# KLT 4.14 FIX — technical specification

Version: 4.14 web fixed build for kurpishev.com
Author: Ivan B. Kurpishev

## Input

Allowed file types:

```text
.doc
.docx
```

All other file extensions are rejected in the browser.

## Processing

1. DOCX text extraction from `word/document.xml`, headers and footers.
2. Legacy DOC fallback extraction using binary text scan. If text is not recovered, user is instructed to save the document as DOCX.
3. Browser-side Reper calculation:
   - U / limU — concept field.
   - I / limI — leading idea.
   - R / limR — real factual content.
   - D — sufficient reason.
4. Lambda calculation:

```text
lambda = ((U-R)*(I-D))/((U-D)*(I-R))
Delta  = |lambda + 1|
Auth   = 1/(1+Delta)
D*     = (2UI-UR-IR)/(U+I-2R)
```

5. On-page result rendering.
6. DOCX checklist generation in browser.
7. Group ZIP generation.
8. Optional email transfer through `/api/k414mail.php`.

## Output

- HTML result on page.
- One DOCX checklist per analyzed file.
- ZIP package with all checklists and `summary.csv`.
- Optional email delivery through PHP.

## External dependency

JSZip is loaded from CDN:

```text
https://cdn.jsdelivr.net/npm/jszip@3.10.1/dist/jszip.min.js
```

For an offline/self-hosted contour, place `jszip.min.js` locally and update the script tag in `ru/klt/k414.html`.