fix: drop AttachGuard only after all GlobalRefs in JReaderInputStream are dropped

extractous-core/Cargo.toml

@@ -16,16 +16,19 @@ readme = "README.md"
 keywords = ["unstructured", "tika", "text", "pdf", "parser"]
 categories = ["parsing", "text-processing"]
 
+[features]
+stream-attachguard = []
+
 [[bench]]
 name = "extractor"
 harness = false
 required-features = []
 
 [dependencies]
 libc = { version = "0.2.158" }
-jni = { version = "0.21.1",features = ["invocation"] }
+jni = { version = "0.21.1", features = ["invocation"] }
 thiserror = { version = "1.0.63" }
-bytemuck =  { version = "1.17.1"}
+bytemuck = { version = "1.17.1" }
 # String enums
 strum = { version = "0.26.2" }
 strum_macros = { version = "0.26.2" }

extractous-core/benches/extractor.rs

@@ -10,7 +10,7 @@ fn extract_to_stream(c: &mut Criterion) {
     c.bench_function("extract_to_stream", |b| {
         b.iter(|| {
             // Extract the provided file content to a stream
-            let stream = extractor.extract_file(file_path).unwrap();
+            let (stream, _) = extractor.extract_file(file_path).unwrap();
             // Because stream implements std::io::Read trait we can perform buffered reading
             // For example we can use it to create a BufReader
             let mut reader = BufReader::new(stream);

extractous-core/src/extractor.rs

@@ -202,7 +202,6 @@ impl Extractor {
             self.xml_output,
         )
     }
-
 }
 
 #[cfg(test)]

extractous-core/src/tika/parse.rs

@@ -18,15 +18,15 @@ pub(crate) fn vm() -> &'static JavaVM {
     GRAAL_VM.get_or_init(create_vm_isolate)
 }
 
-fn get_vm_attach_current_thread<'local>() -> ExtractResult<AttachGuard<'local>> {
+fn get_vm_attach_current_thread() -> ExtractResult<AttachGuard<'static>> {
     // Attaching a thead that is already attached is a no-op. Good to have this in case this method
     // is called from another thread
     let env = vm().attach_current_thread()?;
     Ok(env)
 }
 
 fn parse_to_stream(
-    mut env: AttachGuard,
+    mut env: AttachGuard<'static>,
     data_source_val: JValue,
     char_set: &CharSet,
     pdf_conf: &PdfParserConfig,
@@ -60,7 +60,7 @@ fn parse_to_stream(
 
     // Create and process the JReaderResult
     let result = JReaderResult::new(&mut env, call_result_obj)?;
-    let j_reader = JReaderInputStream::new(&mut env, result.java_reader)?;
+    let j_reader = JReaderInputStream::new(env, result.java_reader)?;
 
     Ok((StreamReader { inner: j_reader }, result.metadata))
 }
@@ -71,7 +71,7 @@ pub fn parse_file(
     pdf_conf: &PdfParserConfig,
     office_conf: &OfficeParserConfig,
     ocr_conf: &TesseractOcrConfig,
-    as_xml: bool
+    as_xml: bool,
 ) -> ExtractResult<(StreamReader, Metadata)> {
     let mut env = get_vm_attach_current_thread()?;
 

extractous-core/src/tika/wrappers.rs

@@ -8,31 +8,31 @@ use crate::{Metadata, OfficeParserConfig, PdfParserConfig, TesseractOcrConfig, D
 use bytemuck::cast_slice_mut;
 use jni::objects::{GlobalRef, JByteArray, JObject, JValue};
 use jni::sys::jsize;
-use jni::JNIEnv;
+use jni::{AttachGuard, JNIEnv};
 
 /// Wrapper for [`JObject`]s that contain `org.apache.commons.io.input.ReaderInputStream`
 /// It saves a GlobalRef to the java object, which is cleared when the last GlobalRef is dropped
 /// Implements [`Drop] trait to properly close the `org.apache.commons.io.input.ReaderInputStream`
-#[derive(Clone)]
 pub struct JReaderInputStream {
     internal: GlobalRef,
     buffer: GlobalRef,
     capacity: jsize,
+    #[cfg(feature = "stream-attachguard")]
+    _guard: AttachGuard<'static>,
 }
 
 impl JReaderInputStream {
-    pub(crate) fn new<'local>(
-        env: &mut JNIEnv<'local>,
-        obj: JObject<'local>,
-    ) -> ExtractResult<Self> {
+    pub(crate) fn new(guard: AttachGuard<'static>, obj: JObject<'_>) -> ExtractResult<Self> {
         // Creates new jbyte array
         let capacity = DEFAULT_BUF_SIZE as jsize;
-        let jbyte_array = env.new_byte_array(capacity)?;
+        let jbyte_array = guard.new_byte_array(capacity)?;
 
         Ok(Self {
-            internal: env.new_global_ref(obj)?,
-            buffer: env.new_global_ref(jbyte_array)?,
+            internal: guard.new_global_ref(obj)?,
+            buffer: guard.new_global_ref(jbyte_array)?,
             capacity,
+            #[cfg(feature = "stream-attachguard")]
+            _guard: guard,
         })
     }