Citation Workflow v0.5.1

Changes

• Added OSCOLA, ACS, AMA 11, Chicago 18 (author-date), IEEE, and MLA 9 to single-paper and batch reference generation.
• Ordered reference styles for practical relevance to biology, with APA first: APA 7, Vancouver, AMA 11, Harvard, ACS, Chicago 18 (author-date), IEEE, MLA 9, and OSCOLA. BibTeX, RIS / EndNote, and custom formatting follow.
• Bundled the six additional CSL styles locally, recorded their upstream source revision, and updated third-party attribution. Generating references requires no extra style download.
• Updated frontend asset URLs so returning visitors load the new citation engine.

Validation

• Production bundle build and the full automated test suite passed (one existing Python test skipped).
• Tested all 11 built-in formats with journal articles, conference papers, and versioned preprints using the actual browser bundle.
• Verified matching single-paper and batch option ordering and distinct journal-reference output for each CSL style.

This release updates the GitHub Pages frontend; no separate Cloudflare Worker deployment is required.

Citation Workflow v0.5.0

What's Changed

Optional browser connector

• Added an optional Chrome/Edge extension for sources requiring human verification or sign-in.
• Verify & continue opens the selected source in a new tab. After verification, the connector attempts to retrieve an available PDF using the browser session.
• Batch ZIP downloads can pause for verification and continue after receiving the PDF.
• Added cancellation, tab-closure handling, a 15-minute waiting limit, and Skip paper & continue.
• Added a connector ZIP download and installation instructions to the website.

Source retrieval

• Connected DOI-based fallback requests to Sci-Hub, Z-Library, and Library Genesis.
• Fixed official Elsevier redirects and added support for matching HTML download intermediary pages.
• Improved diagnostics for verification requirements, access restrictions, unavailable sources, and unmatched records.

Interface improvements

• Connected status now appears in bright green with a subtle glow.
• Connector setup changes to Connector re-setup after connection.
• Removed the “No installation required” status text.

Installation

The connector is optional. Standard cloud downloads remain available without it.

1. Download the connector ZIP from Connector setup and extract it.
2. Open chrome://extensions or edge://extensions.
3. Enable Developer mode, select Load unpacked, and choose the extracted folder.
4. Reload Citation Workflow and confirm Connector connected.

Validation

• Automated checks passed on Windows, Linux, and macOS.
• Tested the actual unpacked extension in Edge with controlled verification pages, including browser-session cookie reuse, single downloads, ZIP continuation, and cancellation.
• Verified standard PDF downloads without the extension and independently checked ZIP integrity.

Notes

• Login cookies and connector-retrieved PDFs are not uploaded to Cloudflare.
• Verification must be completed manually. Some sites may still restrict downloads or use unsupported download hosts.
• Live third-party CAPTCHA completion was not tested.
• The connector is distributed as an unpacked extension and is not listed in the Chrome Web Store.

Citation Workflow v0.4.1

Changes

• Added a progress panel to Batch download, positioned above the results and kept visible when scrolling through long batches.
• Shows the current DOI or paper identifier, paper number, reference-data lookup and full-text search stages, completed steps and elapsed time. Long-running stages display a waiting message, with an activity indicator that respects reduced-motion preferences.
• Individual PDF downloads and Download all as ZIP now show their current task. PDF transfers report received megabytes and the total when available; ZIP creation reports how many files have been packed and yields between files so the browser can repaint.
• Completion messages distinguish processed papers, failed lookups, included PDFs and unavailable or skipped downloads. Failed items do not stop the remaining batch; controls and timers reset for another attempt.
• Updated versioned frontend asset URLs so returning visitors receive the new interface.

Progress measures completed tasks, not an estimated completion time. A finished batch can include unavailable PDFs. Existing limits remain 20 papers per batch, 80 MB per PDF and 100 MB per ZIP.

Validation

• Full automated suite passed: Python metadata/source tests, frontend and citation-format checks, cloud API restrictions, PDF validation and independent ZIP integrity checks.
• Added regression coverage for slow lookup/source-search stages, failure continuation, retry resets, download failure cleanup, PDF byte progress with and without a known total, and ZIP packing progress.
• Local Chrome browser checks at desktop (1365 × 900) and mobile (390 × 844) sizes passed using controlled API responses: elapsed-time updates, stage changes, invalid-item continuation, completed progress and an actual browser ZIP download. No JavaScript errors or mobile page overflow were observed.
• Production bundle build passed. Native Safari and mobile-device downloads were not tested.

This release changes the frontend only; it requires no separate Cloudflare Worker deployment.

Citation Workflow v0.4.0

Changes

• Metadata lookup and full-text retrieval now run on a deployed Cloudflare Worker. No localhost connection, helper, Python, extension or project-file download is required.
• PDFs use normal browser downloads with article-title filenames. Browser settings control save location and duplicates; automatic download is optional and off by default.
• Batch download creates one ZIP with available PDFs and a report. Limits: 20 papers, 80 MB per PDF, 100 MB per ZIP. Failed or invalid items do not stop other papers.
• Citation styles, editable titles, reference-example custom formatting and simple GitHub feedback are preserved.
• Added expiring signed tickets, restricted source hosts/redirects, response limits, CORS and cloud rate limiting.

Basic validation

• Real cloud PDF downloads: PMLR (Haarnoja et al., 2017), arXiv 1706.03762v1, PLOS Biology 10.1371/journal.pbio.3000117, Nature 10.1038/s41586-021-03819-2, Journal of Statistical Software 10.18637/jss.v067.i01.
• Browser: single conference-paper PDF; mixed batch with journal, preprint, conference paper and invalid input; ZIP with available PDFs and a report. Python independently verifies ZIP CRCs and contents.
• Automated legacy metadata/source regression tests, frontend identifiers/capitalisation, custom styles, citation exports, cloud tickets/CORS/host restrictions, PDF validation and ZIP integrity.

Availability notes

The cloud checks original repositories, PMC and supported open-access publishers. Legacy headless-browser mirror discovery and automation for Sci-Hub, ResearchGate, Z-Library, Library Genesis and Anna’s Archive are not deployed. Subscription access, verification pages and cloud-network restrictions can still prevent PDFs; official article links remain available.

Windows browser downloads were tested. macOS is available but has not undergone complete testing; native Safari/mobile downloads remain unverified. No OS-specific installation is needed. Internet and cloud service availability are required.

Backend source/configuration are in cloud/ and are deployed separately from Pages. GitHub Actions handles frontend publication and this release.

Citation Workflow v0.3.2

Feedback now opens one short form for paper download and citation problems.

Open the website · Send feedback

Changes

• Replaced the separate bug-report and suggestion templates with a single Paper feedback form.
• Choose Paper download or Paper citation, then enter the paper DOI or describe the citation problem. A DOI alone is enough for download reports.
• Contact email is optional. The form explains that email addresses are visible in the public GitHub issue.
• The website's Feedback link opens this form directly, avoiding the template chooser. Blank issues are disabled so only the feedback template is offered.

Compatibility

Download and citation behaviour is unchanged. Windows and Linux checks run before deployment; macOS compatibility checks run separately.

macOS support is available, but has not undergone complete testing. Native macOS folder selection, browser permissions and end-to-end downloads remain unverified.

Citation Workflow v0.3.1

This patch includes the cross-platform downloads and feedback features from v0.3.0 and fixes stale browser assets after website updates.

Open the website · Setup guide

Changes

• Versioned CSS and JavaScript URLs ensure a refreshed page loads the matching application files, rather than reusing the previous release's cached assets.
• Feedback uses a muted speech-bubble icon and borderless text; Batch download keeps its red outline. The system selector supports Windows, macOS and Linux, defaulting to Windows on a first visit.
• Windows and Linux checks gate deployment. macOS checks run separately.

Upgrade and testing status

Close the old companion, download and extract this release's Source code (zip) or Source code (tar.gz), then start start.bat on Windows or run sh start.sh on macOS/Linux. Python 3.10 or later is required. Refresh the website afterwards.

macOS support is available, but has not undergone complete testing. Native macOS folder selection, browser permissions and end-to-end PDF downloads remain unverified. Windows and Ubuntu have been verified with actual PDF downloads.

Citation Workflow v0.3.0

Choose Windows, macOS or Linux for your download folder, and send feedback directly through GitHub Issues.

Open the website · Read the setup guide

What's new

• Cross-platform downloads: the homepage, download settings and batch page share a Windows / macOS / Linux selector. Windows is the default on a first visit; each system remembers its own folder.
• macOS and Linux companion support: POSIX absolute paths and ~/... home-directory paths are supported. Start the companion with sh start.sh; Windows continues to use start.bat. The selected system must match the computer running the companion.
• Feedback: a muted speech-bubble icon and text link sits beside the outlined Batch download button on the homepage and remains in the batch page's top-right corner. It opens GitHub Issues, with bug-report and suggestion forms. A GitHub account is required to submit feedback; issues are public.
• Preprints and conferences: DOI, arXiv and supported preprint/conference links retain their paper type and explicit version. Original repositories include arXiv, bioRxiv/medRxiv, ChemRxiv, PMLR, NeurIPS, CVF and ACL where metadata and accessible PDFs are available.
• Full-text routing: source history records availability and verification failures across PMC and the supported databases. Current PMC cloud datasets and public catalogue download routes are supported. Source failures do not mean a paper has no accessible full text.
• Compatibility checks: Windows and Ubuntu checks must pass before deployment. macOS checks run separately, so a waiting macOS runner does not delay website updates.

Upgrade

1. Close the old companion before starting the new version.
2. Download Source code (zip) or Source code (tar.gz) from this release and extract it into a new folder.
3. Install Python 3.10 or later. On Windows, double-click start.bat. On macOS or Linux, open a terminal in the extracted folder and run sh start.sh.
4. Keep the companion running and refresh the website, or open http://127.0.0.1:8765/.
5. Choose your operating system and an existing download folder. The folder picker requires Tkinter and a graphical desktop; a typed path also works.

No Node.js installation or third-party Python packages are required to run the companion. Updating only the website does not update an already-running companion.

Validation and limitations

Python and JavaScript checks cover path validation, legacy settings, per-system folder memory, PDF saving, duplicate filenames, references and paper types. Windows and Ubuntu (WSL) also saved the same real PMLR PDF successfully, with matching SHA-256 hashes. Desktop and narrow-screen browser checks cover the homepage, system selection and batch feedback link.

macOS support is available, but has not undergone complete testing. Its compatibility workflow runs separately and does not block deployment. Native macOS folder selection, browser permissions and end-to-end PDF downloads remain unverified. Full-text access depends on external sources and may require institutional sign-in or manual downloading. This release does not resolve every retrieval or citation-formatting issue identified during testing.

Developed by Marron Zihao Yao.

Citation Workflow v0.2.0

Citation Workflow v0.2.0 — Batch downloads and custom references

Choose a Windows folder with the native folder picker, process multiple DOIs in one table, and learn a reference format from an example.

Open the website · Read the setup guide

What's new

• Folder setup before searching: the save location now appears above Find paper, with a prompt asking where to save your papers. Type a path or select Browse folders to open the native Windows folder picker.
• Batch download: select the top-right button, choose a working folder and reference style, then paste one DOI per line. The results table shows each DOI, a shortened article title, a PDF download button and a citation-copy button.
• Download all PDFs: save the batch to the selected folder with one action. Papers are processed sequentially; failures do not stop the remaining downloads, and PDFs already saved to that folder during the batch are skipped. Failed downloads can be retried individually.
• Custom reference styles: choose Custom · from an example in single-paper or batch mode and paste a complete journal reference. Detection starts automatically and learns author-name conventions, field order, punctuation and page-range formatting. Including the example's DOI gives the most reliable match; references without a DOI can be identified through Crossref, with publication years helping distinguish originals from reprints.
• Journal abbreviations: custom styles follow the example's full or abbreviated journal-name convention and dotted or undotted form. Abbreviations come from PubMed or Crossref metadata. If no verified abbreviation is available, the full name is retained with a note.
• Simpler citation panel: the formatted preview has been removed. The editable plain-text citation is ready to copy directly. Custom examples and detected templates are shared between single and batch mode and saved in your browser.

PDF source changes

• Third-party databases are tried before offering publisher access: PubMed Central, followed by Sci-Hub mirror candidates discovered through web search.
• Mirror URLs no longer need to be entered manually. If plain search pages provide no usable links, the companion retries with an isolated Chrome or Edge browser session.
• When no PDF is retrieved, publisher and search links are offered for you to choose. Nothing opens automatically.
• Additional database adapters can be added to the companion's source list.

Upgrade from v0.1.0

1. Close the old companion's command window.
2. Download Source code (zip) from the release assets and extract it to a new folder.
3. On Windows with Python 3.10 or later, double-click start.bat and keep its command window open. No extra Python packages or Node.js installation are needed. The native folder picker uses Python's standard Tkinter component.
4. Refresh the website, or use the local page at http://127.0.0.1:8765/. The new custom-style detection and folder picker require the updated companion.

Validation and limitations

Python and JavaScript checks, the production build, and desktop/mobile browser checks passed. Custom styles were also checked against a real published reference, both with and without its DOI. Browser interaction checks used simulated API replies for target papers and PDF downloads.

PDF availability depends on the source; search verification, institutional sign-in or manual downloading may still be required. Automatic Sci-Hub retrieval is not guaranteed. A single plain-text example cannot reveal every journal rule, so review generated references against the journal's instructions. Ambiguous examples prompt for a DOI instead of selecting a paper arbitrarily.

Developed by Marron Zihao Yao.

Citation Workflow v0.1.0

The first public release of Citation Workflow brings DOI lookup, PDF retrieval and reference formatting into one workspace for biological research.

Open the website · Read the setup guide

What’s included

• DOI lookup: enter a DOI or a https://doi.org/ link to find the paper in PubMed, with DOI metadata as a fallback.
• PDF retrieval: check PubMed Central and publisher pages, then an optional, user-supplied Sci-Hub mirror.
• Windows downloads: save PDFs to an existing folder using the original article title. Forward slashes, backslashes and surrounding quotation marks are accepted in folder paths; duplicate filenames receive a numeric suffix.
• Reference formats: generate APA 7, Vancouver, Harvard, BibTeX and RIS references.
• Editable APA capitalisation: apply sentence case to article titles and title case to journal names, with controls to preserve proper nouns and scientific terms.
• Preview and copy: review formatted references with italics, then copy them as plain text.
• A focused interface: dark mode, white and red typography, responsive layouts and British English throughout.

Getting started

1. Download Source code (zip) from the assets below and extract it.
2. Install Python 3.10 or later on Windows, with Python available on your PATH.
3. Double-click start.bat and keep its command window open. No additional Python packages or Node.js installation are needed.
4. Use the local page that opens, or visit the website while the companion is running.
5. Enter a DOI, choose an existing save folder and review your reference.

Known limitations

• The Windows companion is required for DOI queries and downloads, including when using the hosted website. GitHub Pages hosts the interface only.
• Browser restrictions may prevent the hosted page from connecting to the companion. If this happens, use http://127.0.0.1:8765/ instead.
• Some publishers require institutional sign-in, CAPTCHA verification or a manual download. Browser sign-in sessions are not shared with the companion.
• No Sci-Hub mirror is included, and mirror availability has not been verified.
• Scientific names and proper nouns still need a final capitalisation check, especially when the supplied title is entirely in capitals.
• Downloads are limited to 80 MB. Windows drive paths are supported; UNC network paths are not.

This is an early release. Feedback on metadata, reference formatting and publisher compatibility is welcome through GitHub Issues.

Developed by Marron Zihao Yao.