[WIP] random text segments

cv_analysis/utils/rectangle.py

@@ -47,6 +47,10 @@ class Rectangle:
     def height(self):
         return abs(self.y2 - self.y1)
 
+    @property
+    def coords(self):
+        return [self.x1, self.y1, self.x2, self.y2]
+
     def __hash__(self):
         return hash((self.x1, self.y1, self.x2, self.y2))
 

test/fixtures/page_generation/page.py

@@ -11,6 +11,8 @@ from PIL import Image, ImageOps, ImageFont, ImageDraw
 from PIL.Image import Transpose
 from faker import Faker
 
+from cv_analysis.utils.conversion import rectangle_to_box
+
 Image_t = Union[Image.Image, np.ndarray]
 #
 # transform = A.Compose(
@@ -68,7 +70,7 @@ Image_t = Union[Image.Image, np.ndarray]
 #     ],
 #     p=0.5,
 # )
-from funcy import juxt, compose, identity, lflatten, lmap
+from funcy import juxt, compose, identity, lflatten, lmap, first
 
 from cv_analysis.locations import TEST_PAGE_TEXTURES_DIR
 
@@ -344,27 +346,56 @@ class ContentGenerator:
         return lmap(generate_random_text_block, boxes)
 
 
+class RandomTextBlock(ContentRectangle):
+    def __init__(self, x1, y1, x2, y2):
+        super().__init__(x1, y1, x2, y2)
+        self.blank_line_percentage = random.uniform(0, 0.5)
+        self.font = ImageFont.load_default()
+
+    def generate_random_text(self, rectangle: Rectangle):
+
+        image = Image.new("RGBA", (rectangle.width, rectangle.height), (0, 255, 255, 0))
+        draw = ImageDraw.Draw(image)
+        text = Faker().paragraph(nb_sentences=1000, variable_nb_sentences=False, ext_word_list=None)
+
+        wrapped_text = textwrap.wrap(text, width=image.width, break_long_words=False)
+        text_size = draw.textsize(first(wrapped_text), font=self.font)[1]
+
+        last_full = True
+        for i, line in enumerate(wrapped_text):
+            if random.random() < self.blank_line_percentage and last_full:
+                line = self.truncate_line(line)
+                last_full = False
+            else:
+                last_full = True
+
+            draw.text((0, i * text_size), line, font=self.font, fill=(0, 0, 0, 200))
+
+        self.content = image
+
+    def f(self, last_full, line):
+        if random.random() < self.blank_line_percentage and last_full:
+            line = self.truncate_line(line)
+            last_full = False
+        else:
+            last_full = True
+        return last_full, line
+
+    def format_line(self, line, full=True):
+        line = self.truncate_line(line) if not full else line
+        return line, full
+
+    def truncate_line(self, line: str):
+        n_trailing_words = random.randint(0, 4)
+        line = " ".join(line.split()[-n_trailing_words - 1 : -1]).replace(".", "")
+        line = line + ".\n" if line else line
+        return line
+
+
 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)
+    block = RandomTextBlock(*rectangle.coords)
+    block.generate_random_text(rectangle)
+    return block
 
 
 def paste_content(page, content_box: ContentRectangle):