From 5fbfb3c0b1f38d6de58fce2c1059265631451377 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Oliver=20B=C3=B8ving?= <oliver@bvng.dk>
Date: Sat, 17 Feb 2024 17:55:17 +0100
Subject: [PATCH] Skip test if test data is not available

---
 crates/core/src/ranking/models/cross_encoder.rs |  8 ++++----
 crates/core/src/summarizer.rs                   |  8 ++++----
 crates/core/src/warc.rs                         |  9 ++++-----
 crates/core/src/widgets/thesaurus.rs            |  7 ++++++-
 crates/zimba/src/lib.rs                         |  6 ++++--
 crates/zimba/src/wiki.rs                        | 16 ++++++++++++++--
 6 files changed, 36 insertions(+), 18 deletions(-)

diff --git a/crates/core/src/ranking/models/cross_encoder.rs b/crates/core/src/ranking/models/cross_encoder.rs
index 2d6c8e51..d13096e8 100644
--- a/crates/core/src/ranking/models/cross_encoder.rs
+++ b/crates/core/src/ranking/models/cross_encoder.rs
@@ -165,12 +165,12 @@ mod tests {
 
     #[test]
     fn sanity_check() {
-        if !Path::new("../../data/cross_encoder").exists() {
+        let data_path = Path::new("../../data/cross_encoder");
+        if !data_path.exists() {
+            // Skip the test if the test data is not available
             return;
         }
-
-        let model = CrossEncoderModel::open("../../data/cross_encoder")
-            .expect("Failed to find cross-encoder model");
+        let model = CrossEncoderModel::open(data_path).expect("Failed to find cross-encoder model");
 
         let s = model.run(
             "how many people live in paris",
diff --git a/crates/core/src/summarizer.rs b/crates/core/src/summarizer.rs
index 901a8c0c..cb4b2591 100644
--- a/crates/core/src/summarizer.rs
+++ b/crates/core/src/summarizer.rs
@@ -545,12 +545,12 @@ mod tests {
 
     #[test]
     fn test_dual_encoder() {
-        if !Path::new("../../data/summarizer/dual_encoder").exists() {
+        let data_path = Path::new("../../data/summarizer/dual_encoder");
+        if !data_path.exists() {
+            // Skip the test if the test data is not available
             return;
         }
-
-        let model =
-            DualEncoder::open("../../data/summarizer/dual_encoder").expect("Failed to load model");
+        let model = DualEncoder::open(data_path).expect("Failed to load model");
         let query = "What is the capital of France?";
         let pos = "The capital of France is Paris.";
         let neg = "The best baguette in Paris can be found at Boulangerie Pichard.";
diff --git a/crates/core/src/warc.rs b/crates/core/src/warc.rs
index 112ed285..c8664dc4 100644
--- a/crates/core/src/warc.rs
+++ b/crates/core/src/warc.rs
@@ -725,16 +725,15 @@ mod tests {
 
     #[test]
     fn internet_archive_parse() {
-        if !Path::new("../../data/internet_archive.warc.gz").exists() {
+        let data_path = Path::new("../../data/internet_archive.warc.gz");
+        if !data_path.exists() {
+            // Skip the test if the test data is not available
             return;
         }
 
         let mut records = 0;
 
-        for record in WarcFile::open("../../data/internet_archive.warc.gz")
-            .unwrap()
-            .records()
-        {
+        for record in WarcFile::open(data_path).unwrap().records() {
             records += 1;
             if let Err(err) = record {
                 panic!("Error: {:?}", err);
diff --git a/crates/core/src/widgets/thesaurus.rs b/crates/core/src/widgets/thesaurus.rs
index 736eeb63..adeed6a2 100644
--- a/crates/core/src/widgets/thesaurus.rs
+++ b/crates/core/src/widgets/thesaurus.rs
@@ -504,7 +504,12 @@ mod tests {
 
     #[test]
     fn build_dict() {
-        let dict = Dictionary::build("../../data/english-wordnet-2022-subset.ttl").unwrap();
+        let data_path = Path::new("../../data/english-wordnet-2022-subset.ttl");
+        if !data_path.exists() {
+            // Skip the test if the test data is not available
+            return;
+        }
+        let dict = Dictionary::build(data_path).unwrap();
 
         let infos = dict.get(Lemma("barely".to_string()));
 
diff --git a/crates/zimba/src/lib.rs b/crates/zimba/src/lib.rs
index 8febd32d..c718b83f 100644
--- a/crates/zimba/src/lib.rs
+++ b/crates/zimba/src/lib.rs
@@ -636,11 +636,13 @@ mod tests {
 
     #[test]
     fn it_works() {
-        if !Path::new("../../data/test.zim").exists() {
+        let data_path = Path::new("../../data/test.zim");
+        if !data_path.exists() {
+            // Skip test if data file is not present
             return;
         }
 
-        let zim = ZimFile::open("../../data/test.zim").unwrap();
+        let zim = ZimFile::open(data_path).unwrap();
 
         assert_eq!(zim.header.magic, 72173914);
         assert_eq!(zim.header.major_version, 5);
diff --git a/crates/zimba/src/wiki.rs b/crates/zimba/src/wiki.rs
index 3dd07e8d..ee5d4a3b 100644
--- a/crates/zimba/src/wiki.rs
+++ b/crates/zimba/src/wiki.rs
@@ -256,11 +256,18 @@ impl<'a> Iterator for ImageIterator<'a> {
 
 #[cfg(test)]
 mod tests {
+    use std::path::Path;
+
     use super::*;
 
     #[test]
     fn test_article_iterator() {
-        let zim = ZimFile::open("../../data/test.zim").unwrap();
+        let data_path = Path::new("../../data/test.zim");
+        if !data_path.exists() {
+            // Skip the test if the test data is not available
+            return;
+        }
+        let zim = ZimFile::open(data_path).unwrap();
         let mut iter = ArticleIterator::new(&zim).unwrap();
 
         let article = iter.next().unwrap();
@@ -271,7 +278,12 @@ mod tests {
 
     #[test]
     fn test_image_iterator() {
-        let zim = ZimFile::open("../../data/test.zim").unwrap();
+        let data_path = Path::new("../../data/test.zim");
+        if data_path.exists() {
+            // Skip the test if the test data is not available
+            return;
+        }
+        let zim = ZimFile::open(data_path).unwrap();
         let mut iter = ImageIterator::new(&zim).unwrap();
 
         let image = iter.next().unwrap();