init

LlamaCli.func.php

@@ -0,0 +1,77 @@
+<?php
+
+require_once __DIR__."/lib/index.php";
+
+function LlamaCli(string $prompt, string $system = "You are a helpful assistant", array $options = []) {	
+
+	// Get the options passed
+	extract($options);
+
+	// Set defaults for what was not explicitly specified
+
+	$llamacpp_bin         = $llamacpp_bin ?? "/home/frederico/llama.cpp/build/bin/llama-cli";
+	$llamacpp_model_path  = $llamacpp_model_path ?? "/home/frederico/llama.cpp/models";
+	$llamacpp_model       = $llamacpp_model ?? "Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf";
+
+	$parseCodeInOutput    = $parseCodeInOutput ?? true;
+	$previousConversation = $previousConversation ?? []; // expects: [["role"=>"assistant","content"=>"How can I help you?"],["role"=>"user","content"=>"Tell me the capital of France"]]
+	$debug_level          = $debug_level ?? 0;
+
+	// 1. Parse the prompt and resolve actions "@"
+	$prompt = ResolveActions($prompt);
+
+	
+	// $llm_stdin = $prompt;
+	// Llama chat template
+	$llm_stdin  = "<|start_header_id|>system<|end_header_id|>$system<|eot_id|>";
+	foreach($previousConversation as $message)
+		$llm_stdin .= "<|start_header_id|>{$message["role"]}<|end_header_id|>{$message["content"]}<|eot_id|>";
+	$llm_stdin .= "<|start_header_id|>user<|end_header_id|>$prompt<|eot_id|>";
+	$llm_stdin .= "<|start_header_id|>assistant<|end_header_id|>";
+	
+	// Create a unique temporary file for STDIN, input data
+	$input_file_name = tempnam(sys_get_temp_dir(), 'tmp_');
+	if ($debug_level >= 4) file_put_contents("php://stderr","input_file_name $input_file_name content: $llm_stdin\n");
+	file_put_contents($input_file_name, $llm_stdin);
+
+	// Create a unique temporary file for STDERR, debug data
+	$debug_file_name = tempnam(sys_get_temp_dir(), 'tmp_');
+	if ($debug_file_name === false) die("Could not create temporary file\n");
+
+
+
+	$command = "$llamacpp_bin -m $llamacpp_model_path/$llamacpp_model -f $input_file_name -no-cnv --simple-io 2> $debug_file_name ";
+	if ($debug_level >= 3) file_put_contents("php://stderr","Attempting command: $command\n");
+	$output = shell_exec($command);
+
+
+	// 
+	$debug_data = file_get_contents($debug_file_name);
+	if ($debug_level >= 5) file_put_contents("php://stderr","debug_file_name $debug_file_name content: $debug_data\n");
+
+	// Clean TEMP files
+	if (file_exists($debug_file_name)) unlink($debug_file_name);
+	if (file_exists($input_file_name)) unlink($input_file_name);
+
+	//
+	// 5. Erase the input from the output
+	//
+
+	// 5.1 Llama specific model. produces in this format:
+	$output = str_replace("system$system", "", $output);
+	$output = str_replace("user$prompt"."assistant", "", $output);
+	foreach($previousConversation as $message)
+		$output = str_replace($message["role"].$message["content"],"",$output);
+	$output = str_replace("[end of text]", "", $output);
+	// 5.2 Generic input 
+	//$output = str_replace($llm_stdin, "", $output);
+	// 5.3 Sanitize output
+	$output = trim($output);
+	if ($parseCodeInOutput)
+		$output = ExtractCodeSections($output);
+
+
+	return $output;
+}
+
+function LlamaCli_raw(string $prompt, string $system, array $options = []) { $options["parseCodeInOutput"] = false; return LlamaCli($prompt, $system,$options); }

lib/ExtractCodeSections.function.php

@@ -0,0 +1,33 @@
+<?php
+
+function ExtractCodeSections(string $raw_output) {
+
+	$output = [];
+	$currentBlock = "";
+	$currentBlockLang = "";
+	foreach(explode("\n",$raw_output) as $line) {
+		if (substr($line,0,3) == "```") {
+			if (empty($currentBlockLang)) {
+				if (!empty($currentBlock)) {$output[] = trim($currentBlock); $currentBlock = "";}
+				$currentBlockLang = str_replace("```","",$line) ?? "code";
+				continue;
+			} else {
+				if (isset($output[$currentBlockLang]) && is_string($output[$currentBlockLang])) $output[$currentBlockLang] = [$output[$currentBlockLang]];
+				if (isset($output[$currentBlockLang]) && is_array($output[$currentBlockLang])) 
+					$output[$currentBlockLang][] = trim($currentBlock);
+				else 
+					$output[$currentBlockLang] = trim($currentBlock); 
+				
+				$currentBlock = "";
+				$currentBlockLang = ""; 
+				continue;
+			}
+		}
+		$currentBlock .= $line."\n";
+	}	
+
+	$output[] = trim($currentBlock);
+
+	return $output;
+	
+}

lib/ResolveActions.function.php

@@ -0,0 +1,47 @@
+<?php 
+function ResolveActions(string $text) {
+
+
+	if (preg_match("/@rss_reader ([\w:\/\.]+)/",$text,$matches)) {
+		$url = $matches[1];
+
+		    // Initialize cURL
+                    $ch = curl_init();
+                    curl_setopt_array($ch, [
+                        CURLOPT_URL => $url,
+                        CURLOPT_RETURNTRANSFER => true,
+                        CURLOPT_FOLLOWLOCATION => true,
+                        CURLOPT_TIMEOUT => 10,
+                        CURLOPT_USERAGENT => "PHP cURL RSS Reader"
+                    ]);
+                    
+                    $response = curl_exec($ch);
+                    
+                    if (curl_errno($ch)) die("cURL error: " . curl_error($ch));
+                    curl_close($ch);
+                    
+                    // Parse as XML
+                    $xml = @simplexml_load_string($response);
+                    
+                    if (!$xml) die("Failed to parse XML.");
+                    
+                    // Find first <title>
+                    $firstTitle = '';
+                    if (isset($xml->channel->item[0]->title)) {
+                        // RSS 2.0 style
+                        $firstTitle = (string)$xml->channel->item[0]->title;
+                    } elseif (isset($xml->entry[0]->title)) {
+                        // Atom style
+                        $firstTitle = (string)$xml->entry[0]->title;
+                    } elseif (isset($xml->title)) {
+                        // fallback
+                        $firstTitle = (string)$xml->title;
+                    }
+                
+
+		$text = str_replace($matches[0],$firstTitle,$text);
+	}
+
+	return $text;
+
+}

lib/index.php

@@ -0,0 +1,6 @@
+<?php
+
+foreach(scandir(__DIR__) as $filename) {
+	if ($filename == "." || $filename == "..") continue;
+	if (substr($filename, -13) == ".function.php") include __DIR__."/$filename";
+}

test.php

@@ -0,0 +1,74 @@
+<?php require_once __DIR__."/LlamaCli.func.php";
+
+// TEST 01: Simple query. 
+
+// $out = LlamaCli("What is the capital of France?"); print_r($out);
+
+// TEST 02: Simple query with previous background conversation
+
+// $out = LlamaCli("Portugal?","",["previousConversation"=>[["role"=>"user","content"=>"What is the capital of France?"],["role"=>"assistant","content"=>"Paris."]]]); print_r($out);
+
+
+// TEST 03: Simple query with SYSTEM PRIMING message
+
+//$summary = LlamaCli(file_get_contents("test_input_data.3.txt"),"You are an executive assistant that will summarize english text in 1 paragraph");
+//$title = LlamaCli($summary[0],"You are an talented journalist that will produce a provocative headline title based on a summary of a text");
+
+
+// TEST 04: RECURSIVE QUERY, to produce long-content from short-one. LONG-WRITING. ( GOAL: Learning from model deep knowledge )
+
+// // -------- BOOK EXAMPLE ---------
+// 
+// // 0. CATEGORY
+// $category = "learning finnish";
+// $expert   = "teacher";
+// $total_no_of_chapters = 3;
+// 
+// 
+// // 1. TITLE
+// $book_title = LlamaCli_raw("Write me a title for my new book about $category. Output title only, no options, no chat.","You are an expert $expert.");
+// 
+// 
+// // 2. CHAPTER TITLES
+// $sys_msg = "You are an expert $expert writing a book with the title $book_title with $total_no_of_chapters chapters";
+// $chapters = []; $conv_hist = [];
+// 
+// $msg = "Write me the title of my first chapter. Output title only, no options, no chat.";
+// $conv_hist[] = ["role"=>"user","content"=>$msg];
+// $conv_hist[] = ["role"=>"assistant","content"=>($chapters[] = LlamaCli_raw($msg,$sys_msg,["debug_level"=>0]))];
+// 
+// for($no=1; $no < $total_no_of_chapters; $no++) {
+// 	$msg = "Write me the title of chapter number $no.";
+// 	$conv_hist[] = ["role"=>"user","content"=>$msg];
+// 	$conv_hist[] = ["role"=>"assistant","content"=>($chapters[] = LlamaCli_raw($msg,$sys_msg,["previousConversation"=>$conv_hist, "debug_level"=>0]))];
+// 
+// }
+// 
+// // 3. CHAPTER CONTENTS
+// $content = [];
+// foreach($chapters as $chapter_title)
+// 	$content[$chapter_title] = LlamaCli_raw(
+// 		"Write 2 paragraphs for a chapter titled $chapter_title in a book called $book_title. Output content only, no chat.",
+// 		"You are an expert $expert."
+// 	);
+// 
+// print_r([$book_title, $content]);
+
+
+
+// TEST 05: CHAINED QUERY, to parse long-content and produce short one. SUMMARIZATION. ( GOAL: )
+
+
+// $summaries = [];
+// $content = file_get_contents("test_input_data.2.txt");
+// $chunks = array_chunk(explode("\n",$content),80);
+// 
+// foreach($chunks as $chunk) {
+// 	$summaries[] = LlamaCli_raw(implode("\n",$chunk),"You are a legal executive assistant that will summarize a conversation in english between ex-husband and wife in 1 paragraph");
+// }
+// print_r($summaries);
+	
+
+// TEST 06: Resolve ACTIONS
+
+echo LlamaCli_raw("translate this title to portuguese and finnish: @rss_reader https://feeds.bbci.co.uk/news/rss.xml cde def","");