []string
+	var lastWord strings.Builder
+	escape := false
+	prev := ' '
+	wordDelim := ' '
+
+	for _, r := range s {
+		if escape {
+			// last char was a backslash, write the byte as-is.
+			lastWord.WriteRune(r)
+			escape = false
+		} else if r == '\\' {
+			escape = true
+		} else if wordDelim == ' ' && unicode.IsSpace(r) {
+			// end of last word
+			if !unicode.IsSpace(prev) {
+				words = append(words, lastWord.String())
+				lastWord.Reset()
+			}
+		} else if r == wordDelim {
+			// wordDelim is either " or ', switch back to
+			// space-delimited words.
+			wordDelim = ' '
+		} else if r == '"' || r == '\'' {
+			if wordDelim == ' ' {
+				// start of (double-)quoted word
+				wordDelim = r
+			} else {
+				// either wordDelim is " and r is ' or vice-versa
+				lastWord.WriteRune(r)
+			}
+		} else {
+			lastWord.WriteRune(r)
+		}
+
+		prev = r
+	}
+
+	if !unicode.IsSpace(prev) {
+		words = append(words, lastWord.String())
+	}
+
+	if wordDelim != ' ' {
+		return nil, fmt.Errorf("unterminated quoted string")
+	}
+	if escape {
+		return nil, fmt.Errorf("unterminated backslash sequence")
+	}
+
+	return words, nil
+}
+
 func handleServicePRIVMSG(dc *downstreamConn, text string) {
-	words, err := shlex.Split(text)
+	words, err := splitWords(text)
 	if err != nil {
-		sendServicePRIVMSG(dc, fmt.Sprintf("error: failed to parse command: %v", err))
+		sendServicePRIVMSG(dc, fmt.Sprintf(`error: failed to parse command: %v`, err))
 		return
 	}
 

service_test.go

@@ -0,0 +1,54 @@
+package soju
+
+import (
+	"testing"
+)
+
+func assertSplit(t *testing.T, input string, expected []string) {
+	actual, err := splitWords(input)
+	if err != nil {
+		t.Errorf("%q: %v", input, err)
+		return
+	}
+	if len(actual) != len(expected) {
+		t.Errorf("%q: expected %d words, got %d\nexpected: %v\ngot: %v", input, len(expected), len(actual), expected, actual)
+		return
+	}
+	for i := 0; i < len(actual); i++ {
+		if actual[i] != expected[i] {
+			t.Errorf("%q: expected word #%d to be %q, got %q\nexpected: %v\ngot: %v", input, i, expected[i], actual[i], expected, actual)
+		}
+	}
+}
+
+func TestSplit(t *testing.T) {
+	assertSplit(t, "  ch 'up' #soju    'relay'-det\"ache\"d  message  ", []string{
+		"ch",
+		"up",
+		"#soju",
+		"relay-detached",
+		"message",
+	})
+	assertSplit(t, "net update \\\"free\\\"node -pass 'political \"stance\" desu!' -realname '' -nick lee", []string{
+		"net",
+		"update",
+		"\"free\"node",
+		"-pass",
+		"political \"stance\" desu!",
+		"-realname",
+		"",
+		"-nick",
+		"lee",
+	})
+	assertSplit(t, "Omedeto,\\ Yui! ''", []string{
+		"Omedeto, Yui!",
+		"",
+	})
+
+	if _, err := splitWords("end of 'file"); err == nil {
+		t.Errorf("expected error on unterminated single quote")
+	}
+	if _, err := splitWords("end of backquote \\"); err == nil {
+		t.Errorf("expected error on unterminated backquote sequence")
+	}
+}