From ed8a83708b387069f31c7b235fe036aedc8198e4 Mon Sep 17 00:00:00 2001
From: Jan Lukas Gernert <jangernert@gmail.com>
Date: Tue, 13 Feb 2024 17:00:45 +0100
Subject: [PATCH] update deps & fix some flaky tests

---
 article_scraper/Cargo.toml                           | 12 ++++++------
 .../tests/readability/hukumusume/source.html         | 10 +++++-----
 article_scraper/src/clean.rs                         |  6 ++----
 3 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/article_scraper/Cargo.toml b/article_scraper/Cargo.toml
index 948fe98..1c7d2ac 100644
--- a/article_scraper/Cargo.toml
+++ b/article_scraper/Cargo.toml
@@ -14,18 +14,18 @@ exclude = ["resources/tests"]
 thiserror = "1.0"
 libxml = "0.3"
 reqwest = { version = "0.11", features = ["json", "native-tls", "gzip", "brotli", "stream"] }
-tokio = { version = "1.28", features = ["macros", "fs", "io-util"] }
-url = "2.3"
-regex = "1.8"
+tokio = { version = "1", features = ["macros", "fs", "io-util"] }
+url = "2.5"
+regex = "1.10"
 encoding_rs = "0.8"
 chrono = "0.4"
 base64 = "0.21"
 image = "0.24"
 log = "0.4"
-rust-embed="6.6"
-once_cell = "1.17"
+rust-embed="8.2"
+once_cell = "1.19"
 escaper = "0.1"
 futures = "0.3"
 
 [dev-dependencies]
-env_logger = "0.10"
+env_logger = "0.11"
diff --git a/article_scraper/resources/tests/readability/hukumusume/source.html b/article_scraper/resources/tests/readability/hukumusume/source.html
index 7034430..7f7afd4 100644
--- a/article_scraper/resources/tests/readability/hukumusume/source.html
+++ b/article_scraper/resources/tests/readability/hukumusume/source.html
@@ -175,7 +175,7 @@
                         </p>
                     </td>
                     <td width="10">
-                        <img src="file:///C|/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" />
+                        <img src="file:///C:/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" />
                     </td>
                     <td width="166" valign="top">
                         <table width="100%" border="0" cellpadding="0" cellspacing="0" bgcolor="#C8FFC8">
@@ -204,14 +204,14 @@
                                 </tr>
                                 <tr>
                                     <td height="50" bgcolor="#ECFFEC">
-                                        <img src="file:///C|/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97" width="1" height="1" /><b><font size="-1">きょうの記念日</font></b><br />
+                                        <img src="file:///C:/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97" width="1" height="1" /><b><font size="-1">きょうの記念日</font></b><br />
                                         <br />
                                         <a href="../../../../366/kinenbi/pc/01gatu/1_01.htm"><font size="-1">元旦</font></a>
                                     </td>
                                 </tr>
                                 <tr>
                                     <td height="50" bgcolor="#ECFFEC">
-                                        <img src="file:///C|/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /><b><font size="-1">きょうの誕生花</font></b><br />
+                                        <img src="file:///C:/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /><b><font size="-1">きょうの誕生花</font></b><br />
                                         <br />
                                         <a href="../../../../366/hana/pc/01gatu/1_01.htm"><font size="-1">松(まつ)</font></a>
                                     </td>
@@ -269,14 +269,14 @@
                                 </tr>
                                 <tr>
                                     <td height="50" bgcolor="#ECFFEC">
-                                        <b><font size="-1">きょうの世界昔話<img src="file:///C|/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /></font></b><br />
+                                        <b><font size="-1">きょうの世界昔話<img src="file:///C:/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /></font></b><br />
                                         <br />
                                         <a href="../../../../douwa/pc/world/01/01a.htm"><font size="-1">モンゴルの十二支話</font></a>
                                     </td>
                                 </tr>
                                 <tr>
                                     <td height="50" bgcolor="#ECFFEC">
-                                        <img src="file:///C|/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /><b><font size="-1">きょうの日本民話</font></b><br />
+                                        <img src="file:///C:/Documents%20and%20Settings/%E7%A6%8F%E5%A8%98note/%E3%83%87%E3%82%B9%E3%82%AF%E3%83%88%E3%83%83%E3%83%97/company_website15/image/spacer.gif" width="1" height="1" /><b><font size="-1">きょうの日本民話</font></b><br />
                                         <br />
                                         <a href="../../../../douwa/pc/minwa/01/01c.html"><font size="-1">仕事の取替えっこ</font></a>
                                     </td>
diff --git a/article_scraper/src/clean.rs b/article_scraper/src/clean.rs
index 81297a2..7ab49e8 100644
--- a/article_scraper/src/clean.rs
+++ b/article_scraper/src/clean.rs
@@ -145,9 +145,7 @@ mod tests {
         let url =
             Url::parse("https://blogs.gnome.org/tbernard/2023/07/26/rethinking-window-management/")
                 .unwrap();
-        let res = clean_html_fragment(html, &url).unwrap();
-
-        std::fs::write("/home/jeanluc/result.html", res.html).unwrap();
+        _ = clean_html_fragment(html, &url).unwrap();
     }
 
     #[test]
@@ -156,7 +154,7 @@ mod tests {
         let url = Url::parse("https://finshots.in").unwrap();
         let res = clean_html_fragment(html, &url).unwrap();
 
-        assert_eq!(res.html.len(), 11989);
+        assert!(res.html.len().abs_diff(12_000) < 200);
         assert_eq!(
             res.thumbnail.as_deref(),
             Some("https://cdn.finshots.app/images/2023/03/Design-8-Amul.jpg")