From ff770ceb2b500831518955431be3852cb9714c1b Mon Sep 17 00:00:00 2001
From: j-mendez <jeff@a11ywatch.com>
Date: Sat, 11 Jan 2025 07:57:59 -0500
Subject: [PATCH] chore(tests): fix transformations build parse

---
 spider/src/page.rs                               | 15 +++++++++++++++
 spider_transformations/src/transformation/mod.rs | 16 ++++++++--------
 2 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/spider/src/page.rs b/spider/src/page.rs
index 046cd7960..7a36ed481 100644
--- a/spider/src/page.rs
+++ b/spider/src/page.rs
@@ -489,6 +489,21 @@ fn get_error_status(
     get_error_status_base(should_retry, error_for_status).map(std::sync::Arc::new)
 }
 
+
+/// Instantiate a new page without scraping it and with the base URL parsed (used for testing purposes).
+#[cfg(not(feature = "decentralized"))]
+pub fn build_with_parse(url: &str, res: PageResponse) -> Page {
+    let mut page = build(url, res);
+    page.set_url_parsed_direct_empty();
+    page
+}
+
+/// Instantiate a new page without scraping it and with the base URL parsed (used for testing purposes).
+#[cfg(feature = "decentralized")]
+pub fn build_with_parse(url: &str, res: PageResponse) -> Page {
+    build(url, res)
+}
+
 /// Instantiate a new page without scraping it (used for testing purposes).
 #[cfg(not(feature = "decentralized"))]
 pub fn build(url: &str, res: PageResponse) -> Page {
diff --git a/spider_transformations/src/transformation/mod.rs b/spider_transformations/src/transformation/mod.rs
index 397d7ef31..3a7275bb4 100644
--- a/spider_transformations/src/transformation/mod.rs
+++ b/spider_transformations/src/transformation/mod.rs
@@ -13,7 +13,7 @@ mod tests {
     use maud::PreEscaped;
     use spider::{
         bytes::Bytes,
-        page::build,
+        page::build_with_parse,
         tokio::{self, fs::File},
         utils::PageResponse,
     };
@@ -51,7 +51,7 @@ mod tests {
         let mut page_response = PageResponse::default();
 
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
 
         conf.return_format = ReturnFormat::Markdown;
 
@@ -111,7 +111,7 @@ mod tests {
         let mut page_response = PageResponse::default();
         conf.return_format = ReturnFormat::XML;
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
         let content = content::transform_content(&page, &conf, &None, &None, &None);
         assert!(
             content
@@ -129,7 +129,7 @@ mod tests {
         let mut page_response = PageResponse::default();
 
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
 
         conf.return_format = ReturnFormat::Markdown;
 
@@ -154,7 +154,7 @@ mod tests {
         let mut page_response = PageResponse::default();
 
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
 
         conf.return_format = ReturnFormat::Markdown;
 
@@ -179,7 +179,7 @@ mod tests {
         let mut page_response = PageResponse::default();
 
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
 
         conf.return_format = ReturnFormat::Text;
 
@@ -204,7 +204,7 @@ mod tests {
         let mut page_response = PageResponse::default();
 
         page_response.content = Some(Bytes::from(markup).into());
-        let page = build(url, page_response);
+        let page = build_with_parse(url, page_response);
 
         conf.return_format = ReturnFormat::Text;
 
@@ -236,7 +236,7 @@ mod tests {
 
         page_response.content = Some(b.into());
 
-        let page = build("https://example.com/example.pdf", page_response);
+        let page = build_with_parse("https://example.com/example.pdf", page_response);
 
         let content = content::transform_content(&page, &conf, &None, &None, &None);