Add endpoint to read and extract text from PDF files without streaming

app/MindWork AI Studio/Tools/Services/RustService.Retrieval.cs

@@ -0,0 +1,16 @@
+namespace AIStudio.Tools.Services;
+
+public sealed partial class RustService
+{
+    public async Task<string> GetPDFText(string filePath)
+    {
+        var response = await this.http.GetAsync($"/retrieval/fs/read/pdf?file_path={filePath}");
+        if (!response.IsSuccessStatusCode)
+        {
+            this.logger!.LogError($"Failed to read the PDF file due to an network error: '{response.StatusCode}'");
+            return string.Empty;
+        }
+
+        return await response.Content.ReadAsStringAsync();
+    }
+}

runtime/src/file_data.rs

@@ -146,6 +146,30 @@ async fn stream_text_file(file_path: &str) -> Result<ChunkStream> {
     Ok(Box::pin(stream))
 }
 
+#[get("/retrieval/fs/read/pdf?<file_path>")]
+pub fn read_pdf(_token: APIToken, file_path: String) -> String {
+    let pdfium = Pdfium::default();
+    let doc = match pdfium.load_pdf_from_file(&file_path, None) {
+        Ok(document) => document,
+        Err(e) => return e.to_string(),
+    };
+
+    let mut pdf_content = String::new();
+    for page in doc.pages().iter() {
+        let content = match page.text().map(|text_content| text_content.all()) {
+            Ok(content) => content,
+            Err(_) => {
+                continue
+            }
+        };
+
+        pdf_content.push_str(&content);
+        pdf_content.push_str("\n\n");
+    }
+
+    pdf_content
+}
+
 async fn stream_pdf(file_path: &str) -> Result<ChunkStream> {
     let path = file_path.to_owned();
     let (tx, rx) = mpsc::channel(10);

runtime/src/runtime_api.rs

@@ -79,6 +79,7 @@ pub fn start_runtime_api() {
                 crate::environment::get_config_directory,
                 crate::environment::read_user_language,
                 crate::file_data::extract_data,
+                crate::file_data::read_pdf,
                 crate::log::get_log_paths,
             ])
             .ignite().await.unwrap()