[
https://issues.apache.org/jira/browse/TIKA-4975?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18128530#comment-18128530
]
ASF GitHub Bot commented on TIKA-4975:
--------------------------------------
Copilot commented on code in PR #3353:
URL: https://github.com/apache/tika/pull/3353#discussion_r4308409878
##########
tika-detectors/tika-detector-magika/pom.xml:
##########
@@ -51,7 +51,7 @@
<groupId>${project.groupId}</groupId>
<artifactId>tika-serialization</artifactId>
<version>${project.version}</version>
- <scope>test</scope>
+ <scope>provided</scope>
Review Comment:
`MagikaDetector` now invokes `TikaObjectMapperFactory` during static
initialization, including for default/programmatic construction. `provided`
does not supply serialization to downstream applications, so consumers using
this detector and `tika-core` without separately adding serialization fail with
`NoClassDefFoundError` before detection starts. Use compile scope.
##########
tika-detectors/tika-detector-siegfried/pom.xml:
##########
@@ -39,7 +39,7 @@
<groupId>${project.groupId}</groupId>
<artifactId>tika-serialization</artifactId>
<version>${project.version}</version>
- <scope>test</scope>
+ <scope>provided</scope>
Review Comment:
`SiegfriedDetector` now invokes `TikaObjectMapperFactory` during static
initialization, including for default/programmatic construction. `provided`
does not supply serialization to downstream applications, so consumers using
this detector and `tika-core` without separately adding serialization fail with
`NoClassDefFoundError` before detection starts. Use compile scope.
##########
tika-grpc/src/main/java/org/apache/tika/pipes/grpc/TikaGrpcServerImpl.java:
##########
@@ -83,9 +83,8 @@
class TikaGrpcServerImpl extends TikaGrpc.TikaImplBase {
private static final Logger LOG =
LoggerFactory.getLogger(TikaGrpcServerImpl.class);
- private static final ObjectMapper OBJECT_MAPPER = JsonMapper.builder()
-
.defaultPropertyInclusion(JsonInclude.Value.construct(JsonInclude.Include.NON_NULL,
JsonInclude.Include.NON_NULL))
- .build();
+ private static final ObjectMapper OBJECT_MAPPER =
TikaObjectMapperFactory.createPlainMapper()
Review Comment:
`buildRequestParseContext` uses this mapper to read `parse_context_json`.
The plain mapper silently collapses duplicate configuration keys during
`readTree`, before `resolveAll` can validate them. Use a strict mapper for that
initial read so gRPC requests follow the new duplicate-key policy, and add a
regression test for this entry point.
##########
tika-langdetect/tika-langdetect-lingo24/pom.xml:
##########
@@ -46,6 +46,12 @@
<groupId>com.fasterxml.jackson.core</groupId>
<artifactId>jackson-databind</artifactId>
</dependency>
+ <dependency>
+ <groupId>${project.groupId}</groupId>
+ <artifactId>tika-serialization</artifactId>
+ <version>${project.version}</version>
+ <scope>provided</scope>
Review Comment:
`Lingo24LangDetector` now calls `TikaObjectMapperFactory` to read service
responses, even without JSON component configuration. `provided` leaves this
dependency out of downstream runtime classpaths, so consumers using this module
and `tika-core` without separately adding serialization encounter
`NoClassDefFoundError`. Use compile scope.
##########
tika-server/tika-server-core/src/main/java/org/apache/tika/server/core/resource/TikaResource.java:
##########
@@ -246,7 +247,7 @@ public static String detectFilename(MultivaluedMap<String,
String> httpHeaders)
* @throws IOException if parsing fails
*/
public static void mergeParseContextFromConfig(String configJson,
ParseContext context) throws IOException, TikaConfigException {
- ObjectMapper mapper = new ObjectMapper();
+ ObjectMapper mapper = TikaObjectMapperFactory.createPlainMapper();
Review Comment:
This reads component configuration with a plain data mapper. Duplicate keys
in a multipart `config` part are silently collapsed by `readTree` before
`readParseContext` or the strict configuration readers can reject them,
contrary to the new duplicate-key policy. Use a strict mapper for this initial
read and add a request-level regression test.
##########
tika-parsers/tika-parsers-ml/tika-vlm/pom.xml:
##########
@@ -81,7 +81,7 @@
<groupId>${project.groupId}</groupId>
<artifactId>tika-serialization</artifactId>
<version>${project.version}</version>
- <scope>test</scope>
+ <scope>provided</scope>
Review Comment:
All three VLM parsers now initialize static mappers through
`TikaObjectMapperFactory`, even when constructed with Java config objects. This
`provided` declaration overrides the ML parent's compile-scoped serialization
dependency, leaving standalone consumers without the required class and causing
`NoClassDefFoundError` during parser initialization. Use compile scope.
##########
tika-translate/pom.xml:
##########
@@ -41,6 +41,12 @@
<version>${project.version}</version>
<scope>provided</scope>
</dependency>
+ <dependency>
+ <groupId>${project.groupId}</groupId>
+ <artifactId>tika-serialization</artifactId>
+ <version>${project.version}</version>
+ <scope>provided</scope>
Review Comment:
The translators now call `TikaObjectMapperFactory` when reading ordinary API
responses. `provided` does not propagate this dependency to downstream
applications, so consumers using `tika-core` and `tika-translate` without
separately adding `tika-serialization` now encounter `NoClassDefFoundError`.
Use compile scope for this new runtime requirement.
##########
tika-langdetect/tika-langdetect-mitll-text/pom.xml:
##########
@@ -41,6 +41,12 @@
<groupId>com.fasterxml.jackson.core</groupId>
<artifactId>jackson-databind</artifactId>
</dependency>
+ <dependency>
+ <groupId>${project.groupId}</groupId>
+ <artifactId>tika-serialization</artifactId>
+ <version>${project.version}</version>
+ <scope>provided</scope>
Review Comment:
`TextLangDetector` now calls `TikaObjectMapperFactory` when reading language
lists and detection responses. With `provided` scope, consumers using this
module and `tika-core` without separately adding `tika-serialization` encounter
`NoClassDefFoundError` on those paths. Use compile scope to propagate the
required runtime dependency.
##########
tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/serialization/JsonFetchEmitTupleList.java:
##########
@@ -40,7 +41,7 @@ public class JsonFetchEmitTupleList {
/** Envelope key: the body must be {"tuples":[...]}; a bare array is
rejected. */
public static final String TUPLES = "tuples";
- private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();
+ private static final ObjectMapper OBJECT_MAPPER =
TikaObjectMapperFactory.createPlainMapper();
Review Comment:
`fromJson` reads the entire tuple envelope, including `parseContext`, into a
tree with this plain mapper. Duplicate configuration keys are already collapsed
when the restricted deserializer receives that tree, so later strict readers
cannot reject them. Use a strict mapper at the initial read and cover duplicate
keys inside a tuple's `parseContext`.
> Read component configuration JSON only through tika-serialization's hardened
> readers
> ------------------------------------------------------------------------------------
>
> Key: TIKA-4975
> URL: https://issues.apache.org/jira/browse/TIKA-4975
> Project: Tika
> Issue Type: Task
> Reporter: Tim Allison
> Priority: Major
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
