<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[toolcin]]></title><description><![CDATA[toolcin]]></description><link>https://toolcin.hashnode.dev</link><generator>RSS for Node</generator><lastBuildDate>Wed, 18 Nov 2026 14:14:34 GMT</lastBuildDate><atom:link href="https://toolcin.hashnode.dev/rss.xml" rel="self" type="application/rss+xml"/><language><![CDATA[en]]></language><ttl>60</ttl><item><title><![CDATA[Compress a PDF to an exact byte target in the browser]]></title><description><![CDATA[Hit a 200 KB portal limit without uploading the file. A binary search over JPEG quality finds the size, inside the browser tab.]]></description><link>https://toolcin.hashnode.dev/compress-a-pdf-to-an-exact-byte-target-in-the-browser-mwizgsje</link><guid isPermaLink="true">https://toolcin.hashnode.dev/compress-a-pdf-to-an-exact-byte-target-in-the-browser-mwizgsje</guid><category><![CDATA[webdev]]></category><category><![CDATA[JavaScript]]></category><category><![CDATA[pdf]]></category><category><![CDATA[showdev]]></category><dc:creator><![CDATA[Github]]></dc:creator><pubDate>Mon, 16 Nov 2026 08:10:06 GMT</pubDate><enclosure url="https://cdn.hashnode.com/uploads/covers/6af6b4c558a644b61ace7721/132f8618-2ccc-435e-94ee-d9948138225e.png" length="0" type="image/jpeg"/><content:encoded><![CDATA[<p>Try it: <a href="https://toolcin.pages.dev/tools/compress-pdf-to-size/?utm_source=hashnode&amp;utm_medium=blog&amp;utm_campaign=pdf200">https://toolcin.pages.dev/tools/compress-pdf-to-size/?utm_source=hashnode&amp;utm_medium=blog&amp;utm_campaign=pdf200</a></p>
<p>Portals do not ask for a quality percentage. They ask for a file under 200 KB, and every compressor on the market answers with a slider that has nothing to do with kilobytes.</p>
<p>The fix is to invert the loop: render each page once, then search for the quality that fits the target size.</p>
<p>Two details matter more than the search itself.</p>
<p>Render once. Rendering a page with pdf.js costs far more than a JPEG encode, so keep the canvas at 2x and reuse it across attempts. Otherwise a two-page document takes a minute instead of four seconds.</p>
<p>Drop the scale before dropping the quality. Quality 32% at full resolution looks like a photocopier running out of toner. Quality 70% at 85% scale is indistinguishable on screen and often smaller.</p>
<p>The honest trade: re-rendering pages as images removes the text layer. The output cannot be selected or searched, and a fillable form becomes flat. If you need both, compress and then add an invisible OCR layer back.</p>
<p>The whole thing runs in the tab, so the file never leaves the laptop.</p>
]]></content:encoded></item><item><title><![CDATA[Compress a PDF to an exact byte target in the browser]]></title><description><![CDATA[Hit a 200 KB portal limit without uploading the file. It searches JPEG quality until the PDF fits, inside your browser tab.]]></description><link>https://toolcin.hashnode.dev/compress-a-pdf-to-an-exact-byte-target-in-the-browser</link><guid isPermaLink="true">https://toolcin.hashnode.dev/compress-a-pdf-to-an-exact-byte-target-in-the-browser</guid><category><![CDATA[JavaScript]]></category><category><![CDATA[pdf]]></category><category><![CDATA[webdev]]></category><category><![CDATA[showdev]]></category><dc:creator><![CDATA[Github]]></dc:creator><pubDate>Sun, 15 Nov 2026 18:39:26 GMT</pubDate><content:encoded><![CDATA[<p>Portals do not ask for a quality percentage. They ask for a file under 200 KB, and every compressor answers with a slider that has nothing to do with kilobytes.</p>
<p>The fix is to invert the loop: render each page once, then search for the quality that fits. Render once, because rendering a page with pdf.js costs far more than a JPEG encode, so keep the canvas at 2x and reuse it across attempts.</p>
<p>Drop the scale before dropping the quality. Quality 32 percent at full resolution looks like a photocopier running out of toner. Quality 70 percent at 85 percent scale is indistinguishable on screen and often smaller.</p>
<p>The honest trade: re-rendering pages as images removes the text layer. The output cannot be selected or searched, and a fillable form becomes flat. If you need both, compress and then add an invisible OCR layer back.</p>
<p>Live tool: <a href="https://toolcin.pages.dev/tools/compress-pdf-to-size/">https://toolcin.pages.dev/tools/compress-pdf-to-size/</a></p>
]]></content:encoded></item><item><title><![CDATA[Building 392 browser tools that never upload your files]]></title><description><![CDATA[Every "free online tool" I tried during a week of paperwork asked for the same thing: upload the file first. A passport scan, a bank statement, a signed…]]></description><link>https://toolcin.hashnode.dev/building-392-browser-tools-that-never-upload-your-files</link><guid isPermaLink="true">https://toolcin.hashnode.dev/building-392-browser-tools-that-never-upload-your-files</guid><dc:creator><![CDATA[Github]]></dc:creator><pubDate>Fri, 13 Nov 2026 07:21:43 GMT</pubDate><content:encoded><![CDATA[<p>Every "free online tool" I tried during a week of paperwork asked for the same thing: upload the file first. A passport scan, a bank statement, a signed contract — all of it goes to a server I have never seen and stays there for a period nobody tells me.</p>
<p>So I built the opposite: <a href="https://toolcin.pages.dev/">Toolcin</a> — 392 small tools where the file never leaves the tab.</p>
<h2><strong>The rule that decided the architecture</strong></h2>
<p>If the work happens in the browser, there is no upload endpoint. No upload endpoint means nothing to log, nothing to leak and nothing to sell. That one rule decided the hosting, the cost and the privacy policy at the same time: static files, no application server, no database, no user table.</p>
<h2><strong>What actually runs in the tab</strong></h2>
<ul>
<li><strong>PDF</strong> — pdf-lib writes and edits documents, pdf.js reads the text layer. Merge, split, rotate, compress, fill a form, or rebuild a table into a real <code>.xlsx</code>.</li>
<li><strong>Video</strong> — ffmpeg compiled to WebAssembly. MKV to MP4 is a stream copy when the codecs already fit, a re-encode when they do not.</li>
<li><strong>Images</strong> — canvas for resize, crop and compress; libheif as WebAssembly for HEIC, because a renamed <code>.heic</code> is still HEVC inside.</li>
</ul>
<h2><strong>The parts that were harder than expected</strong></h2>
<p><strong>PDF text extraction.</strong> A PDF has no paragraphs, only glyphs at coordinates. Rebuilding rows and columns from those coordinates is the whole problem, which is why the <a href="https://toolcin.pages.dev/tools/pdf-to-excel/">PDF to Excel tool</a> exposes the row tolerance instead of pretending the conversion is magic.</p>
<p><strong>Staying fast at 392 pages.</strong> A static site with hundreds of pages still needs a plan: one curated shelf on the home page, a hub per category, and a build step that keeps every page title and description unique.</p>
<h2><strong>Why the tools are free</strong></h2>
<p>No server means no bill that grows with usage, so there is nothing to charge for. There are no ads and no third-party scripts on a normal page; the only counter is first-party and stores a number, not a visitor identity.</p>
<h2><strong>Try it with a file you would not upload anywhere else</strong></h2>
<ul>
<li><a href="https://toolcin.pages.dev/tools/pdf-to-excel/">PDF to Excel</a></li>
<li><a href="https://toolcin.pages.dev/tools/fill-pdf-form/">Fill a PDF form</a></li>
<li><a href="https://toolcin.pages.dev/tools/mkv-to-mp4/">MKV to MP4</a></li>
<li><a href="https://toolcin.pages.dev/tools/heic-to-jpg/">HEIC to JPG</a></li>
</ul>
<p>Open the browser's Network tab while you convert: if you never see the file leave in a request, nothing was uploaded.</p>
]]></content:encoded></item></channel></rss>