lwazi 1.7.0 → 1.7.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "lwazi",
-  "version": "1.7.0",
+  "version": "1.7.2",
   "description": "Lwazi is an AI assistant for Laravel. Install with one command to add an AI assistant to your Laravel app.",
   "main": "bin/lwazi.js",
   "bin": {

package/src/Console/AnalyzeProjectCommand.php

@@ -6,6 +6,9 @@ use Illuminate\Console\Command;
 use Lwazi\Core\Installer\NavigationCrawler;
 use Lwazi\Core\Services\GraphVisualizer;
 use Illuminate\Support\Facades\Storage;
+use Illuminate\Support\Str;
+use DOMDocument;
+use DOMXPath;
 
 class AnalyzeProjectCommand extends Command
 {
@@ -26,6 +29,11 @@ class AnalyzeProjectCommand extends Command
         $crawler = new NavigationCrawler($rootUrl);
         $manifest = $crawler->crawl();
         
+        if (empty($manifest['nodes'] ?? [])) {
+            $this->info("Crawler returned no pages, trying fallback...");
+            $manifest = $this->fallbackExtractLinks($rootUrl);
+        }
+        
         $manifest = $this->mergeWithRoutes($manifest, $rootUrl);
 
         $storagePath = storage_path('lwazi');
@@ -37,7 +45,7 @@ class AnalyzeProjectCommand extends Command
         file_put_contents($manifestFile, json_encode($manifest, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES));
 
         $this->info("Analysis complete. Manifest stored at: {$manifestFile}");
-        $this->info("Pages discovered: " . count($manifest['nodes']));
+        $this->info("Pages discovered: " . count($manifest['nodes'] ?? $manifest['flat'] ?? []));
 
         $this->info("\n" . str_repeat('=', 50));
         $this->info('SITE NAVIGATION GRAPH');
@@ -138,4 +146,90 @@ class AnalyzeProjectCommand extends Command
         $path = preg_replace('/\s+/', ' ', $path);
         return ucwords(trim($path));
     }
+    
+    protected function fallbackExtractLinks(string $url): array
+    {
+        try {
+            $response = \Illuminate\Support\Facades\Http::timeout(10)->get($url);
+            if (!$response->successful()) {
+                return ['nodes' => [], 'adjacency' => [], 'flat' => [], 'root_url' => $url];
+            }
+            
+            $html = $response->body();
+            
+            libxml_use_internal_errors(true);
+            $dom = new DOMDocument();
+            @$dom->loadHTML($html);
+            libxml_clear_errors();
+            
+            $xpath = new DOMXPath($dom);
+            $links = $xpath->query('//a[@href]');
+            
+            $manifest = [
+                'nodes' => [],
+                'adjacency' => [$url => []],
+                'flat' => [],
+                'root_url' => $url,
+            ];
+            
+            $seen = [];
+            
+            foreach ($links as $a) {
+                $href = trim($a->getAttribute('href'));
+                if (!$href || Str::startsWith($href, ['#', 'javascript:', 'mailto:', 'tel:'])) {
+                    continue;
+                }
+                
+                $fullUrl = $this->makeAbsolute($href, $url);
+                if (!$fullUrl || !Str::startsWith($fullUrl, $url)) {
+                    continue;
+                }
+                
+                if (isset($seen[$fullUrl])) {
+                    continue;
+                }
+                $seen[$fullUrl] = true;
+                
+                $text = trim($a->textContent ?? '');
+                
+                $manifest['nodes'][$fullUrl] = [
+                    'url' => $fullUrl,
+                    'title' => $text ?: basename($fullUrl),
+                    'headings' => [$text],
+                ];
+                
+                $manifest['adjacency'][$url][] = $fullUrl;
+                
+                $manifest['flat'][$fullUrl] = [
+                    'label' => $text ?: basename($fullUrl),
+                    'segments' => array_filter(explode('/', parse_url($fullUrl, PHP_PATH) ?? '')),
+                    '_path' => $fullUrl,
+                    '_weight' => 1,
+                ];
+            }
+            
+            return $manifest;
+            
+        } catch (\Exception $e) {
+            return ['nodes' => [], 'adjacency' => [], 'flat' => [], 'root_url' => $url];
+        }
+    }
+    
+    protected function makeAbsolute(string $href, string $base): string
+    {
+        if (Str::startsWith($href, ['http://', 'https://'])) {
+            return $href;
+        }
+        
+        if ($href === '/') {
+            return $base;
+        }
+        
+        if (Str::startsWith($href, '/')) {
+            $parsed = parse_url($base);
+            return ($parsed['scheme'] ?? 'http') . '://' . ($parsed['host'] ?? '') . $href;
+        }
+        
+        return $base . '/' . $href;
+    }
 }

package/src/Console/SetupCommand.php

@@ -108,15 +108,21 @@ class SetupCommand extends Command
     protected function runCrawler(string $url): void
     {
         $this->info("\nCrawling website: {$url}...");
+        
         try {
             $crawler = new NavigationCrawler($url, true);
             $manifest = $crawler->crawl();
             
+            if (empty($manifest['nodes'] ?? [])) {
+                $this->info("Crawler returned no pages, trying fallback extraction...");
+                $manifest = $this->fallbackExtractLinks($url);
+            }
+            
             $manifest = $this->mergeWithRoutes($manifest, $url);
             
             $crawler->saveManifest();
             
-            $this->info("Crawled " . count($manifest['nodes']) . " pages.");
+            $this->info("Crawled " . count($manifest['nodes'] ?? $manifest['flat'] ?? []) . " pages.");
             
             $this->info("\n" . str_repeat('=', 50));
             $this->info('SITE NAVIGATION GRAPH');
@@ -130,8 +136,107 @@ class SetupCommand extends Command
             $this->line($summaryOutput);
         } catch (\Throwable $e) {
             $this->warn('Website crawling failed: ' . $e->getMessage());
+            
+            $this->info("Trying fallback link extraction...");
+            try {
+                $manifest = $this->fallbackExtractLinks($url);
+                $manifest = $this->mergeWithRoutes($manifest, $url);
+                
+                $crawler = new NavigationCrawler($url);
+                $crawler->saveManifest();
+                
+                $this->info("Fallback extraction complete.");
+            } catch (\Throwable $e2) {
+                $this->warn('Fallback also failed: ' . $e2->getMessage());
+            }
         }
     }
+    
+    protected function fallbackExtractLinks(string $url): array
+    {
+        try {
+            $response = \Illuminate\Support\Facades\Http::timeout(10)->get($url);
+            if (!$response->successful()) {
+                return ['nodes' => [], 'adjacency' => [], 'flat' => [], 'root_url' => $url];
+            }
+            
+            $html = $response->body();
+            
+            libxml_use_internal_errors(true);
+            $dom = new DOMDocument();
+            @$dom->loadHTML($html);
+            libxml_clear_errors();
+            
+            $xpath = new DOMXPath($dom);
+            $links = $xpath->query('//a[@href]');
+            
+            $manifest = [
+                'nodes' => [],
+                'adjacency' => [$url => []],
+                'flat' => [],
+                'root_url' => $url,
+            ];
+            
+            $seen = [];
+            
+            foreach ($links as $a) {
+                $href = trim($a->getAttribute('href'));
+                if (!$href || Str::startsWith($href, ['#', 'javascript:', 'mailto:', 'tel:'])) {
+                    continue;
+                }
+                
+                $fullUrl = $this->makeAbsolute($href, $url);
+                if (!$fullUrl || !Str::startsWith($fullUrl, $url)) {
+                    continue;
+                }
+                
+                if (isset($seen[$fullUrl])) {
+                    continue;
+                }
+                $seen[$fullUrl] = true;
+                
+                $text = trim($a->textContent ?? '');
+                
+                $manifest['nodes'][$fullUrl] = [
+                    'url' => $fullUrl,
+                    'title' => $text ?: basename($fullUrl),
+                    'headings' => [$text],
+                ];
+                
+                $manifest['adjacency'][$url][] = $fullUrl;
+                
+                $manifest['flat'][$fullUrl] = [
+                    'label' => $text ?: basename($fullUrl),
+                    'segments' => array_filter(explode('/', parse_url($fullUrl, PHP_PATH) ?? '')),
+                    '_path' => $fullUrl,
+                    '_weight' => 1,
+                ];
+            }
+            
+            return $manifest;
+            
+        } catch (\Exception $e) {
+            return ['nodes' => [], 'adjacency' => [], 'flat' => [], 'root_url' => $url];
+        }
+    }
+    
+    protected function makeAbsolute(string $href, string $base): string
+    {
+        if (Str::startsWith($href, ['http://', 'https://'])) {
+            return $href;
+        }
+        
+        if ($href === '/') {
+            return $base;
+        }
+        
+        if (Str::startsWith($href, '/')) {
+            $parsed = parse_url($base);
+            return ($parsed['scheme'] ?? 'http') . '://' . ($parsed['host'] ?? '') . $href;
+        }
+        
+        return $base . '/' . $href;
+    }
 
     protected function mergeWithRoutes(array $manifest, string $rootUrl): array
     {

package/src/Services/LwaziService.php

@@ -189,8 +189,20 @@ class LwaziService
         ]);
 
         $json = $this->extractJson($response2['content'] ?? '');
-        $terms = is_array($json) ? array_values($json) : [];
-        $terms[] = strtolower($topic);
+        $rawTerms = is_array($json) ? array_values($json) : [];
+        
+        $terms = [];
+        foreach ($rawTerms as $t) {
+            $t = strtolower(trim($t));
+            if (strpos($t, ' ') === false && strlen($t) >= 3) {
+                $terms[] = $t;
+            }
+        }
+        
+        $first = explode(' ', $topic)[0];
+        if (strlen($first) >= 3) {
+            $terms[] = strtolower($first);
+        }
 
         if (empty($terms)) {
             return null;
@@ -358,8 +370,20 @@ class LwaziService
         ]);
 
         $json = $this->extractJson($response2['content'] ?? '');
-        $terms = is_array($json) ? array_values($json) : [];
-        $terms[] = strtolower($topic);
+        $rawTerms = is_array($json) ? array_values($json) : [];
+        
+        $terms = [];
+        foreach ($rawTerms as $t) {
+            $t = strtolower(trim($t));
+            if (strpos($t, ' ') === false && strlen($t) >= 3) {
+                $terms[] = $t;
+            }
+        }
+        
+        $first = explode(' ', $topic)[0];
+        if (strlen($first) >= 3) {
+            $terms[] = strtolower($first);
+        }
 
         if (empty($terms)) {
             return null;