feat: improve markdown formatting and add AI configuration to init command

tommyknocker · tommyknocker · commit 6d8fbbb5df02 · 2025-12-08T16:10:16.000+03:00
- Fix markdown formatter: preserve asterisks and underscores in code blocks
- Fix markdown formatter: prevent underscores in words (e.g. users_created_at) from being formatted as italic
- Fix markdown formatter: process lists before italic formatting to avoid conflicts
- Fix markdown formatter: improve code block border alignment
- Fix: add missing DeepSeek API key loading from environment variables
- feat: add AI configuration to pdodb init command
  - Add askAiConfiguration() method to InitWizard
  - Add loadAiConfigFromEnv() for non-interactive mode
  - Update InitConfigGenerator to include AI settings in .env and config/db.php
  - Support all AI providers (openai, anthropic, google, microsoft, ollama, deepseek)
diff --git a/src/ai/AiConfig.php b/src/ai/AiConfig.php
@@ -108,6 +108,11 @@ protected function loadFromEnvironment(): void
             $this->apiKeys['microsoft'] = $microsoftKey;
         }
 
+        $deepseekKey = getenv('PDODB_AI_DEEPSEEK_KEY');
+        if ($deepseekKey !== false && $deepseekKey !== '') {
+            $this->apiKeys['deepseek'] = $deepseekKey;
+        }
+
         $ollamaUrl = getenv('PDODB_AI_OLLAMA_URL');
         if ($ollamaUrl !== false && $ollamaUrl !== '') {
             $this->ollamaUrl = $ollamaUrl;
diff --git a/src/cli/InitConfigGenerator.php b/src/cli/InitConfigGenerator.php
@@ -80,6 +80,36 @@ public static function generateEnv(array $config, array $structure, string $path
             $lines[] = '';
         }
 
+        // AI configuration
+        if (isset($config['ai']) && is_array($config['ai'])) {
+            $lines[] = '# AI Configuration';
+            if (isset($config['ai']['provider'])) {
+                $lines[] = 'PDODB_AI_PROVIDER=' . $config['ai']['provider'];
+            }
+
+            $providers = ['openai', 'anthropic', 'google', 'microsoft', 'deepseek', 'ollama'];
+            foreach ($providers as $provider) {
+                $keyName = $provider . '_key';
+                if (isset($config['ai'][$keyName])) {
+                    $lines[] = 'PDODB_AI_' . strtoupper($provider) . '_KEY=' . $config['ai'][$keyName];
+                }
+            }
+
+            if (isset($config['ai']['ollama_url'])) {
+                $lines[] = 'PDODB_AI_OLLAMA_URL=' . $config['ai']['ollama_url'];
+            }
+
+            // Provider-specific settings
+            if (isset($config['ai']['providers']) && is_array($config['ai']['providers'])) {
+                foreach ($config['ai']['providers'] as $provider => $settings) {
+                    if (is_array($settings) && isset($settings['model'])) {
+                        $lines[] = 'PDODB_AI_' . strtoupper($provider) . '_MODEL=' . $settings['model'];
+                    }
+                }
+            }
+            $lines[] = '';
+        }
+
         // Project paths
         if (!empty($structure)) {
             $lines[] = '# Project Paths';
@@ -237,6 +267,11 @@ protected static function buildSingleConnectionArray(array $config): array
             $result['enable_regexp'] = $config['enable_regexp'];
         }
 
+        // AI configuration
+        if (isset($config['ai']) && is_array($config['ai'])) {
+            $result['ai'] = $config['ai'];
+        }
+
         return $result;
     }
 
diff --git a/src/cli/InitWizard.php b/src/cli/InitWizard.php
@@ -168,6 +168,13 @@ protected function loadConfigFromEnv(): void
                 }
             }
         }
+
+        // Load AI configuration if enabled
+        $aiProvider = getenv('PDODB_AI_PROVIDER');
+        if ($aiProvider !== false && $aiProvider !== '') {
+            $this->config['ai'] = ['provider' => mb_strtolower($aiProvider, 'UTF-8')];
+            $this->loadAiConfigFromEnv();
+        }
     }
 
     /**
@@ -185,6 +192,45 @@ protected function loadStructureFromEnv(): void
         ];
     }
 
+    /**
+     * Load AI configuration from environment variables.
+     */
+    protected function loadAiConfigFromEnv(): void
+    {
+        if (!isset($this->config['ai']) || !is_array($this->config['ai'])) {
+            $this->config['ai'] = [];
+        }
+
+        $providers = ['openai', 'anthropic', 'google', 'microsoft', 'deepseek', 'ollama'];
+        foreach ($providers as $provider) {
+            $envVar = 'PDODB_AI_' . strtoupper($provider) . '_KEY';
+            $key = getenv($envVar);
+            if ($key !== false && $key !== '') {
+                $this->config['ai'][$provider . '_key'] = $key;
+            }
+        }
+
+        $ollamaUrl = getenv('PDODB_AI_OLLAMA_URL');
+        if ($ollamaUrl !== false && $ollamaUrl !== '') {
+            $this->config['ai']['ollama_url'] = $ollamaUrl;
+        }
+
+        // Load provider-specific settings
+        if (!isset($this->config['ai']['providers'])) {
+            $this->config['ai']['providers'] = [];
+        }
+        foreach ($providers as $provider) {
+            $modelVar = 'PDODB_AI_' . strtoupper($provider) . '_MODEL';
+            $model = getenv($modelVar);
+            if ($model !== false && $model !== '') {
+                if (!isset($this->config['ai']['providers'][$provider])) {
+                    $this->config['ai']['providers'][$provider] = [];
+                }
+                $this->config['ai']['providers'][$provider]['model'] = $model;
+            }
+        }
+    }
+
     /**
      * Ask for basic database connection settings.
      */
@@ -447,6 +493,7 @@ protected function askAdvancedOptions(): void
         $this->askTablePrefix();
         $this->askCaching();
         $this->askPerformance();
+        $this->askAiConfiguration();
         $this->askMultipleConnections();
     }
 
@@ -623,6 +670,73 @@ protected function askPerformance(): void
         }
     }
 
+    /**
+     * Ask AI configuration.
+     */
+    protected function askAiConfiguration(): void
+    {
+        echo "\n  AI Configuration\n";
+        echo "  ----------------\n";
+        $enableAi = static::readConfirmation('  Enable AI-powered database analysis?', false);
+        if (!$enableAi) {
+            return;
+        }
+
+        if (!isset($this->config['ai'])) {
+            $this->config['ai'] = [];
+        }
+
+        $providers = ['openai', 'anthropic', 'google', 'microsoft', 'ollama', 'deepseek'];
+        echo "  Available providers: " . implode(', ', $providers) . "\n";
+        $provider = static::readInput('  Default AI provider', 'openai');
+        $provider = mb_strtolower(trim($provider), 'UTF-8');
+        if (!in_array($provider, $providers, true)) {
+            $provider = 'openai';
+        }
+        $this->config['ai']['provider'] = $provider;
+
+        // Ask for API keys
+        echo "\n  API Keys (leave empty to skip)\n";
+        foreach ($providers as $p) {
+            if ($p === 'ollama') {
+                // Ollama doesn't need API key, but ask for URL
+                $url = static::readInput("  Ollama URL", 'http://localhost:11434');
+                if ($url !== '') {
+                    $this->config['ai']['ollama_url'] = $url;
+                }
+            } else {
+                $key = static::readPassword("  {$p} API key (optional, press Enter to skip)");
+                if ($key !== '') {
+                    $this->config['ai'][$p . '_key'] = $key;
+                }
+            }
+        }
+
+        // Ask for provider-specific settings
+        $configureProviderSettings = static::readConfirmation('  Configure provider-specific settings (model, temperature, etc.)?', false);
+        if ($configureProviderSettings) {
+            if (!isset($this->config['ai']['providers'])) {
+                $this->config['ai']['providers'] = [];
+            }
+
+            foreach ($providers as $p) {
+                $hasKey = isset($this->config['ai'][$p . '_key']) || ($p === 'ollama');
+                if (!$hasKey) {
+                    continue;
+                }
+
+                echo "\n  {$p} settings:\n";
+                $model = static::readInput("    Model (optional)", '');
+                if ($model !== '') {
+                    if (!isset($this->config['ai']['providers'][$p])) {
+                        $this->config['ai']['providers'][$p] = [];
+                    }
+                    $this->config['ai']['providers'][$p]['model'] = $model;
+                }
+            }
+        }
+    }
+
     /**
      * Ask multiple connections configuration.
      */
diff --git a/src/cli/MarkdownFormatter.php b/src/cli/MarkdownFormatter.php
@@ -32,20 +32,58 @@ public function format(string $markdown): string
     {
         $text = $markdown;
 
-        // Remove markdown code blocks but keep content
-        $text = preg_replace_callback('/```(\w+)?\n(.*?)```/s', function (array $matches): string {
+        // First, protect code blocks (```code```) by replacing them with placeholders
+        $codeBlocks = [];
+        $codeBlockIndex = 0;
+        $text = preg_replace_callback('/```(\w+)?\n(.*?)```/s', function (array $matches) use (&$codeBlocks, &$codeBlockIndex): string {
             // preg_replace_callback guarantees indices 1 and 2 exist for this pattern
             $code = $matches[2];
             $lang = $matches[1] !== '' ? $matches[1] : '';
-            return $this->formatCodeBlock($code, $lang);
+            $placeholder = "\x00CODEBLOCK" . $codeBlockIndex . "\x00";
+            $codeBlocks[$placeholder] = $this->formatCodeBlock($code, $lang);
+            $codeBlockIndex++;
+            return $placeholder;
         }, $text);
         if ($text === null) {
             $text = $markdown;
         }
+        $text = (string)$text;
 
-        // Format inline code
-        $replaced = preg_replace('/`([^`]+)`/', $this->useColors ? "\033[36m\$1\033[0m" : '[$1]', $text);
-        $text = $replaced !== null ? $replaced : $text;
+        // Protect inline code blocks (`code`) by replacing them with placeholders
+        // This protects content from markdown formatting (like asterisks and underscores)
+        $inlineCode = [];
+        $inlineCodeIndex = 0;
+        
+        // First, handle double backticks (``code``) - these can contain single backticks
+        // Pattern: `` followed by content (which can include `) followed by ``
+        // Match the shortest possible sequence to avoid greedy matching
+        $text = preg_replace_callback('/``(.*?)``/s', function (array $matches) use (&$inlineCode, &$inlineCodeIndex): string {
+            // preg_replace_callback guarantees index 1 exists
+            $code = $matches[1];
+            $placeholder = "\x00INLINECODE" . $inlineCodeIndex . "\x00";
+            $inlineCode[$placeholder] = $this->useColors ? "\033[36m{$code}\033[0m" : "[{$code}]";
+            $inlineCodeIndex++;
+            return $placeholder;
+        }, $text);
+        if ($text === null) {
+            $text = '';
+        }
+        $text = (string)$text;
+        
+        // Then handle single backticks (`code`) - these cannot contain backticks
+        // Use non-greedy matching to handle cases correctly
+        $text = preg_replace_callback('/`([^`\n]+?)`/', function (array $matches) use (&$inlineCode, &$inlineCodeIndex): string {
+            // preg_replace_callback guarantees index 1 exists
+            $code = $matches[1];
+            $placeholder = "\x00INLINECODE" . $inlineCodeIndex . "\x00";
+            $inlineCode[$placeholder] = $this->useColors ? "\033[36m{$code}\033[0m" : "[{$code}]";
+            $inlineCodeIndex++;
+            return $placeholder;
+        }, $text);
+        if ($text === null) {
+            $text = '';
+        }
+        $text = (string)$text;
 
         // Format headers
         $text = (string)preg_replace('/^### (.*)$/m', $this->formatHeader(3, '$1'), $text);
@@ -56,23 +94,38 @@ public function format(string $markdown): string
         $text = (string)preg_replace('/\*\*(.+?)\*\*/', $this->useColors ? "\033[1m\$1\033[0m" : '**$1**', $text);
         $text = (string)preg_replace('/__(.+?)__/', $this->useColors ? "\033[1m\$1\033[0m" : '__$1__', $text);
 
-        // Format italic text
-        $text = (string)preg_replace('/\*(.+?)\*/', $this->useColors ? "\033[3m\$1\033[0m" : '*$1*', $text);
-        $text = (string)preg_replace('/_(.+?)_/', $this->useColors ? "\033[3m\$1\033[0m" : '_$1_', $text);
-
-        // Format unordered lists
-        $text = preg_replace_callback('/^(\s*)[-*+] (.+)$/m', function (array $matches): string {
+        // Format unordered lists BEFORE italic formatting to avoid conflicts
+        // This ensures list markers are processed before italic markers
+        $replaced = preg_replace_callback('/^(\s*)[-*+] (.+)$/m', function (array $matches): string {
             // preg_replace_callback guarantees these indices exist
             $indent = $matches[1];
             $content = $matches[2];
             $bullet = $this->useColors ? "\033[0;33m•\033[0m" : '•';
             return $indent . $bullet . ' ' . $content;
         }, $text);
-        if ($text === null) {
-            $text = '';
-        }
+        $text = $replaced !== null ? $replaced : $text;
         $text = (string)$text;
 
+        // Format italic text AFTER lists to avoid interfering with list markers
+        // Only format if underscores are around words (not inside words like users_created_at)
+        // Pattern: _word_ but not word_word or _word_word
+        // Exclude asterisks that are list markers (at start of line with optional whitespace)
+        $replaced = preg_replace('/(?<!^|\n)(?<!\S)\*(.+?)\*(?!\S)/', $this->useColors ? "\033[3m\$1\033[0m" : '*$1*', $text);
+        $text = $replaced !== null ? $replaced : $text;
+        // Match _text_ only if not preceded/followed by word characters (letters, digits, underscores)
+        $replaced = preg_replace('/(?<![a-zA-Z0-9_])_(.+?)_(?![a-zA-Z0-9_])/', $this->useColors ? "\033[3m\$1\033[0m" : '_$1_', $text);
+        $text = $replaced !== null ? $replaced : $text;
+
+        // Restore inline code blocks
+        foreach ($inlineCode as $placeholder => $formatted) {
+            $text = str_replace($placeholder, $formatted, $text);
+        }
+
+        // Restore code blocks
+        foreach ($codeBlocks as $placeholder => $formatted) {
+            $text = str_replace($placeholder, $formatted, $text);
+        }
+
         // Format ordered lists
         $text = preg_replace_callback('/^(\s*)(\d+)\. (.+)$/m', function (array $matches): string {
             // preg_replace_callback guarantees these indices exist
@@ -363,26 +416,34 @@ protected function formatCodeBlock(string $code, string $lang = ''): string
             $maxLength = max($maxLength, mb_strlen($line));
         }
 
-        $border = str_repeat('─', min($maxLength + 2, 80));
         $langLabel = $lang !== '' ? " {$lang}" : '';
+        // Calculate border width: max of code width + 2 (for padding) and lang label width + 2
+        // But ensure it doesn't exceed 80 characters
+        $langWidth = $lang !== '' ? mb_strlen($langLabel) : 0;
+        $borderWidth = min(max($maxLength + 2, $langWidth + 2), 80);
+        $border = str_repeat('─', $borderWidth);
 
         $result = [];
         if ($this->useColors) {
             $result[] = "\033[0;90m┌{$border}┐\033[0m";
             if ($lang !== '') {
-                $result[] = "\033[0;90m│\033[0m\033[0;36m{$langLabel}\033[0m" . str_repeat(' ', max(0, $maxLength - mb_strlen($lang) + 1)) . "\033[0;90m│\033[0m";
+                $langPadding = max(0, $borderWidth - mb_strlen($langLabel) - 2);
+                $result[] = "\033[0;90m│\033[0m\033[0;36m{$langLabel}\033[0m" . str_repeat(' ', $langPadding) . "\033[0;90m│\033[0m";
                 $result[] = "\033[0;90m├{$border}┤\033[0m";
             }
         } else {
             $result[] = "┌{$border}┐";
             if ($lang !== '') {
-                $result[] = "│{$langLabel}" . str_repeat(' ', max(0, $maxLength - mb_strlen($lang) + 1)) . '│';
+                $langPadding = max(0, $borderWidth - mb_strlen($langLabel) - 2);
+                $result[] = "│{$langLabel}" . str_repeat(' ', $langPadding) . '│';
                 $result[] = "├{$border}┤";
             }
         }
 
         foreach ($lines as $line) {
-            $padded = $line . str_repeat(' ', max(0, $maxLength - mb_strlen($line)));
+            // Pad line to border width - 2 (for left and right padding spaces)
+            $linePadding = max(0, $borderWidth - mb_strlen($line) - 2);
+            $padded = $line . str_repeat(' ', $linePadding);
             if ($this->useColors) {
                 $result[] = "\033[0;90m│\033[0m \033[36m{$padded}\033[0m \033[0;90m│\033[0m";
             } else {
