Cache Goodreads data and slow down bulk requests

tools/content/books.py

@@ -1,5 +1,6 @@
 import datetime
 
+import book_cache
 import goodreads
 import jsonfile
 
@@ -47,13 +48,21 @@ def short_description(text: str | None) -> str | None:
     return text[:DESCRIPTION_MAX_LENGTH].rsplit(" ", 1)[0] + "..."
 
 
+def local_cover(scraped: dict) -> str | None:
+    try:
+        return book_cache.cache_cover(scraped["goodreadsId"], scraped["cover"]) or scraped["cover"]
+    except OSError:
+        return scraped["cover"]
+
+
 def stored_fields(scraped: dict) -> dict:
     return {
         "goodreadsId": scraped["goodreadsId"],
         "url": scraped["url"],
         "title": scraped["title"],
         "authors": scraped["authors"],
-        "cover": scraped["cover"],
+        "cover": local_cover(scraped),
+        "coverSource": scraped["cover"],
         "description": short_description(scraped["description"]),
         "pages": scraped["pages"],
         "published": scraped["published"],
@@ -107,7 +116,7 @@ def add_book(url: str, shelf: str = "read", rating: int | None = None) -> dict:
     if find_index(load_books(), identifier) is not None:
         raise ValueError(f"{identifier} is already listed")
 
-    scraped = goodreads.scrape_book(url)
+    scraped = book_cache.scrape_book(url)
     if not scraped["title"]:
         raise ValueError("Could not read the book from Goodreads")
 
@@ -130,6 +139,7 @@ def remove_book(text: str) -> dict:
         raise ValueError(f"{text} is not listed")
     removed = books.pop(index)
     save_books(books)
+    book_cache.remove_cover(removed["goodreadsId"])
     return removed
 
 
@@ -156,5 +166,18 @@ def refresh_book(text: str) -> dict:
     index = find_index(books, text)
     if index is None:
         raise ValueError(f"{text} is not listed")
-    scraped = goodreads.scrape_book(books[index]["url"])
+    scraped = book_cache.scrape_book(books[index]["url"], fresh=True)
     return update_book(text, lambda book: {**book, **stored_fields(scraped)})
+
+
+def books_missing_local_covers() -> list[dict]:
+    return [book for book in load_books() if not book_cache.is_local_cover(book.get("cover"))]
+
+
+def cache_book_cover(identifier: str) -> dict:
+    def change(book: dict) -> dict:
+        source = book.get("coverSource") or book.get("cover")
+        cover = book_cache.cache_cover(identifier, source)
+        return {**book, "cover": cover or book.get("cover"), "coverSource": source}
+
+    return update_book(identifier, change)