| 1 | package builtin |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "encoding/json" |
| 7 | "os" |
| 8 | "path/filepath" |
| 9 | "strings" |
| 10 | "testing" |
| 11 | "unicode/utf8" |
| 12 | |
| 13 | "golang.org/x/text/encoding/simplifiedchinese" |
| 14 | "golang.org/x/text/transform" |
| 15 | |
| 16 | "reasonix/internal/tool" |
| 17 | ) |
| 18 | |
| 19 | func gbkBytes(t *testing.T, s string) []byte { |
| 20 | t.Helper() |
| 21 | b, _, err := transform.Bytes(simplifiedchinese.GB18030.NewEncoder(), []byte(s)) |
| 22 | if err != nil { |
| 23 | t.Fatalf("encode GBK: %v", err) |
| 24 | } |
| 25 | return b |
| 26 | } |
| 27 | |
| 28 | func TestE2EGBKRoundTrip(t *testing.T) { |
| 29 | path := filepath.Join(t.TempDir(), "test_gbk.txt") |
| 30 | if err := os.WriteFile(path, gbkBytes(t, "你好世界\n这是第二行\n包含函数的测试\n"), 0o644); err != nil { |
| 31 | t.Fatal(err) |
| 32 | } |
| 33 | |
| 34 | readTL, _ := tool.LookupBuiltin("read_file") |
| 35 | editTL, _ := tool.LookupBuiltin("edit_file") |
| 36 | grepTL, _ := tool.LookupBuiltin("grep") |
| 37 | |
| 38 | args := func(m map[string]any) json.RawMessage { |
| 39 | b, _ := json.Marshal(m) |
| 40 | return json.RawMessage(b) |
| 41 | } |
| 42 | |
| 43 | out, err := readTL.Execute(context.Background(), args(map[string]any{"path": path})) |
| 44 | if err != nil { |
| 45 | t.Fatalf("read_file: %v", err) |
| 46 | } |
| 47 | if !strings.Contains(out, "你好世界") || !strings.Contains(out, "这是第二行") { |
| 48 | t.Errorf("read_file did not decode GBK to readable Chinese:\n%s", out) |
| 49 | } |
| 50 | |
| 51 | if raw, _ := os.ReadFile(path); utf8.Valid(raw) { |
| 52 | t.Error("read_file rewrote GBK file as UTF-8 on disk") |
| 53 | } |
| 54 | |
| 55 | if _, err := editTL.Execute(context.Background(), args(map[string]any{ |
| 56 | "path": path, |
| 57 | "old_string": "这是第二行", |
| 58 | "new_string": "这是新的行", |
| 59 | })); err != nil { |
| 60 | t.Fatalf("edit_file: %v", err) |
| 61 | } |
| 62 | |
| 63 | raw2, _ := os.ReadFile(path) |
| 64 | if utf8.Valid(raw2) { |
| 65 | t.Error("edit_file rewrote GBK file as UTF-8 on disk") |
| 66 | } |
| 67 | decoded, _, _ := transform.Bytes(simplifiedchinese.GB18030.NewDecoder(), raw2) |
| 68 | if s := string(decoded); !strings.Contains(s, "这是新的行") || strings.Contains(s, "这是第二行") { |
| 69 | t.Errorf("edit not applied to GBK file on disk: %q", s) |
| 70 | } |
| 71 | |
| 72 | out2, err := readTL.Execute(context.Background(), args(map[string]any{"path": path})) |
| 73 | if err != nil { |
| 74 | t.Fatalf("read_file after edit: %v", err) |
| 75 | } |
| 76 | if !strings.Contains(out2, "这是新的行") { |
| 77 | t.Errorf("read_file after edit missing new text:\n%s", out2) |
| 78 | } |
| 79 | |
| 80 | grepOut, err := grepTL.Execute(context.Background(), args(map[string]any{ |
| 81 | "pattern": "函数", |
| 82 | "path": path, |
| 83 | })) |
| 84 | if err != nil { |
| 85 | t.Fatalf("grep: %v", err) |
| 86 | } |
| 87 | if !strings.Contains(grepOut, "函数") { |
| 88 | t.Errorf("grep did not match decoded GBK content:\n%s", grepOut) |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | func e2eArgs(m map[string]any) json.RawMessage { |
| 93 | b, _ := json.Marshal(m) |
| 94 | return json.RawMessage(b) |
| 95 | } |
| 96 | |
| 97 | // write_file must preserve the encoding of a file it overwrites rather than |
| 98 | // always writing UTF-8, which would silently corrupt a GBK file. |
| 99 | func TestE2EWriteFilePreservesGBKOnOverwrite(t *testing.T) { |
| 100 | path := filepath.Join(t.TempDir(), "gbk.txt") |
| 101 | if err := os.WriteFile(path, gbkBytes(t, "原始内容\n"), 0o644); err != nil { |
| 102 | t.Fatal(err) |
| 103 | } |
| 104 | writeTL, _ := tool.LookupBuiltin("write_file") |
| 105 | if _, err := writeTL.Execute(context.Background(), e2eArgs(map[string]any{ |
| 106 | "path": path, |
| 107 | "content": "全新中文内容\n第二行\n", |
| 108 | })); err != nil { |
| 109 | t.Fatalf("write_file: %v", err) |
| 110 | } |
| 111 | raw, _ := os.ReadFile(path) |
| 112 | if utf8.Valid(raw) { |
| 113 | t.Error("write_file rewrote an existing GBK file as UTF-8 on disk") |
| 114 | } |
| 115 | decoded, _, _ := transform.Bytes(simplifiedchinese.GB18030.NewDecoder(), raw) |
| 116 | if s := string(decoded); !strings.Contains(s, "全新中文内容") || !strings.Contains(s, "第二行") { |
| 117 | t.Errorf("write_file content wrong after GBK round-trip: %q", s) |
| 118 | } |
| 119 | } |
| 120 | |
| 121 | // A newly created file (no existing encoding to preserve) defaults to UTF-8. |
| 122 | func TestE2EWriteFileNewFileIsUTF8(t *testing.T) { |
| 123 | path := filepath.Join(t.TempDir(), "new.txt") |
| 124 | writeTL, _ := tool.LookupBuiltin("write_file") |
| 125 | if _, err := writeTL.Execute(context.Background(), e2eArgs(map[string]any{ |
| 126 | "path": path, |
| 127 | "content": "新文件中文\n", |
| 128 | })); err != nil { |
| 129 | t.Fatalf("write_file: %v", err) |
| 130 | } |
| 131 | raw, _ := os.ReadFile(path) |
| 132 | if !utf8.Valid(raw) { |
| 133 | t.Error("a newly created file should be written as UTF-8") |
| 134 | } |
| 135 | if string(raw) != "新文件中文\n" { |
| 136 | t.Errorf("new file content = %q, want UTF-8 Chinese", raw) |
| 137 | } |
| 138 | } |
| 139 | |
| 140 | // delete_range must decode a GBK file before matching anchors (a UTF-8 anchor |
| 141 | // would never match the GBK bytes otherwise) and re-encode it on write. |
| 142 | func TestE2EDeleteRangePreservesGBK(t *testing.T) { |
| 143 | path := filepath.Join(t.TempDir(), "gbk.txt") |
| 144 | if err := os.WriteFile(path, gbkBytes(t, "第一行\n第二行\n第三行\n"), 0o644); err != nil { |
| 145 | t.Fatal(err) |
| 146 | } |
| 147 | delTL, _ := tool.LookupBuiltin("delete_range") |
| 148 | if _, err := delTL.Execute(context.Background(), e2eArgs(map[string]any{ |
| 149 | "path": path, |
| 150 | "start_anchor": "第二行", |
| 151 | "end_anchor": "第二行", |
| 152 | })); err != nil { |
| 153 | t.Fatalf("delete_range on GBK file: %v", err) |
| 154 | } |
| 155 | raw, _ := os.ReadFile(path) |
| 156 | if utf8.Valid(raw) { |
| 157 | t.Error("delete_range rewrote a GBK file as UTF-8 on disk") |
| 158 | } |
| 159 | decoded, _, _ := transform.Bytes(simplifiedchinese.GB18030.NewDecoder(), raw) |
| 160 | s := string(decoded) |
| 161 | if strings.Contains(s, "第二行") { |
| 162 | t.Errorf("delete_range did not remove the target line: %q", s) |
| 163 | } |
| 164 | if !strings.Contains(s, "第一行") || !strings.Contains(s, "第三行") { |
| 165 | t.Errorf("delete_range removed the wrong lines: %q", s) |
| 166 | } |
| 167 | } |
| 168 | |
| 169 | // A UTF-8 CJK file carrying one stray invalid byte is not GB18030. Decoding it |
| 170 | // as GB18030 turns the orphaned continuation bytes into U+FFFD, and an edit |
| 171 | // anywhere in the file then rewrites every one of them. |
| 172 | func TestE2EEditKeepsBytesOfUTF8FileWithStrayByte(t *testing.T) { |
| 173 | path := filepath.Join(t.TempDir(), "mixed.go") |
| 174 | before := []byte("// 中文注释：修改前的内容\nvalue := 1\n// 结尾\xff\n") |
| 175 | if err := os.WriteFile(path, before, 0o644); err != nil { |
| 176 | t.Fatal(err) |
| 177 | } |
| 178 | editTL, _ := tool.LookupBuiltin("edit_file") |
| 179 | if _, err := editTL.Execute(context.Background(), e2eArgs(map[string]any{ |
| 180 | "path": path, |
| 181 | "old_string": "value := 1", |
| 182 | "new_string": "value := 2", |
| 183 | })); err != nil { |
| 184 | t.Fatalf("edit_file: %v", err) |
| 185 | } |
| 186 | got, _ := os.ReadFile(path) |
| 187 | want := []byte("// 中文注释：修改前的内容\nvalue := 2\n// 结尾\xff\n") |
| 188 | if !bytes.Equal(got, want) { |
| 189 | t.Fatalf("edit_file changed bytes outside the edit:\n got % x\nwant % x", got, want) |
| 190 | } |
| 191 | } |
| 192 |