Merge pull request 'Capture oversized screenshots instead of dropping them' (#6) from fix/oversized-screenshot-capture into main

docs/TESTING.md

@@ -16,14 +16,14 @@ maCopy has an automated suite. There is no separate “live” Storybook/UI docs
 | `npm run lint` | TypeScript `tsc --noEmit` |
 | `npm run check` | Lint + all tests (CI-style gate) |
 
-Current counts (approx.): **~60 frontend** + **~52 Rust** unit tests.
+Current counts (approx.): **~60 frontend** + **~65 Rust** unit tests.
 
 ## What is covered
 
 ### Frontend (`src/**/*.test.tsx`, `transforms.test.ts`)
 
 - App: load entries, paste invoke, delete via keyboard
-- ClipboardList: selection, multi-select modifiers, image stubs/thumbs, Alt=plain paste
+- ClipboardList: selection, multi-select modifiers, image stubs/thumbs (including truncated thumb previews), Alt=plain paste
 - ContextMenu: Paste / Paste plain / transforms / Pin / Delete
 - SettingsPanel: toggles, history limit, hotkey recorder, **Launch at login → autostart enable**
 - SearchBar basics
@@ -34,18 +34,23 @@ Tauri APIs are mocked in `src/test/setup.ts` (invoke, window, events, clipboard,
 ### Backend (`src-tauri/src/**` `#[cfg(test)]`)
 
 - SQLite CRUD, FTS5, trim + pin preservation, settings
-- List previews: truncate text, omit image blobs, return thumbnails
+- List previews: truncate text, omit image blobs, return thumbnails (including stub `[image WxH]` rows)
+- Search excludes image rows (text-only FTS)
 - Sensitive redaction (card / password / banking) while `get_entry` stays full
 - Hotkey parser (`cmd+\`` etc.)
-- Image thumbnail / PNG helpers
+- Image thumbnail / PNG helpers + `fit_rgba_under_bytes` against production `MAX_IMAGE_BYTES`
+- **Oversized / Retina screenshot prepare** — never skip; downscale + thumb + DB list preview
 - Clipboard content hashing fingerprints
 
 ## What is *not* covered (yet)
 
 - End-to-end UI on a real Mac (no Playwright against the Tauri window)
+- Live pasteboard polling (requires macOS Accessibility / real `NSPasteboard`)
 - Accessibility permission / AppleScript paste in CI
 - Notarized release smoke tests
 - Performance benchmarks in CI
+- Prefer-image-when-text-also-present (poller text-wins path)
+- Exclude-apps / pause-monitoring integration
 
 ## Recommended workflow
 

src-tauri/src/clipboard.rs

@@ -1,5 +1,5 @@
 use crate::db::Database;
-use crate::image_util::{encode_rgba_to_png, make_thumbnail_data_uri};
+use crate::image_util::{encode_rgba_to_png, fit_rgba_under_bytes, make_thumbnail_data_uri};
 use arboard::Clipboard;
 use base64::Engine;
 use sha2::{Digest, Sha256};
@@ -8,10 +8,23 @@ use std::sync::{Arc, Mutex};
 use std::time::Duration;
 use tauri::{AppHandle, Emitter};
 
-/// Skip encoding RGBA buffers larger than this (keeps history DB usable).
-const MAX_IMAGE_BYTES: usize = 16 * 1024 * 1024;
+/// Max RGBA bytes kept for encode/storage (keeps history DB usable).
+/// Larger pasteboard images are downscaled — never silently dropped.
+pub(crate) const MAX_IMAGE_BYTES: usize = 16 * 1024 * 1024;
 /// Cap stored full PNG size; list uses a separate tiny thumbnail.
-const MAX_STORED_PNG_BYTES: usize = 1_500_000;
+pub(crate) const MAX_STORED_PNG_BYTES: usize = 1_500_000;
+
+/// Result of turning a pasteboard RGBA buffer into DB-ready fields.
+#[derive(Debug, Clone)]
+pub(crate) struct PreparedImage {
+    pub content: String,
+    pub thumbnail: Option<String>,
+    pub width: u32,
+    pub height: u32,
+    pub downscaled: bool,
+    /// True when `content` is a stub and only the thumbnail is visual.
+    pub thumbnail_only: bool,
+}
 
 /// Hash arbitrary bytes for deduplication.
 pub fn hash_content(data: &[u8]) -> String {
@@ -21,7 +34,7 @@ pub fn hash_content(data: &[u8]) -> String {
 }
 
 /// Cheap image fingerprint: dimensions + length + edge samples (not full-buffer SHA).
-fn fingerprint_image(width: usize, height: usize, bytes: &[u8]) -> String {
+pub(crate) fn fingerprint_image(width: usize, height: usize, bytes: &[u8]) -> String {
     let mut hasher = Sha256::new();
     hasher.update((width as u64).to_le_bytes());
     hasher.update((height as u64).to_le_bytes());
@@ -37,6 +50,51 @@ fn fingerprint_image(width: usize, height: usize, bytes: &[u8]) -> String {
     hex::encode(hasher.finalize())
 }
 
+/// Downscale if needed, build thumbnail + PNG (or stub). Never returns "skip".
+/// This is the regression lock for Retina screenshots that used to vanish.
+pub(crate) fn prepare_clipboard_image(
+    width: u32,
+    height: u32,
+    rgba: &[u8],
+    max_rgba_bytes: usize,
+    max_png_bytes: usize,
+) -> Result<PreparedImage, String> {
+    let orig_w = width;
+    let orig_h = height;
+    let (w, h, fitted) = fit_rgba_under_bytes(width, height, rgba, max_rgba_bytes);
+    let downscaled = w < orig_w || h < orig_h;
+    let thumb = make_thumbnail_data_uri(w, h, &fitted).ok();
+
+    match encode_rgba_to_png(w, h, &fitted) {
+        Ok(png_data) if png_data.len() <= max_png_bytes => {
+            let b64 = base64::engine::general_purpose::STANDARD.encode(&png_data);
+            Ok(PreparedImage {
+                content: format!("data:image/png;base64,{}", b64),
+                thumbnail: thumb,
+                width: w,
+                height: h,
+                downscaled,
+                thumbnail_only: false,
+            })
+        }
+        Ok(_) => {
+            let stub = format!("[image {}x{}]", w, h);
+            if thumb.is_none() {
+                return Err("PNG too large and thumbnail failed".into());
+            }
+            Ok(PreparedImage {
+                content: stub,
+                thumbnail: thumb,
+                width: w,
+                height: h,
+                downscaled,
+                thumbnail_only: true,
+            })
+        }
+        Err(e) => Err(format!("PNG encode failed: {}", e)),
+    }
+}
+
 /// Shared state so paste_and_refocus can update the hash the polling
 /// thread compares against, preventing re-insertion of pasted content.
 pub type LastHash = Arc<Mutex<String>>;
@@ -115,56 +173,52 @@ pub fn start_polling(app: AppHandle, db: Arc<Database>, paused: Arc<AtomicBool>)
                 }
                 _ => {
                     if let Ok(img) = clipboard.get_image() {
-                        if img.bytes.len() > MAX_IMAGE_BYTES {
-                            log::warn!(
-                                "Skipping oversized clipboard image ({} bytes)",
-                                img.bytes.len()
-                            );
-                            continue;
-                        }
-
+                        // Fingerprint the original pasteboard image for dedup, then
+                        // downscale if needed. Previously we skipped anything over
+                        // MAX_IMAGE_BYTES — Retina screenshots never appeared.
                         let h = fingerprint_image(img.width, img.height, &img.bytes);
                         if h == current_hash {
                             continue;
                         }
                         *hash_for_thread.lock().unwrap() = h.clone();
 
-                        let w = img.width as u32;
-                        let hgt = img.height as u32;
-                        let thumb = make_thumbnail_data_uri(w, hgt, &img.bytes).ok();
-
-                        match encode_rgba_to_png(w, hgt, &img.bytes) {
-                            Ok(png_data) if png_data.len() <= MAX_STORED_PNG_BYTES => {
-                                let b64 =
-                                    base64::engine::general_purpose::STANDARD.encode(&png_data);
-                                let data_uri = format!("data:image/png;base64,{}", b64);
-                                if let Err(e) =
-                                    db.insert_entry(&data_uri, "image", &h, thumb.as_deref())
-                                {
+                        match prepare_clipboard_image(
+                            img.width as u32,
+                            img.height as u32,
+                            &img.bytes,
+                            MAX_IMAGE_BYTES,
+                            MAX_STORED_PNG_BYTES,
+                        ) {
+                            Ok(prepared) => {
+                                if prepared.downscaled {
+                                    log::info!(
+                                        "Downscaled clipboard image {}x{} → {}x{} for storage",
+                                        img.width,
+                                        img.height,
+                                        prepared.width,
+                                        prepared.height
+                                    );
+                                }
+                                if prepared.thumbnail_only {
+                                    log::warn!(
+                                        "Stored thumbnail only for large PNG ({}x{})",
+                                        prepared.width,
+                                        prepared.height
+                                    );
+                                }
+                                if let Err(e) = db.insert_entry(
+                                    &prepared.content,
+                                    "image",
+                                    &h,
+                                    prepared.thumbnail.as_deref(),
+                                ) {
                                     log::error!("DB insert error (image): {}", e);
                                 } else {
                                     let _ = app.emit("clipboard-changed", ());
                                 }
                                 trim_if_needed(&db);
                             }
-                            Ok(png_data) => {
-                                // Still store a thumbnail-only entry so the list can show something.
-                                if let Some(ref t) = thumb {
-                                    let stub = format!("[image {}x{}]", w, hgt);
-                                    if db
-                                        .insert_entry(&stub, "image", &h, Some(t))
-                                        .is_ok()
-                                    {
-                                        let _ = app.emit("clipboard-changed", ());
-                                        trim_if_needed(&db);
-                                    }
-                                }
-                                log::warn!(
-                                    "Stored thumbnail only for large PNG ({} bytes encoded)",
-                                    png_data.len()
-                                );
-                            }
-                            Err(e) => log::error!("PNG encode failed: {}", e),
+                            Err(e) => log::error!("Image prepare failed: {}", e),
                         }
                     }
                 }
@@ -184,6 +238,8 @@ fn trim_if_needed(db: &Database) {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use crate::db::Database;
+    use crate::image_util::data_uri_to_rgba;
 
     #[test]
     fn hash_content_deterministic() {
@@ -241,4 +297,119 @@ mod tests {
         let result = encode_rgba_to_png(0, 0, &[]);
         assert!(result.is_ok() || result.is_err());
     }
+
+    /// Regression: Retina-class screenshots used to be skipped entirely when
+    /// RGBA exceeded MAX_IMAGE_BYTES. They must still produce a listable entry.
+    #[test]
+    fn prepare_retina_screenshot_is_never_skipped() {
+        // ~14" MBP Retina full-screen-ish: far over the 16MB RGBA cap.
+        let w = 3024u32;
+        let h = 1964u32;
+        let rgba_len = (w as usize) * (h as usize) * 4;
+        assert!(
+            rgba_len > MAX_IMAGE_BYTES,
+            "fixture must exceed MAX_IMAGE_BYTES to exercise the bug"
+        );
+        let rgba = vec![60u8; rgba_len];
+
+        let prepared =
+            prepare_clipboard_image(w, h, &rgba, MAX_IMAGE_BYTES, MAX_STORED_PNG_BYTES)
+                .expect("oversized screenshot must prepare, not skip");
+
+        assert!(prepared.downscaled);
+        assert!(
+            (prepared.width as usize) * (prepared.height as usize) * 4 <= MAX_IMAGE_BYTES
+        );
+        assert!(prepared.thumbnail.is_some());
+        let thumb = prepared.thumbnail.as_ref().unwrap();
+        assert!(
+            thumb.starts_with("data:image/png;base64,"),
+            "list preview needs a real data URI thumb"
+        );
+        assert!(!prepared.content.is_empty());
+    }
+
+    #[test]
+    fn prepare_small_image_keeps_full_png() {
+        let w = 32u32;
+        let h = 24u32;
+        let mut rgba = vec![0u8; (w * h * 4) as usize];
+        for px in rgba.chunks_exact_mut(4) {
+            px.copy_from_slice(&[10, 200, 80, 255]);
+        }
+        let prepared =
+            prepare_clipboard_image(w, h, &rgba, MAX_IMAGE_BYTES, MAX_STORED_PNG_BYTES).unwrap();
+        assert!(!prepared.downscaled);
+        assert!(!prepared.thumbnail_only);
+        assert!(prepared.content.starts_with("data:image/png;base64,"));
+        let (rw, rh, _) = data_uri_to_rgba(&prepared.content).unwrap();
+        assert_eq!((rw, rh), (w, h));
+    }
+
+    #[test]
+    fn prepare_thumbnail_only_when_png_cap_tiny() {
+        let w = 200u32;
+        let h = 200u32;
+        // High-entropy noise → PNG won't shrink under a tiny cap.
+        let mut rgba = vec![0u8; (w * h * 4) as usize];
+        for (i, px) in rgba.chunks_exact_mut(4).enumerate() {
+            let v = (i % 251) as u8;
+            px.copy_from_slice(&[v, v.wrapping_mul(3), v.wrapping_mul(7), 255]);
+        }
+        let prepared = prepare_clipboard_image(w, h, &rgba, MAX_IMAGE_BYTES, 200).unwrap();
+        assert!(prepared.thumbnail_only);
+        assert!(prepared.content.starts_with("[image "));
+        assert!(prepared
+            .thumbnail
+            .as_ref()
+            .unwrap()
+            .starts_with("data:image/png;base64,"));
+    }
+
+    #[test]
+    fn prepare_then_insert_shows_thumb_in_list() {
+        let w = 2500u32;
+        let h = 1700u32;
+        let rgba = vec![90u8; (w * h * 4) as usize];
+        let prepared =
+            prepare_clipboard_image(w, h, &rgba, MAX_IMAGE_BYTES, MAX_STORED_PNG_BYTES).unwrap();
+
+        let db = Database::in_memory().unwrap();
+        let hash = fingerprint_image(w as usize, h as usize, &rgba);
+        let id = db
+            .insert_entry(
+                &prepared.content,
+                "image",
+                &hash,
+                prepared.thumbnail.as_deref(),
+            )
+            .unwrap();
+
+        let list = db.get_entries(10).unwrap();
+        assert_eq!(list.len(), 1);
+        assert_eq!(list[0].id, id);
+        assert_eq!(list[0].content_type, "image");
+        assert!(
+            list[0].content.starts_with("data:image/png;base64,"),
+            "UI renders <img> only when list content is a data URI; got {:?}",
+            &list[0].content[..list[0].content.len().min(40)]
+        );
+    }
+
+    #[test]
+    fn prepare_preserves_landscape_aspect_roughly() {
+        let w = 4800u32;
+        let h = 1600u32; // 29.3 MB RGBA → must downscale
+        let rgba = vec![1u8; (w * h * 4) as usize];
+        assert!((w as usize) * (h as usize) * 4 > MAX_IMAGE_BYTES);
+        let prepared =
+            prepare_clipboard_image(w, h, &rgba, MAX_IMAGE_BYTES, MAX_STORED_PNG_BYTES).unwrap();
+        assert!(prepared.downscaled);
+        let orig_aspect = w as f64 / h as f64;
+        let new_aspect = prepared.width as f64 / prepared.height as f64;
+        assert!(
+            (orig_aspect - new_aspect).abs() < 0.05,
+            "aspect {orig_aspect} vs {new_aspect}"
+        );
+    }
 }

src-tauri/src/db.rs

@@ -754,6 +754,38 @@ mod tests {
         assert_eq!(full.content, blob);
     }
 
+    #[test]
+    fn list_shows_thumb_for_stub_image_entry() {
+        // Oversized-PNG path stores "[image WxH]" + thumbnail — list must still
+        // expose a data URI so the UI can render a preview.
+        let db = test_db();
+        let thumb = "data:image/png;base64,stubthumb";
+        db.insert_entry("[image 2800x1600]", "image", "hstub", Some(thumb))
+            .unwrap();
+
+        let entries = db.get_entries(10).unwrap();
+        assert_eq!(entries[0].content, thumb);
+        assert!(entries[0].truncated);
+        assert_eq!(entries[0].content_type, "image");
+    }
+
+    #[test]
+    fn search_entries_excludes_images() {
+        let db = test_db();
+        db.insert_entry("findme text", "text", "ht1", None).unwrap();
+        db.insert_entry(
+            "data:image/png;base64,findme",
+            "image",
+            "hi1",
+            Some("data:image/png;base64,thumb"),
+        )
+        .unwrap();
+
+        let hits = db.search_entries("findme", 50).unwrap();
+        assert_eq!(hits.len(), 1);
+        assert_eq!(hits[0].content_type, "text");
+    }
+
     #[test]
     fn list_entries_truncate_long_text() {
         let db = test_db();

src-tauri/src/image_util.rs

@@ -90,6 +90,27 @@ pub fn make_thumbnail_data_uri(width: u32, height: u32, rgba: &[u8]) -> Result<S
     rgba_to_data_uri(tw, th, &thumb)
 }
 
+/// Shrink RGBA so `width * height * 4 <= max_bytes`, preserving aspect ratio.
+/// Returns the original buffer unchanged when already under the limit.
+pub fn fit_rgba_under_bytes(
+    width: u32,
+    height: u32,
+    rgba: &[u8],
+    max_bytes: usize,
+) -> (u32, u32, Vec<u8>) {
+    let need = (width as usize)
+        .saturating_mul(height as usize)
+        .saturating_mul(4);
+    if need == 0 || need <= max_bytes {
+        return (width.max(1), height.max(1), rgba.to_vec());
+    }
+    let max_pixels = (max_bytes / 4).max(1) as f64;
+    let pixels = (width as f64) * (height as f64);
+    let scale = (max_pixels / pixels).sqrt();
+    let max_edge = ((width.max(height) as f64) * scale).floor().max(1.0) as u32;
+    downscale_rgba(width, height, rgba, max_edge)
+}
+
 pub fn data_uri_to_rgba(data_uri: &str) -> Result<(u32, u32, Vec<u8>), String> {
     let b64 = data_uri
         .split_once(',')
@@ -139,4 +160,46 @@ mod tests {
         let uri = make_thumbnail_data_uri(50, 50, &rgba).unwrap();
         assert!(uri.starts_with("data:image/png;base64,"));
     }
+
+    #[test]
+    fn fit_rgba_leaves_small_images_alone() {
+        let rgba = vec![1u8; 10 * 10 * 4];
+        let (w, h, out) = fit_rgba_under_bytes(10, 10, &rgba, 16 * 1024 * 1024);
+        assert_eq!((w, h), (10, 10));
+        assert_eq!(out.len(), rgba.len());
+    }
+
+    #[test]
+    fn fit_rgba_shrinks_oversized_buffer() {
+        let w = 2500u32;
+        let h = 1700u32;
+        let rgba = vec![40u8; (w * h * 4) as usize];
+        let max = 16 * 1024 * 1024;
+        assert!(rgba.len() > max);
+        let (tw, th, out) = fit_rgba_under_bytes(w, h, &rgba, max);
+        assert!(out.len() <= max);
+        assert!(tw < w || th < h);
+        assert_eq!(out.len(), (tw * th * 4) as usize);
+    }
+
+    #[test]
+    fn fit_rgba_uses_production_max_image_bytes_cap() {
+        // Lock the constant the clipboard poller uses — if someone raises the
+        // skip threshold again without downscale, this fixture still exceeds it.
+        let max = crate::clipboard::MAX_IMAGE_BYTES;
+        let w = 3024u32;
+        let h = 1964u32;
+        let rgba = vec![0u8; (w * h * 4) as usize];
+        assert!(rgba.len() > max);
+        let (tw, th, out) = fit_rgba_under_bytes(w, h, &rgba, max);
+        assert!(out.len() <= max);
+        assert!(tw > 0 && th > 0);
+    }
+
+    #[test]
+    fn downscale_handles_zero_dimensions() {
+        let (w, h, out) = downscale_rgba(0, 0, &[], 96);
+        assert!(w >= 1 && h >= 1);
+        assert_eq!(out.len(), 4); // single transparent pixel path
+    }
 }

src/components/ClipboardList.test.tsx

@@ -105,6 +105,22 @@ describe("ClipboardList", () => {
     expect(img).toHaveAttribute("src", "data:image/png;base64,abc");
   });
 
+  it("renders thumbnail data URI for truncated oversized-image list rows", () => {
+    // Backend list preview swaps full PNG for thumbnail; truncated stays true.
+    const entries = [
+      makeEntry({
+        content: "data:image/png;base64,thumbpreview",
+        content_type: "image",
+        truncated: true,
+      }),
+    ];
+    render(<ClipboardList {...defaultProps} entries={entries} showImages={true} />);
+    expect(screen.getByAltText("Clipboard image")).toHaveAttribute(
+      "src",
+      "data:image/png;base64,thumbpreview"
+    );
+  });
+
   it("renders image stub when blob was omitted from list payload", () => {
     const entries = [
       makeEntry({ content: "", content_type: "image", truncated: true }),