feat: detect image format from magic bytes, not file extension

- Fixes misnamed files (e.g. JPEG saved as .png) being skipped
- Uses image::ImageReader with guessed format from content
- Fixes Android screenshots with wrong extension being skipped
- New test: misnamed_jpeg_as_png_still_scanned
- 22 tests passing

src/lib.rs

@@ -1,8 +1,10 @@
 use anyhow::Result;
 use image::imageops::FilterType;
+use image::ImageReader;
 use sha2::{Digest, Sha256};
 use std::collections::{HashMap, HashSet};
 use std::fs;
+use std::io::Cursor;
 use std::path::{Path, PathBuf};
 use walkdir::WalkDir;
 
@@ -48,8 +50,14 @@ pub fn scan_images(root: &Path) -> Result<Vec<ImageEntry>> {
             }
         };
         let sha256 = format!("{:x}", Sha256::digest(&bytes));
-        let img = match image::open(path) {
-            Ok(i) => i,
+        let img = match ImageReader::new(Cursor::new(&bytes)).with_guessed_format() {
+            Ok(reader) => match reader.decode() {
+                Ok(i) => i,
+                Err(e) => {
+                    eprintln!("warning: skipping {}: {e}", path.display());
+                    continue;
+                }
+            },
             Err(e) => {
                 eprintln!("warning: skipping {}: {e}", path.display());
                 continue;

tests/image_phase.rs

@@ -133,3 +133,16 @@ fn cli_binary_reports_duplicates() {
     assert!(stdout.contains("[exact]"), "output should contain exact groups: {stdout}");
     assert!(stdout.contains("[similar]"), "output should contain similar groups: {stdout}");
 }
+
+#[test]
+fn misnamed_jpeg_as_png_still_scanned() {
+    // fake_png.png is actually JPEG data with .png extension
+    let dir = Path::new("/a0/usr/projects/deduper/.a0proj/test_media/images");
+    let entries = scan_images(dir).expect("scan");
+    let fake = entries.iter().find(|e| e.path.ends_with("fake_png.png"));
+    assert!(fake.is_some(), "misnamed JPEG-as-PNG should be scanned via magic bytes");
+    // should have same hash as orig.jpg since it's a copy
+    let orig = entries.iter().find(|e| e.path.ends_with("orig.jpg")).unwrap();
+    let fake = fake.unwrap();
+    assert_eq!(orig.sha256, fake.sha256, "same content = same sha256");
+}