package lang import ( "os" "path/filepath" "testing" ) // TestALanguageAbsentFromTheRepositoryCutsFromDataAlone is the project's default review question, asked as a // test: "will a pair that is NOT in this repository work without editing Go?" // // Korean is the witness because it answers BOTH halves. It is a CJK-script language, so with no data file it // falls to the embedded default and gets that default's marker 第 — which never appears in a Korean book, so // nothing is cut. Drop a ko/structure.txt naming 제 and 장, change no Go, and the same text cuts. The data // file is the only difference between the two halves. func TestALanguageAbsentFromTheRepositoryCutsFromDataAlone(t *testing.T) { root := t.TempDir() // Half one: no file. The language resolves to the embedded CJK default, whose marker is not Korean. bare, err := LoadSourceStructure(root, "ko") if err != nil { t.Fatal(err) } if bare == nil { t.Fatal("a CJK-script language must still resolve to the embedded default") } if bare.Units['장'] { t.Fatal("the embedded default must not already know the Korean section unit — the test would prove nothing") } // Half two: the language ships its own grammar. No Go changes, no new package, no registry entry. if err := os.MkdirAll(filepath.Join(root, "ko"), 0o755); err != nil { t.Fatal(err) } data := "# Korean chapter grammar\nmarker\t제\nunits\t장\n" if err := os.WriteFile(filepath.Join(root, "ko", structureFile), []byte(data), 0o644); err != nil { t.Fatal(err) } st, err := LoadSourceStructure(root, "ko") if err != nil { t.Fatal(err) } if len(st.Markers) != 1 || st.Markers[0] != "제" { t.Fatalf("markers = %q, want [제]", st.Markers) } if !st.Units['장'] { t.Fatal("the declared section unit did not load") } if !st.HeaderNumeralRE().MatchString("제1장 시작") { t.Fatal("the compiled header pattern does not match the language's own header") } // The fingerprint must come from the FILE, not from the default: it is what re-cuts the book when the // grammar is edited. if st.Fingerprint() == bare.Fingerprint() { t.Fatal("a book cut by its own grammar must not share the default's cut-tag component") } }