rework inference & page options

Author: sebastianMindee

src/main/java/com/mindee/CommonClient.java

@@ -29,10 +29,6 @@ public final class InferenceParameters {
    * IDs of webhooks to propagate the API response to (may be empty).
    */
   private final List<String> webhookIds;
-  /**
-   * Page options to apply to the document.
-   */
-  private final PageOptions pageOptions;
   /*
    * Asynchronous polling options.
    */
@@ -57,7 +53,6 @@ public static final class Builder {
     private boolean rag = false;
     private String alias;
     private List<String> webhookIds = Collections.emptyList();
-    private PageOptions pageOptions = null;
     private AsyncPollingOptions pollingOptions = AsyncPollingOptions.builder().build();
 
     private Builder(String modelId) {
@@ -82,10 +77,6 @@ public Builder webhookIds(List<String> webhookIds) {
       return this;
     }
 
-    public Builder pageOptions(PageOptions pageOptions) {
-      this.pageOptions = pageOptions;
-      return this;
-    }
 
     public Builder pollingOptions(AsyncPollingOptions pollingOptions) {
       this.pollingOptions = pollingOptions;
@@ -99,7 +90,6 @@ public InferenceParameters build() {
           rag,
           alias,
           webhookIds,
-          pageOptions,
           pollingOptions
       );
     }

src/main/java/com/mindee/InferenceParameters.java

@@ -16,6 +16,7 @@
 import com.mindee.parsing.common.WorkflowResponse;
 import com.mindee.pdf.PdfBoxApi;
 import com.mindee.pdf.PdfOperation;
+import com.mindee.pdf.SplitQuery;
 import com.mindee.product.custom.CustomV1;
 import com.mindee.product.generated.GeneratedV1;
 import java.io.IOException;
@@ -24,8 +25,9 @@
 /**
  * Main entrypoint for Mindee operations.
  */
-public class MindeeClient extends CommonClient {
+public class MindeeClient {
 
+  protected PdfOperation pdfOperation;
   private final MindeeApi mindeeApi;
 
   /**
@@ -118,6 +120,28 @@ public <T extends Inference> AsyncPredictResponse<T> enqueue(
         null);
   }
 
+  /**
+   * Retrieves the file after applying page operations to it.
+   * @param localInputSource Local input source to apply operations to.
+   * @param pageOptions Options to apply.
+   * @return A byte array of the file after applying page operations.
+   * @throws IOException Throws if the file can't be accessed.
+   */
+  protected byte[] getSplitFile(
+      LocalInputSource localInputSource,
+      PageOptions pageOptions
+  ) throws IOException {
+    byte[] splitFile;
+    if (pageOptions == null || !localInputSource.isPdf()) {
+      splitFile = localInputSource.getFile();
+    } else {
+      splitFile = pdfOperation.split(
+          new SplitQuery(localInputSource.getFile(), pageOptions)
+      ).getFile();
+    }
+    return splitFile;
+  }
+
   /**
    * Send a local file to an async queue.
    * @param <T> Type of inference.

src/main/java/com/mindee/MindeeClient.java

@@ -9,34 +9,27 @@
 import com.mindee.parsing.v2.ErrorResponse;
 import com.mindee.parsing.v2.InferenceResponse;
 import com.mindee.parsing.v2.JobResponse;
-import com.mindee.pdf.PdfBoxApi;
-import com.mindee.pdf.PdfOperation;
 import java.io.IOException;
 
 /**
  * Entry point for the Mindee **V2** API features.
  */
-public class MindeeClientV2 extends CommonClient {
+public class MindeeClientV2 {
   private final MindeeApiV2 mindeeApi;
 
   /** Uses an API-key read from the environment variables. */
   public MindeeClientV2() {
-    this(new PdfBoxApi(), createDefaultApiV2(""));
+    this(createDefaultApiV2(""));
   }
 
   /** Uses the supplied API-key. */
   public MindeeClientV2(String apiKey) {
-    this(new PdfBoxApi(), createDefaultApiV2(apiKey));
+    this(createDefaultApiV2(apiKey));
   }
 
-  /** Directly inject an already configured {@link MindeeApiV2}. */
-  public MindeeClientV2(MindeeApiV2 mindeeApi) {
-    this(new PdfBoxApi(), mindeeApi);
-  }
 
   /** Inject both a PDF implementation and a HTTP implementation. */
-  public MindeeClientV2(PdfOperation pdfOperation, MindeeApiV2 mindeeApi) {
-    this.pdfOperation = pdfOperation;
+  public MindeeClientV2(MindeeApiV2 mindeeApi) {
     this.mindeeApi = mindeeApi;
   }
 
@@ -46,16 +39,9 @@ public MindeeClientV2(PdfOperation pdfOperation, MindeeApiV2 mindeeApi) {
   public JobResponse enqueue(
       LocalInputSource inputSource,
       InferenceParameters params) throws IOException {
-    LocalInputSource finalInput;
-    if (params.getPageOptions() != null) {
-      finalInput = new LocalInputSource(getSplitFile(inputSource, params.getPageOptions()), inputSource.getFilename());
-    } else {
-      finalInput = inputSource;
-    }
-    return mindeeApi.reqPostInferenceEnqueue(finalInput, params);
+    return mindeeApi.reqPostInferenceEnqueue(inputSource, params);
   }
 
-
   /**
    * Poll queue for a previously enqueued document.
    */
@@ -103,7 +89,7 @@ public InferenceResponse enqueueAndParse(
       if (resp.getJob().getStatus().equals("Failed")) {
         break;
       }
-      else if(resp.getJob().getStatus().equals("Processed")) {
+      else if (resp.getJob().getStatus().equals("Processed")) {
         return parseQueued(resp.getJob().getId());
       }
       attempts++;

src/main/java/com/mindee/MindeeClientV2.java

@@ -1,7 +1,10 @@
 package com.mindee.input;
 
 import com.mindee.image.ImageCompressor;
+import com.mindee.pdf.PdfBoxApi;
 import com.mindee.pdf.PdfCompressor;
+import com.mindee.pdf.PdfOperation;
+import com.mindee.pdf.SplitQuery;
 import java.io.File;
 import java.io.IOException;
 import java.io.InputStream;
@@ -45,6 +48,22 @@ public LocalInputSource(String fileAsBase64, String filename) {
     this.filename = filename;
   }
 
+
+  /**
+   * Applies PDF-specific operations on the current file based on the specified {@code PageOptions}.
+   *
+   * @param pageOptions The options specifying which pages to modify or retain in the PDF file.
+   * @throws IOException If an I/O error occurs during the PDF operation.
+   */
+  public void applyOperations (PageOptions pageOptions) throws IOException {
+    if (pageOptions != null && this.isPdf()) {
+      PdfOperation pdfOperation = new PdfBoxApi();
+      this.file = pdfOperation.split(
+          new SplitQuery(this.file, pageOptions)
+      ).getFile();
+    }
+  }
+
   public boolean isPdf() {
     return InputSourceUtils.isPdf(this.file);
   }