|
| 1 | +package org.springframework.ai.writer; |
| 2 | + |
| 3 | +import org.junit.jupiter.api.BeforeEach; |
| 4 | +import org.junit.jupiter.api.Test; |
| 5 | +import org.junit.jupiter.api.io.TempDir; |
| 6 | +import org.springframework.ai.document.Document; |
| 7 | +import org.springframework.ai.document.MetadataMode; |
| 8 | + |
| 9 | +import java.io.IOException; |
| 10 | +import java.nio.file.Files; |
| 11 | +import java.nio.file.Path; |
| 12 | +import java.util.List; |
| 13 | + |
| 14 | +import static org.junit.jupiter.api.Assertions.*; |
| 15 | + |
| 16 | +/** |
| 17 | + * @author Jemin Huh |
| 18 | + */ |
| 19 | +public class FileDocumentWriterTest { |
| 20 | + |
| 21 | + @TempDir |
| 22 | + Path tempDir; |
| 23 | + |
| 24 | + private String testFileName; |
| 25 | + |
| 26 | + private List<Document> testDocuments; |
| 27 | + |
| 28 | + @BeforeEach |
| 29 | + void setUp() { |
| 30 | + testFileName = tempDir.resolve("file-document-test-output.txt").toString(); |
| 31 | + testDocuments = List.of( |
| 32 | + Document.builder() |
| 33 | + .text("Document one introduces the core functionality of Spring AI.") |
| 34 | + .metadata("page_number", "1") |
| 35 | + .metadata("end_page_number", "2") |
| 36 | + .metadata("source", "intro.pdf") |
| 37 | + .metadata("title", "Spring AI Overview") |
| 38 | + .metadata("author", "QA Team") |
| 39 | + .build(), |
| 40 | + Document.builder() |
| 41 | + .text("Document two illustrates multi-line handling and line breaks.\nEnsure preservation of formatting.") |
| 42 | + .metadata("page_number", "3") |
| 43 | + .metadata("end_page_number", "4") |
| 44 | + .metadata("source", "formatting.pdf") |
| 45 | + .build(), |
| 46 | + Document.builder() |
| 47 | + .text("Document three checks metadata inclusion and output formatting behavior.") |
| 48 | + .metadata("page_number", "5") |
| 49 | + .metadata("end_page_number", "6") |
| 50 | + .metadata("version", "v1.2") |
| 51 | + .build()); |
| 52 | + } |
| 53 | + |
| 54 | + @Test |
| 55 | + void testBasicWrite() throws IOException { |
| 56 | + var writer = new FileDocumentWriter(testFileName); |
| 57 | + writer.accept(testDocuments); |
| 58 | + |
| 59 | + List<String> lines = Files.readAllLines(Path.of(testFileName)); |
| 60 | + assertEquals("", lines.get(0)); |
| 61 | + assertEquals("", lines.get(1)); |
| 62 | + assertEquals("Document one introduces the core functionality of Spring AI.", lines.get(2)); |
| 63 | + assertEquals("", lines.get(3)); |
| 64 | + assertEquals("Document two illustrates multi-line handling and line breaks.", lines.get(4)); |
| 65 | + assertEquals("Ensure preservation of formatting.", lines.get(5)); |
| 66 | + assertEquals("", lines.get(6)); |
| 67 | + assertEquals("Document three checks metadata inclusion and output formatting behavior.", lines.get(7)); |
| 68 | + } |
| 69 | + |
| 70 | + @Test |
| 71 | + void testWriteWithDocumentMarkers() throws IOException { |
| 72 | + var writer = new FileDocumentWriter(testFileName, true, MetadataMode.NONE, false); |
| 73 | + writer.accept(testDocuments); |
| 74 | + |
| 75 | + List<String> lines = Files.readAllLines(Path.of(testFileName)); |
| 76 | + assertEquals("", lines.get(0)); |
| 77 | + assertEquals("### Doc: 0, pages:[1,2]", lines.get(1)); |
| 78 | + assertEquals("", lines.get(2)); |
| 79 | + assertEquals("", lines.get(3)); |
| 80 | + assertEquals("Document one introduces the core functionality of Spring AI.", lines.get(4)); |
| 81 | + assertEquals("### Doc: 1, pages:[3,4]", lines.get(5)); |
| 82 | + assertEquals("", lines.get(6)); |
| 83 | + assertEquals("", lines.get(7)); |
| 84 | + assertEquals("Document two illustrates multi-line handling and line breaks.", lines.get(8)); |
| 85 | + assertEquals("Ensure preservation of formatting.", lines.get(9)); |
| 86 | + assertEquals("### Doc: 2, pages:[5,6]", lines.get(10)); |
| 87 | + assertEquals("", lines.get(11)); |
| 88 | + assertEquals("", lines.get(12)); |
| 89 | + assertEquals("Document three checks metadata inclusion and output formatting behavior.", lines.get(13)); |
| 90 | + } |
| 91 | + |
| 92 | + @Test |
| 93 | + void testMetadataModeAllWithDocumentMarkers() throws IOException { |
| 94 | + var writer = new FileDocumentWriter(testFileName, true, MetadataMode.ALL, false); |
| 95 | + writer.accept(testDocuments); |
| 96 | + |
| 97 | + List<String> lines = Files.readAllLines(Path.of(testFileName)); |
| 98 | + assertEquals("", lines.get(0)); |
| 99 | + assertEquals("### Doc: 0, pages:[1,2]", lines.get(1)); |
| 100 | + String subListToString = lines.subList(2, 7).toString(); |
| 101 | + assertTrue(subListToString.contains("page_number: 1")); |
| 102 | + assertTrue(subListToString.contains("end_page_number: 2")); |
| 103 | + assertTrue(subListToString.contains("source: intro.pdf")); |
| 104 | + assertTrue(subListToString.contains("title: Spring AI Overview")); |
| 105 | + assertTrue(subListToString.contains("author: QA Team")); |
| 106 | + assertEquals("", lines.get(7)); |
| 107 | + assertEquals("Document one introduces the core functionality of Spring AI.", lines.get(8)); |
| 108 | + |
| 109 | + assertEquals("### Doc: 1, pages:[3,4]", lines.get(9)); |
| 110 | + subListToString = lines.subList(10, 13).toString(); |
| 111 | + assertTrue(subListToString.contains("page_number: 3")); |
| 112 | + assertTrue(subListToString.contains("source: formatting.pdf")); |
| 113 | + assertTrue(subListToString.contains("end_page_number: 4")); |
| 114 | + assertEquals("", lines.get(13)); |
| 115 | + assertEquals("Document two illustrates multi-line handling and line breaks.", lines.get(14)); |
| 116 | + assertEquals("Ensure preservation of formatting.", lines.get(15)); |
| 117 | + |
| 118 | + assertEquals("### Doc: 2, pages:[5,6]", lines.get(16)); |
| 119 | + subListToString = lines.subList(17, 20).toString(); |
| 120 | + assertTrue(subListToString.contains("page_number: 5")); |
| 121 | + assertTrue(subListToString.contains("end_page_number: 6")); |
| 122 | + assertTrue(subListToString.contains("version: v1.2")); |
| 123 | + assertEquals("", lines.get(20)); |
| 124 | + assertEquals("Document three checks metadata inclusion and output formatting behavior.", lines.get(21)); |
| 125 | + } |
| 126 | + |
| 127 | + @Test |
| 128 | + void testAppendWrite() throws IOException { |
| 129 | + Files.writeString(Path.of(testFileName), "Test String\n"); |
| 130 | + |
| 131 | + var writer = new FileDocumentWriter(testFileName, false, MetadataMode.NONE, true); |
| 132 | + writer.accept(testDocuments.subList(0, 2)); |
| 133 | + |
| 134 | + List<String> lines = Files.readAllLines(Path.of(testFileName)); |
| 135 | + assertEquals("Test String", lines.get(0)); |
| 136 | + assertEquals("", lines.get(1)); |
| 137 | + assertEquals("", lines.get(2)); |
| 138 | + assertEquals("Document one introduces the core functionality of Spring AI.", lines.get(3)); |
| 139 | + assertEquals("", lines.get(4)); |
| 140 | + assertEquals("Document two illustrates multi-line handling and line breaks.", lines.get(5)); |
| 141 | + assertEquals("Ensure preservation of formatting.", lines.get(6)); |
| 142 | + assertEquals(7, lines.size()); |
| 143 | + } |
| 144 | + |
| 145 | +} |
0 commit comments