ferrero-opentext/src/FilenameParser.php

<?php

/**
 * FilenameParser - V2 Naming Convention Parser
 *
 * Parses filenames according to Ferrero V2 naming convention:
 * [OMG_JOB_NUMBER]_[BRAND_CODE]_[COUNTRY_CODE]_[LANGUAGE_CODE]_[SUBJECT_TITLE]_[ASSET_TYPE]_[SPOT_VERSION]_[SECONDS]S_[ASPECT_RATIO]_[TRACKING_ID]
 *
 * Example: 1234567_RAF_CH_de_TEST_FILE_OLV_001_15S_16x9_a7K9mP.mp4
 *
 * On upload to DAM, the OMG Job Number and Tracking ID are stripped:
 * Final: RAF_CH_de_TEST_FILE_OLV_001_15S_16x9.mp4
 */
class FilenameParser
{
    private $validationErrors = [];
    private $warnings = [];

    /**
     * Parse a filename according to V2 naming convention
     *
     * @param string $filename The filename to parse (with or without extension)
     * @return array Parsed components or null if invalid
     */
    public function parseFilename($filename)
    {
        $this->validationErrors = [];
        $this->warnings = [];

        // Remove extension
        $pathInfo = pathinfo($filename);
        $filenameWithoutExt = $pathInfo['filename'];
        $extension = isset($pathInfo['extension']) ? '.' . $pathInfo['extension'] : '';

        // Split by underscore
        $parts = explode('_', $filenameWithoutExt);

        // V2 naming convention has minimum 10 parts (with tracking ID)
        // Without tracking ID: 9 parts minimum
        // Allow fewer parts for better error messages, but still validate
        if (count($parts) < 8) {
            $this->validationErrors[] = "Invalid filename structure. Expected minimum 9 parts, got " . count($parts);
            // Continue parsing to provide detailed error messages
        }

        // Parse components
        $parsed = [
            'original_filename' => $filename,
            'filename_without_ext' => $filenameWithoutExt,
            'extension' => $extension,
            'omg_job_number' => null,
            'brand_code' => null,
            'country_code' => null,
            'language_code' => null,
            'subject_title' => null,
            'asset_type' => null,
            'spot_version' => null,
            'has_master' => false,
            'seconds' => null,
            'aspect_ratio' => null,
            'tracking_id' => null,
            'validation_errors' => [],
            'warnings' => []
        ];

        $index = 0;

        // 1. OMG Job Number (must be all digits, max 10 digits)
        if (isset($parts[$index]) && ctype_digit($parts[$index])) {
            $omgJobNumber = $parts[$index];
            if (strlen($omgJobNumber) > 10) {
                $this->validationErrors[] = "OMG Job Number too long: $omgJobNumber (max 10 digits)";
            } else {
                $parsed['omg_job_number'] = $omgJobNumber;
            }
            $index++;
        } else {
            // Try to proceed without OMG Job Number (allow parsing but mark as invalid)
            if (isset($parts[$index]) && !ctype_digit($parts[$index])) {
                $this->validationErrors[] = "OMG Job Number missing or invalid (must be numbers only). Found: {$parts[$index]}";
                // Don't increment index - treat current part as brand code
            } else {
                $this->validationErrors[] = "OMG Job Number missing";
            }
        }

        // 2. Brand Code (2-5 characters, uppercase)
        if (isset($parts[$index])) {
            $brandCode = $parts[$index];
            if (strlen($brandCode) >= 2 && strlen($brandCode) <= 5) {
                $parsed['brand_code'] = strtoupper($brandCode);
            } else {
                $this->validationErrors[] = "Brand Code invalid: $brandCode (must be 2-5 characters)";
            }
            $index++;
        } else {
            $this->validationErrors[] = "Brand Code missing";
        }

        // 3. Country Code (2 characters, uppercase)
        if (isset($parts[$index])) {
            $countryCode = $parts[$index];
            if (strlen($countryCode) === 2) {
                $parsed['country_code'] = strtoupper($countryCode);
            } else {
                $this->validationErrors[] = "Country Code invalid: $countryCode (must be 2 characters)";
            }
            $index++;
        } else {
            $this->validationErrors[] = "Country Code missing";
        }

        // 4. Language Code (2-3 characters, lowercase)
        if (isset($parts[$index])) {
            $languageCode = $parts[$index];
            if (strlen($languageCode) >= 2 && strlen($languageCode) <= 3) {
                $parsed['language_code'] = strtolower($languageCode);
            } else {
                $this->validationErrors[] = "Language Code invalid: $languageCode (must be 2-3 characters)";
            }
            $index++;
        } else {
            $this->validationErrors[] = "Language Code missing";
        }

        // 5. Subject Title (can be multiple parts until we hit a 3-char asset type)
        // Asset type is always 3 characters, so we need to find it
        $subjectTitleParts = [];
        $foundAssetType = false;
        $assetTypeIndex = $index;

        // Look ahead to find asset type (3 characters uppercase)
        for ($i = $index; $i < count($parts); $i++) {
            if (strlen($parts[$i]) === 3 && ctype_alpha($parts[$i]) && ctype_upper($parts[$i])) {
                // Potential asset type - check if it's followed by spot version pattern
                if (isset($parts[$i + 1]) && (strlen($parts[$i + 1]) === 3 || $parts[$i + 1] === 'MST')) {
                    // This is likely the asset type
                    $assetTypeIndex = $i;
                    $foundAssetType = true;
                    break;
                }
            }
        }

        if ($foundAssetType) {
            // Everything between language code and asset type is subject title
            for ($i = $index; $i < $assetTypeIndex; $i++) {
                $subjectTitleParts[] = $parts[$i];
            }
            $parsed['subject_title'] = implode('_', $subjectTitleParts);

            if (strlen($parsed['subject_title']) > 15) {
                $this->warnings[] = "Subject Title exceeds 15 characters: {$parsed['subject_title']}";
            }

            // Move index to asset type
            $index = $assetTypeIndex;
        } else {
            $this->validationErrors[] = "Could not locate Asset Type (must be 3 uppercase letters)";
            // Assume next part is subject title
            if (isset($parts[$index])) {
                $parsed['subject_title'] = $parts[$index];
                $index++;
            }
        }

        // 6. Asset Type (3 characters, uppercase)
        if (isset($parts[$index]) && strlen($parts[$index]) === 3) {
            $parsed['asset_type'] = strtoupper($parts[$index]);
            $index++;
        } else {
            $this->validationErrors[] = "Asset Type missing or invalid (must be 3 characters)";
        }

        // 7. Spot Version (3 characters or 'MST' for master)
        if (isset($parts[$index])) {
            $spotVersion = $parts[$index];
            if ($spotVersion === 'MST' || strtoupper($spotVersion) === 'MST') {
                $parsed['has_master'] = true;
                $parsed['spot_version'] = 'MST';
            } else if (strlen($spotVersion) === 3) {
                $parsed['spot_version'] = strtoupper($spotVersion);
                // Check if it contains MST
                if (strpos(strtoupper($spotVersion), 'MST') !== false) {
                    $parsed['has_master'] = true;
                }
            } else {
                $this->validationErrors[] = "Spot Version invalid: $spotVersion (must be 3 characters)";
                $parsed['spot_version'] = $spotVersion;
            }
            $index++;
        } else {
            $this->validationErrors[] = "Spot Version missing";
        }

        // 8. Duration (Seconds) - format: 15S or 6S
        if (isset($parts[$index])) {
            $durationPart = $parts[$index];
            if (preg_match('/^(\d+)S$/i', $durationPart, $matches)) {
                $parsed['seconds'] = $matches[1];
            } else {
                $this->validationErrors[] = "Duration invalid: $durationPart (must be format: 15S)";
            }
            $index++;
        } else {
            $this->validationErrors[] = "Duration missing";
        }

        // 9. Aspect Ratio (3-4 characters) - format: 16x9, 4x3, 1x1
        if (isset($parts[$index])) {
            $aspectRatio = $parts[$index];
            if (preg_match('/^\d+x\d+$/i', $aspectRatio)) {
                $parsed['aspect_ratio'] = $aspectRatio;
            } else {
                $this->validationErrors[] = "Aspect Ratio invalid: $aspectRatio (must be format: 16x9)";
                $parsed['aspect_ratio'] = $aspectRatio;
            }
            $index++;
        } else {
            $this->validationErrors[] = "Aspect Ratio missing";
        }

        // 10. Tracking ID (optional, 6 alphanumeric characters)
        if (isset($parts[$index])) {
            $trackingId = $parts[$index];
            if (strlen($trackingId) === 6 && ctype_alnum($trackingId)) {
                $parsed['tracking_id'] = $trackingId;
            } else {
                $this->warnings[] = "Tracking ID format invalid: $trackingId (should be 6 alphanumeric characters)";
                $parsed['tracking_id'] = $trackingId;
            }
            $index++;
        }

        // Check for extra parts
        if ($index < count($parts)) {
            $extraParts = array_slice($parts, $index);
            $this->warnings[] = "Extra parts in filename: " . implode('_', $extraParts);
        }

        // Add validation results
        $parsed['validation_errors'] = $this->validationErrors;
        $parsed['warnings'] = $this->warnings;
        $parsed['is_valid'] = empty($this->validationErrors);

        return $parsed;
    }

    /**
     * Validate filename structure strictly
     *
     * @param string $filename
     * @return bool
     */
    public function validateStructure($filename)
    {
        $parsed = $this->parseFilename($filename);
        return $parsed && $parsed['is_valid'];
    }

    /**
     * Strip upload components (OMG Job Number and Tracking ID)
     *
     * @param string $filename
     * @return string|null Clean filename for upload, or null if parsing failed
     */
    public function stripUploadComponents($filename)
    {
        $parsed = $this->parseFilename($filename);

        if (!$parsed) {
            return null;
        }

        // Build clean filename without OMG Job Number and Tracking ID
        $cleanParts = [];

        if ($parsed['brand_code']) $cleanParts[] = $parsed['brand_code'];
        if ($parsed['country_code']) $cleanParts[] = $parsed['country_code'];
        if ($parsed['language_code']) $cleanParts[] = $parsed['language_code'];
        if ($parsed['subject_title']) $cleanParts[] = $parsed['subject_title'];
        if ($parsed['asset_type']) $cleanParts[] = $parsed['asset_type'];
        if ($parsed['spot_version']) $cleanParts[] = $parsed['spot_version'];
        if ($parsed['seconds']) $cleanParts[] = $parsed['seconds'] . 'S';
        if ($parsed['aspect_ratio']) $cleanParts[] = $parsed['aspect_ratio'];

        $cleanFilename = implode('_', $cleanParts);

        // Add extension back
        if ($parsed['extension']) {
            $cleanFilename .= $parsed['extension'];
        }

        return $cleanFilename;
    }

    /**
     * Get clean filename for upload (same as stripUploadComponents)
     *
     * @param string $filename
     * @return string|null
     */
    public function getCleanFilename($filename)
    {
        return $this->stripUploadComponents($filename);
    }

    /**
     * Extract tracking ID from filename
     *
     * @param string $filename
     * @return string|null Tracking ID or null if not found
     */
    public function extractTrackingId($filename)
    {
        $parsed = $this->parseFilename($filename);
        return $parsed ? $parsed['tracking_id'] : null;
    }

    /**
     * Get validation errors
     *
     * @return array
     */
    public function getValidationErrors()
    {
        return $this->validationErrors;
    }

    /**
     * Get warnings
     *
     * @return array
     */
    public function getWarnings()
    {
        return $this->warnings;
    }

    /**
     * Format parsed data for display
     *
     * @param array $parsed
     * @return string
     */
    public function formatForDisplay($parsed)
    {
        if (!$parsed) {
            return "Invalid filename";
        }

        $output = "Filename: {$parsed['original_filename']}\n";
        $output .= "Valid: " . ($parsed['is_valid'] ? 'YES' : 'NO') . "\n\n";

        $output .= "Components:\n";
        $output .= "  OMG Job Number: " . ($parsed['omg_job_number'] ?? 'N/A') . "\n";
        $output .= "  Brand Code: " . ($parsed['brand_code'] ?? 'N/A') . "\n";
        $output .= "  Country Code: " . ($parsed['country_code'] ?? 'N/A') . "\n";
        $output .= "  Language Code: " . ($parsed['language_code'] ?? 'N/A') . "\n";
        $output .= "  Subject Title: " . ($parsed['subject_title'] ?? 'N/A') . "\n";
        $output .= "  Asset Type: " . ($parsed['asset_type'] ?? 'N/A') . "\n";
        $output .= "  Spot Version: " . ($parsed['spot_version'] ?? 'N/A') . "\n";
        $output .= "  Has Master: " . ($parsed['has_master'] ? 'YES' : 'NO') . "\n";
        $output .= "  Duration: " . ($parsed['seconds'] ?? 'N/A') . " seconds\n";
        $output .= "  Aspect Ratio: " . ($parsed['aspect_ratio'] ?? 'N/A') . "\n";
        $output .= "  Tracking ID: " . ($parsed['tracking_id'] ?? 'N/A') . "\n";

        if (!empty($parsed['validation_errors'])) {
            $output .= "\nValidation Errors:\n";
            foreach ($parsed['validation_errors'] as $error) {
                $output .= "  - $error\n";
            }
        }

        if (!empty($parsed['warnings'])) {
            $output .= "\nWarnings:\n";
            foreach ($parsed['warnings'] as $warning) {
                $output .= "  - $warning\n";
            }
        }

        return $output;
    }
}