mindee · sebastianMindee · May 28, 2026 · May 28, 2026
diff --git a/bin/MindeeCliCommand.php b/bin/MindeeCliCommand.php
@@ -5,14 +5,14 @@
 namespace Mindee\Cli;
 
 use Exception;
-use Mindee\Error\V1\MindeeV1HttpException;
 use Mindee\Input\InputSource;
 use Mindee\Input\PageOptions;
 use Mindee\Input\PathInput;
 use Mindee\Input\UrlInputSource;
 use Mindee\V1\Client;
 use Mindee\V1\ClientOptions\PredictMethodOptions;
 use Mindee\V1\ClientOptions\PredictOptions;
+use Mindee\V1\Error\MindeeV1HttpException;
 use Mindee\V1\Parsing\Common\AsyncPredictResponse;
 use Mindee\V1\Parsing\Common\PredictResponse;
 use Symfony\Component\Console\Command\Command;

diff --git a/src/Image/ImageExtractor.php b/src/Image/ImageExtractor.php
@@ -9,15 +9,13 @@
 use ImagickException;
 use Mindee\Dependency\DependencyChecker;
 use Mindee\Error\ErrorCode;
-use Mindee\Error\MindeeGeometryException;
 use Mindee\Error\MindeeImageException;
 use Mindee\Error\MindeePdfException;
 use Mindee\Geometry\BBox;
 use Mindee\Geometry\BBoxUtils;
 use Mindee\Geometry\Point;
 use Mindee\Geometry\Polygon;
 use Mindee\Input\LocalInputSource;
-use Mindee\V1\Parsing\Standard\BaseField;
 
 use function count;
 use function sprintf;
@@ -128,20 +126,6 @@ public function getPageCount(): int
         return count($this->pageImages);
     }
 
-    /**
-     * Extract multiple images on a given page from a list of fields having position data.
-     *
-     * @param array<BaseField<string|float|integer|boolean|Polygon>> $fields List of Fields to extract.
-     * @param integer $pageIndex The page index to extract, begins at 0.
-     * @param null|string $outputName The base output filename, must have an image extension.
-     *
-     * @return array<ExtractedImage> a list of extracted images
-     */
-    public function extractImagesFromPage(array $fields, int $pageIndex, ?string $outputName = null): array
-    {
-        $outputName ??= $this->filename;
-        return $this->extractFromPage($fields, $pageIndex, $outputName);
-    }
 
     /**
      * Extracts images from a page.
@@ -212,47 +196,6 @@ public function extractPolygonFromPage(
         return new ExtractedImage($extractedImageData, $filename, $format, $pageIndex, $index);
     }
 
-    /**
-     * Extracts a single image from a Position field.
-     *
-     * @param BaseField<string|float|integer|boolean|Polygon> $field The field to extract.
-     * @param integer $pageIndex The page index to extract, begins at 0.
-     * @param integer $index The index to use for naming the extracted image.
-     * @param string $filename The output filename.
-     * @param string $format The output format.
-     *
-     * @return null|ExtractedImage The extracted image, or null if the field does not have valid position data.
-     *
-     * @throws MindeeGeometryException Throws if a field does not contain positional data.
-     */
-    public function extractImage(
-        BaseField $field,
-        int $pageIndex,
-        int $index,
-        string $filename,
-        string $format
-    ): ?ExtractedImage {
-        $polygon = null;
-
-        if (!empty($field->polygon)) {
-            $polygon = $field->polygon;
-        } elseif (!empty($field->boundingBox)) {
-            $polygon = $field->boundingBox;
-        } elseif (!empty($field->quadrangle)) {
-            $polygon = $field->quadrangle;
-        } elseif (!empty($field->rectangle)) {
-            $polygon = $field->rectangle;
-        }
-
-        if (null === $polygon) {
-            throw new MindeeGeometryException(
-                'Provided field has no valid position data.',
-                ErrorCode::GEOMETRIC_OPERATION_FAILED
-            );
-        }
-
-        return $this->extractPolygonFromPage($polygon, $pageIndex, $index, $filename, $format);
-    }
 
     /**
      * Getter for the local input source.
@@ -262,33 +205,6 @@ public function getInputSource(): LocalInputSource
         return $this->inputSource;
     }
 
-    /**
-     * Extracts images from a page.
-     *
-     * @param array<BaseField<string|float|integer|boolean|Polygon>> $fields List of Fields to extract.
-     * @param integer $pageIndex The page index to extract, begins at 0.
-     * @param string $outputName Name of the created file.
-     * @param string $format The output format.
-     *
-     * @return array<ExtractedImage> An array of created images
-     */
-    protected function extractFromPage(array $fields, int $pageIndex, string $outputName, string $format = 'jpg'): array
-    {
-        $extractedImages = [];
-
-        $i = 0;
-        foreach ($fields as $field) {
-            $filename = sprintf('%s_page%d-%d.%s', $outputName, $pageIndex, $i, $format);
-            $extractedImage = $this->extractImage($field, $pageIndex, $i, $filename, $format);
-            if (null !== $extractedImage) {
-                $extractedImages[] = $extractedImage;
-            }
-            ++$i;
-        }
-
-        return $extractedImages;
-    }
-
     /**
      * Extracts an image from a set of coordinates.
      *

diff --git a/src/Pdf/PdfExtractor.php b/src/Pdf/PdfExtractor.php
@@ -10,15 +10,13 @@
 use Mindee\Dependency\DependencyChecker;
 use Mindee\Error\MindeePdfException;
 use Mindee\Input\LocalInputSource;
-use Mindee\V1\Product\InvoiceSplitter\InvoiceSplitterV1InvoicePageGroups;
 use setasign\Fpdi\Fpdi;
 use setasign\Fpdi\PdfParser\CrossReference\CrossReferenceException;
 use setasign\Fpdi\PdfParser\Filter\FilterException;
 use setasign\Fpdi\PdfParser\PdfParserException;
 use setasign\Fpdi\PdfReader\PdfReaderException;
 
 use function count;
-use function is_array;
 use function sprintf;
 
 /**
@@ -39,8 +37,8 @@ class PdfExtractor
     /**
      * @param LocalInputSource $localInput Local Input, accepts all compatible formats.
      *
-     * @throws MindeePdfException Throws if PDF operations aren't supported, or if the file
-     *                            can't be read, respectively.
+     * @throws MindeePdfException|ImagickException Throws if PDF operations aren't supported, or if the file
+     *                                             can't be read, respectively.
      */
     public function __construct(LocalInputSource $localInput)
     {
@@ -86,14 +84,14 @@ public function getPageCount(): int
     /**
      * Extracts sub-documents from the source document using list of page indexes.
      *
-     * @param array<array<integer>>|InvoiceSplitterV1InvoicePageGroups $pageIndexes List of sub-lists of pages to keep.
+     * @param array<array<integer>> $pageIndexes List of sub-lists of pages to keep.
      *
      * @return ExtractedPdf[] list of extracted documents
      *
      * @throws MindeePdfException Throws if FDPF/FPDI wasn't able to handle the pdf during the extraction.
      * @throws InvalidArgumentException Throws if invalid indexes are provided.
      */
-    public function extractSubDocuments(array|InvoiceSplitterV1InvoicePageGroups $pageIndexes): array
+    public function extractSubDocuments(array $pageIndexes): array
     {
         $extractedPdfs = [];
 
@@ -141,58 +139,17 @@ public function extractSubDocuments(array|InvoiceSplitterV1InvoicePageGroups $pa
     /**
      * Extracts invoices as complete PDFs from the document.
      *
-     * @param array<array<integer>>|InvoiceSplitterV1InvoicePageGroups $pageIndexes List of sub-lists of pages to keep.
+     * @param array<array<integer>> $pageIndexes List of sub-lists of pages to keep.
      * @param boolean $strict Whether to trust confidence scores or not.
      *
      * @return ExtractedPdf[] a list of extracted invoices
      */
-    public function extractInvoices(array|InvoiceSplitterV1InvoicePageGroups $pageIndexes, bool $strict = false): array
+    public function extractInvoices(array $pageIndexes, bool $strict = false): array
     {
         if (empty($pageIndexes)) {
             return [];
         }
-        if (!$strict) {
-            $indexes = array_map(static fn($invoicePageIndexes) => $invoicePageIndexes->pageIndexes, (array) $pageIndexes);
-
-            return $this->extractSubDocuments($indexes);
-        }
-        if (is_array($pageIndexes[0])) {
-            return $this->extractSubDocuments($pageIndexes);
-        }
-
-        $correctPageIndexes = [];
-        $currentList = [];
-        $previousConfidence = null;
-
-        $i = 0;
-        foreach ($pageIndexes as $pageIndex) {
-            $confidence = $pageIndex->confidence;
-            $pageList = $pageIndex->pageIndexes;
-
-            if ($confidence >= 0.5 && null === $previousConfidence) {
-                $currentList = $pageList;
-            } elseif ($confidence >= 0.5 && $i !== count($pageIndexes) - 1) {
-                if (!empty($currentList)) {
-                    $correctPageIndexes[] = $currentList;
-                }
-                $currentList = $pageList;
-            } elseif ($confidence < 0.5 && $i === count($pageIndexes) - 1) {
-                $currentList = array_merge($currentList, $pageList);
-                if (!empty($currentList)) {
-                    $correctPageIndexes[] = $currentList;
-                }
-            } else {
-                if (!empty($currentList)) {
-                    $correctPageIndexes[] = $currentList;
-                }
-                $correctPageIndexes[] = $pageList;
-            }
-
-            $previousConfidence = $confidence;
-            ++$i;
-        }
-
-        return $this->extractSubDocuments($correctPageIndexes);
+        return $this->extractSubDocuments($pageIndexes);
     }
 
     /**

diff --git a/src/V1/Client.php b/src/V1/Client.php
@@ -16,14 +16,14 @@
 use Mindee\Error\ErrorCode;
 use Mindee\Error\MindeeApiException;
 use Mindee\Error\MindeeException;
-use Mindee\Error\V1\MindeeV1ClientException;
-use Mindee\Error\V1\MindeeV1HttpException;
 use Mindee\Input\InputSource;
 use Mindee\Input\LocalInputSource;
 use Mindee\Input\LocalResponse;
 use Mindee\Input\PageOptions;
 use Mindee\V1\ClientOptions\PredictMethodOptions;
 use Mindee\V1\ClientOptions\WorkflowOptions;
+use Mindee\V1\Error\MindeeV1ClientException;
+use Mindee\V1\Error\MindeeV1HttpException;
 use Mindee\V1\Http\Endpoint;
 use Mindee\V1\Http\MindeeApi;
 use Mindee\V1\Http\MindeeWorkflowApi;

diff --git a/src/Error/V1/MindeeV1ClientException.php → src/V1/Error/MindeeV1ClientException.php b/src/Error/V1/MindeeV1ClientException.php → src/V1/Error/MindeeV1ClientException.php
@@ -7,7 +7,7 @@
  * Mindee Client Exceptions.
  */
 
-namespace Mindee\Error\V1;
+namespace Mindee\V1\Error;
 
 use Mindee\Error\MindeeException;
 

diff --git a/src/Error/V1/MindeeV1HttpException.php → src/V1/Error/MindeeV1HttpException.php b/src/Error/V1/MindeeV1HttpException.php → src/V1/Error/MindeeV1HttpException.php
@@ -7,7 +7,7 @@
  * Mindee HTTP Exceptions.
  */
 
-namespace Mindee\Error\V1;
+namespace Mindee\V1\Error;
 
 use Mindee\Error\ErrorCode;
 use Mindee\Error\MindeeException;

diff --git a/src/V1/Image/ImageExtractor.php b/src/V1/Image/ImageExtractor.php
@@ -4,9 +4,102 @@
 
 namespace Mindee\V1\Image;
 
+use Mindee\Error\ErrorCode;
+use Mindee\Error\MindeeGeometryException;
+use Mindee\Geometry\Polygon;
+use Mindee\Image\ExtractedImage;
 use Mindee\Image\ImageExtractor as BaseImageExtractor;
+use Mindee\V1\Parsing\Standard\BaseField;
+
+use function sprintf;
 
 /**
  * Wrapper class for V1 of the BaseImageExtractor.
  */
-class ImageExtractor extends BaseImageExtractor {}
+class ImageExtractor extends BaseImageExtractor
+{
+    /**
+     * Extract multiple images on a given page from a list of fields having position data.
+     *
+     * @param array<BaseField<string|float|integer|boolean|Polygon>> $fields List of Fields to extract.
+     * @param integer $pageIndex The page index to extract, begins at 0.
+     * @param null|string $outputName The base output filename, must have an image extension.
+     *
+     * @return array<ExtractedImage> a list of extracted images
+     */
+    public function extractImagesFromPage(array $fields, int $pageIndex, ?string $outputName = null): array
+    {
+        $outputName ??= $this->filename;
+        return $this->extractFromPage($fields, $pageIndex, $outputName);
+    }
+
+    /**
+     * Extracts a single image from a Position field.
+     *
+     * @param BaseField<string|float|integer|boolean|Polygon> $field The field to extract.
+     * @param integer $pageIndex The page index to extract, begins at 0.
+     * @param integer $index The index to use for naming the extracted image.
+     * @param string $filename The output filename.
+     * @param string $format The output format.
+     *
+     * @return null|ExtractedImage The extracted image, or null if the field does not have valid position data.
+     *
+     * @throws MindeeGeometryException Throws if a field does not contain positional data.
+     */
+    public function extractImage(
+        BaseField $field,
+        int $pageIndex,
+        int $index,
+        string $filename,
+        string $format
+    ): ?ExtractedImage {
+        $polygon = null;
+
+        if (!empty($field->polygon)) {
+            $polygon = $field->polygon;
+        } elseif (!empty($field->boundingBox)) {
+            $polygon = $field->boundingBox;
+        } elseif (!empty($field->quadrangle)) {
+            $polygon = $field->quadrangle;
+        } elseif (!empty($field->rectangle)) {
+            $polygon = $field->rectangle;
+        }
+
+        if (null === $polygon) {
+            throw new MindeeGeometryException(
+                'Provided field has no valid position data.',
+                ErrorCode::GEOMETRIC_OPERATION_FAILED
+            );
+        }
+
+        return $this->extractPolygonFromPage($polygon, $pageIndex, $index, $filename, $format);
+    }
+
+    /**
+     * Extracts images from a page.
+     *
+     * @param array<BaseField<string|float|integer|boolean|Polygon>> $fields List of Fields to extract.
+     * @param integer $pageIndex The page index to extract, begins at 0.
+     * @param string $outputName Name of the created file.
+     * @param string $format The output format.
+     *
+     * @return array<ExtractedImage> An array of created images
+     */
+    protected function extractFromPage(array $fields, int $pageIndex, string $outputName, string $format = 'jpg'): array
+    {
+        $extractedImages = [];
+
+        $i = 0;
+        foreach ($fields as $field) {
+            $filename = sprintf('%s_page%d-%d.%s', $outputName, $pageIndex, $i, $format);
+            $extractedImage = $this->extractImage($field, $pageIndex, $i, $filename, $format);
+            if (null !== $extractedImage) {
+                $extractedImages[] = $extractedImage;
+            }
+            ++$i;
+        }
+
+        return $extractedImages;
+    }
+
+}