Enforce configurable height and pixel limits for tablet documents

copienator/annotation_documents.py

@@ -1,12 +1,19 @@
 """Locate tablet documents, including the parts of a large checked copy."""
 from pathlib import Path
 
-from copienator.configuration import ANNOTATION_MAX_HEIGHT_PX
+from copienator.configuration import ANNOTATION_MAX_HEIGHT_PX, ANNOTATION_MAX_PIXELS
 from copienator.json_io import read_json
 
 if type(ANNOTATION_MAX_HEIGHT_PX) is not int or ANNOTATION_MAX_HEIGHT_PX <= 0:
     raise ValueError("ANNOTATION_MAX_HEIGHT_PX must be a positive integer (pixels)")
 MAX_HEIGHT_PX = ANNOTATION_MAX_HEIGHT_PX
+if type(ANNOTATION_MAX_PIXELS) is not int or ANNOTATION_MAX_PIXELS <= 0:
+    raise ValueError("ANNOTATION_MAX_PIXELS must be a positive integer (pixels)")
+MAX_PIXELS = ANNOTATION_MAX_PIXELS
+
+
+def document_fits(width: int, height: int, max_height: int, max_pixels: int) -> bool:
+    return height <= max_height and width * height <= max_pixels
 
 
 def copy_documents(directory: Path) -> list[Path]:

copienator/commands/annotating_by_label.py

@@ -23,7 +23,7 @@ from copienator import (
     workspace_from_args,
 )
 from copienator.annotation_data import load_annotation_data
-from copienator.annotation_documents import MAX_HEIGHT_PX
+from copienator.annotation_documents import MAX_HEIGHT_PX, MAX_PIXELS, document_fits
 from copienator.commands import annotating, annotating_with_checks
 from copienator.commands.import_annotations import clear_import_directory
 from copienator.filesystem import staged_directory
@@ -64,10 +64,11 @@ def save_batch(batch, prefix, group_id, output_root: Path) -> None:
     print(f"Generating group PDF: {prefix} G{group_id} ({len(batch)} elements)")
     max_width = max(item[2].width for item in batch)
     total_height = sum(item[2].height for item in batch)
-    if total_height > MAX_HEIGHT_PX and len({item[0] for item in batch}) > 1:
+    if not document_fits(max_width, total_height, MAX_HEIGHT_PX, MAX_PIXELS):
         raise CliError(
-            f"{prefix} G{group_id} would be {total_height} px high; "
-            f"only a single student's block may exceed {MAX_HEIGHT_PX} px."
+            f"{prefix} G{group_id} would be {max_width} × {total_height} px "
+            f"({max_width * total_height} pixels); document limits are "
+            f"{MAX_HEIGHT_PX} px high and {MAX_PIXELS} pixels."
         )
     concatenated = Image.new("RGB", (max_width, total_height), "white")
     draw = ImageDraw.Draw(concatenated)
@@ -281,6 +282,26 @@ def _existing_group_state(
 def split_batches(rendered):
     batches = []
     heights = []
+    widths = []
+
+    def fits(width, height):
+        return document_fits(width, height, MAX_HEIGHT_PX, MAX_PIXELS)
+
+    def add_block(items):
+        width = max(item[2].width for item in items)
+        height = sum(item[2].height for item in items)
+        # Reuse space, accounting for padding to the widest answer in a group.
+        for batch_index in range(len(batches)):
+            combined_width = max(widths[batch_index], width)
+            combined_height = heights[batch_index] + height
+            if fits(combined_width, combined_height):
+                batches[batch_index].extend(items)
+                widths[batch_index] = combined_width
+                heights[batch_index] = combined_height
+                return
+        batches.append(list(items))
+        widths.append(width)
+        heights.append(height)
 
     index = 0
     while index < len(rendered):
@@ -291,22 +312,31 @@ def split_batches(rendered):
         student_items = rendered[index:end]
         student_height = sum(item[2].height for item in student_items)
 
-        if student_height > MAX_HEIGHT_PX:
-            # Oversized blocks remain intact in a group of their own.
-            batches.append(list(student_items))
-            heights.append(student_height)
-            index = end
-            continue
-
-        # Reuse space in earlier groups, keeping each student's block intact.
-        for batch_index, height in enumerate(heights):
-            if height + student_height <= MAX_HEIGHT_PX:
-                batches[batch_index].extend(student_items)
-                heights[batch_index] += student_height
-                break
+        student_width = max(item[2].width for item in student_items)
+        if fits(student_width, student_height):
+            add_block(student_items)
         else:
-            batches.append(list(student_items))
-            heights.append(student_height)
+            # Split oversized student blocks only between complete answers.
+            block = []
+            width = height = 0
+            for item in student_items:
+                image = item[2]
+                if not fits(image.width, image.height):
+                    raise CliError(
+                        f"Copie{student_id} ({item[1]}) is {image.width} × "
+                        f"{image.height} px ({image.width * image.height} pixels); "
+                        f"a single answer cannot be split. Document limits are "
+                        f"{MAX_HEIGHT_PX} px high and {MAX_PIXELS} pixels."
+                    )
+                if block and not fits(max(width, image.width), height + image.height):
+                    add_block(block)
+                    block = []
+                    width = height = 0
+                block.append(item)
+                width = max(width, image.width)
+                height += image.height
+            if block:
+                add_block(block)
         index = end
 
     return batches

copienator/commands/annotating_with_checks.py

@@ -26,7 +26,7 @@ from copienator import (
     workspace_from_target,
 )
 from copienator.annotation_data import load_annotation_data
-from copienator.annotation_documents import MAX_HEIGHT_PX, copy_documents
+from copienator.annotation_documents import MAX_HEIGHT_PX, MAX_PIXELS, copy_documents, document_fits
 from copienator.commands import annotating
 from copienator.commands.import_annotations import clear_import_directory
 from copienator.filesystem import staged_directory
@@ -212,18 +212,23 @@ def _render_student(
     batches = []
     start = 0
     height = 0
+    width = 0
     for index, image in enumerate(label_images):
-        if image.height > MAX_HEIGHT_PX:
+        if not document_fits(image.width, image.height, MAX_HEIGHT_PX, MAX_PIXELS):
             raise CliError(
                 f"Copie{student_id} ({bnote_entries[index]['label']}) is "
-                f"{image.height} px high; a single answer cannot be split and "
-                f"the maximum document height is {MAX_HEIGHT_PX} px."
+                f"{image.width} × {image.height} px ({image.width * image.height} pixels); "
+                f"a single answer cannot be split. Document limits are "
+                f"{MAX_HEIGHT_PX} px high and {MAX_PIXELS} pixels."
             )
-        if height + image.height > MAX_HEIGHT_PX:
+        if not document_fits(max(width, image.width), height + image.height,
+                             MAX_HEIGHT_PX, MAX_PIXELS):
             batches.append((start, index))
             start = index
             height = 0
+            width = 0
         height += image.height
+        width = max(width, image.width)
     batches.append((start, len(label_images)))
 
     with staged_directory(output_dir) as staging:
@@ -251,6 +256,12 @@ def _save_document(
 ) -> None:
     max_width = max(image.width for image in label_images)
     total_height = sum(image.height for image in label_images)
+    if not document_fits(max_width, total_height, MAX_HEIGHT_PX, MAX_PIXELS):
+        raise CliError(
+            f"Document would be {max_width} × {total_height} px "
+            f"({max_width * total_height} pixels); document limits are "
+            f"{MAX_HEIGHT_PX} px high and {MAX_PIXELS} pixels."
+        )
     concatenated = Image.new("RGB", (max_width, total_height), "white")
     checkbox_map: list[dict[str, Any]] = []
     current_y = 0

copienator/configuration.py

@@ -41,7 +41,8 @@ THINKING_LEVEL_PRO = "high"
 ALWAYS_CROP = False
 ANNOTATION_MARGIN_LEFT = 300
 ANNOTATION_LOW_MARGIN_LEFT = 400
-ANNOTATION_MAX_HEIGHT_PX = 32000
+ANNOTATION_MAX_HEIGHT_PX = 33000
+ANNOTATION_MAX_PIXELS = 67_108_864
 SHOW_PERSONAL_STEPS = False
 RETURN_JPEG_ENABLED = True
 RETURN_PDF_ENABLED = True