issues/completed/8-028-clean-output-directory.md

Issue 8-028: Clean Output Directory of Test/Demo Files

Current Behavior

The output/ directory contains files that were manually created during development and testing, rather than being generated by the pipeline. This violates the principle that output/ should be clearable and fully regenerable.

Files That Should Not Be Present:

  1. corner-box-demo.html (20KB)
  • Demo file created for Issue 8-007 testing
  • Contains inline CSS: style="text-align: left; display: inline-block;"
  • Not generated by any pipeline script
  1. test_similar_1001.html (12MB)
  • Test file created during interactive mode testing
  • Contains stale CSS from before Issue 8-003 fixes
  • Generated by manual test option, not automated pipeline
  1. archive/ directory (~6MB)
  • Contains generated-site/ - old site structure
  • Contains test-outputs/ - various test HTML files
  • Contains output5.pdf - unrelated PDF file
  • Historical artifacts, not part of current generation
  1. test/ directory
  • Contains similar/ and different/ test subdirectories
  • Created during development testing
  • Duplicates structure of production directories

Additional Issue: Stale Generated Files

Files in similar/ and different/ were generated before Issue 8-003 CSS removal:

  • similar/001.html contains 255 style= attributes
  • similar/001.html contains 1 <style> block
  • These need regeneration after cleanup

Intended Behavior

The output/ directory should contain ONLY files generated by the pipeline:

  • index.html - main chronological index
  • chronological.html - all poems by date
  • chronological.txt - plain text export
  • explore.html - discovery instructions
  • numeric-index.html - numeric search index
  • similar/ - similarity-sorted poem pages (generated)
  • different/ - diversity-sorted poem pages (generated)
  • centroid/ - mood-based entry points (generated)

All test/demo files should live in a separate location or be cleaned up after use.

Implementation Steps

Step 1: Remove test/demo files from output directory ✅ COMPLETED

  • [x] Remove corner-box-demo.html
  • [x] Remove test_similar_1001.html
  • [x] Remove archive/ directory
  • [x] Remove test/ directory

Result: Output directory reduced from 241MB/78 files to 165MB/25 files.

Step 2: Verify cleanup ✅ COMPLETED

  • [x] Confirm only pipeline-generated files remain
  • [x] Document remaining file structure

Remaining structure:

output/
├── centroid/           # 11 mood-based entry pages
├── chronological.html  # 12.1 MB, all poems by date
├── different/          # 4 diversity-sorted pages (stale)
├── explore.html        # Discovery instructions
├── index.html          # Copy of chronological.html
├── numeric-index.html  # 288 KB numeric index
└── similar/            # 6 similarity-sorted pages (stale)

Step 3: Create validation test page ✅ COMPLETED

  • [x] Generate a fresh test page using current pipeline
  • [x] Verify zero style= attributes
  • [x] Verify zero <style> blocks in template
  • [x] Verify pure HTML structure (no CSS, no JavaScript in template)

Validation results:

  • Regenerated chronological.html with 7,797 poems
  • 0 style= attributes in HTML template
  • 0 <style> blocks in HTML template (1 match is poem content, not template)
  • 0 <script> tags in HTML template (1 match is poem content, not template)
  • Template uses only: <center>, <pre>, <font color="">, <b>, <a href="">

Note: Poem content may contain <style> or <script> as literal text (e.g., someone
posted Vim's :TOhtml output as a poem). These are inside <pre> blocks and render as
text, not executed HTML. This is acceptable.

Step 4: Update .gitignore if needed

  • [x] Test files removed; no pattern needed as they shouldn't be recreated
  • [x] Document that test files should be cleaned before commits (noted in issue)

Quality Assurance Criteria

  • [x] ls output/ shows only pipeline-generated files
  • [x] grep -c "style=" output/chronological.html returns 0 (template has no inline CSS)
  • [x] No <style> blocks in HTML headers (only in poem content)
  • [x] output/ can be safely deleted and fully regenerated by run.sh

Related Issues

  • Issue 8-003: Remove remaining CSS from HTML generation (COMPLETED)
  • Issue 8-007: Box-drawing borders demo (demo file created here)
  • Issue 8-001: Unified website generation pipeline

Notes

The output directory is meant to be the deployment target for Neocities. It should contain only production-ready files that can be uploaded directly. Test artifacts pollute this space and may accidentally be deployed.

ISSUE STATUS: COMPLETED

Created: 2026-01-04
Completed: 2026-01-04
Phase: 8 (Website Completion)
Priority: Medium (cleanup/hygiene)

Summary

Cleaned output directory of 4 test/demo artifacts (76MB total):

  • Removed corner-box-demo.html (Issue 8-007 demo)
  • Removed test_similar_1001.html (interactive test output)
  • Removed archive/ directory (old generated content)
  • Removed test/ directory (test subdirectories)

Validated HTML generation produces CSS-free output:

  • Template uses pure HTML: <center>, <pre>, <font color="">, <b>, <a>
  • No style= attributes in generated markup
  • No <style> or <script> blocks in template (only in poem content)