From 453cb5087744a42547f8b0081dd149a254a32fa2 Mon Sep 17 00:00:00 2001
From: Tiago Ribeiro <tfnribeiro@gmail.com>
Date: Tue, 17 Sep 2024 10:59:45 +0200
Subject: [PATCH] Improved Article formatting to contain stats

- The emails are now sent with information about the difficulty, source and read time.
---
 tools/send_subscription_emails.py          | 13 ++++++++++---
 zeeguu/core/util/__init__.py               |  1 +
 zeeguu/core/util/reading_time_estimator.py |  9 +++++++++
 3 files changed, 20 insertions(+), 3 deletions(-)
 create mode 100644 zeeguu/core/util/reading_time_estimator.py

diff --git a/tools/send_subscription_emails.py b/tools/send_subscription_emails.py
index b0fd90a9..13986989 100644
--- a/tools/send_subscription_emails.py
+++ b/tools/send_subscription_emails.py
@@ -5,10 +5,16 @@
 from zeeguu.core.emailer.zeeguu_mailer import ZeeguuMailer
 from zeeguu.core.model.search_subscription import SearchSubscription
 from zeeguu.api.app import create_app
+from zeeguu.core.util.reading_time_estimator import estimate_read_time
 
 app = create_app()
 app.app_context().push()
 
+def format_article_info(article):
+    art_info = article.article_info()
+    return f"""<b>{art_info["title"]}</b>\n
+                <p>&nbsp;&nbsp;{estimate_read_time(art_info["metrics"]["word_count"])}min, {art_info["metrics"]["cefr_level"]}, <a href="{art_info["url"]}">read</a> on <a href="{article.feed.url.domain.domain_name}">{article.feed.url.domain.domain_name}</a></p>\n"""
+
 
 def send_mail_new_articles_search(to_email, new_content_dict):
     body = "\r\n".join(
@@ -19,8 +25,8 @@ def send_mail_new_articles_search(to_email, new_content_dict):
             " "
         ])
     
-    for keyword, titles in new_content_dict.items():
-        body += "\r\n".join([" ", f"Search: '{keyword}': "] + [f"""- <a href="{url}">{t}</a>""" for t, url in titles]) + "\n"
+    for keyword, articles in new_content_dict.items():
+        body += "\r\n".join([" ", f"<h3>{keyword}</h3>", " "] + [format_article_info(a) for a in articles]) + "\n"
         
     body += "\r\n".join([
             " ",
@@ -31,6 +37,7 @@ def send_mail_new_articles_search(to_email, new_content_dict):
     )
     
     subject = f"New articles for {"'" + "','".join(new_content_dict.keys()) + "'"}"
+    print(body)
     emailer = ZeeguuMailer(subject, body, to_email)
     emailer.send()
 
@@ -59,7 +66,7 @@ def send_subscription_emails():
         ]
         if new_articles_found:
             updated_dict = user_subscriptions.get(user.email, {})
-            updated_dict[subscription.search.keywords] = [(article.title, article.url) for article in new_articles_found]
+            updated_dict[subscription.search.keywords] = [article for article in new_articles_found]
             user_subscriptions[user.email] = updated_dict
     for user_email, new_content_dict in user_subscriptions.items():
         send_mail_new_articles_search(user_email, new_content_dict)
diff --git a/zeeguu/core/util/__init__.py b/zeeguu/core/util/__init__.py
index dc00aeac..e594112f 100644
--- a/zeeguu/core/util/__init__.py
+++ b/zeeguu/core/util/__init__.py
@@ -4,3 +4,4 @@
 from zeeguu.core.util.encoding import JSONSerializable, encode, encode_error
 from zeeguu.core.util.hash import text_hash, password_hash
 from zeeguu.core.util.time import get_server_time_utc
+from zeeguu.core.util.reading_time_estimator import estimate_read_time
diff --git a/zeeguu/core/util/reading_time_estimator.py b/zeeguu/core/util/reading_time_estimator.py
new file mode 100644
index 00000000..feee43e3
--- /dev/null
+++ b/zeeguu/core/util/reading_time_estimator.py
@@ -0,0 +1,9 @@
+import math
+
+
+def estimate_read_time(word_count: int):
+    """
+    Returns the estimated reading time in minutes, assuming
+    a reading rate of 160WPM.
+    """
+    return math.ceil(word_count / 160)