Search in sources :

Example 6 with ReadImportManifest

use of com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest in project DataflowTemplates by GoogleCloudPlatform.

the class TextImportTransformTest method readImportManifestColumnListMustBeProvidedForGeneratedColumn.

@Test
public void readImportManifestColumnListMustBeProvidedForGeneratedColumn() throws Exception {
    Path f31 = Files.createTempFile("table3-file", "1");
    Path manifestFile = Files.createTempFile("import-manifest", ".json");
    Charset charset = Charset.forName("UTF-8");
    try (BufferedWriter writer = Files.newBufferedWriter(manifestFile, charset)) {
        String jsonString = String.format("{\"tables\": [" + "{\"table_name\": \"table3\"," + "\"file_patterns\": [\"%s\"]}" + "]}", f31.toString());
        writer.write(jsonString, 0, jsonString.length());
    } catch (IOException e) {
        e.printStackTrace();
    }
    ValueProvider<String> importManifest = ValueProvider.StaticValueProvider.of(manifestFile.toString());
    PCollectionView<Ddl> ddlView = pipeline.apply("ddl", Create.of(getTestDdl())).apply(View.asSingleton());
    PCollection<KV<String, String>> tableAndFiles = pipeline.apply("Read manifest file", new ReadImportManifest(importManifest)).apply("Resolve data files", new ResolveDataFiles(importManifest, ddlView));
    try {
        pipeline.run();
    } catch (PipelineExecutionException e) {
        assertThat(e.getMessage(), equalTo("java.lang.RuntimeException: DB table table3 has one or more generated columns. An" + " explicit column list that excludes the generated columns must be provided in" + " the manifest."));
    }
}
Also used : Path(java.nio.file.Path) Charset(java.nio.charset.Charset) IOException(java.io.IOException) KV(org.apache.beam.sdk.values.KV) Ddl(com.google.cloud.teleport.spanner.ddl.Ddl) BufferedWriter(java.io.BufferedWriter) ResolveDataFiles(com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles) PipelineExecutionException(org.apache.beam.sdk.Pipeline.PipelineExecutionException) ReadImportManifest(com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest) Test(org.junit.Test)

Example 7 with ReadImportManifest

use of com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest in project DataflowTemplates by GoogleCloudPlatform.

the class TextImportTransformTest method readImportManifestPartialMatching.

@Test
public void readImportManifestPartialMatching() throws Exception {
    Path f11 = Files.createTempFile("table1-file", "1");
    Path f12 = Files.createTempFile("table1-file", "2");
    Path manifestFile = Files.createTempFile("import-manifest", ".json");
    Charset charset = Charset.forName("US-ASCII");
    try (BufferedWriter writer = Files.newBufferedWriter(manifestFile, charset)) {
        String jsonString = String.format("{\"tables\": [" + "{\"table_name\": \"table1\"," + "\"file_patterns\":[\"%s\",\"%s\"]}," + "{\"table_name\": \"table2\"," + "\"file_patterns\":[\"NOT_FOUND_FILE_PATTERN_\"]}" + "]}", f11.toString(), f12.toString());
        writer.write(jsonString, 0, jsonString.length());
    } catch (IOException e) {
        e.printStackTrace();
    }
    ValueProvider<String> importManifest = ValueProvider.StaticValueProvider.of(manifestFile.toString());
    PCollectionView<Ddl> ddlView = pipeline.apply("ddl", Create.of(getTestDdl())).apply(View.asSingleton());
    PCollection<KV<String, String>> tableAndFiles = pipeline.apply("Read manifest file", new ReadImportManifest(importManifest)).apply("Resolve data files", new ResolveDataFiles(importManifest, ddlView));
    PAssert.that(tableAndFiles).containsInAnyOrder(KV.of("table1", f11.toString()), KV.of("table1", f12.toString()));
    try {
        pipeline.run();
    } catch (Exception e) {
        e.printStackTrace();
    }
}
Also used : Path(java.nio.file.Path) Charset(java.nio.charset.Charset) IOException(java.io.IOException) KV(org.apache.beam.sdk.values.KV) Ddl(com.google.cloud.teleport.spanner.ddl.Ddl) PipelineExecutionException(org.apache.beam.sdk.Pipeline.PipelineExecutionException) IOException(java.io.IOException) BufferedWriter(java.io.BufferedWriter) ResolveDataFiles(com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles) ReadImportManifest(com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest) Test(org.junit.Test)

Example 8 with ReadImportManifest

use of com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest in project DataflowTemplates by GoogleCloudPlatform.

the class TextImportTransformTest method readPgImportManifestTypeMismatch.

@Test
public void readPgImportManifestTypeMismatch() throws Exception {
    Path f11 = Files.createTempFile("table1-file", "1");
    Path manifestFile = Files.createTempFile("import-manifest", ".json");
    Charset charset = Charset.forName("UTF-8");
    try (BufferedWriter writer = Files.newBufferedWriter(manifestFile, charset)) {
        String jsonString = String.format("{\"tables\": [" + "{\"table_name\": \"table1\"," + "\"file_patterns\": [\"%s\"]," + "\"columns\": [{\"column_name\": \"int_col\", \"type_name\": \"text\"}]}" + "]}", f11.toString());
        writer.write(jsonString, 0, jsonString.length());
    }
    ValueProvider<String> importManifest = ValueProvider.StaticValueProvider.of(manifestFile.toString());
    PCollectionView<Ddl> ddlView = pipeline.apply("ddl", Create.of(getPgTestDdl())).apply(View.asSingleton());
    PCollection<KV<String, String>> tableAndFiles = pipeline.apply("Read manifest file", new ReadImportManifest(importManifest)).apply("Resolve data files", new ResolveDataFiles(importManifest, ddlView));
    PipelineExecutionException thrown = assertThrows(PipelineExecutionException.class, () -> pipeline.run());
    assertThat(thrown.getMessage(), equalTo("java.lang.RuntimeException: Mismatching type: Table table1 Column int_col [PG_INT8" + " from DB and text from manifest]"));
}
Also used : Path(java.nio.file.Path) ResolveDataFiles(com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles) PipelineExecutionException(org.apache.beam.sdk.Pipeline.PipelineExecutionException) ReadImportManifest(com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest) Charset(java.nio.charset.Charset) KV(org.apache.beam.sdk.values.KV) Ddl(com.google.cloud.teleport.spanner.ddl.Ddl) BufferedWriter(java.io.BufferedWriter) Test(org.junit.Test)

Example 9 with ReadImportManifest

use of com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest in project DataflowTemplates by GoogleCloudPlatform.

the class TextImportTransformTest method readPgImportManifestTypeMustMatch.

@Test
public void readPgImportManifestTypeMustMatch() throws Exception {
    Path f11 = Files.createTempFile("table1-file", "1");
    Path manifestFile = Files.createTempFile("import-manifest", ".json");
    Charset charset = Charset.forName("UTF-8");
    try (BufferedWriter writer = Files.newBufferedWriter(manifestFile, charset)) {
        String jsonString = String.format("{\"tables\": [{\"table_name\": \"table1\",\"file_patterns\": [\"%s\"],\"columns\":" + " [{\"column_name\": \"int_col\", \"type_name\": \"bigint\"}," + " {\"column_name\":\"str_10_col\", \"type_name\": \"character varying(10)\"}," + " {\"column_name\":\"float_col\", \"type_name\": \"double precision\"}," + " {\"column_name\":\"bool_col\", \"type_name\": \"boolean\"}," + " {\"column_name\": \"byte_col\", \"type_name\": \"bytea\"}," + " {\"column_name\": \"timestamp_col\"," + " \"type_name\":\"timestamp with time zone\"}," + " {\"column_name\": \"numeric_col\", \"type_name\": \"numeric\"}," + " {\"column_name\": \"date_col\", \"type_name\": \"date\"}]}]}", f11.toString());
        writer.write(jsonString, 0, jsonString.length());
    }
    ValueProvider<String> importManifest = ValueProvider.StaticValueProvider.of(manifestFile.toString());
    PCollectionView<Ddl> ddlView = pipeline.apply("ddl", Create.of(getPgTestDdl())).apply(View.asSingleton());
    PCollection<KV<String, String>> tableAndFiles = pipeline.apply("Read manifest file", new ReadImportManifest(importManifest)).apply("Resolve data files", new ResolveDataFiles(importManifest, ddlView));
    pipeline.run();
}
Also used : Path(java.nio.file.Path) ResolveDataFiles(com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles) ReadImportManifest(com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest) Charset(java.nio.charset.Charset) KV(org.apache.beam.sdk.values.KV) Ddl(com.google.cloud.teleport.spanner.ddl.Ddl) BufferedWriter(java.io.BufferedWriter) Test(org.junit.Test)

Example 10 with ReadImportManifest

use of com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest in project DataflowTemplates by GoogleCloudPlatform.

the class TextImportTransformTest method readImportManifestGeneratedColumnCannotBeImported.

@Test
public void readImportManifestGeneratedColumnCannotBeImported() throws Exception {
    Path f31 = Files.createTempFile("table3-file", "1");
    Path manifestFile = Files.createTempFile("import-manifest", ".json");
    Charset charset = Charset.forName("UTF-8");
    try (BufferedWriter writer = Files.newBufferedWriter(manifestFile, charset)) {
        String jsonString = String.format("{\"tables\": [" + "{\"table_name\": \"table3\"," + "\"file_patterns\": [\"%s\"]," + "\"columns\": [{\"column_name\": \"gen_col\", \"type_name\": \"INT64\"}]}" + "]}", f31.toString());
        writer.write(jsonString, 0, jsonString.length());
    } catch (IOException e) {
        e.printStackTrace();
    }
    ValueProvider<String> importManifest = ValueProvider.StaticValueProvider.of(manifestFile.toString());
    PCollectionView<Ddl> ddlView = pipeline.apply("ddl", Create.of(getTestDdl())).apply(View.asSingleton());
    PCollection<KV<String, String>> tableAndFiles = pipeline.apply("Read manifest file", new ReadImportManifest(importManifest)).apply("Resolve data files", new ResolveDataFiles(importManifest, ddlView));
    try {
        pipeline.run();
    } catch (PipelineExecutionException e) {
        assertThat(e.getMessage(), equalTo("java.lang.RuntimeException: Column gen_col in manifest is a generated column " + "in DB table table3. Generated columns cannot be imported."));
    }
}
Also used : Path(java.nio.file.Path) Charset(java.nio.charset.Charset) IOException(java.io.IOException) KV(org.apache.beam.sdk.values.KV) Ddl(com.google.cloud.teleport.spanner.ddl.Ddl) BufferedWriter(java.io.BufferedWriter) ResolveDataFiles(com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles) PipelineExecutionException(org.apache.beam.sdk.Pipeline.PipelineExecutionException) ReadImportManifest(com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest) Test(org.junit.Test)

Aggregations

ReadImportManifest (com.google.cloud.teleport.spanner.TextImportTransform.ReadImportManifest)10 ResolveDataFiles (com.google.cloud.teleport.spanner.TextImportTransform.ResolveDataFiles)10 Ddl (com.google.cloud.teleport.spanner.ddl.Ddl)10 BufferedWriter (java.io.BufferedWriter)10 Charset (java.nio.charset.Charset)10 Path (java.nio.file.Path)10 KV (org.apache.beam.sdk.values.KV)10 Test (org.junit.Test)10 IOException (java.io.IOException)8 PipelineExecutionException (org.apache.beam.sdk.Pipeline.PipelineExecutionException)5