src/centroid-html-generator.lua

11local function setup_dir_path(provided_dir)

12 if provided_dir then

13 return provided_dir

14 end

15 return "/mnt/mtwo/programming/ai-stuff/neocities-modernization"

40local function cosine_similarity(vec1, vec2)

41 if not vec1 or not vec2 or #vec1 ~= #vec2 then

42 return 0

43 end

45 local dot_product = 0

46 local norm1 = 0

47 local norm2 = 0

49 for i = 1, #vec1 do

50 dot_product = dot_product + (vec1[i] * vec2[i])

51 norm1 = norm1 + (vec1[i] * vec1[i])

52 norm2 = norm2 + (vec2[i] * vec2[i])

53 end

55 norm1 = math.sqrt(norm1)

56 norm2 = math.sqrt(norm2)

58 if norm1 == 0 or norm2 == 0 then

59 return 0

60 end

62 return dot_product / (norm1 * norm2)

68function M.load_centroids(model_name)

69 model_name = model_name or current_model_dir()

70 local centroids_file = utils.embeddings_dir(model_name) .. "/centroids.json"

72 local content, err = utils.read_file(centroids_file)

73 if not content then

74 utils.log_warn("Could not load centroids: " .. (err or "file not found"))

75 utils.log_warn("Run centroid-generator.lua first to create centroid embeddings")

76 return nil

77 end

79 local data = dkjson.decode(content)

80 if not data or not data.centroids then

81 utils.log_error("Invalid centroids file format")

82 return nil

83 end

85 local count = 0

86 for _ in pairs(data.centroids) do count = count + 1 end

87 utils.log_info(string.format("Loaded %d centroid embeddings", count))

89 return data

95local function build_embeddings_lookup(embeddings_data)

96 local lookup = {}

98 if not embeddings_data or not embeddings_data.embeddings then

99 return lookup

100 end

102 -- Handle array format: [{id: 1, embedding: [...]}, {id: 2, embedding: [...]}, ...]

103 if type(embeddings_data.embeddings) == "table" then

104 for _, entry in ipairs(embeddings_data.embeddings) do

105 if entry.id and entry.embedding then

106 lookup[tostring(entry.id)] = entry.embedding

107 end

108 end

109 end

111 return lookup

117function M.generate_centroid_similarity_ranking(centroid_data, poems_data, embeddings_data)

118 local centroid_embedding = centroid_data.embedding

120 if not centroid_embedding then

121 utils.log_error("Centroid has no embedding: " .. (centroid_data.name or "unknown"))

122 return nil

123 end

125 -- Build lookup table from array-format embeddings

126 local embeddings_lookup = build_embeddings_lookup(embeddings_data)

128 local ranked_poems = {}

130 -- Calculate similarity for each poem

131 for poem_id, poem in ipairs(poems_data.poems) do

132 if poem.id then

133 local poem_embedding = embeddings_lookup[tostring(poem.id)]

135 local similarity = 0

136 if poem_embedding and type(poem_embedding) == "table" and #poem_embedding > 0 then

137 similarity = cosine_similarity(centroid_embedding, poem_embedding)

138 end

140 table.insert(ranked_poems, {

141 id = poem.id,

142 poem = poem,

143 similarity = similarity

144 })

145 end

146 end

148 -- Sort by similarity (descending = most similar first)

149 table.sort(ranked_poems, function(a, b)

150 return a.similarity > b.similarity

151 end)

153 -- Add rank numbers

154 for i, poem_info in ipairs(ranked_poems) do

155 poem_info.rank = i

156 end

158 return ranked_poems

164function M.generate_centroid_diversity_ranking(centroid_data, poems_data, embeddings_data)

165 -- Get similarity ranking first

166 local ranked_poems = M.generate_centroid_similarity_ranking(centroid_data, poems_data, embeddings_data)

168 if not ranked_poems then

169 return nil

170 end

172 -- Reverse the order (least similar first)

173 local reversed = {}

174 for i = #ranked_poems, 1, -1 do

175 table.insert(reversed, ranked_poems[i])

176 end

178 -- Re-assign rank numbers

179 for i, poem_info in ipairs(reversed) do

180 poem_info.rank = i

181 -- Convert similarity to "diversity" score (1 - similarity)

182 poem_info.diversity = 1 - poem_info.similarity

183 end

185 return reversed

191local function generate_centroid_header(centroid_data, page_type, total_poems)

192 local title = string.format("%s - %s exploration",

193 centroid_data.name,

194 page_type == "similar" and "Similar" or "Different")

196 local description = centroid_data.description or ""

198 local header = string.format([[<!DOCTYPE html>

199<html lang="en">

200<head>

201<meta charset="UTF-8">

202<meta name="viewport" content="width=device-width, initial-scale=1.0">

203<title>%s</title>

204</head>

205<body bgcolor="#000000" text="#FFFFFF" link="#6699FF" vlink="#9966FF">

206<pre>

207================================================================================

208 %s

209================================================================================

211Theme: %s

212Description: %s

213Mode: %s poems %s

214Total poems: %d

216Navigation: <a href="../chronological.html">chronological</a> │ <a href="../explore.html">explore</a> │ <a href="../centroid/">all moods</a>

217This page: <a href="%s-similar.html">similar</a> │ <a href="%s-different.html">different</a>

219================================================================================

221]],

222 title,

223 string.upper(centroid_data.name),

224 centroid_data.name,

225 description,

226 page_type == "similar" and "most" or "least",

227 page_type == "similar" and "like this mood" or "like this mood",

228 total_poems,

229 centroid_data.output_slug,

230 centroid_data.output_slug

231 )

233 return header

238local function generate_centroid_footer()

239 return [[

240================================================================================

241 END OF COLLECTION

242================================================================================

243</pre>

244</body>

245</html>

246]]

253local function get_unique_poem_filename_id(poem)

254 local category = poem.category or "unknown"

255 local id = poem.id or 0

256 return string.format("%s-%04d", category, id)

262local function format_poem_for_centroid_page(poem_info, page_type)

263 local poem = poem_info.poem

264 local content = poem.content or poem.text or "[No content]"

266 -- Clean up content

267 content = content:gsub("\r\n", "\n"):gsub("\r", "\n")

269 local score_label = page_type == "similar" and "similarity" or "diversity"

270 local score = page_type == "similar" and poem_info.similarity or poem_info.diversity

272 -- Use category prefix for unique filenames

273 local unique_id = get_unique_poem_filename_id(poem)

275 local entry = string.format([[

277#%d │ Poem %s │ %s: %.4f

279<a href="../similar/%s.html">[similar]</a> <a href="../different/%s.html">[different]</a>

281%s

283]],

284 poem_info.rank,

285 poem.id,

286 score_label,

287 score or 0,

288 unique_id,

289 unique_id,

290 content

291 )

293 return entry

299function M.generate_centroid_html_page(centroid_data, ranked_poems, page_type)

300 local parts = {}

302 -- Header

303 table.insert(parts, generate_centroid_header(centroid_data, page_type, #ranked_poems))

305 -- Poem entries

306 for _, poem_info in ipairs(ranked_poems) do

307 table.insert(parts, format_poem_for_centroid_page(poem_info, page_type))

308 end

310 -- Footer

311 table.insert(parts, generate_centroid_footer())

313 return table.concat(parts)

319function M.generate_centroid_index_page(centroids_data)

320 local parts = {}

322 table.insert(parts, [[<!DOCTYPE html>

323<html lang="en">

324<head>

325<meta charset="UTF-8">

326<meta name="viewport" content="width=device-width, initial-scale=1.0">

327<title>Mood Exploration - Centroids</title>

328</head>

329<body bgcolor="#000000" text="#FFFFFF" link="#6699FF" vlink="#9966FF">

330<pre>

331================================================================================

332 MOOD-BASED EXPLORATION

333================================================================================

335Browse poems by emotional tone or thematic cluster. Each mood has two views:

336- Similar: poems that resonate with this mood

337- Different: poems that contrast with this mood

339Navigation: <a href="../chronological.html">chronological</a> │ <a href="../explore.html">explore</a>

341================================================================================

342 AVAILABLE MOODS

343================================================================================

345]])

347 -- Sort centroids by name for consistent ordering

348 local sorted_slugs = {}

349 for slug, _ in pairs(centroids_data.centroids) do

350 table.insert(sorted_slugs, slug)

351 end

352 table.sort(sorted_slugs)

354 for _, slug in ipairs(sorted_slugs) do

355 local centroid = centroids_data.centroids[slug]

356 local entry = string.format([[

358%s

360 <a href="%s-similar.html">[similar poems]</a> <a href="%s-different.html">[different poems]</a>

362]],

363 string.upper(centroid.name),

364 centroid.description or "",

365 slug,

366 slug

367 )

368 table.insert(parts, entry)

369 end

371 table.insert(parts, [[

372================================================================================

373 ABOUT MOOD CENTROIDS

374================================================================================

376These mood pages are generated from custom "centroid" embeddings - semantic

377fingerprints created by combining keywords and source texts that represent

378each mood. When you visit a mood's "similar" page, you see all poems ranked

379by how closely they match that emotional signature.

381To add new moods, edit: assets/centroids.json

382Then run: lua src/centroid-generator.lua

384================================================================================

385</pre>

386</body>

387</html>

388]])

390 return table.concat(parts)

396function M.generate_all_centroid_pages(poems_data, embeddings_data, output_dir)

397 local model_name = current_model_dir()

399 -- Load centroids

400 local centroids_data = M.load_centroids(model_name)

401 if not centroids_data then

402 utils.log_warn("No centroids available - skipping centroid page generation")

403 return nil

404 end

406 -- Create output directory

407 local centroid_dir = output_dir .. "/centroid"

408 os.execute("mkdir -p " .. centroid_dir)

410 local results = {

411 similar_pages = {},

412 different_pages = {},

413 index_page = nil

414 }

416 -- Generate pages for each centroid

417 for slug, centroid_data in pairs(centroids_data.centroids) do

418 utils.log_info(string.format("Generating centroid pages: %s", centroid_data.name))

420 -- Generate similarity ranking

421 local similar_ranking = M.generate_centroid_similarity_ranking(

422 centroid_data, poems_data, embeddings_data)

424 if similar_ranking then

425 -- Generate similar page

426 local similar_html = M.generate_centroid_html_page(centroid_data, similar_ranking, "similar")

427 local similar_file = centroid_dir .. "/" .. slug .. "-similar.html"

429 if utils.write_file(similar_file, similar_html) then

430 table.insert(results.similar_pages, similar_file)

431 utils.log_info(string.format(" Created: %s-similar.html", slug))

432 end

434 -- Generate different page (reverse order)

435 local different_ranking = M.generate_centroid_diversity_ranking(

436 centroid_data, poems_data, embeddings_data)

438 if different_ranking then

439 local different_html = M.generate_centroid_html_page(centroid_data, different_ranking, "different")

440 local different_file = centroid_dir .. "/" .. slug .. "-different.html"

442 if utils.write_file(different_file, different_html) then

443 table.insert(results.different_pages, different_file)

444 utils.log_info(string.format(" Created: %s-different.html", slug))

445 end

446 end

447 else

448 utils.log_warn(string.format(" Could not generate ranking for: %s", centroid_data.name))

449 end

450 end

452 -- Generate index page

453 local index_html = M.generate_centroid_index_page(centroids_data)

454 local index_file = centroid_dir .. "/index.html"

456 if utils.write_file(index_file, index_html) then

457 results.index_page = index_file

458 utils.log_info("Created: centroid/index.html")

459 end

461 utils.log_info(string.format("Centroid generation complete: %d similar, %d different pages",

462 #results.similar_pages, #results.different_pages))

464 return results

469if arg and arg[0] and arg[0]:match("centroid%-html%-generator%.lua$") then

470 utils.log_info("=== Centroid HTML Generator ===")

471 utils.log_info("Generating mood-based exploration pages")

472 utils.log_info("")

474 -- Load poems data

475 local poems_file = utils.get_assets_root() .. "/poems.json"

476 local poems_content = utils.read_file(poems_file)

477 if not poems_content then

478 utils.log_error("Could not load poems.json")

479 os.exit(1)

480 end

481 local poems_data = dkjson.decode(poems_content)

483 -- Load embeddings data

484 local embeddings_file = utils.embeddings_dir() .. "/embeddings.json"

485 local embeddings_content = utils.read_file(embeddings_file)

486 if not embeddings_content then

487 utils.log_error("Could not load embeddings.json")

488 os.exit(1)

489 end

490 local embeddings_data = dkjson.decode(embeddings_content)

492 -- Generate pages

493 local output_dir = DIR .. "/output"

494 local results = M.generate_all_centroid_pages(poems_data, embeddings_data, output_dir)

496 if results then

497 utils.log_info("\n=== Summary ===")

498 utils.log_info(string.format("Similar pages: %d", #results.similar_pages))

499 utils.log_info(string.format("Different pages: %d", #results.different_pages))

500 utils.log_info(string.format("Index page: %s", results.index_page and "created" or "failed"))

501 else

502 utils.log_error("Centroid page generation failed")

503 os.exit(1)

504 end