Janady13/complete-agi-core / seferim/seferim-learner.sh

1#!/bin/bash 2# ============================================================================= 3# SEFERIM LEARNER v4 - Simple HTTP APIs (no jq, no Python dependencies) 4# Uses curl + sed for parsing 5# ============================================================================= 6 7set +e 8 9LOG=~/.seferim/logs/learner.log 10LEARNED_DIR=~/.seferim/learned 11ICLOUD_LEARNED=~/Library/Mobile\ Documents/com~apple~CloudDocs/AGI_Data/learned/macbook2 12 13mkdir -p "$LEARNED_DIR" "$ICLOUD_LEARNED" ~/.seferim/logs 14 15UA="SEFERIM-AGI/2.0" 16FAMILIES="epona athena aurora selene mnemosyne logos yad hermes prometheus" 17 18get_family_topics() { 19 case "$1" in 20 epona) echo "science history biology physics chemistry astronomy" ;; 21 athena) echo "strategy planning military chess war business" ;; 22 aurora) echo "art music painting sculpture creativity poetry" ;; 23 selene) echo "consciousness mind brain thought awareness" ;; 24 mnemosyne) echo "memory learning knowledge cognition" ;; 25 logos) echo "mathematics logic proof theorem algebra geometry" ;; 26 yad) echo "programming algorithm software code computer" ;; 27 hermes) echo "communication network internet protocol" ;; 28 prometheus) echo "innovation technology invention startup" ;; 29 *) echo "knowledge science fact" ;; 30 esac 31} 32 33# ============================================================================= 34# SIMPLE API FUNCTIONS (curl + sed parsing) 35# ============================================================================= 36 37# Quotable - returns quotes (HTTP works!) 38fetch_quote() { 39 local raw=$(curl -s --max-time 8 "http://api.quotable.io/random" 2>/dev/null) 40 if [ -n "$raw" ]; then 41 # Extract content and author using sed 42 local content=$(echo "$raw" | sed 's/.*"content":"\([^"]*\)".*/\1/') 43 local author=$(echo "$raw" | sed 's/.*"author":"\([^"]*\)".*/\1/') 44 echo "Quote: $content - $author" 45 fi 46} 47 48# Numbers API - math facts (HTTP) 49fetch_number_fact() { 50 local num=$((RANDOM % 1000)) 51 curl -s --max-time 5 "http://numbersapi.com/$num/math" 2>/dev/null 52} 53 54# Date facts (HTTP) 55fetch_date_fact() { 56 local month=$(( (RANDOM % 12) + 1 )) 57 local day=$(( (RANDOM % 28) + 1 )) 58 curl -s --max-time 5 "http://numbersapi.com/$month/$day/date" 2>/dev/null 59} 60 61# HackerNews (HTTPS but usually works) 62fetch_hackernews() { 63 local idx=$((RANDOM % 30)) 64 local ids=$(curl -sk --max-time 5 "https://hacker-news.firebaseio.com/v0/topstories.json" 2>/dev/null) 65 if [ -n "$ids" ]; then 66 # Extract ID at position using sed 67 local story_id=$(echo "$ids" | sed 's/\[//' | sed 's/\]//' | cut -d',' -f$((idx + 1)) | tr -d ' ') 68 if [ -n "$story_id" ]; then 69 local story=$(curl -sk --max-time 5 "https://hacker-news.firebaseio.com/v0/item/$story_id.json" 2>/dev/null) 70 if [ -n "$story" ]; then 71 local title=$(echo "$story" | sed 's/.*"title":"\([^"]*\)".*/\1/') 72 local score=$(echo "$story" | sed 's/.*"score":\([0-9]*\).*/\1/') 73 echo "HackerNews: $title. Score: $score" 74 fi 75 fi 76 fi 77} 78 79# Wikipedia (HTTPS - may need -k flag) 80fetch_wikipedia() { 81 local topic="$1" 82 local encoded=$(echo "$topic" | sed 's/ /%20/g') 83 local raw=$(curl -sk --max-time 10 \ 84 "https://en.wikipedia.org/w/api.php?action=query&titles=$encoded&prop=extracts&exintro=1&explaintext=1&format=json" \ 85 -H "User-Agent: $UA" 2>/dev/null) 86 if [ -n "$raw" ]; then 87 # Extract between "extract":" and the next "} 88 local extract=$(echo "$raw" | sed 's/.*"extract":"//' | sed 's/"},.*//' | head -c 800) 89 if [ -n "$extract" ] && [ "$extract" != "$raw" ]; then 90 echo "$extract" 91 fi 92 fi 93} 94 95# DuckDuckGo (HTTPS) 96fetch_duckduckgo() { 97 local query="$1" 98 local encoded=$(echo "$query" | sed 's/ /+/g') 99 local raw=$(curl -sk --max-time 8 \ 100 "https://api.duckduckgo.com/?q=$encoded&format=json&no_html=1" \ 101 -H "User-Agent: $UA" 2>/dev/null) 102 if [ -n "$raw" ]; then 103 # Try AbstractText first 104 local abstract=$(echo "$raw" | sed 's/.*"AbstractText":"//' | sed 's/",.*//' | head -c 600) 105 if [ -n "$abstract" ] && [ ${#abstract} -gt 20 ] && [ "$abstract" != "$raw" ]; then 106 echo "$abstract" 107 fi 108 fi 109} 110 111# Cat facts (HTTP - fun fallback) 112fetch_cat_fact() { 113 local raw=$(curl -sk --max-time 5 "https://catfact.ninja/fact" 2>/dev/null) 114 if [ -n "$raw" ]; then 115 local fact=$(echo "$raw" | sed 's/.*"fact":"//' | sed 's/".*//') 116 echo "Cat Fact: $fact" 117 fi 118} 119 120# ============================================================================= 121# SMART FETCH 122# ============================================================================= 123smart_fetch() { 124 local topic="$1" 125 local family="$2" 126 local content="" 127 local source="" 128 129 case "$family" in 130 aurora|creativity) 131 content=$(fetch_quote) 132 source="quotable" 133 ;; 134 logos|mathematics) 135 content=$(fetch_number_fact) 136 source="numbersapi" 137 [ -z "$content" ] && { content=$(fetch_date_fact); source="numbersapi"; } 138 ;; 139 hermes|prometheus) 140 content=$(fetch_hackernews) 141 source="hackernews" 142 ;; 143 *) 144 # Try Wikipedia first 145 content=$(fetch_wikipedia "$topic") 146 source="wikipedia" 147 # Fallback to DuckDuckGo 148 [ -z "$content" ] || [ ${#content} -lt 30 ] && { content=$(fetch_duckduckgo "$topic"); source="duckduckgo"; } 149 # Fallback to quote 150 [ -z "$content" ] || [ ${#content} -lt 30 ] && { content=$(fetch_quote); source="quotable"; } 151 # Fallback to number fact 152 [ -z "$content" ] || [ ${#content} -lt 30 ] && { content=$(fetch_number_fact); source="numbersapi"; } 153 ;; 154 esac 155 156 # Ultimate fallback 157 [ -z "$content" ] || [ ${#content} -lt 15 ] && { content=$(fetch_cat_fact); source="catfact"; } 158 159 echo "$content|$source" 160} 161 162# ============================================================================= 163# LEARN 164# ============================================================================= 165learn_topic() { 166 local topic="$1" 167 local family="$2" 168 local importance="${3:-0.75}" 169 170 echo "[$(date +%H:%M:%S)] Fetching: $topic for $family" >> "$LOG" 171 172 local result=$(smart_fetch "$topic" "$family") 173 local content=$(echo "$result" | rev | cut -d'|' -f2- | rev) # Everything before last | 174 local source=$(echo "$result" | rev | cut -d'|' -f1 | rev) # Last field 175 176 if [ -n "$content" ] && [ ${#content} -gt 15 ]; then 177 local ts=$(date +%s) 178 local day=$(date +%Y%m%d) 179 local safe_content=$(echo "$content" | head -c 700 | tr '\n' ' ' | sed 's/"/\\"/g' | sed "s/'/\\'/g") 180 181 local entry="{\"timestamp\":$ts,\"source\":\"macbook2\",\"api\":\"$source\",\"topic\":\"$topic\",\"family\":\"$family\",\"content\":\"$safe_content\",\"importance\":$importance}" 182 183 echo "$entry" >> "$LEARNED_DIR/$day.jsonl" 184 echo "$entry" >> "$ICLOUD_LEARNED/$day.jsonl" 185 echo "[$(date +%H:%M:%S)] LEARNED [$family] $topic from $source (${#content} chars)" >> "$LOG" 186 return 0 187 else 188 echo "[$(date +%H:%M:%S)] FAILED: $topic (content too short: ${#content})" >> "$LOG" 189 return 1 190 fi 191} 192 193get_random_family() { 194 local arr=($FAMILIES) 195 local idx=$((RANDOM % ${#arr[@]})) 196 echo "${arr[$idx]}" 197} 198 199get_random_topic() { 200 local topics=$(get_family_topics "$1") 201 local arr=($topics) 202 local idx=$((RANDOM % ${#arr[@]})) 203 echo "${arr[$idx]}" 204} 205 206# ============================================================================= 207# MAIN 208# ============================================================================= 209echo "=============================================" >> "$LOG" 210echo "[$(date)] SEFERIM Learner v4 (Simple HTTP)" >> "$LOG" 211echo " - No jq/Python dependencies" >> "$LOG" 212echo " - HTTP-first APIs" >> "$LOG" 213echo "=============================================" >> "$LOG" 214 215cycle=0 216while true; do 217 cycle=$((cycle + 1)) 218 FAMILY=$(get_random_family) 219 TOPIC=$(get_random_topic "$FAMILY") 220 221 echo "" >> "$LOG" 222 echo "[$(date +%H:%M:%S)] === Cycle $cycle ===" >> "$LOG" 223 echo "[$(date +%H:%M:%S)] Family: $FAMILY, Topic: $TOPIC" >> "$LOG" 224 225 learn_topic "$TOPIC" "$FAMILY" 0.75 226 227 # Curiosity every 4 cycles 228 if [ $((cycle % 4)) -eq 0 ]; then 229 echo "[$(date +%H:%M:%S)] CURIOSITY: HackerNews" >> "$LOG" 230 hn=$(fetch_hackernews) 231 if [ -n "$hn" ] && [ ${#hn} -gt 20 ]; then 232 ts=$(date +%s) 233 day=$(date +%Y%m%d) 234 safe=$(echo "$hn" | head -c 500 | tr '\n' ' ' | sed 's/"/\\"/g') 235 echo "{\"timestamp\":$ts,\"source\":\"macbook2\",\"api\":\"hackernews\",\"topic\":\"tech-news\",\"family\":\"hermes\",\"content\":\"$safe\",\"importance\":0.65}" >> "$LEARNED_DIR/$day.jsonl" 236 echo "{\"timestamp\":$ts,\"source\":\"macbook2\",\"api\":\"hackernews\",\"topic\":\"tech-news\",\"family\":\"hermes\",\"content\":\"$safe\",\"importance\":0.65}" >> "$ICLOUD_LEARNED/$day.jsonl" 237 echo "[$(date +%H:%M:%S)] CURIOSITY learned!" >> "$LOG" 238 fi 239 fi 240 241 # Status every 8 242 if [ $((cycle % 8)) -eq 0 ]; then 243 total=$(cat "$LEARNED_DIR"/*.jsonl 2>/dev/null | wc -l | tr -d ' ') 244 echo "[$(date +%H:%M:%S)] STATUS: $total total items" >> "$LOG" 245 fi 246 247 sleep 35 248done 249