ebook: fix ellipsis spacing

scripts/ebook/step_6.py

@@ -16,6 +16,21 @@
 target_file = Path("hpmor.html")
 
 
+def fix_ellipsis(s: str) -> str:
+    """
+    Fix ellipsis spacing for ebooks.
+    """
+    # 1. remove all spaces around ellipsis
+    s = re.sub(r" *… *", "…", s)
+    # 2. recreate spaces around ellipsis
+    # between words
+    s = re.sub(r"(?<=[\w])…(?=[\w])", "… ", s)
+    # after punctuation
+    s = re.sub(r"(?<=[\.\?!:])…(?=[\w])", " …", s)
+    # before punctuation : no space, so governed by 1.
+    return s
+
+
 if __name__ == "__main__":
     print("=== 6. HTML modifications ===")
 
@@ -61,13 +76,15 @@
     # remove training slashes to satisfy https://validator.w3.org
     cont = cont.replace("<br />", "<br>")
     cont = cont.replace("<hr />", "<hr>")
-
     cont = re.sub(
         r"(<meta [^>]*) />",
         r"\1>",
         cont,
     )
 
+    # fix spaces around ellipsis
+    cont = fix_ellipsis(cont)
+
     # remove bad span ids (containing spaces) from newspaper spans
     cont = re.sub(r'<span id="[^"]+" label="[^"]+">', r"<span>", cont, count=5)
 

scripts/ebook/step_6_test.py

@@ -0,0 +1,24 @@
+"""Unit Tests."""  # noqa: INP001
+# ruff: noqa: S101
+
+from step_6 import fix_ellipsis
+
+# quotations
+assert fix_ellipsis("foo…”") == "foo…”"
+assert fix_ellipsis("“…foo") == "“…foo"
+# html
+assert fix_ellipsis("foo…</p>") == "foo…</p>"
+assert fix_ellipsis("<p>…foo") == "<p>…foo"
+# between 2 words
+assert fix_ellipsis("foo…bar") == "foo… bar"
+assert fix_ellipsis("foo …bar") == "foo… bar"
+assert fix_ellipsis("foo … bar") == "foo… bar"
+assert fix_ellipsis("foo… bar") == "foo… bar"
+# start of sentence
+assert fix_ellipsis("foo.…bar") == "foo. …bar"
+assert fix_ellipsis("foo!…bar") == "foo! …bar"
+assert fix_ellipsis("foo?…bar") == "foo? …bar"
+# end of sentence
+assert fix_ellipsis("foo…. bar") == "foo…. bar"
+assert fix_ellipsis("foo…! bar") == "foo…! bar"
+assert fix_ellipsis("foo…? bar") == "foo…? bar"