## Summary
Use `std::string_view` and case-insensitive comparisons to avoid string
allocations during CSS parsing.
**Hot path:** `resolveStyle` (called per HTML start tag during chapter
rendering) now does zero heap allocations. Previously it allocated a
normalized tag string, a vector of class strings, and a composite key
per class. For a chapter with ~2000 tags × 2 classes each, that's ~12
000 small short-lived allocations eliminated per page render — primarily
a heap-fragmentation win on the ESP32-C3's ~380KB RAM.
**Cold path:** CSS load no longer allocates per-rule selector vectors or
per-token strings; `splitOnChar`/`splitWhitespace` are gone, replaced
with callback-based tokenization (`forEachDelimitedToken`).
**Behavioral notes:**
- The selector `unordered_map` now uses an ASCII-case-insensitive
hash/equal. Selectors are stored with their original case rather than
pre-lowercased; the observable lookup result is unchanged.
- `stripTrailingImportant` is now case-insensitive (per CSS spec;
previously matched only lowercase `!important`).
**Cache compatibility:** `CSS_CACHE_VERSION` unchanged. Old caches
(lowercase selectors) load correctly under the new lookup; new caches
will contain verbatim-case selectors — both forms work.
---
### AI Usage
Did you use AI tools to help write this code? _**PARTIALLY**_
## Summary
This PR optimizes the text rendering process by skipping underline style
calculations and measurements during the initial font cache scanning
phase. This prevents excessive and unnecessary SD card reads on pages
with heavy use of underlines (e.g., Table of Contents pages).
### **The Problem**
During the first rendering pass (the font cache scan pass used to
collect text for prewarming), `GfxRenderer::drawText()` early-returns
after recording text as expected. However, `TextBlock::render()`
continued past this point to execute the underline decoration logic.
Because underline calculation calls `getTextWidth()` and
`getTextAdvanceX()`, it triggered immediate glyph lookups via
`EpdFont::getGlyph()`. Since the SD card font had not been prewarmed yet
at this stage, the lookups fell back to the `glyphMissHandler`,
resulting in hundreds of individual, slow SD card reads into a limited
8-slot ring buffer.
### **The Fix**
1. **Exposed Scan State:** Added `GfxRenderer::isFontCacheScanning()` to
safely check if the font cache manager is currently in
text-collection/scan mode.
2. **Bypassed Underline Logic:** Modified `TextBlock::render()` to check
this state and skip underline measurement and drawing entirely while
scanning is active.
> [!NOTE]
> The text itself is still properly captured for prewarming via
`drawText()`. The underlines will be safely and efficiently calculated
and drawn during the actual render pass after the fonts have been
completely prewarmed.
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**< YES >**_
## Summary
* **What is the goal of this PR?**
* This PR ports over Crossink's handling of large EPUBs that helps
prevent crashes during open after the book metadata cache is built.
* **What changes are included?**
* Removes the post-indexing ZIP-wide CSS discovery pass that built an
in-memory map of every ZIP entry.
* Reuses `content.opf` parsing to collect declared CSS files without
writing spine entries again.
* Temporarily releases the loaded book metadata cache while rebuilding
CSS for cached books.
* Parses CSS before reloading `book.bin` after a fresh cache build,
leaving more heap available during CSS rule parsing.
## Additional Context
* User reported their EPUB opening fine on Crossink but would crash on
Crosspoint. Verified this claim on my own devices.
* The crash this addresses happened after `book.bin` was successfully
built, when CSS discovery allocated a large `unordered_map` for ~3k EPUB
ZIP entries.
* Tradeoff: CSS files not declared in `content.opf` are no longer
discovered by scanning the full ZIP. This avoids the high-risk memory
allocation but improperly formatted EPUBs (ones that don't declare their
CSS styles in `content.opf` will render without styling and fallback to
inline styles.
* User provided epub that was crashing prior to this change:
https://www.mediafire.com/file/g57ea4mj13iunvh/Quang+%C3%82m+Chi+Ngo%E1%BA%A1i+-+Nh%C4%A9+C%C4%83n.epub/file
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**< YES >**_
## Summary
* **What is the goal of this PR?**
Mainly fixes#2196, but also allows users to set a sleep time of 1~30
min or to never sleep.
* **What changes are included?**
1. Revert changes from #1948 and #2137 (Sorry @Uri-Tauber)
2. Cherrypick changes from
https://github.com/uxjulia/CrossInk/commit/0cb91c3dd1c63f0559d4392fadc4a8340af5bdde
and
https://github.com/uxjulia/CrossInk/commit/4e6bd634b6ede51ff1a7107ba3e8efcd152bc6ee
for a timeout interval picker UI, the custom sleep time feature, and
migrate the existing setting in v1.3.0, `sleepTimeout`, to
`sleepTimeoutMinutes` (Thanks! @uxjulia)
3. Add a "Never" at the far right of the timeout interval picker
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**PARTIALLY**_
---------
Co-authored-by: Julia <julia@uxj.io>
Chapter selection previously only used the spine index, causing
navigation to always land on page 0 of the spine item. Sub-chapters that
share a spine file but differ by anchor (e.g. `chapter.xhtml#sec2`) were
silently ignored. Now the TOC anchor is passed through `ChapterResult`
and applied via the existing `pendingAnchor` mechanism.
## Summary
* This PR implements navigation to sub-chapters which didn't work
correctly previously. If a sub-chapter of the current top-level chapter
was selected, nothing happened. If a sub-chapter of another top-level
chapter was selected, reader switched to the beginning of the top-level
chapter.
* In addition, chapters now always start from a new page. This fixes
anchor to page calculation for the cases when the actual chapter content
doesn't fit on the page where the corresponding ToC anchor was found.
## Additional Context
* I might misuse `pendingAnchor` here which was previously used for
footnote navigation, please double check. I'm open to suggestions for
improvements.
* Note that the chapter selected by default when
`EpubReaderChapterSelectionActivity` opens is still wrong. I'm going to
fix this separately. This PR addresses only navigation to the selected
chapter.
* I tested this PR on my X4 and verified that navigation to a different
sub-chapter works correctly, both inside and outside the current spine.
* Some of the changes were borrowed from
https://github.com/crosspoint-reader/crosspoint-reader/pull/1455
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**PARTIALLY**_
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Tiled grayscale rendering to drop the storeBwBuffer peak largest
contiguous free block (the value that actually drives OOM on the C3)
from ~114 KB to ~82-90 KB.
This renders each grayscale plane band-by-band into a small (~8 KB)
scratch and
streams each band straight to controller RAM (community-sdk
writeGrayscalePlaneStrip), leaving the BW framebuffer intact. No save,
no
restore; controller RAM is re-synced for the next differential turn
directly
from the live framebuffer.
Three writers honor the active band target so per-band re-rendering
stays cheap
and correct:
- drawPixel (text) redirects writes to the band scratch and clips to it.
- renderCharImpl skips glyphs whose physical y-extent is outside the
band before
the bitmap decode (glyphIntersectsStrip), so the per-band re-render
doesn't
pay N x glyph decode.
- DirectPixelWriter (images) writes the band scratch via getWriteTarget
instead
of the framebuffer. Without this, image pixels wrote the live BW frame
directly and cleanup re-synced that corruption, leaving thin outlines
after
navigating away from an image.
Controller specifics live in the SDK (X4 setRamArea windowing, X3 PTL);
the
reader checks supportsStripGrayscale() and is otherwise
controller-agnostic.
Measured on hardware (X4 and X3, text and images, visually correct):
- Grayscale scratch ~8 KB vs ~50 KB save; largest contiguous free block
held at
full size during grayscale instead of dropping ~25-32 KB.
- X4 text page about +25 ms/page; X3 page time is dominated by its
intrinsic
grayscale refresh, not tiling.
Depends on community-sdk #13 (the writeGrayscalePlaneStrip API). The
submodule
bump here points at that branch, so until #13 merges the submodule won't
resolve
from upstream and CI will fail there; keeping this a draft until then.
Will
rebase onto master and re-point the submodule to the merged SDK commit
once #13
lands.
Did you use AI tools to help write this code? partial
## Summary
Minor cleanup flagged by newer gcc:
- Removed unused variables
- Removed unimplemented function declaration
- `static` -> `inline` to avoid per-TU duplication
---
### AI Usage
Did you use AI tools to help write this code? _**NO**_
SdFat's SdSpiCard tracks SPI bus state with an unsynchronized
m_spiActive bool. When two tasks call into SdFat concurrently they can
confuse that state machine, ending with one task calling
SPIClass::endTransaction() against a paramLock the other task holds.
That trips FreeRTOS's xTaskPriorityDisinherit assert (tasks.c:5156,
pxTCB == pxCurrentTCBs[0]) and panics the system.
HalStorage already serialized every explicit method call via
storageMutex, but HalFile's destructor was `= default`, which let the
underlying SdFat FsFile destructor run close() outside any lock
(DESTRUCTOR_CLOSES_FILE=1). Any task that destructed a HalFile while
another task was mid-SD-op would race the unsynchronized state.
Move the locking discipline into HalFile::Impl::~Impl: an explicit
close() under StorageLock, then the FsFile member destructor's redundant
close() is a no-op. HalFile's special members can stay = default.
Switch storageMutex to xSemaphoreCreateRecursiveMutex so openFileForRead
and openFileForWrite can hold the lock while assigning to a HalFile&
out-param whose prior Impl needs locked teardown. Priority inheritance
still applies to recursive mutexes.
Also documented the no-bypass rule in CLAUDE.md: never call SdFat /
SdSpiCard / FsBaseFile / SDCardManager directly, never define
HAL_STORAGE_IMPL outside HalStorage.cpp.
Addresses my admittedly synthetic repro for #2047
Did you use AI tools to help write this code? partial
## Summary
**What is the goal of this PR?**
Adds a clearer Quick Resume sleep-screen flow. The previous “Page as
Sleep Screen” behavior is now exposed as a dedicated `Sleep Screen >
Quick Resume `option, with the timeout-only behavior controlled by a
renamed `Quick Resume on Timeout `setting.
**What changes are included?**
- Adds `Quick Resume` as a new `Sleep Screen` option.
- Renames the old `Page as Sleep Screen` setting to `Quick Resume on
Timeout`.
- Changes that setting’s choices from `Never / After Timeout / Always`
to `OFF / ON`.
- Makes `Quick Resume on Timeout = ON` equivalent to the old `After
Timeout` behavior.
- Makes `Sleep Screen > Quick Resume` equivalent to the old `Always`
behavior.
- Automatically forces `Quick Resume on Timeout` to `ON` when `Sleep
Screen` is set to `Quick Resume`.
- Renames internal setting references from `seamlessSleepScreen` to
`quickResumeSleepScreen`.
- Updates translations for the renamed setting label.
**Additional Context**
- This is mostly a settings/labeling restructure around existing
behavior, not a new rendering path.
- The runtime quick-resume behavior still uses the existing saved
framebuffer / last-screen sleep flow.
- Review focus areas:
- Sleep entry behavior from manual sleep vs timeout sleep.
- The automatic dependency where selecting `Sleep Screen > Quick Resume`
sets `Quick Resume on Timeout` to `ON`.
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**< YES >**_
---
**New `Quick Resume` option for `Sleep Screen` will automatically set
`Quick Resume on Timeout` to `ON`**:
<img width="480" height="800" alt="quick resume"
src="https://github.com/user-attachments/assets/94c553fd-a122-47a8-add9-f29694f55566"
/>
**Example where a different sleep screen setting like `Cover` can be
used in combination with the `Quick Resume on Timeout` setting**:
<img width="480" height="800" alt="cover + quick resume"
src="https://github.com/user-attachments/assets/dd18ce18-230b-4b78-808b-ac85f5e7d5d8"
/>
## Summary
* **What is the goal of this PR?** Prefer epub format over kepub or
other formats offered from an OPDS server
* **What changes are included?**
## Additional Context
Should address #1419
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? _**< NO >**_
---------
Co-authored-by: Arthur Tazhitdinov <lisnake@gmail.com>
## Summary
After a silent reboot, there was a small window where the esp32 would
listen for button presses but the full refresh would hold the event
loop. This gave a UX experience where the silent reboot had completed to
the home screen, a user taps select (at any time during the process),
and they find themselves unexpectedly in a book.
## Additional Context
This must land after
https://github.com/crosspoint-reader/community-sdk/pull/11 and will need
the submodule SHA changes included in.
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? partially
<img width="605" height="454" alt="image"
src="https://github.com/user-attachments/assets/bfd84afe-3b58-436e-9a5d-539af3ec3d4e"
/>
Actualized "Last" sleep screen setting from previous PRs, rebranded as a
~~`Seamless Sleep`~~ `Page as Sleep Screen` option with more
improvements.
https://github.com/user-attachments/assets/59029ba6-007e-4841-abfa-f680d7e98b79
---
New option: `Page as Sleep Screen` - `Never (default)`, `After Timeout`,
`Always`
When enabled, it seamlessly sleeps on timeout or power off, making a
fast refresh for the moon icon. When waking up, we still show the last
page, instead of the boot screen, making it fully seamless.
I tried different icons such as "refresh arrow" and others, but they
looked not as nice as 3 simple dots.
With this mode, the device turns off 4 seconds faster. And has 6 seconds
less delay when turning back on. Much more responsive.
Previously, even a 10-minute timeout sometimes wasn't enough, and I'd
worry about seeing the book cover. It's now easier to use a shorter
sleep timeout: if I get distracted during a reading session but don't
want to stop, the new screen is much more inviting to come back to.
---
Did you use AI tools to help write this code? _**PARTIALLY**_.
---
Test v1.3.0 firmware.bin file
[download](https://github.com/user-attachments/files/28015193/firmware.zip)
Based on PRs #410 and #495Closes#400, #1649
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
When the device went into deep sleep while plugged into USB, a
power-button press would occasionally not wake it. The display held its
last frame, the chip stayed in deep sleep, and recovery required an
unplug + reset + hold-power cycle. On battery the symptom never surfaced
because the power button physically re-energises the chip.
Two peripherals were holding power domains alive across the deep sleep
boundary and interfering with the configured GPIO wake on the power
button:
1. HWCDC. Once Serial is initialized, the USB Serial/JTAG peripheral
keeps its power domain configured even with TX timeout at zero and no
host draining. Tear it down with Serial.end() in
HalPowerManager::startDeepSleep, gated by ENABLE_SERIAL_LOG to match the
Serial.begin site. This hit me if I was charging off my computer.
2. WiFi. enterDeepSleep had no WiFi teardown, so sleeping from any
network-using activity left the modem domain alive. Call
WiFi.disconnect(true) + WiFi.mode(WIFI_OFF) when WiFi is active. Wake
from deep sleep is effectively a chip reset, so no WiFi state needs to
survive. While this doesn't cause higher power drain, it apparently was
causing issues where I'd occasionally have the chip hang on sleep
transition from a wifi activity.
Confirmed on device.
Did you use AI tools to help write this code? partial
On-device repro showed the cover snapshot pinning ~52KB of contiguous
heap (cloning the full 48KB framebuffer with malloc overhead). MaxAlloc
on Home was 61KB; nothing was leaving headroom for HTTPS, which needs
30-50KB contiguous for the mbedTLS handshake.
Add region-aware framebuffer helpers to GfxRenderer that translate a
logical rect through rotateCoordinates and copy only the byte range that
contains the rotated rect. HomeActivity records the tile rect it passes
to drawRecentBookCover and caches only that subregion.
Measured on device (X3, Portrait):
Idle on Home | Free 102K -> 139K | MaxAlloc 61K -> 115K
Mid-EPUB-read | Free 81K -> 134K | MaxAlloc 70K -> 115K
Cover cache | ~52K -> ~16K (per allocation)
Works in all four orientations because the bounds helper samples the
four logical corners through the existing rotation, so the cached byte
range always covers the pixels the theme could have drawn into.
Savings will vary with theme, but should be significant across all.
## Summary
Add ability to long press 'confirm' on a book in the recent books list
to be prompted to remove it from the list.
---
### AI Usage
While CrossPoint doesn't have restrictions on AI tools in contributing,
please be transparent about their usage as it
helps set the right context for reviewers.
Did you use AI tools to help write this code? *Yes, Claude*
---------
Zach Nelson
