From 60a1f3a27ab01f082ba114bf1498e7b27282108e Mon Sep 17 00:00:00 2001
From: msramalho <19508417+msramalho@users.noreply.github.com>
Date: Mon, 31 Jul 2023 16:08:48 +0100
Subject: [PATCH] minor fixes

---
 src/auto_archiver/enrichers/pdq_hash_enricher.py   | 7 ++++++-
 src/auto_archiver/formatters/templates/macros.html | 2 +-
 src/auto_archiver/utils/url.py                     | 5 +++++
 3 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/src/auto_archiver/enrichers/pdq_hash_enricher.py b/src/auto_archiver/enrichers/pdq_hash_enricher.py
index da99375..ff88bab 100644
--- a/src/auto_archiver/enrichers/pdq_hash_enricher.py
+++ b/src/auto_archiver/enrichers/pdq_hash_enricher.py
@@ -26,11 +26,16 @@ class PdqHashEnricher(Enricher):
     def enrich(self, to_enrich: Metadata) -> None:
         url = to_enrich.get_url()
         logger.debug(f"calculating perceptual hashes for {url=}")
+        media_with_hashes = []
 
         for m in to_enrich.media:
             for media in m.all_inner_media(True):
-                if media.is_image() and "screenshot" not in media.get("id") and "warc-file-" not in media.get("id") and len(hd := self.calculate_pdq_hash(media.filename)):
+                media_id = media.get("id", "")
+                if media.is_image() and "screenshot" not in media_id and "warc-file-" not in media_id and len(hd := self.calculate_pdq_hash(media.filename)):
                     media.set("pdq_hash", hd)
+                    media_with_hashes.append(media.filename)
+
+        logger.debug(f"calculated '{len(media_with_hashes)}' perceptual hashes for {url=}: {media_with_hashes}")
 
     def calculate_pdq_hash(self, filename):
         # returns a hexadecimal string with the perceptual hash for the given filename
diff --git a/src/auto_archiver/formatters/templates/macros.html b/src/auto_archiver/formatters/templates/macros.html
index d03a36c..50c3673 100644
--- a/src/auto_archiver/formatters/templates/macros.html
+++ b/src/auto_archiver/formatters/templates/macros.html
@@ -16,7 +16,7 @@ No URL available for {{ m.key }}.
         <a href="https://lens.google.com/uploadbyurl?url={{ url | quote }}">Google Lens</a>,&nbsp;
         <a href="https://yandex.ru/images/touch/search?rpt=imageview&url={{ url | quote }}">Yandex</a>,&nbsp;
         <a href="https://www.bing.com/images/search?view=detailv2&iss=sbi&form=SBIVSP&sbisrc=UrlPaste&q=imgurl:{{ url | quote }}">Bing</a>,&nbsp;
-        <a href="https://www.tineye.com/search/?url={{ url | quote }}">Tineye</a>,&nbsp;
+        <a href="https://www.tineye.com/search/?url={{ url | quote }}">Tineye</a>
     </div>
     <p></p>
 </div>
diff --git a/src/auto_archiver/utils/url.py b/src/auto_archiver/utils/url.py
index 6c43909..ce96330 100644
--- a/src/auto_archiver/utils/url.py
+++ b/src/auto_archiver/utils/url.py
@@ -52,6 +52,11 @@ class UrlUtil:
         # telegram
         if "https://telegram.org/img/emoji/" in url: return False
 
+        # youtube
+        if "https://www.youtube.com/s/gaming/emoji/" in url: return False
+        if "https://yt3.ggpht.com" in url and "default-user=" in url: return False
+        if "https://www.youtube.com/s/search/audio/" in url: return False
+
         return True
 
     @staticmethod