Fix issue #86 by upgrade to MuPDF 1.24.4 (#97)

Co-authored-by: vincent <[email protected]>

mupdf-sys/build.rs

@@ -351,6 +351,13 @@ fn build_libmupdf() {
                 profile
             );
         }
+
+        if profile == "Debug" {
+            println!("cargo:rustc-link-lib=dylib=ucrtd");
+            println!("cargo:rustc-link-lib=dylib=vcruntimed");
+            println!("cargo:rustc-link-lib=dylib=msvcrtd");
+        }
+
         println!("cargo:rustc-link-lib=dylib=libmupdf");
         println!("cargo:rustc-link-lib=dylib=libthirdparty");
     } else {

mupdf-sys/wrapper.c

@@ -3012,7 +3012,7 @@ void mupdf_pdf_page_set_crop_box(fz_context *ctx, pdf_page *page, fz_rect rect,
 {
     fz_try(ctx)
     {
-        fz_rect mediabox = pdf_bound_page(ctx, page);
+        fz_rect mediabox = pdf_bound_page(ctx, page, FZ_MEDIA_BOX);
         pdf_obj *obj = pdf_dict_get_inheritable(ctx, page->obj, PDF_NAME(MediaBox));
         if (obj)
         {

tests/test_issues.rs

@@ -80,3 +80,21 @@ fn test_issue_60_display_list() {
         })
         .collect();
 }
+
+#[test]
+fn test_issue_86_invalid_utf8() {
+    let doc = PdfDocument::open("tests/files/utf8-error-on-this-file.pdf").unwrap();
+    for (idx, page) in doc.pages().unwrap().enumerate() {
+        let page = page.unwrap();
+        let text = page.to_text();
+        assert!(text.is_ok());
+        println!("page: {idx}, text: {}", text.unwrap());
+
+        let json = page.stext_page_as_json_from_page(1.0);
+        assert!(json.is_ok());
+
+        // Validate JSON parsing
+        let parsed_json: Result<serde_json::Value, _> = serde_json::from_str(&json.unwrap());
+        assert!(parsed_json.is_ok());
+    }
+}