Complete digest system

This commit is contained in:
2026-01-06 10:01:24 +01:00
parent 3f2d58b257
commit 252ba9f5cc
6 changed files with 751 additions and 224 deletions

542
publish_digest.rb Normal file
View File

@@ -0,0 +1,542 @@
#!/usr/bin/env ruby
# -*- coding: utf-8 -*-
#
# Zprávobot.news - AI Daily Digest Publisher
# Version: 1.0.1 (Cloudron - Direct HTTP)
#
# Generates and publishes daily digest posts to Mastodon bots:
# - @zpravobot (7:30) - neutral overview
# - @pozitivni (12:00) - positive news
# - @sarkasticky (19:00) - sarcastic commentary
require 'csv'
require 'json'
require 'time'
require 'net/http'
require 'uri'
require 'optparse'
# ==========================================
# CONFIGURATION
# ==========================================
MASTODON_URL = 'https://zpravobot.news'
CSV_PATH = '/app/data/posts-latest.csv'
ANTHROPIC_API_URL = 'https://api.anthropic.com/v1/messages'
BOTS = {
'zpravobot' => {
token: ENV['ZPRAVOBOT_TOKEN'],
style: 'neutral',
time_slot: 'morning',
hashtags: '#zpravobot #trendydne'
},
'pozitivni' => {
token: ENV['POZITIVNI_TOKEN'],
style: 'positive',
time_slot: 'noon',
hashtags: '#dobréZprávy #zpravobot'
},
'sarkasticky' => {
token: ENV['SARKASTICKY_TOKEN'],
style: 'sarcastic',
time_slot: 'evening',
hashtags: '#realita #zpravobot'
}
}
# ==========================================
# COMMAND LINE PARSING
# ==========================================
options = {}
OptionParser.new do |opts|
opts.banner = "Usage: publish_digest.rb [options]"
opts.on("--bot BOT", String, "Bot name (zpravobot, pozitivni, sarkasticky)") do |b|
options[:bot] = b
end
opts.on("--dry-run", "Test mode - don't actually publish") do
options[:dry_run] = true
end
opts.on("--date DATE", String, "Process specific date (YYYY-MM-DD)") do |d|
options[:date] = d
end
opts.on("-h", "--help", "Show this help") do
puts opts
exit
end
end.parse!
bot_name = options[:bot]
unless bot_name && BOTS.key?(bot_name)
puts "❌ ERROR: Invalid bot name. Use: zpravobot, pozitivni, or sarkasticky"
exit 1
end
config = BOTS[bot_name]
# Validate environment
unless config[:token]
puts "❌ ERROR: Missing token for @#{bot_name}"
puts " Set environment variable: #{bot_name.upcase}_TOKEN"
exit 1
end
unless ENV['ANTHROPIC_API_KEY']
puts "❌ ERROR: Missing ANTHROPIC_API_KEY"
exit 1
end
# ==========================================
# UTILITIES
# ==========================================
def log(message)
timestamp = Time.now.strftime('%Y-%m-%d %H:%M:%S')
puts "[#{timestamp}] #{message}"
end
def extract_url(text)
text[/https?:\/\/[^\s<>"]+/]
end
# ==========================================
# DATA LOADING
# ==========================================
def load_posts_from_csv(date = nil)
target_date = date || (Time.now - 86400).strftime('%Y-%m-%d')
unless File.exist?(CSV_PATH)
log "❌ CSV file not found: #{CSV_PATH}"
exit 1
end
posts = []
CSV.foreach(CSV_PATH, headers: true, encoding: 'utf-8') do |row|
begin
created = Time.parse(row['created_at'])
if created.strftime('%Y-%m-%d') == target_date
posts << {
'text' => row['text'],
'url' => row['url'] || '',
'created_at' => row['created_at']
}
end
rescue => e
# Skip problematic rows
next
end
end
log "📊 Loaded #{posts.size} posts from #{target_date}"
if posts.empty?
log "⚠️ No posts found for #{target_date}"
exit 1
end
posts
end
# ==========================================
# TOPIC EXTRACTION
# ==========================================
def extract_topics(posts)
topics = Hash.new { |h, k| h[k] = [] }
posts.each do |post|
text = post['text'].downcase
# Add URL to post if not present
post['extracted_url'] = extract_url(post['text']) || post['url']
# Categorize by topic
if text.match?(/trump|venezuela|maduro|grónsko|greenland|usa|bílý dům/)
topics['🌍 Zahraniční politika'] << post
elsif text.match?(/hokej|extraliga|nhl|ms u20/)
topics['🏒 Hokej'] << post
elsif text.match?(/fotbal|chelsea|liga|gól|penalty/)
topics['⚽ Fotbal'] << post
elsif text.match?(/film|seriál|stranger things|hudba|koncert|festival|netflix/)
topics['🎬 Kultura'] << post
elsif text.match?(/počasí|teplota|mráz|sníh|déšť/)
topics['❄️ Počasí'] << post
elsif text.match?(/politika|parlament|vláda|ministr/)
topics['🏛️ Politika'] << post
elsif text.match?(/ekonomika|koruna|inflace|mzdy|ceny/)
topics['💼 Ekonomika'] << post
end
end
# Sort by post count
topics = topics.sort_by { |_, posts| -posts.size }.to_h
log "🔍 Found #{topics.size} topics:"
topics.each { |topic, posts| log " #{topic}: #{posts.size} posts" }
topics
end
# ==========================================
# CONTENT FILTERING BY STYLE
# ==========================================
def filter_topics_by_style(topics, style)
case style
when 'neutral'
topics
when 'positive'
positive_topics = {}
topics.each do |topic, posts|
next if topic.include?('Politika') || topic.include?('Zahraniční')
positive_posts = posts.select do |post|
text = post['text'].downcase
has_positive = text.match?(/úspěch|vítěz|rekord|festival|koncert|ocenění|talent/)
no_negative = !text.match?(/nehoda|smrt|tragédie|havárie|konflikt|krize/)
has_positive && no_negative
end
positive_topics[topic] = positive_posts unless positive_posts.empty?
end
log "💚 Filtered to #{positive_topics.size} positive topics"
positive_topics
when 'sarcastic'
sarcastic_topics = {}
topics.each do |topic, posts|
if topic.include?('Zahraniční') || topic.include?('Politika')
sarcastic_topics[topic] = posts
end
end
if sarcastic_topics.size < 3
topics.each do |topic, posts|
break if sarcastic_topics.size >= 5
sarcastic_topics[topic] = posts unless sarcastic_topics.key?(topic)
end
end
log "😏 Selected #{sarcastic_topics.size} topics for sarcasm"
sarcastic_topics
else
topics
end
end
# ==========================================
# CLAUDE API ANALYSIS
# ==========================================
def analyze_with_claude(posts, topics)
log "🤖 Analyzing with Claude API..."
topic_summary = topics.map { |topic, posts| "#{topic}: #{posts.size}" }.join(', ')
sample_texts = posts[0..49].map { |p| p['text'][0..150] }
prompt = <<~PROMPT
Analyzuj #{posts.size} českých/slovenských zpráv z Mastodon instance Zprávobot.news.
Témata: #{topic_summary}
Ukázka textů:
#{sample_texts[0..9].join("\n---\n")}
Vrať POUZE JSON (žádný markdown):
{
"main_topics": ["téma1", "téma2", "téma3"],
"sentiment": "neutral|positive|negative",
"notable_events": ["událost1", "událost2"]
}
PROMPT
uri = URI(ANTHROPIC_API_URL)
request = Net::HTTP::Post.new(uri)
request['anthropic-version'] = '2023-06-01'
request['content-type'] = 'application/json'
request['x-api-key'] = ENV['ANTHROPIC_API_KEY']
request.body = {
model: 'claude-sonnet-4-20250514',
max_tokens: 1000,
messages: [
{ role: 'user', content: prompt }
]
}.to_json
response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
http.request(request)
end
if response.code != '200'
log "⚠️ Claude API error: #{response.code}"
return default_analysis(topics)
end
data = JSON.parse(response.body)
text = data['content'][0]['text']
analysis = JSON.parse(text.gsub(/```json|```/, '').strip)
log "✅ Claude analysis complete"
analysis
rescue => e
log "⚠️ Claude API error: #{e.message}"
default_analysis(topics)
end
def default_analysis(topics)
{
'main_topics' => topics.keys[0..2],
'sentiment' => 'neutral',
'notable_events' => []
}
end
# ==========================================
# TOOT GENERATION
# ==========================================
def generate_summary_toot(posts_count, topics, style, hashtags)
date = (Time.now - 86400).strftime('%d.%m.%Y')
topic_lines = topics.keys[0..4].map do |topic|
count = topics[topic].size
"#{topic} (#{count}#{style == 'sarcastic' ? '×' : ' postů'})"
end
case style
when 'neutral'
summary = <<~TOOT
📊 TRENDY DNE (#{date})
Zpracováno #{posts_count} postů:
#{topic_lines.join("\n")}
#{hashtags}
👇 Odkazy na vybrané články
TOOT
when 'positive'
summary = <<~TOOT
DOBRÉ ZPRÁVY DNE (#{date})
Z dnešních #{posts_count} zpráv vybrané momenty:
#{topic_lines[0..3].join("\n")}
#{hashtags}
👇 Inspirace na čtení
TOOT
when 'sarcastic'
summary = <<~TOOT
😏 DNEŠNÍ REALITA (#{date})
#{posts_count} postů = co se stalo?
#{topic_lines[0..3].join("\n")}
#{hashtags}
👇 Důkazy zmaru
TOOT
end
if summary.length > 500
summary = summary[0..496] + "..."
end
summary.strip
end
def generate_links_toot(topics, style)
links = []
max_topics = 5
max_links_per_topic = 2
topics.keys[0...max_topics].each do |topic|
posts = topics[topic]
links << "\n#{topic}:"
selected = []
selected << posts[0] if posts[0]
selected << posts[posts.size / 2] if posts.size > 1
selected[0...max_links_per_topic].each do |post|
title = post['text'].split("\n")[0][0..50].strip
title = title.gsub(/\s+/, ' ')
url = post['extracted_url']
next unless url && !url.empty?
short_url = url.gsub(/https?:\/\//, '')
short_url = short_url[0..37] + '...' if short_url.length > 40
links << "#{title}..."
links << " 🔗 #{short_url}"
end
end
case style
when 'neutral'
header = "📌 VYBRANÉ ČLÁNKY DNE:"
footer = "\n#články #zprávy"
when 'positive'
header = "💚 POZITIVNÍ PŘÍBĚHY DNE:"
footer = "\n💙 Máte skvělý den!\n#inspirace"
when 'sarcastic'
header = "🤡 \"BREAKING NEWS\" DNE:"
footer = "\n🙃 Zítra: repeat\n#sarkasmus"
end
toot = header + links.join("\n") + footer
if toot.length > 500
truncated_links = links[0..(links.size * 2 / 3)]
toot = header + truncated_links.join("\n") + footer
if toot.length > 500
toot = toot[0..496] + "..."
end
end
toot.strip
end
# ==========================================
# MASTODON PUBLISHING (DIRECT HTTP)
# ==========================================
def publish_thread(bot_name, summary_toot, links_toot, dry_run: false)
config = BOTS[bot_name]
log "📤 Publishing thread for @#{bot_name}..."
if dry_run
log "🧪 DRY RUN MODE - Not actually publishing"
log "\n--- TOOT 1/2 (#{summary_toot.length} chars) ---"
log summary_toot
log "\n--- TOOT 2/2 (#{links_toot.length} chars) ---"
log links_toot
log "\n✅ Dry run complete"
return [nil, nil]
end
# Publish toot 1
uri = URI("#{MASTODON_URL}/api/v1/statuses")
request = Net::HTTP::Post.new(uri)
request['Authorization'] = "Bearer #{config[:token]}"
request['Content-Type'] = 'application/json'
request.body = { status: summary_toot, visibility: 'public' }.to_json
response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
http.request(request)
end
unless response.code == '200'
log "❌ ERROR: #{response.body}"
exit 1
end
toot1_data = JSON.parse(response.body)
toot1_url = toot1_data['url']
toot1_id = toot1_data['id']
log "✅ Toot 1/2 published: #{toot1_url}"
# Publish toot 2 as reply
request2 = Net::HTTP::Post.new(uri)
request2['Authorization'] = "Bearer #{config[:token]}"
request2['Content-Type'] = 'application/json'
request2.body = {
status: links_toot,
in_reply_to_id: toot1_id,
visibility: 'public'
}.to_json
response2 = Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
http.request(request2)
end
log "✅ Toot 2/2 published (thread)"
[toot1_data, JSON.parse(response2.body)]
rescue => e
log "❌ ERROR publishing thread: #{e.message}"
exit 1
end
# ==========================================
# MAIN EXECUTION
# ==========================================
def main(bot_name, options = {})
log "🚀 Starting Daily Digest for @#{bot_name}"
log "=" * 60
config = BOTS[bot_name]
posts = load_posts_from_csv(options[:date])
log "\n🔍 Extracting topics..."
all_topics = extract_topics(posts)
topics = filter_topics_by_style(all_topics, config[:style])
if topics.empty?
log "⚠️ No suitable topics found for style: #{config[:style]}"
exit 1
end
log "\n🤖 Analyzing with Claude..."
analysis = analyze_with_claude(posts, topics)
log "\n📝 Generating content..."
summary = generate_summary_toot(posts.size, topics, config[:style], config[:hashtags])
links = generate_links_toot(topics, config[:style])
log " Summary: #{summary.length} chars"
log " Links: #{links.length} chars"
log "\n📤 Publishing to Mastodon..."
toot1, toot2 = publish_thread(bot_name, summary, links, dry_run: options[:dry_run])
log "\n" + "=" * 60
log "✅ Digest complete for @#{bot_name}"
unless options[:dry_run]
log "🔗 Thread: #{toot1['url']}" if toot1
end
end
# Run main
begin
main(bot_name, options)
rescue Interrupt
log "\n⚠️ Interrupted by user"
exit 130
rescue => e
log "❌ FATAL ERROR: #{e.message}"
log " #{e.backtrace[0..4].join("\n ")}"
exit 1
end