caddy/caddyconfig/caddyfile/lexer_test.go
Matthew Holt db62942d63
Make file modes consistent
No need to have executable bit on .go or .txt files
2022-04-21 15:06:55 -06:00

281 lines
6.5 KiB
Go

// Copyright 2015 Matthew Holt and The Caddy Authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package caddyfile
import (
"testing"
)
type lexerTestCase struct {
input []byte
expected []Token
}
func TestLexer(t *testing.T) {
testCases := []lexerTestCase{
{
input: []byte(`host:123`),
expected: []Token{
{Line: 1, Text: "host:123"},
},
},
{
input: []byte(`host:123
directive`),
expected: []Token{
{Line: 1, Text: "host:123"},
{Line: 3, Text: "directive"},
},
},
{
input: []byte(`host:123 {
directive
}`),
expected: []Token{
{Line: 1, Text: "host:123"},
{Line: 1, Text: "{"},
{Line: 2, Text: "directive"},
{Line: 3, Text: "}"},
},
},
{
input: []byte(`host:123 { directive }`),
expected: []Token{
{Line: 1, Text: "host:123"},
{Line: 1, Text: "{"},
{Line: 1, Text: "directive"},
{Line: 1, Text: "}"},
},
},
{
input: []byte(`host:123 {
#comment
directive
# comment
foobar # another comment
}`),
expected: []Token{
{Line: 1, Text: "host:123"},
{Line: 1, Text: "{"},
{Line: 3, Text: "directive"},
{Line: 5, Text: "foobar"},
{Line: 6, Text: "}"},
},
},
{
input: []byte(`host:123 {
# hash inside string is not a comment
redir / /some/#/path
}`),
expected: []Token{
{Line: 1, Text: "host:123"},
{Line: 1, Text: "{"},
{Line: 3, Text: "redir"},
{Line: 3, Text: "/"},
{Line: 3, Text: "/some/#/path"},
{Line: 4, Text: "}"},
},
},
{
input: []byte("# comment at beginning of file\n# comment at beginning of line\nhost:123"),
expected: []Token{
{Line: 3, Text: "host:123"},
},
},
{
input: []byte(`a "quoted value" b
foobar`),
expected: []Token{
{Line: 1, Text: "a"},
{Line: 1, Text: "quoted value"},
{Line: 1, Text: "b"},
{Line: 2, Text: "foobar"},
},
},
{
input: []byte(`A "quoted \"value\" inside" B`),
expected: []Token{
{Line: 1, Text: "A"},
{Line: 1, Text: `quoted "value" inside`},
{Line: 1, Text: "B"},
},
},
{
input: []byte("An escaped \"newline\\\ninside\" quotes"),
expected: []Token{
{Line: 1, Text: "An"},
{Line: 1, Text: "escaped"},
{Line: 1, Text: "newline\\\ninside"},
{Line: 2, Text: "quotes"},
},
},
{
input: []byte("An escaped newline\\\noutside quotes"),
expected: []Token{
{Line: 1, Text: "An"},
{Line: 1, Text: "escaped"},
{Line: 1, Text: "newline"},
{Line: 1, Text: "outside"},
{Line: 1, Text: "quotes"},
},
},
{
input: []byte("line1\\\nescaped\nline2\nline3"),
expected: []Token{
{Line: 1, Text: "line1"},
{Line: 1, Text: "escaped"},
{Line: 3, Text: "line2"},
{Line: 4, Text: "line3"},
},
},
{
input: []byte("line1\\\nescaped1\\\nescaped2\nline4\nline5"),
expected: []Token{
{Line: 1, Text: "line1"},
{Line: 1, Text: "escaped1"},
{Line: 1, Text: "escaped2"},
{Line: 4, Text: "line4"},
{Line: 5, Text: "line5"},
},
},
{
input: []byte(`"unescapable\ in quotes"`),
expected: []Token{
{Line: 1, Text: `unescapable\ in quotes`},
},
},
{
input: []byte(`"don't\escape"`),
expected: []Token{
{Line: 1, Text: `don't\escape`},
},
},
{
input: []byte(`"don't\\escape"`),
expected: []Token{
{Line: 1, Text: `don't\\escape`},
},
},
{
input: []byte(`un\escapable`),
expected: []Token{
{Line: 1, Text: `un\escapable`},
},
},
{
input: []byte(`A "quoted value with line
break inside" {
foobar
}`),
expected: []Token{
{Line: 1, Text: "A"},
{Line: 1, Text: "quoted value with line\n\t\t\t\t\tbreak inside"},
{Line: 2, Text: "{"},
{Line: 3, Text: "foobar"},
{Line: 4, Text: "}"},
},
},
{
input: []byte(`"C:\php\php-cgi.exe"`),
expected: []Token{
{Line: 1, Text: `C:\php\php-cgi.exe`},
},
},
{
input: []byte(`empty "" string`),
expected: []Token{
{Line: 1, Text: `empty`},
{Line: 1, Text: ``},
{Line: 1, Text: `string`},
},
},
{
input: []byte("skip those\r\nCR characters"),
expected: []Token{
{Line: 1, Text: "skip"},
{Line: 1, Text: "those"},
{Line: 2, Text: "CR"},
{Line: 2, Text: "characters"},
},
},
{
input: []byte("\xEF\xBB\xBF:8080"), // test with leading byte order mark
expected: []Token{
{Line: 1, Text: ":8080"},
},
},
{
input: []byte("simple `backtick quoted` string"),
expected: []Token{
{Line: 1, Text: `simple`},
{Line: 1, Text: `backtick quoted`},
{Line: 1, Text: `string`},
},
},
{
input: []byte("multiline `backtick\nquoted\n` string"),
expected: []Token{
{Line: 1, Text: `multiline`},
{Line: 1, Text: "backtick\nquoted\n"},
{Line: 3, Text: `string`},
},
},
{
input: []byte("nested `\"quotes inside\" backticks` string"),
expected: []Token{
{Line: 1, Text: `nested`},
{Line: 1, Text: `"quotes inside" backticks`},
{Line: 1, Text: `string`},
},
},
{
input: []byte("reverse-nested \"`backticks` inside\" quotes"),
expected: []Token{
{Line: 1, Text: `reverse-nested`},
{Line: 1, Text: "`backticks` inside"},
{Line: 1, Text: `quotes`},
},
},
}
for i, testCase := range testCases {
actual, err := Tokenize(testCase.input, "")
if err != nil {
t.Errorf("%v", err)
}
lexerCompare(t, i, testCase.expected, actual)
}
}
func lexerCompare(t *testing.T, n int, expected, actual []Token) {
if len(expected) != len(actual) {
t.Errorf("Test case %d: expected %d token(s) but got %d", n, len(expected), len(actual))
}
for i := 0; i < len(actual) && i < len(expected); i++ {
if actual[i].Line != expected[i].Line {
t.Errorf("Test case %d token %d ('%s'): expected line %d but was line %d",
n, i, expected[i].Text, expected[i].Line, actual[i].Line)
break
}
if actual[i].Text != expected[i].Text {
t.Errorf("Test case %d token %d: expected text '%s' but was '%s'",
n, i, expected[i].Text, actual[i].Text)
break
}
}
}