revert to previous pdf-extract; remove test for encrypted pdf support (spiceai#5355)

Author: kczimm

Cargo.lock

@@ -10,7 +10,7 @@ version.workspace = true
 
 [dependencies]
 docx-rs = { git = "https://github.com/spiceai/docx-rs", rev="2d86a60b58afb02b157f96b42a92626417e47f71"}
-pdf-extract = { git = "https://github.com/spiceai/pdf-extract.git", rev = "3989f973c061cd805ba6fe905f14b01074844dfc" }
+pdf-extract = { git = "https://github.com/spiceai/pdf-extract", rev = "3ca5e2c904cdf5897f3d671e3b2c4b62b1612b0c" }
 snafu.workspace = true
 bytes.workspace = true
 tokio = { workspace = true, features = ["sync"] }

crates/document_parse/Cargo.toml

@@ -127,61 +127,6 @@ async fn s3_federation() -> Result<(), anyhow::Error> {
         .await
 }
 
-#[tokio::test]
-async fn s3_pdfs() -> Result<(), anyhow::Error> {
-    let _tracing = init_tracing(Some("integration=debug,info"));
-
-    let mut dataset = Dataset::new("s3://spiceai-public-datasets/test_pdf_files", "pdfs");
-    dataset.params = Some(Params::from_string_map(
-        vec![
-            ("file_format".to_string(), "pdf".to_string()),
-            ("client_timeout".to_string(), "120s".to_string()),
-        ]
-        .into_iter()
-        .collect(),
-    ));
-
-    test_request_context()
-        .scope(async {
-            let app = AppBuilder::new("s3_pdfs").with_dataset(dataset).build();
-
-            let status = status::RuntimeStatus::new();
-            let df = get_test_datafusion(Arc::clone(&status));
-
-            let rt = Runtime::builder()
-                .with_datafusion(df)
-                .with_app(app)
-                .build()
-                .await;
-            let cloned_rt = Arc::new(rt.clone());
-
-            // Set a timeout for the test
-            tokio::select! {
-                () = tokio::time::sleep(std::time::Duration::from_secs(60)) => {
-                    return Err(anyhow::anyhow!("Timed out waiting for datasets to load"));
-                }
-                () = cloned_rt.load_components() => {}
-            }
-
-            let mut query_result = rt
-                .datafusion()
-                .query_builder("SELECT * FROM pdfs")
-                .build()
-                .run()
-                .await
-                .map_err(|e| anyhow::anyhow!(e))?;
-            let mut batches = vec![];
-            while let Some(batch) = query_result.data.next().await {
-                batches.push(batch?);
-            }
-
-            assert_eq!(batches.len(), 2);
-
-            Ok(())
-        })
-        .await
-}
-
 #[tokio::test]
 async fn s3_hive_partitioning() -> Result<(), anyhow::Error> {
     let _tracing = init_tracing(Some("integration=debug,info"));