only upload if gameloot changed, remove seen_last in feed

README.md

@@ -26,7 +26,7 @@ This project is in ongoing development, not yet usable for the average user and
 - [x] Script for data migration in case of updates
 - [x] Store all dates in UTC
 - [x] Support start and end dates of offers
-- [ ] Only upload if hash of gameloot.xml has changed
+- [x] Only upload if hash of gameloot.xml has changed
 - [ ] Add tests (pytest?)
 - [ ] Dynamically generate ATOM feeds split by source and type (e.g. only amazon ingame loot) in addition to the full feed
 - [ ] Error handling

app/feed.py

@@ -5,12 +5,10 @@
 
 from feedgen.feed import FeedGenerator
 
-from app.configparser import Config
-
 from .common import TIMESTAMP_LONG, TIMESTAMP_READABLE_WITH_HOUR, LootOffer
 
 
-def generate_feed(offers: list[LootOffer]) -> None:
+def generate_feed(offers: list[LootOffer], out_file: Path) -> None:
     last_updated = datetime.now(timezone.utc)
 
     # Generate Feed Info
@@ -54,7 +52,7 @@ def generate_feed(offers: list[LootOffer]) -> None:
             title += f": {offer.subtitle}"
         title += f" ({offer.type})"
         feed_entry.title(title)
-        feed_entry.updated(offer.seen_last)
+        feed_entry.updated(offer.seen_first)
         # Atom Recommended
         # - Author
         feed_entry.author(
@@ -80,7 +78,7 @@ def generate_feed(offers: list[LootOffer]) -> None:
         if offer.publisher:
             content += f"<li>Publisher: {html.escape(offer.publisher)}</li>"
         content += "</ul><p>"
-        content += f"<small>Source: {html.escape(offer.source)}, Seen first: {offer.seen_first.strftime(TIMESTAMP_LONG)}, Seen last: {offer.seen_last.strftime(TIMESTAMP_LONG)}</small>"
+        content += f"<small>Source: {html.escape(offer.source)}, Seen first: {offer.seen_first.strftime(TIMESTAMP_LONG)}</small>"
         content += "</p>"
         feed_entry.content(content, type="xhtml")
         # - Link
@@ -96,7 +94,5 @@ def generate_feed(offers: list[LootOffer]) -> None:
 
         # feed_entry.link(href="http://lernfunk.de/feed")
 
-    out_file = Config.data_path() / Path(Config.config()["common"]["FeedFile"])
-
     # Write the ATOM feed to a file
     feed_generator.atom_file(filename=str(out_file), pretty=True)

lootscraper.py

@@ -1,3 +1,4 @@
+import hashlib
 import logging
 import shutil
 import sys
@@ -128,15 +129,23 @@ def job() -> None:
     cfg_generate_feed: bool = Config.config().getboolean("actions", "GenerateFeed")  # type: ignore
     cfg_upload: bool = Config.config().getboolean("actions", "UploadFtp")  # type: ignore
 
+    feed_file = Config.data_path() / Path(Config.config()["common"]["FeedFile"])
+
+    old_hash = hash_file(feed_file)
+
     if cfg_generate_feed:
-        generate_feed(all_offers)
+        generate_feed(all_offers, feed_file)
     else:
-        logging.info("Skipping feed generation")
+        logging.info("Skipping feed generation, disabled")
+
+    new_hash = hash_file(feed_file)
 
-    if cfg_upload:
+    if new_hash == old_hash:
+        logging.info("Skipping upload, no changes to feed file")
+    elif cfg_upload:
         upload_to_server()
     else:
-        logging.info("Skipping upload")
+        logging.info("Skipping upload, disabled")
 
 
 def log_new_offer(offer: LootOffer) -> None:
@@ -149,6 +158,22 @@ def log_new_offer(offer: LootOffer) -> None:
     logging.info(res)
 
 
+def hash_file(file: Path) -> str:
+    if not file.exists():
+        return ""
+
+    hash = hashlib.sha256()
+
+    with open(file, "rb") as f:
+        while True:
+            data = f.read(65536)
+            if not data:
+                break
+            hash.update(data)
+
+    return hash.hexdigest()
+
+
 if __name__ == "__main__":
     import signal