Add SVG generation

Author: janek-cosmose

parse_qwantz/cli.py

@@ -22,6 +22,7 @@ class Inner:
     debug: bool
     show_boxes: bool
     unambiguous_words: bool
+    generate_svg: bool
 
     def __call__(self, image_path: Path):
         try:
@@ -31,6 +32,7 @@ def __call__(self, image_path: Path):
                 debug=self.debug,
                 show_boxes=self.show_boxes,
                 unambiguous_words=self.unambiguous_words,
+                svg=self.generate_svg,
             )
         except ImageError:
             pass
@@ -44,6 +46,7 @@ def cli(
     debug: bool = typer.Option(False, help="Enable debug features."),
     show_boxes: bool = typer.Option(False, help="Show character boxes (for debug)"),
     unambiguous_words: bool = typer.Option(False, help="Print only unambiguous words"),
+    generate_svg: bool = typer.Option(False, help="Generate SVG file"),
 ):
     """Generate transcripts for Ryan North's Dinosaur Comics from https://qwantz.com"""
     set_logging_formatter()
@@ -60,6 +63,7 @@ def cli(
             debug=debug,
             show_boxes=show_boxes,
             unambiguous_words=unambiguous_words,
+            generate_svg=generate_svg,
         )
 
         with Pool(4) as pool:

parse_qwantz/colors.py

@@ -1,19 +1,23 @@
-from enum import Enum
-from typing import Optional
+from typing import Optional, NamedTuple
 
 COLOR_THRESHOLD = 250
 
 
-class Color(Enum):
-    WHITE = (255, 255, 255)
-    BLACK = (0, 0, 0)
-    RED = (128, 0, 0)
+class Color(NamedTuple):
+    red: int
+    green: int
+    blue: int
 
-    @classmethod
-    def get_with_threshold(cls, rgb: tuple[int, int, int]) -> Optional["Color"]:
-        for color in cls:
-            if square_distance(color.value, rgb) < COLOR_THRESHOLD:
-                return color
+
+WHITE = Color(255, 255, 255)
+BLACK = Color(0, 0, 0)
+RED = Color(128, 0, 0)
+
+
+def get_color_with_threshold(rgb: tuple[int, int, int]) -> Optional["Color"]:
+    for color in [WHITE, BLACK, RED]:
+        if square_distance(color, rgb) < COLOR_THRESHOLD:
+            return color
 
 
 def square_distance(rgb1: tuple[int, int, int], rgb2: tuple[int, int, int]):

parse_qwantz/data/panel_overrides.json

@@ -978,6 +978,10 @@
     "panels": {
       "1": [
         "Heading: SUBJECT: hee hee hee hee MOOD: ☹ Groggy MUSIC: Cibo Matto - Birthday Cake"
+      ],
+      "4": [
+        "Heading: 4 COMMENTS | 〚blue〛 POST A COMMENT",
+        "Utahraptor: *hugses*"
       ]
     }
   },

parse_qwantz/elements.py

@@ -16,9 +16,10 @@
 
 def get_elements(
     image: SimpleImage
-) -> tuple[list[Line], list[Box], list[TextLine], list[Character], list[list[Pixel]]]:
+) -> tuple[list[Line], list[int], list[Box], list[TextLine], list[Character], list[list[Pixel]]]:
     text_lines: list[TextLine] = []
     lines: list[Line] = []
+    line_widths: list[int] = []
     thoughts: list[Box] = []
     unmatched: list[list[Pixel]] = []
     sorted_pixels = sorted(image.pixels)
@@ -50,8 +51,9 @@ def get_elements(
         else:
             result = get_line(pixel, tmp_image)
             if result:
-                line, line_pixels = result
+                line, line_pixels, width = result
                 lines.append(line)
+                line_widths.append(width)
                 sorted_pixels = remove_subsequence(sorted_pixels, line_pixels)
             elif result := get_batman(pixel, tmp_image):
                 batman_box, batman_pixels = result
@@ -69,7 +71,7 @@ def get_elements(
                 if len(unmatched) == 5:
                     logger.warning("At least five unmatched objects detected, aborting")
                     break
-    return lines, thoughts, cleanup_text_lines(text_lines), extra_characters, unmatched
+    return lines, line_widths, thoughts, cleanup_text_lines(text_lines), extra_characters, unmatched
 
 
 def get_batman(pixel: Pixel, image: SimpleImage) -> tuple[Box, list[Pixel]] | None: