This is an automated email from the ASF dual-hosted git repository. tallison pushed a commit to branch TIKA-4559 in repository https://gitbox.apache.org/repos/asf/tika.git
commit 49810cd5661300d11b6bcaf309b1831d3d6fa779 Author: tallison <[email protected]> AuthorDate: Tue Dec 9 12:11:40 2025 -0500 TIKA-4559 -- refactor config structure for fetchers and emitters --- .../resources/configs/config-fetch-emit-only.json | 10 ++-- .../test/resources/configs/config-template.json | 13 +++-- .../src/test/resources/kafka/plugins-template.json | 8 +-- .../resources/opensearch/plugins-template.json | 8 +-- .../opensearch/tika-config-opensearch.json | 8 +-- .../src/test/resources/solr/plugins-template.json | 8 +-- .../src/test/resources/tika-config-solr-urls.json | 12 ++--- .../src/main/resources/config-template.json | 8 +-- .../test/resources/configs/config-template.json | 19 ++++--- .../tika/pipes/core/AbstractComponentManager.java | 63 ++++++++++++++-------- .../pipes/core/emitter/EmitterManagerTest.java | 34 +++++++----- .../pipes/core/fetcher/FetcherManagerTest.java | 34 +++++++----- .../resources/configs/tika-config-bad-class.json | 8 +-- .../configs/tika-config-bad-java-path.json | 8 +-- .../configs/tika-config-bad-jvm-args.json | 9 ++-- .../test/resources/configs/tika-config-basic.json | 8 +-- .../configs/tika-config-crashing-detector.json | 16 +++--- .../resources/configs/tika-config-emit-all.json | 12 +++-- .../resources/configs/tika-config-passback.json | 8 +-- .../configs/tika-config-timeout-lt-heartbeat.json | 13 +++-- .../resources/configs/tika-config-truncate.json | 8 +-- .../test/resources/configs/tika-config-http.json | 4 +- .../resources/configs/cxf-test-base-template.json | 14 ++--- .../tika-config-server-fetchers-emitters.json | 8 +-- .../resources/configs/tika-config-server-tls.json | 8 +-- .../test/resources/configs/tika-config-server.json | 8 +-- .../test/resources/configs/tika-pipes-config.json | 8 +-- .../resources/configs/cxf-test-base-template.json | 14 ++--- 28 files changed, 219 insertions(+), 160 deletions(-) diff --git a/tika-app/src/test/resources/configs/config-fetch-emit-only.json b/tika-app/src/test/resources/configs/config-fetch-emit-only.json index 7fb0553da..d7a3da37f 100644 --- a/tika-app/src/test/resources/configs/config-fetch-emit-only.json +++ b/tika-app/src/test/resources/configs/config-fetch-emit-only.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "jsn", "onExists": "EXCEPTION", @@ -18,4 +18,4 @@ } }, "plugin-roots": "PLUGIN_ROOTS" -} \ No newline at end of file +} diff --git a/tika-app/src/test/resources/configs/config-template.json b/tika-app/src/test/resources/configs/config-template.json index 783012094..d2a7c029a 100644 --- a/tika-app/src/test/resources/configs/config-template.json +++ b/tika-app/src/test/resources/configs/config-template.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "jsn", "onExists": "EXCEPTION", @@ -52,7 +52,10 @@ "maxFilesProcessedPerProcess": 10000, "staleFetcherTimeoutSeconds": 600, "staleFetcherDelaySeconds": 60, - "forkedJvmArgs": ["-Xmx1g", "-XX:+UseG1GC"], + "forkedJvmArgs": [ + "-Xmx1g", + "-XX:+UseG1GC" + ], "javaPath": "java" }, "plugin-roots": "PLUGIN_ROOTS" diff --git a/tika-integration-tests/tika-pipes-kafka-integration-tests/src/test/resources/kafka/plugins-template.json b/tika-integration-tests/tika-pipes-kafka-integration-tests/src/test/resources/kafka/plugins-template.json index 5cf6b03d1..c39590ec2 100644 --- a/tika-integration-tests/tika-pipes-kafka-integration-tests/src/test/resources/kafka/plugins-template.json +++ b/tika-integration-tests/tika-pipes-kafka-integration-tests/src/test/resources/kafka/plugins-template.json @@ -40,15 +40,15 @@ } ], "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH" } } }, "emitters": { - "kafka-emitter": { - "ke": { + "ke": { + "kafka-emitter": { "topic": "EMITTER_TOPIC", "bootstrapServers": "BOOTSTRAP_SERVERS", "acks": "all", diff --git a/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/plugins-template.json b/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/plugins-template.json index 781e2cfa8..eeac6c2b6 100644 --- a/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/plugins-template.json +++ b/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/plugins-template.json @@ -1,14 +1,14 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH" } } }, "emitters": { - "opensearch-emitter": { - "ose": { + "ose": { + "opensearch-emitter": { "openSearchUrl": "OPEN_SEARCH_URL", "updateStrategy": "UPDATE_STRATEGY", "attachmentStrategy": "ATTACHMENT_STRATEGY", diff --git a/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/tika-config-opensearch.json b/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/tika-config-opensearch.json index b15e7c437..5665c3419 100644 --- a/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/tika-config-opensearch.json +++ b/tika-integration-tests/tika-pipes-opensearch-integration-tests/src/test/resources/opensearch/tika-config-opensearch.json @@ -41,15 +41,15 @@ } ], "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH" } } }, "emitters": { - "opensearch-emitter": { - "ose": { + "ose": { + "opensearch-emitter": { "openSearchUrl": "OPEN_SEARCH_URL", "updateStrategy": "UPDATE_STRATEGY", "attachmentStrategy": "ATTACHMENT_STRATEGY", diff --git a/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/solr/plugins-template.json b/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/solr/plugins-template.json index 94f1ec99b..eed18d2c6 100644 --- a/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/solr/plugins-template.json +++ b/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/solr/plugins-template.json @@ -40,15 +40,15 @@ } ], "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH" } } }, "emitters": { - "solr-emitter": { - "se": { + "se": { + "solr-emitter": { "solrCollection": "testcol", "solrUrls": SOLR_URLS, "solrZkHosts": SOLR_ZK_HOSTS, diff --git a/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/tika-config-solr-urls.json b/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/tika-config-solr-urls.json index 134f77962..9b6aadf08 100644 --- a/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/tika-config-solr-urls.json +++ b/tika-integration-tests/tika-pipes-solr-integration-tests/src/test/resources/tika-config-solr-urls.json @@ -55,15 +55,15 @@ "timeoutMillis": 60000 }, "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "{PATH_TO_DOCS}" } } }, "emitters": { - "solr-emitter": { - "se": { + "se": { + "solr-emitter": { "solrUrls": "{SOLR_CONNECTION}", "updateStrategy": "{UPDATE_STRATEGY}", "solrCollection": "testcol", @@ -74,8 +74,8 @@ "socketTimeout": 60000 } }, - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "/path/to/extracts" } } diff --git a/tika-pipes/tika-async-cli/src/main/resources/config-template.json b/tika-pipes/tika-async-cli/src/main/resources/config-template.json index 678e12fd7..e295290dd 100644 --- a/tika-pipes/tika-async-cli/src/main/resources/config-template.json +++ b/tika-pipes/tika-async-cli/src/main/resources/config-template.json @@ -25,16 +25,16 @@ } ], "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" diff --git a/tika-pipes/tika-async-cli/src/test/resources/configs/config-template.json b/tika-pipes/tika-async-cli/src/test/resources/configs/config-template.json index 41db19628..41790ff9b 100644 --- a/tika-pipes/tika-async-cli/src/test/resources/configs/config-template.json +++ b/tika-pipes/tika-async-cli/src/test/resources/configs/config-template.json @@ -1,19 +1,21 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "JSON_EMITTER_BASE_PATH", "fileExtension": "", "onExists": "EXCEPTION" - }, + } + }, + "file-system-emitter": { "fse-bytes": { "basePath": "BYTES_EMITTER_BASE_PATH", "fileExtension": "", @@ -36,8 +38,11 @@ "maxFilesProcessedPerProcess": 10000, "staleFetcherTimeoutSeconds": 600, "staleFetcherDelaySeconds": 60, - "forkedJvmArgs": ["-Xmx1g", "-XX:+UseG1GC"], + "forkedJvmArgs": [ + "-Xmx1g", + "-XX:+UseG1GC" + ], "javaPath": "java" }, "plugin-roots": "PLUGIN_ROOTS" -} \ No newline at end of file +} diff --git a/tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/AbstractComponentManager.java b/tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/AbstractComponentManager.java index 7dab511e0..603248ea0 100644 --- a/tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/AbstractComponentManager.java +++ b/tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/AbstractComponentManager.java @@ -92,36 +92,55 @@ public abstract class AbstractComponentManager<T extends TikaExtension, Map<String, ExtensionConfig> configs = new HashMap<>(); if (configNode != null && !configNode.isNull()) { - // Outer loop: iterate over type names - Iterator<Map.Entry<String, JsonNode>> typeFields = configNode.fields(); - while (typeFields.hasNext()) { + // Outer loop: iterate over instance IDs + Iterator<Map.Entry<String, JsonNode>> instanceFields = configNode.fields(); + while (instanceFields.hasNext()) { + Map.Entry<String, JsonNode> instanceEntry = instanceFields.next(); + String instanceId = instanceEntry.getKey(); + JsonNode typeNode = instanceEntry.getValue(); + + // Check for duplicate IDs (should not happen due to JSON parsing, but validate) + if (configs.containsKey(instanceId)) { + throw new TikaConfigException("Duplicate " + getComponentName() + + " id: " + instanceId); + } + + // Inner loop: extract the type name + // The structure should be: { "type-name": { config } } + // We expect exactly ONE type per instance + Iterator<Map.Entry<String, JsonNode>> typeFields = typeNode.fields(); + + if (!typeFields.hasNext()) { + throw new TikaConfigException( + "Invalid " + getComponentName() + " configuration for id '" + instanceId + + "': missing type specification. Expected format: {\"" + instanceId + + "\": {\"type-name\": {...}}}"); + } + Map.Entry<String, JsonNode> typeEntry = typeFields.next(); String typeName = typeEntry.getKey(); - JsonNode instancesNode = typeEntry.getValue(); + JsonNode config = typeEntry.getValue(); - // Validate that factory exists + // Validate that there's only one type per instance + if (typeFields.hasNext()) { + Map.Entry<String, JsonNode> extraTypeEntry = typeFields.next(); + throw new TikaConfigException( + "Invalid " + getComponentName() + " configuration for id '" + instanceId + + "': multiple types specified ('" + typeName + "', '" + + extraTypeEntry.getKey() + "'). Each instance can only have one type."); + } + + // Validate that factory exists for this type F factory = factories.get(typeName); if (factory == null) { throw new TikaConfigException( - "Unknown " + getComponentName() + " type: " + typeName + - ". Available: " + factories.keySet()); + "Unknown " + getComponentName() + " type: '" + typeName + + "' for instance id '" + instanceId + "'. Available types: " + + factories.keySet()); } - // Inner loop: iterate over instances of this type - Iterator<Map.Entry<String, JsonNode>> instanceFields = instancesNode.fields(); - while (instanceFields.hasNext()) { - Map.Entry<String, JsonNode> instanceEntry = instanceFields.next(); - String instanceId = instanceEntry.getKey(); - JsonNode config = instanceEntry.getValue(); - - if (configs.containsKey(instanceId)) { - throw new TikaConfigException("Duplicate " + getComponentName() + - " id: " + instanceId); - } - - configs.put(instanceId, new ExtensionConfig(instanceId, typeName, - toJsonString(config))); - } + configs.put(instanceId, new ExtensionConfig(instanceId, typeName, + toJsonString(config))); } } diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/emitter/EmitterManagerTest.java b/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/emitter/EmitterManagerTest.java index 8d98865c9..5b7ce9ced 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/emitter/EmitterManagerTest.java +++ b/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/emitter/EmitterManagerTest.java @@ -95,12 +95,14 @@ public class EmitterManagerTest { String configJson = String.format(Locale.ROOT, """ { "emitters": { - "file-system-emitter": { - "fse1": { + "fse1": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" - }, - "fse2": { + } + }, + "fse2": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" } @@ -225,8 +227,8 @@ public class EmitterManagerTest { String configJson = """ { "emitters": { - "non-existent-emitter-type": { - "emitter1": { + "emitter1": { + "non-existent-emitter-type": { "someProp": "value" } } @@ -255,12 +257,14 @@ public class EmitterManagerTest { String configJson = String.format(Locale.ROOT, """ { "emitters": { - "file-system-emitter": { - "fse1": { + "fse1": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" - }, - "fse1": { + } + }, + "fse1": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" } @@ -304,12 +308,14 @@ public class EmitterManagerTest { String configJson = String.format(Locale.ROOT, """ { "emitters": { - "file-system-emitter": { - "fse1": { + "fse1": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" - }, - "fse2": { + } + }, + "fse2": { + "file-system-emitter": { "basePath": "%s", "onExists": "REPLACE" } diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/fetcher/FetcherManagerTest.java b/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/fetcher/FetcherManagerTest.java index 7e206257c..b52677547 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/fetcher/FetcherManagerTest.java +++ b/tika-pipes/tika-pipes-integration-tests/src/test/java/org/apache/tika/pipes/core/fetcher/FetcherManagerTest.java @@ -95,11 +95,13 @@ public class FetcherManagerTest { String configJson = String.format(Locale.ROOT, """ { "fetchers": { - "file-system-fetcher": { - "fsf1": { + "fsf1": { + "file-system-fetcher": { "basePath": "%s" - }, - "fsf2": { + } + }, + "fsf2": { + "file-system-fetcher": { "basePath": "%s" } } @@ -223,8 +225,8 @@ public class FetcherManagerTest { String configJson = """ { "fetchers": { - "non-existent-fetcher-type": { - "fetcher1": { + "fetcher1": { + "non-existent-fetcher-type": { "someProp": "value" } } @@ -253,11 +255,13 @@ public class FetcherManagerTest { String configJson = String.format(Locale.ROOT, """ { "fetchers": { - "file-system-fetcher": { - "fsf1": { + "fsf1": { + "file-system-fetcher": { "basePath": "%s" - }, - "fsf1": { + } + }, + "fsf1": { + "file-system-fetcher": { "basePath": "%s" } } @@ -300,11 +304,13 @@ public class FetcherManagerTest { String configJson = String.format(Locale.ROOT, """ { "fetchers": { - "file-system-fetcher": { - "fsf1": { + "fsf1": { + "file-system-fetcher": { "basePath": "%s" - }, - "fsf2": { + } + }, + "fsf2": { + "file-system-fetcher": { "basePath": "%s" } } diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-class.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-class.json index 51e22b0ad..780180804 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-class.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-class.json @@ -1,7 +1,7 @@ { "fetchers": { - "non-existent-fetcher-plugin": { - "bad-fetcher": { + "bad-fetcher": { + "non-existent-fetcher-plugin": { "basePath": "FETCHER_BASE_PATH" } } @@ -9,7 +9,9 @@ "pipes": { "numClients": 1, "timeoutMillis": 5000, - "forkedJvmArgs": ["-Xmx256m"], + "forkedJvmArgs": [ + "-Xmx256m" + ], "directEmitThresholdBytes": 1000000 }, "plugin-roots": "PLUGINS_PATHS" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-java-path.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-java-path.json index cf7f051c2..4035660ff 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-java-path.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-java-path.json @@ -1,7 +1,7 @@ { "fetchers": { - "non-existent-fetcher-plugin": { - "bad-fetcher": { + "bad-fetcher": { + "non-existent-fetcher-plugin": { "basePath": "FETCHER_BASE_PATH" } } @@ -10,7 +10,9 @@ "numClients": 1, "timeoutMillis": 5000, "javaPath": "thisIsntJava", - "forkedJvmArgs": ["-Xmx256m"], + "forkedJvmArgs": [ + "-Xmx256m" + ], "directEmitThresholdBytes": 1000000 }, "plugin-roots": "PLUGINS_PATHS" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-jvm-args.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-jvm-args.json index 5c63904cd..2faecfed5 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-jvm-args.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-bad-jvm-args.json @@ -1,7 +1,7 @@ { "fetchers": { - "non-existent-fetcher-plugin": { - "bad-fetcher": { + "bad-fetcher": { + "non-existent-fetcher-plugin": { "basePath": "FETCHER_BASE_PATH" } } @@ -9,7 +9,10 @@ "pipes": { "numClients": 1, "timeoutMillis": 5000, - "forkedJvmArgs": ["-Xmx256m", "-ThisIsntAThing"], + "forkedJvmArgs": [ + "-Xmx256m", + "-ThisIsntAThing" + ], "directEmitThresholdBytes": 1000000 }, "plugin-roots": "PLUGINS_PATHS" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-basic.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-basic.json index a8fbe4f10..6c1c454dd 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-basic.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-basic.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-crashing-detector.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-crashing-detector.json index 61fcd5032..7daa902ee 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-crashing-detector.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-crashing-detector.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" @@ -19,11 +19,15 @@ "pipes": { "numClients": 1, "timeoutMillis": 5000, - "forkedJvmArgs": ["-Xmx256m"], + "forkedJvmArgs": [ + "-Xmx256m" + ], "directEmitThresholdBytes": 1000000 }, "detectors": [ - {"org.apache.tika.pipes.core.CrashingDetector": {}} + { + "org.apache.tika.pipes.core.CrashingDetector": {} + } ], "plugin-roots": "PLUGINS_PATHS" } diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-emit-all.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-emit-all.json index 6389562eb..7d517f842 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-emit-all.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-emit-all.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" @@ -19,7 +19,9 @@ "pipes": { "numClients": 1, "timeoutMillis": 60000, - "forkedJvmArgs": ["-Xmx256m"], + "forkedJvmArgs": [ + "-Xmx256m" + ], "directEmitThresholdBytes": 0 }, "plugin-roots": "PLUGINS_PATHS" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-passback.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-passback.json index 3f81bdd19..9c48a2e5c 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-passback.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-passback.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-timeout-lt-heartbeat.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-timeout-lt-heartbeat.json index c40ff1407..1efc06129 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-timeout-lt-heartbeat.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-timeout-lt-heartbeat.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" @@ -21,7 +21,10 @@ "timeoutMillis": 60000, "socketTimeoutMs": 3000, "heartbeatIntervalMs": 10000, - "forkedJvmArgs": ["-Xmx256m", "-Dtika.pipes.allowInvalidHeartbeat=true"] + "forkedJvmArgs": [ + "-Xmx256m", + "-Dtika.pipes.allowInvalidHeartbeat=true" + ] }, "plugin-roots": "PLUGINS_PATHS" } diff --git a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-truncate.json b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-truncate.json index e13838e1a..0d86c4b14 100644 --- a/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-truncate.json +++ b/tika-pipes/tika-pipes-integration-tests/src/test/resources/configs/tika-config-truncate.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse": { + "fse": { + "file-system-emitter": { "basePath": "EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" diff --git a/tika-pipes/tika-pipes-plugins/tika-pipes-http/src/test/resources/configs/tika-config-http.json b/tika-pipes/tika-pipes-plugins/tika-pipes-http/src/test/resources/configs/tika-config-http.json index 8a6af535d..dd8ac7ad0 100644 --- a/tika-pipes/tika-pipes-plugins/tika-pipes-http/src/test/resources/configs/tika-config-http.json +++ b/tika-pipes/tika-pipes-plugins/tika-pipes-http/src/test/resources/configs/tika-config-http.json @@ -1,7 +1,7 @@ { "fetchers": { - "http-fetcher": { - "http-fetcher-1": { + "http-fetcher-1": { + "http-fetcher": { "httpHeaders": [ "Connection", "Expires", diff --git a/tika-server/tika-server-core/src/test/resources/configs/cxf-test-base-template.json b/tika-server/tika-server-core/src/test/resources/configs/cxf-test-base-template.json index 475441885..db24bf2be 100644 --- a/tika-server/tika-server-core/src/test/resources/configs/cxf-test-base-template.json +++ b/tika-server/tika-server-core/src/test/resources/configs/cxf-test-base-template.json @@ -1,20 +1,22 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "JSON_EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" - }, - "fse-bytes": { + } + }, + "fse-bytes": { + "file-system-emitter": { "basePath": "BYTES_EMITTER_BASE_PATH", "fileExtension": "", "onExists": "EXCEPTION" diff --git a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-fetchers-emitters.json b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-fetchers-emitters.json index 0ead781e4..baeacf06c 100644 --- a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-fetchers-emitters.json +++ b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-fetchers-emitters.json @@ -9,16 +9,16 @@ ] }, "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "BASE_PATH", "onExists": "EXCEPTION" } diff --git a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-tls.json b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-tls.json index 029e45e96..17c72590c 100644 --- a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-tls.json +++ b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server-tls.json @@ -19,16 +19,16 @@ } }, "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "BASE_PATH", "onExists": "EXCEPTION" } diff --git a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server.json b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server.json index 8c327e9fa..89a13e92e 100644 --- a/tika-server/tika-server-core/src/test/resources/configs/tika-config-server.json +++ b/tika-server/tika-server-core/src/test/resources/configs/tika-config-server.json @@ -10,16 +10,16 @@ ] }, "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "BASE_PATH", "onExists": "EXCEPTION" } diff --git a/tika-server/tika-server-core/src/test/resources/configs/tika-pipes-config.json b/tika-server/tika-server-core/src/test/resources/configs/tika-pipes-config.json index afe9f8efb..1417029ba 100644 --- a/tika-server/tika-server-core/src/test/resources/configs/tika-pipes-config.json +++ b/tika-server/tika-server-core/src/test/resources/configs/tika-pipes-config.json @@ -1,15 +1,15 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "BASE_PATH", "onExists": "EXCEPTION" } diff --git a/tika-server/tika-server-standard/src/test/resources/configs/cxf-test-base-template.json b/tika-server/tika-server-standard/src/test/resources/configs/cxf-test-base-template.json index 475441885..db24bf2be 100644 --- a/tika-server/tika-server-standard/src/test/resources/configs/cxf-test-base-template.json +++ b/tika-server/tika-server-standard/src/test/resources/configs/cxf-test-base-template.json @@ -1,20 +1,22 @@ { "fetchers": { - "file-system-fetcher": { - "fsf": { + "fsf": { + "file-system-fetcher": { "basePath": "FETCHER_BASE_PATH", "extractFileSystemMetadata": false } } }, "emitters": { - "file-system-emitter": { - "fse-json": { + "fse-json": { + "file-system-emitter": { "basePath": "JSON_EMITTER_BASE_PATH", "fileExtension": "json", "onExists": "EXCEPTION" - }, - "fse-bytes": { + } + }, + "fse-bytes": { + "file-system-emitter": { "basePath": "BYTES_EMITTER_BASE_PATH", "fileExtension": "", "onExists": "EXCEPTION"
