Upload bilingual_book_maker/book_maker/backmatter.py with huggingface_hub

bilingual_book_maker/book_maker/backmatter.py

@@ -0,0 +1,73 @@
+from __future__ import annotations
+
+import re
+from collections.abc import Iterable
+
+
+DEFAULT_BACKMATTER_TITLES = (
+    "about the author",
+    "notes about the author",
+    "notes on the author",
+    "biography",
+    "biographical note",
+    "credits",
+    "acknowledgments",
+    "acknowledgements",
+    "copyright",
+    "bibliography",
+    "index",
+    "endnotes",
+    "notes",
+    "further reading",
+    "also by the author",
+    "about this book",
+    "about this ebook",
+)
+
+NON_ALNUM_RE = re.compile(r"[^a-z0-9]+")
+
+
+def normalize_heading(text: str | None) -> str:
+    value = (text or "").strip().lower()
+    if not value:
+        return ""
+    value = NON_ALNUM_RE.sub(" ", value)
+    return " ".join(value.split())
+
+
+def split_backmatter_titles(value: str | Iterable[str] | None) -> tuple[str, ...]:
+    if value is None:
+        return DEFAULT_BACKMATTER_TITLES
+    if isinstance(value, str):
+        candidates = value.split(",")
+    else:
+        candidates = list(value)
+
+    normalized = [normalize_heading(item) for item in candidates]
+    filtered = [item for item in normalized if item]
+    return tuple(filtered) or DEFAULT_BACKMATTER_TITLES
+
+
+def matches_backmatter_title(title: str | None, keywords: str | Iterable[str] | None = None) -> bool:
+    normalized_title = normalize_heading(title)
+    if not normalized_title:
+        return False
+
+    normalized_keywords = split_backmatter_titles(keywords)
+    return any(
+        normalized_title == keyword
+        or normalized_title.startswith(f"{keyword} ")
+        for keyword in normalized_keywords
+    )
+
+
+def should_skip_backmatter(
+    *,
+    title: str | None,
+    current_percent: float,
+    skip_after_percent: int | float,
+    keywords: str | Iterable[str] | None = None,
+) -> bool:
+    if current_percent <= skip_after_percent:
+        return False
+    return matches_backmatter_title(title, keywords)