Implement Preprocessing of Pdfs Into Images + Setup Pdfium

libs/chonky/Cargo.toml

@@ -17,6 +17,8 @@ edition.workspace = true
 error-stack = { workspace = true, public = true }
 
 # Public third-party dependencies
+pdfium-render = { version = "0.8.25", public = true }
+image = { version = "0.24.9", public = true }
 
 # Private workspace dependencies
 

libs/chonky/src/lib.rs

@@ -1,26 +1,90 @@
 #![doc = include_str!("../README.md")]
 
-/// Adds two numbers together
-///
-/// # Example
-///
-/// ```rust
-/// use chonky::add;
+use image::DynamicImage;
+use pdfium_render::prelude::*;
+
+use error_stack::Report;
+
+
+pub fn load_pdf<'a>(pdfium: &'a Pdfium, file_path: &str)-> Result<PdfDocument<'a>, Report<PdfiumError>>{
+    let pdf = pdfium
+        .load_pdf_from_file(file_path, None)?;
+    Ok(pdf)
+}
+
+/// Takes in a pdf document and returns a vector list where each page
+/// is processed into a raw image that can be later converted to any image format
 ///
-/// assert_eq!(add(1, 3), 4);
-/// ```
 #[must_use]
-pub const fn add(left: u64, right: u64) -> u64 {
-    left + right
+pub fn pdf_to_images(pdf: &PdfDocument) -> Result<Vec<DynamicImage>,Report<PdfiumError>>{
+    let mut images: Vec<DynamicImage> = Vec::new();
+
+    for page in pdf.pages().iter() {
+
+        let resolution_width = 1000; //may adjust resolution depending on need
+
+        // Render the entire page to an image
+        let rendered_page = page
+            .render_with_config(
+                &PdfRenderConfig::new()
+                    .set_target_width(resolution_width),
+                )?; // Renders the page to a PdfBitmap
+
+        // Convert PdfBitmap to DynamicImage
+        let dynamic_image = rendered_page.as_image();
+        images.push(dynamic_image);
+    }
+
+    Ok(images)
 }
 
+
 #[cfg(test)]
 mod tests {
     use super::*;
 
     #[test]
-    fn it_works() {
-        let result = add(2, 2);
-        assert_eq!(result, 4);
+    fn pdf_load_success() {
+        let pdfium = Pdfium::new(
+            Pdfium::bind_to_library(Pdfium::pdfium_platform_library_name_at_path("./libs/")).unwrap()
+        ); // creates instance so must be global
+
+        let test_pdf_string = "tests/docs/test-doc.pdf";
+
+        let pdf = load_pdf(&pdfium, test_pdf_string);
+
+        assert!(pdf.is_ok());
+    }
+
+    #[test]
+    fn pdf_load_failure() {
+        let pdfium = Pdfium::new(
+            Pdfium::bind_to_library(Pdfium::pdfium_platform_library_name_at_path("./libs/")).unwrap()
+        ); // creates instance so must be global
+
+        let test_pdf_string = "tests/docs/invalid.pdf";
+
+        let pdf = load_pdf(&pdfium, test_pdf_string);
+
+        assert!(pdf.is_err());
     }
+
+    #[test]
+    fn pdf_image_conversion() {
+        let pdfium = Pdfium::new(
+            Pdfium::bind_to_library(Pdfium::pdfium_platform_library_name_at_path("./libs/")).unwrap()
+        ); // creates instance so must be global
+
+        let test_pdf_string =  "tests/docs/test-doc.pdf";
+
+        let pdf = load_pdf(&pdfium, test_pdf_string).unwrap();
+
+        let preprocessed_pdf = pdf_to_images(&pdf).unwrap();
+
+        let num_pages = 38; //number of pages of pdf
+
+        assert_eq!(preprocessed_pdf.len(), num_pages) //length of vector should be number of pages
+
+    }
+
 }

libs/chonky/src/main.rs

@@ -0,0 +1,35 @@
+use std::env;
+use chonky::*;
+use pdfium_render::prelude::*;
+use error_stack::Report;
+
+fn main() -> Result<(), Report<PdfiumError>>{
+    let args: Vec<String> = env::args().collect();// read file path arguments
+
+    if args.len() < 2 {
+        panic!("Path to PDF not inputted")
+    }
+
+    let pdfium = Pdfium::new(
+        Pdfium::bind_to_library(Pdfium::pdfium_platform_library_name_at_path("./libs/")).unwrap()
+    ); // creates instance so must be global
+
+    let pdf = load_pdf(&pdfium, &args[1])?;
+
+    let preprocessed_pdf = pdf_to_images(&pdf)?;
+
+    //for now we will print all these images to a folder
+    // this will be a seperate function in the future once knowledge about error-stack increases
+
+    let output_folder = "./out";
+
+    for (index, image) in preprocessed_pdf.iter().enumerate() {
+        // Generate a unique filename for each page image
+        let file_path = format!("{}/page_{}.png", output_folder, index + 1);
+
+        // Save the image as a PNG file
+        image.save(&file_path).unwrap();
+    }
+
+    Ok(())
+}