[WIP] random text segments

cv_analysis/utils/utils.py

@@ -69,3 +69,7 @@ def make_merger_sentinel():
     number_of_records_so_far = -1
 
     return no_new_mergers
+
+
+def rconj(xs, x):
+    return [*xs, x]

test/fixtures/page_generation/page.py

@@ -1,6 +1,5 @@
 import random
 import textwrap
-from functools import lru_cache
 from typing import Tuple, Union, Iterable, List
 
 import albumentations as A
@@ -11,7 +10,7 @@ from PIL import Image, ImageOps, ImageFont, ImageDraw
 from PIL.Image import Transpose
 from faker import Faker
 
-from cv_analysis.utils import star
+from cv_analysis.utils import star, rconj
 
 Image_t = Union[Image.Image, np.ndarray]
 #
@@ -70,7 +69,7 @@ Image_t = Union[Image.Image, np.ndarray]
 #     ],
 #     p=0.5,
 # )
-from funcy import juxt, compose, identity, lflatten, lmap, first, iterate, take, last, rest
+from funcy import juxt, compose, identity, lflatten, lmap, first, iterate, take, last, rest, rcompose
 
 from cv_analysis.locations import TEST_PAGE_TEXTURES_DIR
 
@@ -353,6 +352,8 @@ class RandomTextBlock(ContentRectangle):
         self.font = ImageFont.load_default()
 
     def generate_random_text(self, rectangle: Rectangle):
+        def write_line(line, line_number):
+            draw.text((0, line_number * text_size), line, font=self.font, fill=(0, 0, 0, 200))
 
         image = Image.new("RGBA", (rectangle.width, rectangle.height), (0, 255, 255, 0))
         draw = ImageDraw.Draw(image)
@@ -361,24 +362,30 @@ class RandomTextBlock(ContentRectangle):
         wrapped_text = textwrap.wrap(text, width=image.width, break_long_words=False)
         text_size = draw.textsize(first(wrapped_text), font=self.font)[1]
 
-        lines = last(take(len(wrapped_text), iterate(star(self.f), (True, wrapped_text))))[1]
+        lines = last(take(len(wrapped_text), iterate(star(self.format_lines), (True, wrapped_text))))[1]
 
-        for i, line in enumerate(lines):
-            draw.text((0, i * text_size), line, font=self.font, fill=(0, 0, 0, 200))
+        for line_number, line in enumerate(lines):
+            write_line(line, line_number)
 
         self.content = image
 
-    def f(self, last_full, lines):
-        @lru_cache(maxsize=None)
-        def truncate():
+    def format_lines(self, last_full, lines):
+        def truncate_current_line():
             return random.random() < self.blank_line_percentage and last_full
 
-        line_processor = self.truncate_line if truncate() else identity
-        append = star(rconj)
-
-        fn = compose(append, juxt(rest, compose(line_processor, first)))
-
-        return (False, fn(lines)) if truncate() else (True, fn(lines))
+        # This is meant to be read from the bottom up.
+        current_line_shall_not_be_a_full_line = truncate_current_line()
+        line_processor = self.truncate_line if current_line_shall_not_be_a_full_line else identity
+        process_current_line = compose(line_processor, first)
+        move_current_line_to_back = star(rconj)
+        split_first_line_from_lines_and_process_the_former = juxt(rest, process_current_line)
+        split_off_current_line_then_process_it_then_move_it_to_the_back = rcompose(
+            split_first_line_from_lines_and_process_the_former,
+            move_current_line_to_back,
+        )
+        current_line_is_a_full_line = not current_line_shall_not_be_a_full_line
+        # Start reading here and move up.
+        return current_line_is_a_full_line, split_off_current_line_then_process_it_then_move_it_to_the_back(lines)
 
     def format_line(self, line, full=True):
         line = self.truncate_line(line) if not full else line
@@ -391,10 +398,6 @@ class RandomTextBlock(ContentRectangle):
         return line
 
 
-def rconj(xs, x):
-    return [*xs, x]
-
-
 def generate_random_text_block(rectangle: Rectangle) -> ContentRectangle:
     block = RandomTextBlock(*rectangle.coords)
     block.generate_random_text(rectangle)
@@ -402,10 +405,8 @@ def generate_random_text_block(rectangle: Rectangle) -> ContentRectangle:
 
 
 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