pdfnet coordinate transformer wip

image_prediction/transformer/transformers/coordinate/coordinate_transformer.py

@@ -0,0 +1,15 @@
+import abc
+
+from image_prediction.transformer.transformer import Transformer
+
+
+class PDFNetCoordinateTransformer(Transformer):
+
+    @abc.abstractmethod
+    def forward(self, metadata):
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def backward(self, metadata):
+        raise NotImplementedError
+

image_prediction/transformer/transformers/coordinate/pdfnet.py

@@ -0,0 +1,18 @@
+from operator import itemgetter
+
+from funcy import omit
+
+from image_prediction.info import Info
+from image_prediction.transformer.transformer import Transformer
+
+
+class PDFNetCoordinateTransformer(Transformer):
+
+    def transform(self, metadata: dict):
+        y1, y2, page_height = itemgetter(Info.Y1, Info.Y2, Info.PAGE_HEIGHT)(metadata)
+        y1_t = page_height - y2
+        y2_t = page_height - y1
+        return {**omit(metadata, [Info.Y1, Info.Y2]), **{Info.Y1: y1_t, Info.Y2: y2_t}}
+
+
+

test/unit_tests/coordinate_transformer_test.py

@@ -1,14 +1,21 @@
+import os
+import tempfile
+import time
+
 import numpy as np
 import pytest
 from fpdf import fpdf
 from funcy import project, second
 from pdf2image import pdf2image
 
+from image_prediction.formatter.formatters.enum import EnumFormatter
 from image_prediction.image_extractor.extractor import ImageMetadataPair
 from image_prediction.image_extractor.extractors.parsable import ParsablePDFImageExtractor
 from image_prediction.info import Info
 from image_prediction.transformer.transformers.coordinate.fitz import FitzCoordinateTransformer
 from image_prediction.transformer.transformers.coordinate.fpdf import FPDFCoordinateTransformer
+from image_prediction.transformer.transformers.coordinate.pdfnet import PDFNetCoordinateTransformer
+from image_prediction.utils.pdf_annotation import annotate_pdf, draw_metadata_box
 from test.conftest import array_to_image, add_image
 
 
@@ -19,7 +26,7 @@ def coordinate_test_image():
 
 @pytest.fixture
 def coordinate_test_fpdf():
-    pdf = fpdf.FPDF(unit="pt", format=(200, 200))
+    pdf = fpdf.FPDF(unit="pt", format=(300, 300))
     return pdf
 
 
@@ -27,20 +34,38 @@ def get_fpdf_coordinates(corner):
     metadata = base_position_metadata()
 
     if corner == "top_left":
-        metadata.update({Info.X1: 0, Info.Y1: 0})
+        metadata.update({Info.X1: 0, Info.Y1: 0, Info.X2: 100, Info.Y2: 100})
 
     elif corner == "bottom_left":
-        metadata.update({Info.X1: 0, Info.Y1: 100})
+        metadata.update({Info.X1: 0, Info.Y1: 200, Info.X2: 100, Info.Y2: 300})
 
     return metadata
 
 
 def get_fitz_coordinates(corner):
-    return get_fpdf_coordinates(corner)  # Identical coordinate systems
+    metadata = base_position_metadata()
+
+    if corner == "top_left":
+        metadata.update({Info.X1: 0, Info.Y1: 0, Info.X2: 100, Info.Y2: 100})
+
+    elif corner == "bottom_left":
+        metadata.update({Info.X1: 0, Info.Y1: 200, Info.X2: 100, Info.Y2: 300})
+
+    return metadata
 
 
-def base_position_metadata():
-    return {Info.X1: 0, Info.Y1: 0, Info.WIDTH: 100, Info.HEIGHT: 100, Info.PAGE_IDX: 0}
+def base_position_metadata(width=100, height=100):
+    return {
+        Info.X1: 0,
+        Info.Y1: 0,
+        Info.X2: width,
+        Info.Y2: height,
+        Info.WIDTH: width,
+        Info.HEIGHT: height,
+        Info.PAGE_IDX: 0,
+        Info.PAGE_WIDTH: 300,
+        Info.PAGE_HEIGHT: 300,
+    }
 
 
 @pytest.fixture
@@ -79,10 +104,10 @@ def test_fpdf_coordinate_transformer(position_metadata_in_given_system, position
 
 @pytest.mark.parametrize("corner", ["top_left", "bottom_left"])
 @pytest.mark.parametrize("coordinate_system", ["fitz"])
-def test_coordinate_transformer(
-        position_metadata_in_given_system, position_metadata_in_reference_system, coordinate_test_pdf
+def test_fitz_coordinate_transformer(
+        position_metadata_in_given_system, position_metadata_in_reference_system, coordinate_test_pdf, coordinate_test_fpdf
 ):
-    # coordinate_test_pdf.output("/tmp/x.pdf")
+    # coordinate_test_fpdf.output("/tmp/x.pdf")
     # coordinate_test_page_image.show()
 
     metadata_extracted = next(map(second, ParsablePDFImageExtractor()(coordinate_test_pdf)))
@@ -90,3 +115,22 @@ def test_coordinate_transformer(
 
     assert position_extracted == position_metadata_in_given_system
     assert FitzCoordinateTransformer()(position_metadata_in_given_system) == position_metadata_in_reference_system
+
+
+@pytest.mark.parametrize("corner", ["bottom_left"])
+@pytest.mark.parametrize("coordinate_system", ["fitz"])
+def test_pdfnet_coordinate_transformer(
+        position_metadata_in_given_system, position_metadata_in_reference_system, coordinate_test_fpdf, coordinate_test_page_image
+):
+    # coordinate_test_page_image.show()
+    # coordinate_test_fpdf.output("/tmp/xyz.pdf")
+
+    position_metadata_transformed = PDFNetCoordinateTransformer()(position_metadata_in_given_system)
+
+    with tempfile.NamedTemporaryFile(suffix=".pdf") as f:
+        coordinate_test_fpdf.output(f.name)
+        draw_metadata_box(f.name, EnumFormatter()(position_metadata_transformed),
+        # os.path.join("/tmp", os.path.basename(f.name.replace(".pdf", "_annotated.pdf")))
+                          "/tmp/bla.pdf"
+                          )
+        input()