scripts/test-html-generation

#!/usr/bin/env lua

-- Test script for validating HTML generation
-- Generates similar/different pages for specified poem IDs to validate calculations and visuals

-- {{{ local function setup_dir_path
local function setup_dir_path(provided_dir)
if provided_dir then
return provided_dir
end
return "/mnt/mtwo/programming/ai-stuff/neocities-modernization"
end
-- }}}

-- Script configuration
local DIR = setup_dir_path(arg and arg[1])

-- Load required libraries
package.path = DIR .. "/libs/?.lua;" .. DIR .. "/src/?.lua;" .. package.path
local utils = require("utils")
local flat_html_generator = require("flat-html-generator")

-- {{{ local function relative_path
-- Issue 7-003: Show project name instead of "./" when path equals DIR
local function relative_path(absolute_path)
if absolute_path == DIR or absolute_path == DIR .. "/" then
local dir_name = DIR:match("([^/]+)/?$")
return dir_name .. "/"
end
if absolute_path:sub(1, #DIR) == DIR then
local rel = absolute_path:sub(#DIR + 1)
if rel:sub(1, 1) == "/" then rel = rel:sub(2) end
return "./" .. rel
end
return absolute_path
end
-- }}}

-- {{{ function generate_test_pages
local function generate_test_pages(poem_ids)
local poems_file = DIR .. "/assets/poems.json"
local similarity_file = DIR .. "/assets/embeddings/embeddinggemma_latest/similarity_matrix.json"
local embeddings_file = DIR .. "/assets/embeddings/embeddinggemma_latest/embeddings.json"
local output_dir = DIR .. "/output/test"

print("🔄 Loading data files...")
local poems_data = utils.read_json_file(poems_file)
local similarity_data = utils.read_json_file(similarity_file)
local embeddings_data = utils.read_json_file(embeddings_file)

if not poems_data then
print("❌ Error: Could not load poems.json")
return false
end

if not similarity_data then
print("❌ Error: Could not load similarity_matrix.json")
return false
end

if not embeddings_data then
print("❌ Error: Could not load embeddings.json")
return false
end

print("✅ Data files loaded successfully")
print(string.format(" 📄 Poems: %d", #poems_data.poems))
print(string.format(" 🧮 Embeddings: %d", embeddings_data.total_embeddings or 0))

-- Create output directories
os.execute("mkdir -p " .. output_dir .. "/similar")
os.execute("mkdir -p " .. output_dir .. "/different")

-- Build poem lookup table
local poem_lookup = {}
for i, poem in ipairs(poems_data.poems) do
if poem.id then
poem_lookup[poem.id] = poem
end
end

print("\n📝 Generating test pages for poem IDs: " .. table.concat(poem_ids, ", "))

for _, poem_id in ipairs(poem_ids) do
local poem_data = poem_lookup[poem_id]

if not poem_data then
print(string.format("⚠️ Poem ID %d not found, skipping", poem_id))
else
print(string.format("\n🔍 Processing poem %d: %s/%s.txt",
poem_id, poem_data.category or "unknown", poem_id))

-- Show first 100 chars of content
local preview = (poem_data.content or ""):sub(1, 100):gsub("\n", " ")
print(string.format(" Preview: %s...", preview))

-- Generate similarity page
print(" 📊 Generating similarity-ranked page...")
local similar_ranking = flat_html_generator.generate_similarity_ranked_list(
poem_id, poems_data, similarity_data.similarities or similarity_data)

if #similar_ranking > 0 then
local similar_html = flat_html_generator.generate_flat_poem_list_html(
poem_data, similar_ranking, "similar", poem_id)
local similar_file = string.format("%s/similar/%03d.html", output_dir, poem_id)

if utils.write_file(similar_file, similar_html) then
print(string.format(" ✅ Similar page: %s", relative_path(similar_file)))
print(string.format(" Top 3 similar: %s",
table.concat({
similar_ranking[2] and similar_ranking[2].id or "?",
similar_ranking[3] and similar_ranking[3].id or "?",
similar_ranking[4] and similar_ranking[4].id or "?"
}, ", ")))
else
print(" ❌ Failed to write similar page")
end
else
print(" ⚠️ No similarity data found for this poem")
end

-- Generate difference page
print(" 📊 Generating difference-ranked page...")
local diverse_sequence = flat_html_generator.generate_maximum_diversity_sequence(
poem_id, poems_data, embeddings_data)

if #diverse_sequence > 0 then
local diverse_html = flat_html_generator.generate_flat_poem_list_html(
poem_data, diverse_sequence, "different", poem_id)
local diverse_file = string.format("%s/different/%03d.html", output_dir, poem_id)

if utils.write_file(diverse_file, diverse_html) then
print(string.format(" ✅ Different page: %s", relative_path(diverse_file)))
print(string.format(" Top 3 different: %s",
table.concat({
diverse_sequence[2] and diverse_sequence[2].id or "?",
diverse_sequence[3] and diverse_sequence[3].id or "?",
diverse_sequence[4] and diverse_sequence[4].id or "?"
}, ", ")))
else
print(" ❌ Failed to write different page")
end
else
print(" ⚠️ No embedding data found for this poem")
end
end
end

print("\n✅ Test generation complete!")
print(string.format(" 📁 Output directory: %s", relative_path(output_dir)))
print("\n📖 To view the test pages:")
print(string.format(" firefox %s/similar/%03d.html", relative_path(output_dir), poem_ids[1]))
print(string.format(" firefox %s/different/%03d.html", relative_path(output_dir), poem_ids[1]))

return true
end
-- }}}

-- Main execution
print("=" .. string.rep("=", 60))
print("HTML Generation Test - Validation Script")
print("=" .. string.rep("=", 60))
print("Project directory: " .. relative_path(DIR))

-- Default test poem IDs - one early, one middle
local test_poem_ids = {1, 100}

-- Allow custom IDs from command line
if arg and arg[2] then
test_poem_ids = {}
for i = 2, #arg do
local id = tonumber(arg[i])
if id then
table.insert(test_poem_ids, id)
end
end
end

generate_test_pages(test_poem_ids)