Compress PDF

Compress PDF 4.0 — Smaller PDFs, Entirely On Your Own Computer Compress PDF makes PDF files smaller. Version 4.0 is a complete rewrite that replaces the old extension from top to bottom. Everything now happens locally in your browser: your documents are never uploaded, never queued on a server, and never seen by anyone but you. The extension makes no network requests at all during normal use. WHAT'S NEW A real PDF compression engine. The extension now parses your PDF, finds the images inside it, and downsamples each one to the resolution it is actually drawn at on the page — the single biggest source of savings in scanned documents and image-heavy reports. Each image is then re-encoded three ways (JPEG, Flate, and an indexed colour palette) and the smallest result wins. Photographs usually come out as JPEG; screenshots, charts, diagrams and logos usually come out as a palette, which is both smaller and visibly sharper on hard edges than JPEG would be. Beyond images, the engine removes objects nothing in the document points to any more — old revisions left behind by incremental edits are frequently most of a "bloated" PDF — de-duplicates repeated streams such as logos and embedded fonts, packs the remaining objects into object streams, and writes a compressed cross-reference table. On text-heavy files that last step alone is worth 5–15%. Four presets: Light, Recommended, Strong, and Custom with your own resolution and quality sliders. Your choice is remembered and synced across the devices you sign in to. Target size mode. Tell it "get this under 5 MB" and it binary-searches image quality until the file fits, instead of making you guess at settings and try again. Before and after sizes, per file and for the whole batch, with the percentage saved shown plainly. A batch queue with drag and drop, per-file progress, per-file status, per-file error messages and a retry button. Progress is real progress, and the Cancel button actually cancels. Password-protected PDFs are detected and unlocked — RC4 40-bit and 128-bit, AES-128 and AES-256, with either the user or the owner password. Documents that carry only an owner password (restricted printing or copying, but no password needed to open) now open with no prompt at all. Grayscale conversion, and optional removal of metadata and annotations. Download the whole batch as a single .zip. An options page, a keyboard shortcut (Alt+Shift+P), and a "Compress this PDF" item in the right-click menu on PDF links. Dark mode. Full right-to-left layout for Arabic, Hebrew and Persian. Labelled controls, visible keyboard focus, live status announcements for screen readers, a keyboard-operable drop zone, and respect for your reduced-motion setting. WHAT WAS FIXED Version 3.0 called itself a PDF compressor in all 43 languages but contained no PDF code whatsoever. It uploaded whatever you selected to a third-party archiving service and handed back a .7z file. Because the content inside a PDF is already compressed, that saved close to nothing — and what came back was not a PDF at all. That entire upload path is gone. Nothing leaves your machine. Committed Google credentials — an API key, an OAuth client ID and a project number — were embedded in the shipped source and have been removed. jQuery 3.3.1 (three known vulnerabilities: CVE-2019-11358, CVE-2020-11022, CVE-2020-11023) and Bootstrap 4.1.3 are gone. There are now no third-party runtime dependencies at all. Error handling was broken: a failure handler chained in the wrong order meant any problem surfaced as the useless message "There was an error zipping: undefined". Errors are now specific, shown inline next to the file that produced them, and never in a browser alert box. No more indefinite hangs. The old upload had no timeout, so a stalled server left every button disabled forever with no way out but reloading the page. Downloads no longer navigate the page away mid-batch; results are saved properly through the browser's download manager. The default output is a PDF, not an archive format that Windows and macOS cannot open without extra software. The file picker is no longer disabled by default, so a script error can no longer leave the page permanently unusable. Roughly ninety lines of dead Google Drive code, referencing buttons that did not exist on the page, have been deleted. Manifest errors were cleaned up. A translation bug in which two different sentences shared one key — losing one of them in every single language — is fixed, and a missing translation now falls back to English instead of blanking the control. Review links on Edge now point to the right store. All 43 locales were rewritten to describe compression rather than zipping. KNOWN LIMITS Font subsetting is not implemented; identical embedded fonts are de-duplicated instead. JPEG 2000 images cannot be re-encoded, because no browser can decode them. Files above 100 MB are refused rather than risking a crash. PERMISSIONS Downloads, to save your compressed file. Storage, to remember your preferences. Context menus, for the right-click item. Access to a website is requested only if you use the right-click item on a link, and only for that one site. There are no standing host permissions.