taskman_config/cache.inc.php

<?php
namespace taskman;
use Exception;

/**
 * @property object $config
 * @property string $payload
 */
class ConfigCacheEntry
{
  const FMT_BINARY     = 0;
  const FMT_LZ4        = 1;
  const FMT_FILE_REF   = 2;

  public string $class;
  public int $class_id;
  public int $id;
  public string $strid;
  public string $cache_file;
  //NOTE: actual payload is stored in a separate file for faster incremental retrievals
  public string $payload_file;
  public string $file;
  public array $includes = array();
  public array $refs = array();
  public ?ConfigCacheEntryExtras $extras;

  public $_config;
  public $_payload;

  static function serialize(ConfigCacheEntry $ce) : string
  {
    $d = $ce->export();
    return serialize($d);
  }

  static function unserialize(string $str) : ?ConfigCacheEntry
  {
    $d = @unserialize($str);
    if(!is_array($d))
      return null;
    $ce = new ConfigCacheEntry();
    $ce->import($d);
    return $ce;
  }

  function __construct()
  { 
    $this->extras = ConfigCacheEntryExtras::create();
  }

  function __get(string $name)
  {
    if($name === "config")
    {
      if($this->_config === null)
      {
        $klass = $this->class;
        $data = config_msgpack_unpack($this->payload);
        $this->_config = new $klass();
        $this->_config->import($data);
      }
      return $this->_config;
    }
    else if($name === "payload")
    {
      if($this->_payload !== null)
        return $this->_payload;
      //if payload not set directly not storing it
      return ensure_read($this->payload_file);
    }
    else
      throw new Exception("No such property '$name'");
  }

  function __set(string $name, $v)
  {
    if($name === "config")
      $this->_config = $v;
    else if($name === "payload")
      $this->_payload = $v;
    else
      throw new Exception("No such property '$name'");
  }

  function export() : array
  {
    $d = array();
    $d[] = $this->class;
    $d[] = $this->class_id;
    $d[] = $this->id;
    $d[] = $this->strid;
    $d[] = $this->cache_file;
    $d[] = $this->payload_file;
    $d[] = $this->file;
    $d[] = $this->includes;
    $d[] = $this->refs;
    $d[] = $this->extras->export();
    return $d;
  }

  function import(array $d)
  {
    $extras = array();

    list(
      $this->class,
      $this->class_id,
      $this->id,
      $this->strid,
      $this->cache_file,
      $this->payload_file,
      $this->file,
      $this->includes,
      $this->refs,
      $extras,
    ) = $d;

    $this->extras->import($extras);
  }
}

class ConfigCacheEntryExtras extends \stdClass
{
  private static string $klass = ConfigCacheEntryExtras::class;

  static function init(string $project_specific_klass)
  {
    self::$klass = $project_specific_klass;
  }

  static function create() : object
  {
    return new self::$klass();
  }

  function export() : array
  {
    $as_array = get_object_vars($this);
    return $as_array;
  }

  function import(array $as_array)
  {
    foreach($as_array as $field_name => $field_value)
      $this->$field_name = $field_value;
  }
}

class ConfigFetchParams
{
  public ConfigGlobals $globals;
  public ConfigScanResult $scanned;
  public bool $force_stale = false;
  public bool $verbose = false;
  public ?int $max_workers = null;
  public bool $touch_files_with_includes = true;
  public bool $check_junk = true;

  function __construct(ConfigGlobals $globals, ConfigScanResult $scanned, 
    bool $force_stale = false, bool $verbose = false, ?int $max_workers = null)
  {
    $this->globals = $globals;
    $this->scanned = $scanned;
    $this->force_stale = $force_stale;
    $this->verbose = $verbose;
    $this->max_workers = $max_workers;
  }

  function calcMaxWorkers() : int
  {
    $max_workers = $this->max_workers;
    if($max_workers === null)
      $max_workers = $this->scanned->count() < 100 ? 1 : 5;
    return $max_workers;
  }

  function splitFilesByChunks(int $max_workers, bool $sort = true) : array
  {
    $flat = $this->scanned->getFlatArray();
    if($sort)
      usort($flat, fn($a, $b) => $a[1] <=> $b[1]);

    $chunk_size = (int)ceil(count($flat)/$max_workers);
    return array_chunk($flat, $chunk_size);
  }

  //returns [[idx, time, [[[base_dir, file1], [base_dir, file2], ..]]], ]
  function splitJobs(bool $sort = true) : array
  {
    $max_workers = $this->calcMaxWorkers();
    $jobs = array();
    $chunks = $this->splitFilesByChunks($max_workers, $sort); 
    foreach($chunks as $idx => $chunk)
      $jobs[] = array($idx, microtime(true), $chunk);
    return $jobs;
  }
}

class ConfigFetchResult
{
  /** @var ConfigCacheEntry[] */
  public array $all = array();
  /** @var array<int, ConfigCacheEntry> */
  public array $by_id = array();
  /** @var array<string, ConfigCacheEntry> */
  public array $by_path = array();
  /** @var array<string, ConfigCacheEntry> */
  public array $by_alias = array();
  /** @var array<string, string> */
  public array $errors = array();
  public int $stales = 0;
  public int $corruptions = 0;
  public int $fast_jsons = 0;

  function getByAlias(string $strid) : ConfigCacheEntry
  {
    if(array_key_exists($strid, $this->by_alias))
      return $this->by_alias[$strid];
    throw new Exception("Failed to find config by alias '$strid'!");
  }
  
  function getById(int $id) : ConfigCacheEntry
  {
    if(array_key_exists($id, $this->by_id))
      return $this->by_id[$id];
    throw new Exception("Failed to find config by id '$id'!");
  }
  
  function getByPath(string $path) : ConfigCacheEntry
  {
    if(array_key_exists($path, $this->by_path))
      return $this->by_path[$path];
    throw new Exception("Failed to find config by path '$path'!");
  }
}

function config_cache_fetch(ConfigFetchParams $params) : ConfigFetchResult
{
  $t = microtime(true);

  $result = _config_cache_fetch($params);

  if($result->errors)
  {
    $errors = array();
    foreach($result->errors as $file => $error)
    {
      $errors[] = (count($errors) + 1) . ") Error in file '$file': $error";
      if(count($errors) > 15)
        break;
    }

    throw new Exception(implode("\n", $errors));
  }

  if($params->verbose)
    config_log("Fetch from cache: " . round(microtime(true) - $t,2) . " sec.");

  return $result;
}

function _config_cache_fetch(ConfigFetchParams $params) : ConfigFetchResult
{
  if($params->scanned->isEmpty())
    return new ConfigFetchResult();

  $jobs = $params->splitJobs(sort: true);

  $serial = sizeof($jobs) == 1;

  $results_by_job = _config_worker_run_procs($params, $jobs, $serial);

  if(!$serial)
  {
    //NOTE: in case result unserialize error try serial processing
    if(array_search(false, $results_by_job, true/*strict*/) !== false)
    {
      if($params->verbose)
        config_log("Corrupted result, trying serial processing...");
      $results_by_job = _config_worker_run_procs($params, $jobs, true);
    }
  }

  $t = microtime(true);
  $result = _config_merge_fetch_results($params, $results_by_job);
  if($params->verbose)
    config_log("Merge results: " . round(microtime(true) - $t,2) . " sec.");

  if($params->verbose)
    config_log("Miss(Fast)/Total: {$result->stales}($result->fast_jsons)/" . sizeof($result->all));

  return $result;
}

function config_cache_fetch_by_path(ConfigGlobals $globals, string $path, bool $force_stale = false) : ConfigCacheEntry
{
  $path = realpath($path);
  $base_dir = config_map_base_dir($globals->base_dirs, $path);

  $scanned = new ConfigScanResult();
  $scanned->add($base_dir, $path);

  $ces = config_cache_fetch(new ConfigFetchParams(
    globals: $globals, 
    scanned: $scanned,
    force_stale: $force_stale
  ));

  if(empty($ces->all))
    throw new Exception("Config not found at path '$path'");
  return $ces->all[0];
}

function _config_merge_fetch_results(ConfigFetchParams $params, array $results_by_job) : ConfigFetchResult
{
  $result = new ConfigFetchResult();
  $total_stales = 0;
  $total_fast_jsons = 0;
  
  foreach($results_by_job as $results)
  {
    foreach($results as $item) 
    {
      list($base_dir, $file, $cache_file, $is_stale, $parser_type, $error) = $item;

      if($error !== null)
      {
        $result->errors[$file] = $error;
        continue;
      }

      if($parser_type === 1)
        ++$total_fast_jsons;

      $cache_entry = ConfigCacheEntry::unserialize(ensure_read($cache_file));

      //NOTE: handling any cache corruption errors
      if($cache_entry === null)
      {
        $is_stale = true;
        $cache_entry = _config_invalidate_cache($params, $base_dir, $file, $cache_file);
        ++$result->corruptions;
      }

      $includes = $cache_entry->includes;

      if(!$is_stale && count($includes) > 0 && need_to_regen($file, $includes))
      {
        $is_stale = true;
        $cache_entry = _config_invalidate_cache($params, $base_dir, $file, $cache_file);
        //NOTE: let's change the mtime of the file which include other files,
        //      so that on tne next build it will be 'older' than its includes
        //      and won't trigger rebuild
        if($params->touch_files_with_includes)
          touch($file);
      }

      if($is_stale)
        ++$total_stales;

      //we want results to be returned in the same order, so 
      //we store entries by the file key and later retrieve array values
      $result->by_path[$file] = $cache_entry;
      $result->by_id[$cache_entry->id] = $cache_entry;
      $result->by_alias[$cache_entry->strid] = $cache_entry;
    }
  }

  $result->all = array_values($result->by_path);
  $result->stales = $total_stales;
  $result->fast_jsons = $total_fast_jsons;
  return $result;
}

function _config_invalidate_cache(
  ConfigFetchParams $params, string $base_dir, string $file, 
  string $cache_file, ?int &$parser_type = null) : ConfigCacheEntry
{
  list($conf_id, $conf_strid) = config_ensure_header($base_dir, $file);
  if(!$conf_id)
    throw new Exception("Bad conf id: {$conf_id}");

  //TODO: this is a bit ugly but kinda works for now
  $GLOBALS['CONFIG_CURRENT_FILE'] = $file; 
  $GLOBALS['CONFIG_CURRENT_PROTO_ID'] = $conf_id; 
  $GLOBALS['CONFIG_EXTRAS'] = ConfigCacheEntryExtras::create();

  $pres = config_parse($params->globals->base_dirs, $file);
  if($pres->error !== 0)
    throw new Exception("Parse error({$pres->error}):\n" . $pres->error_descr);

  $parser_type = $pres->parser_type;
  $parsed_arr = $pres->parsed_arr;
  //TODO: these hardcoded keys below probably shouldn't be part of data?
  $parsed_arr['id'] = $conf_id;
  $parsed_arr['strid'] = $conf_strid;
  list($class, $class_id, $normalized_data) 
    = config_apply_class_normalization($parsed_arr, $params->check_junk);

  $cache_payload_file = config_get_cache_payload_path($params->globals, $file);

  $cache_entry = new ConfigCacheEntry();
  $cache_entry->id = $conf_id;
  $cache_entry->strid = $conf_strid;
  $cache_entry->cache_file = $cache_file;
  $cache_entry->class = $class;
  $cache_entry->class_id = $class_id;
  $cache_entry->payload_file = $cache_payload_file;
  //TODO: pass flag if file is actually normalized?
  $cache_entry->file = normalize_path($file);
  $cache_entry->includes = config_get_module_includes($pres->jsm_module);
  $cache_entry->extras = $GLOBALS['CONFIG_EXTRAS'];
  //TODO: do we really need these refs?
  $cache_entry->refs = config_content_match_refs($pres->normalized_jzon);

  ensure_write($cache_file, ConfigCacheEntry::serialize($cache_entry));
  ensure_write($cache_payload_file, config_msgpack_pack($normalized_data));

  return $cache_entry;
}