[WIP] random text segments

poetry.lock

@@ -776,6 +776,17 @@ python-versions = ">=3.7"
 [package.extras]
 test = ["pytest (>=6)"]
 
+[[package]]
+name = "faker"
+version = "16.4.0"
+description = "Faker is a Python package that generates fake data for you."
+category = "dev"
+optional = false
+python-versions = ">=3.7"
+
+[package.dependencies]
+python-dateutil = ">=2.4"
+
 [[package]]
 name = "filelock"
 version = "3.9.0"
@@ -2195,7 +2206,7 @@ testing = ["flake8 (<5)", "func-timeout", "jaraco.functools", "jaraco.itertools"
 [metadata]
 lock-version = "1.1"
 python-versions = "~3.8"
-content-hash = "87e394dd1a2a230f27fec0d1e634dc4c398fbebe5ae1c3baceee9c93d190945b"
+content-hash = "4d220326bd29d9f6b306adc71fe66ec201ca74fc7d277b052195953f8a063037"
 
 [metadata.files]
 aiohttp = [
@@ -2767,6 +2778,10 @@ exceptiongroup = [
     {file = "exceptiongroup-1.1.0-py3-none-any.whl", hash = "sha256:327cbda3da756e2de031a3107b81ab7b3770a602c4d16ca618298c526f4bec1e"},
     {file = "exceptiongroup-1.1.0.tar.gz", hash = "sha256:bcb67d800a4497e1b404c2dd44fca47d3b7a5e5433dbab67f96c1a685cdfdf23"},
 ]
+faker = [
+    {file = "Faker-16.4.0-py3-none-any.whl", hash = "sha256:5420467fad3fa582094057754e5e81326cb1f51ab822bf9df96c077cfb35ae49"},
+    {file = "Faker-16.4.0.tar.gz", hash = "sha256:dcffdca8ec9a715982bcd5f53ee688dc4784cd112f9910f8f7183773eb3ec276"},
+]
 filelock = [
     {file = "filelock-3.9.0-py3-none-any.whl", hash = "sha256:f58d535af89bb9ad5cd4df046f741f8553a418c01a7856bf0d173bbc9f6bd16d"},
     {file = "filelock-3.9.0.tar.gz", hash = "sha256:7b319f24340b51f55a2bf7a12ac0755a9b03e718311dac567a0f4f7fabd2f5de"},

pyproject.toml

@@ -39,6 +39,7 @@ pytest = "^7.0.1"
 
 [tool.poetry.group.test.dependencies]
 albumentations = "^1.3.0"
+faker = "^16.4.0"
 
 [build-system]
 requires = ["poetry-core"]

test/fixtures/page_generation/page.py

@@ -1,12 +1,15 @@
 import random
+import string
+import textwrap
 from typing import Tuple, Union, Iterable, List
 
 import albumentations as A
 import cv2 as cv
 import numpy as np
 import pytest
-from PIL import Image, ImageOps
+from PIL import Image, ImageOps, ImageFont, ImageDraw
 from PIL.Image import Transpose
+from faker import Faker
 
 Image_t = Union[Image.Image, np.ndarray]
 #
@@ -65,7 +68,7 @@ Image_t = Union[Image.Image, np.ndarray]
 #     ],
 #     p=0.5,
 # )
-from funcy import juxt, compose, identity, lflatten
+from funcy import juxt, compose, identity, lflatten, lmap
 
 from cv_analysis.locations import TEST_PAGE_TEXTURES_DIR
 
@@ -316,13 +319,70 @@ def blank_page(texture, texture_fn) -> np.ndarray:
     """Creates a blank page with a given orientation and dpi."""
     page = random_flip(texture)
     page = texture_fn(page)
-    content_box_generator = ContentBoxGenerator()
-    boxes = content_box_generator(page)
-    content_box_generator.draw_boxes(page, boxes)
+    page_partitioner = PagePartitioner()
+    boxes = page_partitioner(page)
+    content_generator = ContentGenerator()
+    boxes = content_generator(boxes)
+    page = paste_contents(page, boxes)
+    page_partitioner.draw_boxes(page, boxes)
+
+    page = np.array(page)
     return page
 
 
-class ContentBoxGenerator:
+class ContentRectangle(Rectangle):
+    def __init__(self, x1, y1, x2, y2, content=None):
+        super().__init__(x1, y1, x2, y2)
+        self.content = content
+
+
+class ContentGenerator:
+    def __init__(self):
+        pass
+
+    def __call__(self, boxes: Iterable[Rectangle]) -> Image:
+        return lmap(generate_random_text_block, boxes)
+
+
+def generate_random_text_block(rectangle: Rectangle) -> ContentRectangle:
+
+    font = ImageFont.load_default()
+    image = Image.new("RGBA", (rectangle.width, rectangle.height), (0, 255, 255, 0))
+    draw = ImageDraw.Draw(image)
+    fake = Faker()
+    text = fake.paragraph(nb_sentences=300, variable_nb_sentences=True, ext_word_list=None)
+
+    wrapped_text = textwrap.wrap(text, width=image.width, break_long_words=False)
+
+    # Set the initial position for the text
+    x, y = 0, 0
+
+    # Iterate through the lines of wrapped text
+    for line in wrapped_text:
+        # Draw the line of text at the current position
+        draw.text((x, y), line, font=font, fill=(0, 0, 0, 200))
+        # Update the y-coordinate for the next line
+        y += draw.textsize(line, font=font)[1]
+
+    return ContentRectangle(rectangle.x1, rectangle.y1, rectangle.x2, rectangle.y2, content=image)
+
+
+def paste_content(page, content_box: ContentRectangle):
+
+    assert page.mode == "RGB"
+    assert content_box.content.mode == "RGBA"
+
+    page.paste(content_box.content, (content_box.x1, content_box.y1), content_box.content)
+    return page
+
+
+def paste_contents(page, contents: Iterable[ContentRectangle]):
+    for content in contents:
+        paste_content(page, content)
+    return page
+
+
+class PagePartitioner:
     def __init__(self):
         self.left_margin_percentage = 0.05
         self.right_margin_percentage = 0.05
@@ -343,6 +403,7 @@ class ContentBoxGenerator:
         box = Rectangle(left_margin, top_margin, page.width - right_margin, page.height - bottom_margin)
         boxes = lflatten(self.generate_content_boxes(box))
         boxes = self.drop_small_boxes(boxes, *page.size)
+        boxes = list(boxes)
         return boxes
 
     def draw_boxes(self, page: Image, boxes: Iterable[Rectangle]):