TIKA-4809: Remove the writeLimit and throwOnWriteLimitReached headers

Author: tballison

tika-server/tika-server-core/src/main/java/org/apache/tika/server/core/resource/RecursiveMetadataResource.java

@@ -17,7 +17,6 @@
 package org.apache.tika.server.core.resource;
 
 import static org.apache.tika.server.core.resource.TikaResource.fillMetadata;
-import static org.apache.tika.server.core.resource.TikaResource.getWriteLimit;
 import static org.apache.tika.server.core.resource.TikaResource.setupContentHandlerFactory;
 import static org.apache.tika.server.core.resource.TikaResource.setupContentHandlerFactoryIfNeeded;
 
@@ -74,8 +73,7 @@ public List<Metadata> parseMetadata(TikaInputStream tis, Metadata metadata,
         TikaResource.logRequest(LOG, "/rmeta", metadata);
 
         // Set up handler factory in context using shared utility
-        setupContentHandlerFactory(context, handlerConfig.type().toString(), handlerConfig.writeLimit(),
-                handlerConfig.throwOnWriteLimitReached());
+        setupContentHandlerFactory(context, handlerConfig.type().toString());
 
         // Set up embedded limits if specified
         if (handlerConfig.maxEmbeddedCount() >= 0) {
@@ -96,8 +94,8 @@ static ServerHandlerConfig buildHandlerConfig(MultivaluedMap<String, String> htt
         } else if (httpHeaders.containsKey("maxEmbeddedCount")) {
             maxEmbeddedCount = Integer.parseInt(httpHeaders.getFirst("maxEmbeddedCount"));
         }
-        return new ServerHandlerConfig(BasicContentHandlerFactory.parseHandlerType(handlerTypeName, DEFAULT_HANDLER_TYPE), parseMode,
-                getWriteLimit(httpHeaders), maxEmbeddedCount, TikaResource.getThrowOnWriteLimitReached(httpHeaders));
+        return new ServerHandlerConfig(BasicContentHandlerFactory.parseHandlerType(handlerTypeName, DEFAULT_HANDLER_TYPE),
+                parseMode, maxEmbeddedCount);
     }
 
     /**
@@ -169,8 +167,7 @@ public Response getMetadataWithConfig(
     private MetadataList parseMetadataWithContext(TikaInputStream tis, Metadata metadata, MultivaluedMap<String, String> httpHeaders,
                                                   ServerHandlerConfig handlerConfig, ParseContext context) throws Exception {
         // Set up handler factory in context if not already set using shared utility
-        setupContentHandlerFactoryIfNeeded(context, handlerConfig.type().toString(),
-                handlerConfig.writeLimit(), handlerConfig.throwOnWriteLimitReached());
+        setupContentHandlerFactoryIfNeeded(context, handlerConfig.type().toString());
 
         // Filtering is done in child process, no need to filter again
         List<Metadata> metadataList = tikaResource.parseWithPipes(tis, metadata, context, ParseMode.RMETA);

tika-server/tika-server-core/src/main/java/org/apache/tika/server/core/resource/ServerHandlerConfig.java

@@ -30,8 +30,6 @@
 public record ServerHandlerConfig(
         BasicContentHandlerFactory.HANDLER_TYPE type,
         ParseMode parseMode,
-        int writeLimit,
-        int maxEmbeddedCount,
-        boolean throwOnWriteLimitReached
+        int maxEmbeddedCount
 ) {
 }

tika-server/tika-server-core/src/main/java/org/apache/tika/server/core/resource/TikaResource.java

@@ -407,65 +407,18 @@ public static void logRequest(Logger logger, String endpoint, Metadata metadata)
         }
     }
 
-    public static boolean getThrowOnWriteLimitReached(MultivaluedMap<String, String> httpHeaders) {
-        if (httpHeaders.containsKey("throwOnWriteLimitReached")) {
-            String val = httpHeaders.getFirst("throwOnWriteLimitReached");
-            if ("true".equalsIgnoreCase(val)) {
-                return true;
-            } else if ("false".equalsIgnoreCase(val)) {
-                return false;
-            } else {
-                throw new IllegalArgumentException("'throwOnWriteLimitReached' must be either 'true' or 'false'");
-            }
-        }
-        // Default: throw on write limit reached
-        return true;
-    }
-
-    /**
-     * Parses the writeLimit header value from HTTP headers.
-     *
-     * @param httpHeaders the HTTP headers
-     * @return the write limit value, or -1 if not specified
-     */
-    public static int getWriteLimit(MultivaluedMap<String, String> httpHeaders) {
-        if (httpHeaders.containsKey("writeLimit")) {
-            return Integer.parseInt(httpHeaders.getFirst("writeLimit"));
-        }
-        return -1;
-    }
-
     /**
-     * Sets up the ContentHandlerFactory in the ParseContext based on handler type and HTTP headers.
-     * This is a shared utility method used by both /tika and /rmeta endpoints.
+     * Sets up the ContentHandlerFactory in the ParseContext, taking the write limits from
+     * {@link org.apache.tika.config.OutputLimits} in the context.
      *
      * @param context the ParseContext to configure
      * @param handlerTypeName the handler type name (text, html, xml, ignore), may be null for default
-     * @param httpHeaders the HTTP headers containing writeLimit and throwOnWriteLimitReached
      */
-    public static void setupContentHandlerFactory(ParseContext context, String handlerTypeName,
-                                                   MultivaluedMap<String, String> httpHeaders) {
-        int writeLimit = getWriteLimit(httpHeaders);
-        boolean throwOnWriteLimitReached = getThrowOnWriteLimitReached(httpHeaders);
-        setupContentHandlerFactory(context, handlerTypeName, writeLimit, throwOnWriteLimitReached);
-    }
-
-    /**
-     * Sets up the ContentHandlerFactory in the ParseContext based on explicit parameters.
-     * This overload is used when the values have already been parsed (e.g., from ServerHandlerConfig).
-     *
-     * @param context the ParseContext to configure
-     * @param handlerTypeName the handler type name (text, html, xml, ignore), may be null for default
-     * @param writeLimit the write limit, or -1 for unlimited
-     * @param throwOnWriteLimitReached whether to throw when write limit is reached
-     */
-    public static void setupContentHandlerFactory(ParseContext context, String handlerTypeName,
-                                                   int writeLimit, boolean throwOnWriteLimitReached) {
+    public static void setupContentHandlerFactory(ParseContext context, String handlerTypeName) {
         BasicContentHandlerFactory.HANDLER_TYPE type = BasicContentHandlerFactory.parseHandlerType(
                 handlerTypeName, DEFAULT_HANDLER_TYPE);
-        ContentHandlerFactory factory = new BasicContentHandlerFactory(type, writeLimit,
-                throwOnWriteLimitReached, context);
-        context.set(ContentHandlerFactory.class, factory);
+        context.set(ContentHandlerFactory.class,
+                BasicContentHandlerFactory.newInstance(type, context));
     }
 
     /**
@@ -474,28 +427,10 @@ public static void setupContentHandlerFactory(ParseContext context, String handl
      *
      * @param context the ParseContext to configure
      * @param handlerTypeName the handler type name
-     * @param httpHeaders the HTTP headers
-     */
-    public static void setupContentHandlerFactoryIfNeeded(ParseContext context, String handlerTypeName,
-                                                           MultivaluedMap<String, String> httpHeaders) {
-        if (context.get(ContentHandlerFactory.class) == null) {
-            setupContentHandlerFactory(context, handlerTypeName, httpHeaders);
-        }
-    }
-
-    /**
-     * Sets up the ContentHandlerFactory in the ParseContext if not already set.
-     * This overload is used when the values have already been parsed.
-     *
-     * @param context the ParseContext to configure
-     * @param handlerTypeName the handler type name
-     * @param writeLimit the write limit, or -1 for unlimited
-     * @param throwOnWriteLimitReached whether to throw when write limit is reached
      */
-    public static void setupContentHandlerFactoryIfNeeded(ParseContext context, String handlerTypeName,
-                                                           int writeLimit, boolean throwOnWriteLimitReached) {
+    public static void setupContentHandlerFactoryIfNeeded(ParseContext context, String handlerTypeName) {
         if (context.get(ContentHandlerFactory.class) == null) {
-            setupContentHandlerFactory(context, handlerTypeName, writeLimit, throwOnWriteLimitReached);
+            setupContentHandlerFactory(context, handlerTypeName);
         }
     }
 
@@ -765,7 +700,7 @@ private Response produceRawOutput(TikaInputStream tis, Metadata metadata,
                                               String handlerTypeName) throws IOException {
         fillMetadata(null, metadata, httpHeaders);
         ParseContext context = createParseContext();
-        setupContentHandlerFactory(context, handlerTypeName, httpHeaders);
+        setupContentHandlerFactory(context, handlerTypeName);
         return produceRawOutputWithContext(tis, metadata, context, handlerTypeName);
     }
 
@@ -781,7 +716,7 @@ private Response produceRawOutputWithContext(TikaInputStream tis, Metadata metad
         logRequest(LOG, "/tika", metadata);
 
         // Ensure content handler factory is set (config may have set it)
-        setupContentHandlerFactoryIfNeeded(context, handlerTypeName, -1, true);
+        setupContentHandlerFactoryIfNeeded(context, handlerTypeName);
 
         LOG.debug("produceRawOutput: handlerType={}, contentHandlerFactory={}",
                 handlerTypeName, context.get(ContentHandlerFactory.class));
@@ -849,7 +784,7 @@ private Metadata produceJson(TikaInputStream tis, Metadata metadata,
                                   String handlerTypeName) throws IOException {
         fillMetadata(null, metadata, headers);
         ParseContext context = createParseContext();
-        setupContentHandlerFactory(context, handlerTypeName, headers);
+        setupContentHandlerFactory(context, handlerTypeName);
         return produceJsonWithContext(tis, metadata, context, handlerTypeName);
     }
 
@@ -871,7 +806,7 @@ private Metadata produceJsonWithContext(TikaInputStream tis, Metadata metadata,
         logRequest(LOG, "/tika", metadata);
 
         // Ensure content handler factory is set (config may have set it)
-        setupContentHandlerFactoryIfNeeded(context, handlerTypeName, -1, true);
+        setupContentHandlerFactoryIfNeeded(context, handlerTypeName);
 
         List<Metadata> metadataList;
         try {

tika-server/tika-server-core/src/test/java/org/apache/tika/server/core/RecursiveMetadataResourceTest.java

@@ -91,66 +91,5 @@ private String rmetaContent(String handlerSuffix) throws Exception {
         Reader reader = new InputStreamReader((InputStream) response.getEntity(), UTF_8);
         return JsonMetadataList.fromJson(reader).get(0).get(TikaCoreProperties.TIKA_CONTENT);
     }
-    /*
-    @Test
-    public void testWriteLimitInAll() throws Exception {
-        //specify your file directory here
-        Path testDocs = Paths.get("..../tika-parsers/src/test/resources/test-documents");
-        for (File f : testDocs.toFile().listFiles()) {
-            if (f.isDirectory()) {
-                continue;
-            }
-            testWriteLimit(f);
-        }
-    }
-    private void testWriteLimit(File f) throws Exception {
-        Response response = WebClient.create(endPoint + META_PATH+"/text").accept(
-                "application/json")
-                .put(f);
-        assertEquals(200, response.getStatus());
-        Reader reader = new InputStreamReader((InputStream) response.getEntity(), UTF_8);
-        List<Metadata> metadataList = JsonMetadataList.fromJson(reader);
-        int totalLen = 0;
-        StringBuilder sb = new StringBuilder();
-        for (Metadata m : metadataList) {
-            String txt = m.get(AbstractRecursiveParserWrapperHandler.TIKA_CONTENT);
-            sb.append(txt);
-            totalLen += (txt == null) ? 0 : txt.length();
-        }
-        String fullText = sb.toString();
-        Random r = new Random();
-        for (int i = 0; i < 20; i++) {
-            int writeLimit = r.nextInt(totalLen+100);
-            response = WebClient.create(endPoint + META_PATH+"/text").accept(
-                    "application/json")
-                    .header("writeLimit", Integer.toString(writeLimit)).put(f);
-            assertEquals(200, response.getStatus());
-            reader = new InputStreamReader((InputStream) response.getEntity(), UTF_8);
-            List<Metadata> writeLimitMetadataList = JsonMetadataList.fromJson(reader);
-            int len = 0;
-            StringBuilder extracted = new StringBuilder();
-            for (Metadata m : writeLimitMetadataList) {
-                String txt = m.get(AbstractRecursiveParserWrapperHandler.TIKA_CONTENT);
-                len += (txt == null) ? 0 : txt.length();
-                extracted.append(txt);
-            }
-            if (totalLen > len) {
-                boolean wlr = false;
-                for (Metadata m : writeLimitMetadataList) {
-                    if ("true".equals(m.get(AbstractRecursiveParserWrapperHandler.WRITE_LIMIT_REACHED))) {
-                        wlr = true;
-                    }
-                }
-                assertTrue(f.getName() + ": writelimit: " + writeLimit + " len: "+len,
-                        len <= writeLimit);
-                assertEquals(f.getName() +" writeLimit: " + writeLimit +
-                                " : fullLen:" + totalLen + " limitedLen: " +len,
-                        true, wlr);
-            } else if (len > totalLen) {
-                fail("len should never be > totalLen "+len + "  : "+ totalLen);
-            }
-        }
-    }
-    */
 
 }

tika-server/tika-server-core/src/test/java/org/apache/tika/server/core/TikaResourceTest.java

@@ -116,24 +116,6 @@ public void testJsonNPE() throws Exception {
         assertNotFound("null pointer message", metadata.get(TikaCoreProperties.CONTAINER_EXCEPTION));
     }
 
-    @Test
-    public void testJsonWriteLimit() throws Exception {
-        Response response = WebClient
-                .create(endPoint + TIKA_PATH + "/json")
-                .header("writeLimit", "100")
-                .put(ClassLoader.getSystemResourceAsStream(TEST_HELLO_WORLD_LONG));
-        Metadata metadata = JsonMetadata.fromJson(new InputStreamReader(((InputStream) response.getEntity()), StandardCharsets.UTF_8));
-
-        assertEquals("Nikolai Lobachevsky", metadata.get("author"));
-        assertEquals("application/mock+xml", metadata.get(Metadata.CONTENT_TYPE));
-        assertContains("Hello world", metadata.get(TikaCoreProperties.TIKA_CONTENT));
-        assertNotFound("dissolve", metadata.get(TikaCoreProperties.TIKA_CONTENT));
-        assertTrue(metadata
-                .get(TikaCoreProperties.CONTAINER_EXCEPTION)
-                .startsWith("org.apache.tika.exception.WriteLimitReachedException"));
-        assertEquals("true", metadata.get(TikaCoreProperties.WRITE_LIMIT_REACHED));
-    }
-
     @Test
     public void testJsonHandlerType() throws Exception {
         // Default /tika/json uses text handler
@@ -161,18 +143,6 @@ public void testJsonHandlerType() throws Exception {
     }
 
     /*
-    @Test
-    public void testWriteLimitInAll() throws Exception {
-        //specify your file directory here
-        Path testDocs = Paths.get("..../tika-parsers/src/test/resources/test-documents");
-        for (File f : testDocs.toFile().listFiles()) {
-            if (f.isDirectory()) {
-                continue;
-            }
-            testWriteLimit(f);
-        }
-    }
-
     private void testWriteLimit(File f) throws Exception {
         Response response =
                 WebClient.create(endPoint + TIKA_PATH + "/text").accept("application/json").put(f);

tika-server/tika-server-core/src/test/java/org/apache/tika/server/core/benchmark/TikaServerBenchmark.java

@@ -339,7 +339,6 @@ private boolean verifyMockParserInUse() {
                     .uri(URI.create(baseUrl + "/rmeta"))
                     .header("Content-Type", "application/mock+xml")
                     .header("Accept", "application/json")
-                    .header("writeLimit", "-1")
                     .PUT(HttpRequest.BodyPublishers.ofString(testXml))
                     .timeout(Duration.ofSeconds(10))
                     .build();