替代使用dryRun区分BigQuery中的标准和传统SQL查询吗?

时间:2019-03-05 12:04:21

标签: google-api google-bigquery standard-sql legacy-sql

什么

有人知道使用BigQuery API更好地识别/区分传统或标准SQL编写的视图或查询的方法吗?

我想到的唯一方法是在将dryRun属性设置为true的情况下运行SQL(这将产生最小的处理开销),并且如果它失败并显示一条包含短语的消息, “尝试使用标准SQL”我可以假设它是旧版SQL,否则它可以工作并且是标准的。例如

{
...
"code": 400,
"message": 
    "Invalid table name: `my-project.my_dataset.hello_world`
    [Try using standard SQL (https://cloud.google.com/bigquery/docs/reference/standard-sql/enabling-standard-sql)]."
}

原因

我正在维护一个集成层(用Cloud Functions-Node.js 8编写),该集成层使用BigQuery API迁移数据。

不幸的是,用户社区被允许以标准或传统SQL编写视图和查询。由于处理旧版SQL的诸多限制,我想检测使用它编写的查询和视图,从而使我能够相应地调整处理方法。

1 个答案:

答案 0 :(得分:2)

  

有人知道使用BigQuery API更好地识别/区分传统或标准SQL编写的视图或查询的方法吗?

您可以尝试使用javascript正则表达式来识别SQL的类型。

您可以使用以下代码段作为基准

isStandardSql(idString) {
  let isStandard, fullId, partialId, projectId = '';
  // This 'if' checks if the provided idString is of type standard and makes sure there is only one ':' in the expression (as in legacy syntax)
  const splitted = idString.split(/[:.]/g);
  if (splitted.length > 3) {
    const __ret = this.try2findProjectId(idString, projectId);
    idString = __ret.idString;
    projectId = __ret.projectId;
  }
  if ((idString.match(/:/g))) {
    // Regex that checks if the format of the id match legacy
    let matched = idString.match(/([\[]([^[]|[\[][\]])*[\]])/g);
    if (matched && matched[0]) {
    fullId = projectId + matched[0].substring(1, idString.length - 1);
    isStandard = false;
    } else {
    this.errorMessage("First Regex", idString);
    }
    // Same as the first only that here instead of ':' we are looking for '.' and we want to make sure there is more than 1 (as in standard syntax)
  } else if ((idString.match(/\./g) && idString.match(/\./g).length === 2)) {
    // Regex that checks if the format of the id match standard
    let matched = idString.match(/(`([^`]|``)*`)/g);// ? idString.match(/(`([^`]|``)*`)/g) : [idString];
    if (matched && matched[0]) {
    fullId = projectId + matched[0].substring(1, idString.length - 1);
    isStandard = true
    } else if(!matched && idString){
    fullId = projectId + idString;
    isStandard = true;
    }
    else {
    this.errorMessage("Second Regex", idString);
    }
  }
  else {//projectID.dataset
    // In case of id without projectId of proxy "use" project.dataset
    if(splitted.length === 2) {
      fullId = '';
      if (idString[0] === '[' && idString[idString.length - 1] === ']') {
        isStandard = false;
      }
      else if (idString[0] === '`' && idString[idString.length - 1] === '`') {
        isStandard = true;
      }
      partialId = idString.replace(/`|\[|\]/g, '')
    }
    else {
      this.errorMessage("Third Regex", idString);
    }
  }
  // Return values is flag the determine the type (standard or legacy) and id without staring/ ending chars (``, [])
  return {
    isStandard,
    fullId: fullId,
    partialId: partialId
  };
  }


try2findProjectId(idString, projectId)
    {
        let numOfInstances = 0
        for (let i = idString.length; i > 0; i--) {
            const char = idString[i - 1]
            if (char === ':' || char === '.') {
                numOfInstances++
                if (numOfInstances === 2) {
                    projectId = idString.substring(1, i - 1)
                    idString = idString.substring(i - 1, idString.length)
                    idString = idString[idString.length - 1] === '`' ? '`' + idString : idString
                    idString = idString[idString.length - 1] === ']' ? '[' + idString : idString
                }
            }
        }
        return {idString, projectId}
    }