[WIP] random font selection

test/fixtures/page_generation/page.py

@@ -1,13 +1,16 @@
 import io
 import itertools
+import os
 import random
 import string
 import textwrap
 from functools import lru_cache
+from pathlib import Path
 from typing import Tuple, Union, Iterable, List
 
 import albumentations as A
 import cv2 as cv
+import loguru
 import numpy as np
 import pandas as pd
 import pytest
@@ -92,6 +95,9 @@ from funcy import (
     rcompose,
     lsplit,
     lfilter,
+    lzip,
+    keep,
+    lkeep,
 )
 
 from cv_analysis.locations import TEST_PAGE_TEXTURES_DIR
@@ -178,8 +184,8 @@ def dpi(request):
 @pytest.fixture(
     params=[
         # "brown",
-        "sepia",
-        # "gray",
+        # "sepia",
+        "gray",
         # "white",
         # "light_red",
         # "light_blue",
@@ -202,8 +208,8 @@ def texture_name(request):
 
 @pytest.fixture(
     params=[
-        # 30,
-        70,
+        30,
+        # 70,
         # 150,
     ]
 )
@@ -362,7 +368,7 @@ class ContentRectangle(Rectangle):
 
 class ContentGenerator:
     def __init__(self):
-        self.constrain_layouts = False
+        self.constrain_layouts = True
 
     def __call__(self, boxes: List[Rectangle]) -> Image:
         random.shuffle(boxes)
@@ -375,7 +381,7 @@ class ContentGenerator:
             text_boxes = merge_related_rectangles(text_boxes)
 
         text_boxes = lmap(generate_random_text_block, text_boxes)
-        plots = lmap(generate_random_plot, every_nth(2, char_boxes))
+        plots = lmap(generate_random_table, every_nth(2, char_boxes))
         tables = lmap(generate_random_table, every_nth(2, char_boxes[1:]))
 
         boxes = text_boxes + plots + tables
@@ -479,6 +485,74 @@ def dump_plt_to_image(rectangle):
     return image
 
 
+class RandomFontPicker:
+    def __init__(self, font_dir=None):
+        self.fonts = get_fonts(font_dir)
+        self.fonts_lower = [font.lower() for font in self.fonts]
+
+        self.test_image = Image.new("RGB", (200, 200), (255, 255, 255))
+        self.draw = ImageDraw.Draw(self.test_image)
+
+    def pick_random_font_available_on_system(self, includes=None, excludes=None) -> ImageFont:  # FIXME: Slow!
+
+        self.shuffle_fonts()
+
+        includes_pattern = (lambda f: includes.lower() in f) if includes else (lambda f: True)
+        excludes_pattern = (lambda f: excludes.lower() not in f) if excludes else (lambda f: True)
+
+        mask = lmap(lambda f: includes_pattern(f) and excludes_pattern(f), self.fonts_lower)
+        fonts = itertools.compress(self.fonts, mask)
+        fonts = keep(map(self.load_font, fonts))
+        fonts = filter(self.font_is_renderable, fonts)  # FIXME: this does not work
+
+        font = first(fonts)
+        loguru.logger.debug(f"Using font: {font}")
+        return font
+
+    def shuffle_fonts(self):
+        l = lzip(self.fonts, self.fonts_lower)
+        random.shuffle(l)
+        self.fonts, self.fonts_lower = lzip(*l)
+
+    def pick_random_mono_space_font_available_on_system(self) -> ImageFont:
+        return self.pick_random_font_available_on_system(includes="mono", excludes="oblique")
+
+    @lru_cache(maxsize=None)
+    def load_font(self, font: str):
+        loguru.logger.trace(f"Loading font: {font}")
+        try:
+            return ImageFont.truetype(font, size=11)
+        except OSError:
+            return None
+
+    @lru_cache(maxsize=None)
+    def font_is_renderable(self, font):
+        text_size = self.draw.textsize("Test String", font=font)
+        return text_size[0] > 0 and text_size[1]
+
+
+def get_fonts(path: Path = None) -> List[str]:
+    path = path or Path("/usr/share/fonts")
+    fonts = list(path.rglob("*.ttf"))
+    fonts = [font.name for font in fonts]
+    return fonts
+
+
+@lru_cache(maxsize=None)
+def get_font_picker(font_dir=None):
+    return RandomFontPicker(font_dir=font_dir)
+
+
+def pick_random_mono_space_font_available_on_system():
+    font_picker = get_font_picker()
+    return font_picker.pick_random_mono_space_font_available_on_system()
+
+
+def pick_random_font_available_on_system():
+    font_picker = get_font_picker()
+    return font_picker.pick_random_font_available_on_system(includes="mono")
+
+
 class RandomPlot(RandomContentRectangle):
     def __init__(self, x1, y1, x2, y2, seed=None):
         super().__init__(x1, y1, x2, y2, seed=seed)
@@ -600,7 +674,7 @@ def write_lines_to_image(lines: List[str], rectangle: Rectangle, font=None) -> I
     def write_line(line, line_number):
         draw.text((0, line_number * text_size), line, font=font, fill=(0, 0, 0, 200))
 
-    font = font or ImageFont.load_default()
+    font = font or pick_random_mono_space_font_available_on_system()
 
     image = Image.new("RGBA", (rectangle.width, rectangle.height), (0, 255, 255, 0))
     draw = ImageDraw.Draw(image)
@@ -616,7 +690,7 @@ 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()
+        self.font = pick_random_font_available_on_system()
 
     def __call__(self, *args, **kwargs):
         pass
@@ -688,7 +762,7 @@ class PagePartitioner:
         self.top_margin_percentage = 0.1
         self.bottom_margin_percentage = 0.1
 
-        self.margin_percentage = 0.005
+        self.margin_percentage = 0.007
         self.max_depth = 3
         self.initial_recursion_probability = 1
         self.recursion_probability_decay = 0.1