|
| 1 | +package com.metaobjects.render.prompt; |
| 2 | + |
| 3 | +import com.fasterxml.jackson.databind.JsonNode; |
| 4 | +import com.fasterxml.jackson.databind.ObjectMapper; |
| 5 | +import com.metaobjects.render.recover.FieldKind; |
| 6 | +import com.metaobjects.render.recover.FieldRecovery; |
| 7 | +import com.metaobjects.render.recover.FieldSpec; |
| 8 | +import com.metaobjects.render.recover.Format; |
| 9 | +import com.metaobjects.render.recover.Recover; |
| 10 | +import com.metaobjects.render.recover.RecoverOptions; |
| 11 | +import com.metaobjects.render.recover.RecoverOutcome; |
| 12 | +import com.metaobjects.render.recover.RecoverSchema; |
| 13 | +import org.junit.BeforeClass; |
| 14 | +import org.junit.Test; |
| 15 | +import org.junit.runner.RunWith; |
| 16 | +import org.junit.runners.Parameterized; |
| 17 | + |
| 18 | +import java.io.IOException; |
| 19 | +import java.nio.charset.StandardCharsets; |
| 20 | +import java.nio.file.Files; |
| 21 | +import java.nio.file.Path; |
| 22 | +import java.nio.file.Paths; |
| 23 | +import java.util.ArrayList; |
| 24 | +import java.util.List; |
| 25 | +import java.util.Map; |
| 26 | +import java.util.stream.Collectors; |
| 27 | +import java.util.stream.Stream; |
| 28 | + |
| 29 | +import static org.junit.Assert.assertEquals; |
| 30 | +import static org.junit.Assert.assertFalse; |
| 31 | + |
| 32 | +/** |
| 33 | + * Cross-port byte-identity gate for the FR-010 output-format prompt fragment. |
| 34 | + * |
| 35 | + * <p>Reproduces every reference {@code expected.<style>.txt} in the shared corpus at |
| 36 | + * {@code fixtures/output-prompt-conformance/} using Java's {@link OutputFormatRenderer}, |
| 37 | + * proving the Java renderer is byte-for-byte identical to the TS reference (and the C# port). |
| 38 | + * For {@code roundTrip} cases, also asserts the emitted example reads back cleanly through |
| 39 | + * {@link Recover} (no MALFORMED / LOST_* states).</p> |
| 40 | + * |
| 41 | + * <p>The corpus is the oracle: assertions are ordinal {@link String} equality on raw UTF-8 |
| 42 | + * bytes with no newline translation.</p> |
| 43 | + */ |
| 44 | +@RunWith(Parameterized.class) |
| 45 | +public class OutputPromptConformanceTest { |
| 46 | + |
| 47 | + private static final ObjectMapper JSON = new ObjectMapper(); |
| 48 | + |
| 49 | + /** Count guard: a port silently skipping cases must fail. Bump when adding cases. */ |
| 50 | + private static final int EXPECTED_CASE_COUNT = 10; |
| 51 | + |
| 52 | + private static final Path CORPUS; |
| 53 | + static { |
| 54 | + Path p = Paths.get(System.getProperty("user.dir")).toAbsolutePath(); |
| 55 | + while (p != null && !Files.exists(p.resolve("fixtures/output-prompt-conformance"))) p = p.getParent(); |
| 56 | + CORPUS = p == null ? null : p.resolve("fixtures/output-prompt-conformance"); |
| 57 | + } |
| 58 | + |
| 59 | + private record StyleSpec(String key, PromptStyle style) {} |
| 60 | + |
| 61 | + private static final List<StyleSpec> STYLES = List.of( |
| 62 | + new StyleSpec("guide", PromptStyle.GUIDE), |
| 63 | + new StyleSpec("inline", PromptStyle.INLINE), |
| 64 | + new StyleSpec("exampleOnly", PromptStyle.EXAMPLE_ONLY)); |
| 65 | + |
| 66 | + private final Path dir; |
| 67 | + public OutputPromptConformanceTest(String name, Path dir) { this.dir = dir; } |
| 68 | + |
| 69 | + @Parameterized.Parameters(name = "{0}") |
| 70 | + public static List<Object[]> cases() throws IOException { |
| 71 | + if (CORPUS == null || !Files.isDirectory(CORPUS)) return List.of(); |
| 72 | + try (Stream<Path> s = Files.list(CORPUS)) { |
| 73 | + return s.filter(Files::isDirectory) |
| 74 | + .filter(d -> Files.exists(d.resolve("spec.json"))) |
| 75 | + .sorted() |
| 76 | + .map(d -> new Object[]{ d.getFileName().toString(), d }) |
| 77 | + .collect(Collectors.toList()); |
| 78 | + } |
| 79 | + } |
| 80 | + |
| 81 | + @BeforeClass |
| 82 | + public static void countGuard() throws IOException { |
| 83 | + assertEquals("output-prompt-conformance case count", EXPECTED_CASE_COUNT, cases().size()); |
| 84 | + } |
| 85 | + |
| 86 | + @Test |
| 87 | + public void reproducesCorpusBytes() throws IOException { |
| 88 | + JsonNode spec = JSON.readTree(dir.resolve("spec.json").toFile()); |
| 89 | + OutputFormatSpec ofs = buildOutputSpec(spec); |
| 90 | + |
| 91 | + for (StyleSpec style : STYLES) { |
| 92 | + PromptOverrides overrides = new PromptOverrides(style.style(), Map.of(), Map.of()); |
| 93 | + String actual = OutputFormatRenderer.render(ofs, overrides); |
| 94 | + String expected = readUtf8(dir.resolve("expected." + style.key() + ".txt")); |
| 95 | + assertEquals(dir + " style[" + style.key() + "]", expected, actual); |
| 96 | + // determinism: identical across runs |
| 97 | + assertEquals(dir + " style[" + style.key() + "] (determinism)", |
| 98 | + actual, OutputFormatRenderer.render(ofs, overrides)); |
| 99 | + } |
| 100 | + |
| 101 | + if (spec.has("roundTrip") && spec.get("roundTrip").asBoolean()) { |
| 102 | + String exampleFragment = readUtf8(dir.resolve("expected.exampleOnly.txt")); |
| 103 | + RecoverOutcome outcome = Recover.recover(exampleFragment, buildRecoverSchema(spec), RecoverOptions.defaults()); |
| 104 | + // Skew guard: a field the renderer emitted must read back cleanly. Assert NO state |
| 105 | + // is MALFORMED or LOST_* (robust to DEFAULTED and to how a nested OBJECT-container |
| 106 | + // path classifies); any such state means the renderer emitted an example the recover |
| 107 | + // parser cannot read. |
| 108 | + for (Map.Entry<String, FieldRecovery> e : outcome.report().states().entrySet()) { |
| 109 | + FieldRecovery state = e.getValue(); |
| 110 | + boolean bad = state == FieldRecovery.MALFORMED |
| 111 | + || state == FieldRecovery.LOST_REQUIRED |
| 112 | + || state == FieldRecovery.LOST_OPTIONAL; |
| 113 | + assertFalse(dir + " roundTrip state[" + e.getKey() + "]=" + state, bad); |
| 114 | + } |
| 115 | + } |
| 116 | + } |
| 117 | + |
| 118 | + private static String readUtf8(Path p) throws IOException { |
| 119 | + return new String(Files.readAllBytes(p), StandardCharsets.UTF_8); |
| 120 | + } |
| 121 | + |
| 122 | + // ----- descriptor (spec.json) -> renderer/recover model ----- |
| 123 | + |
| 124 | + private static OutputFormatSpec buildOutputSpec(JsonNode c) { |
| 125 | + Format fmt = toFormat(c.get("format").asText()); |
| 126 | + String root = c.get("rootName").asText(); |
| 127 | + List<PromptField> fields = new ArrayList<>(); |
| 128 | + for (JsonNode f : c.get("fields")) fields.add(buildPromptField(f)); |
| 129 | + // style is a placeholder; each render overrides it per style. |
| 130 | + return new OutputFormatSpec(fmt, root, PromptStyle.GUIDE, fields); |
| 131 | + } |
| 132 | + |
| 133 | + private static PromptField buildPromptField(JsonNode f) { |
| 134 | + String name = f.get("name").asText(); |
| 135 | + FieldKind kind = FieldKind.valueOf(f.get("kind").asText()); |
| 136 | + boolean required = f.has("required") && f.get("required").asBoolean(); |
| 137 | + boolean array = f.has("array") && f.get("array").asBoolean(); |
| 138 | + List<String> enumValues = null; |
| 139 | + if (f.has("enumValues") && !f.get("enumValues").isNull()) { |
| 140 | + enumValues = new ArrayList<>(); |
| 141 | + for (JsonNode v : f.get("enumValues")) enumValues.add(v.asText()); |
| 142 | + } |
| 143 | + Map<String, String> enumDoc = null; |
| 144 | + if (f.has("enumDoc") && !f.get("enumDoc").isNull()) { |
| 145 | + Map<String, String> m = new java.util.LinkedHashMap<>(); |
| 146 | + f.get("enumDoc").fields().forEachRemaining(e -> m.put(e.getKey(), e.getValue().asText())); |
| 147 | + enumDoc = m; |
| 148 | + } |
| 149 | + String example = f.has("example") && !f.get("example").isNull() ? f.get("example").asText() : null; |
| 150 | + String instruction = f.has("instruction") && !f.get("instruction").isNull() ? f.get("instruction").asText() : null; |
| 151 | + OutputFormatSpec nested = f.has("nested") && !f.get("nested").isNull() ? buildOutputSpec(f.get("nested")) : null; |
| 152 | + return new PromptField(name, kind, required, array, enumValues, enumDoc, example, instruction, nested); |
| 153 | + } |
| 154 | + |
| 155 | + private static RecoverSchema buildRecoverSchema(JsonNode c) { |
| 156 | + Format fmt = toFormat(c.get("format").asText()); |
| 157 | + String root = c.get("rootName").asText(); |
| 158 | + List<FieldSpec> fields = new ArrayList<>(); |
| 159 | + for (JsonNode f : c.get("fields")) fields.add(buildFieldSpec(f)); |
| 160 | + return new RecoverSchema(fmt, root, fields); |
| 161 | + } |
| 162 | + |
| 163 | + private static FieldSpec buildFieldSpec(JsonNode f) { |
| 164 | + String name = f.get("name").asText(); |
| 165 | + FieldKind kind = FieldKind.valueOf(f.get("kind").asText()); |
| 166 | + boolean required = f.has("required") && f.get("required").asBoolean(); |
| 167 | + boolean array = f.has("array") && f.get("array").asBoolean(); |
| 168 | + if (kind == FieldKind.ENUM) { |
| 169 | + List<String> vals = null; |
| 170 | + if (f.has("enumValues") && !f.get("enumValues").isNull()) { |
| 171 | + vals = new ArrayList<>(); |
| 172 | + for (JsonNode v : f.get("enumValues")) vals.add(v.asText()); |
| 173 | + } |
| 174 | + return FieldSpec.enumField(name, required, vals, Map.of()); |
| 175 | + } |
| 176 | + if (kind == FieldKind.OBJECT) { |
| 177 | + RecoverSchema nested = f.has("nested") && !f.get("nested").isNull() ? buildRecoverSchema(f.get("nested")) : null; |
| 178 | + return FieldSpec.object(name, required, array, nested); |
| 179 | + } |
| 180 | + return FieldSpec.scalar(name, kind, required); |
| 181 | + } |
| 182 | + |
| 183 | + private static Format toFormat(String f) { |
| 184 | + return "json".equals(f) ? Format.JSON : Format.XML; |
| 185 | + } |
| 186 | +} |
0 commit comments