data_goes 0.1.0-alpha.3.52

Biblioteca experimental para demonstração.
use std::{collections::HashMap};
use mysql_async::prelude::Queryable;
use tokio::sync::Mutex;
pub use crate::config::ConfigClient;
use mssql_client::{Ready};
use anyhow::{Ok};
use polars::{prelude::*};
use std::fs;
use std::sync::Arc;
pub enum DbClient {
    SqlServer(mssql_client::Client<Ready>),
    MySql(mysql_async::Conn),
    ClickHouse{client:clickhouse::Client, url: String, user: String, pass: String},
}
pub struct ClientSql{   
    pub(crate) clients: HashMap<String, Mutex<DbClient>>,   
}

pub struct TempParquetFile(String);
impl Drop for TempParquetFile {
    fn drop(&mut self) {
        let _=std::fs::remove_file(&self.0);
    }
}
#[derive(Clone)]
pub struct TabGoes{
    pub(crate) lazy: polars::prelude::LazyFrame,
    pub(crate) _files: Vec<Arc<TempParquetFile>>
}

impl ClientSql {
    pub async fn new() -> anyhow::Result<Self>{
        let toml_str = std::fs::read_to_string("database.toml").expect("Arquivo database.toml");
        let toml_str = toml_str.replace("\r\n", "\n");
        let app_config: crate::config::AppConfig = toml::from_str(&toml_str)?;
        let mut clients = std::collections::HashMap::new();
        for server_cfg in app_config.servers{
            let server_key = server_cfg.alias.clone().unwrap_or_else(|| server_cfg.host.clone());
            let conn_mssql = format!("Server={};User Id={};Password={};TrustServerCertificate=true", server_cfg.host, server_cfg.user, server_cfg.password);
            if let std::result::Result::Ok(config) = mssql_client::Config::from_connection_string(&conn_mssql){
                if let std::result::Result::Ok(client) = mssql_client::Client::connect(config).await{
                    clients.insert(server_key.clone(), Mutex::new(DbClient::SqlServer(client)));
                    continue;
                }
            }
            let conn_mysql = format!("mysql://{}:{}@{}:3306", server_cfg.user, server_cfg.password, server_cfg.host);
            if let std::result::Result::Ok(opts)= mysql_async::Opts::from_url(&conn_mysql){
                if let std::result::Result::Ok(client)= mysql_async::Conn::new(opts).await{
                    clients.insert(server_key.clone(), Mutex::new(DbClient::MySql(client)));
                    continue;
                }
            }
            let ch_url = format!("http://{}:8123", server_cfg.host);
            let client_ch = clickhouse::Client::default()
                .with_url(&ch_url)
                .with_user(server_cfg.user.to_string())
                .with_password(server_cfg.password.to_string());
            if let std::result::Result::Ok(_) = client_ch.query("SELECT 1").fetch_one::<u8>().await{
                clients.insert(server_key.clone(), Mutex::new(DbClient::ClickHouse{
                    client: client_ch, 
                    url: ch_url, 
                    user: server_cfg.user.clone(), 
                    pass: server_cfg.password.clone()
                }));
                continue;
            }
        }
        Ok(Self{clients})
    }

    pub async fn query_to_parquet(&self,server_name: &str, sql_gerado: &str, expected_cols: &[&str], expected_types: &[&str])-> anyhow::Result<TabGoes>{
        
        let client_mutex =self.clients.get(server_name)
        .ok_or_else(||anyhow::anyhow!("Falha de conexão com o servidor: {}",server_name))?;
        
        let mut client_lock = client_mutex.lock().await;

        let file_path = format!("temp_query_{}.parquet", crate::uuid::Uuid::new_v4());

        let mut df = match &mut *client_lock{
            DbClient::SqlServer(client)=>{
                Self::fetch_mssql_to_df(client, sql_gerado, expected_cols, expected_types).await?
            },
            DbClient::MySql(client)=>{
                Self::fetch_mysql_to_df(client, sql_gerado, expected_cols, expected_types).await?
            },
            DbClient::ClickHouse{url,user,pass, ..}=>{ 
                let req_client = reqwest::Client::new();
                let sql_parquet = format!("{} FORMAT Parquet", sql_gerado);
                let resp = req_client.post(url.as_str())
                .basic_auth(user.as_str(), Some(pass.as_str()))
                .body(sql_parquet)
                .send()
                .await?;
            if !resp.status().is_success(){
                anyhow::bail!("Erro ClickHouse: {}", resp.text().await?);
            }
            std::fs::write(&file_path, resp.bytes().await?)?;
            let lf = polars::prelude::LazyFrame::scan_parquet(&file_path, polars::prelude::ScanArgsParquet::default())?;
            return Ok(TabGoes { 
                    lazy: lf, 
                    _files: vec![Arc::new(TempParquetFile(file_path))] 
                });
            }
        };  

        let mut file = std::fs::File::create(&file_path)?;
        polars::prelude::ParquetWriter::new(&mut file).finish(&mut df)?;
        drop(df);
        let lf = polars::prelude::LazyFrame::scan_parquet(&file_path, polars::prelude::ScanArgsParquet::default())?;
        Ok(TabGoes  { 
            lazy: lf,
            _files: vec![Arc::new(TempParquetFile(file_path))],
         })
    }

    async fn fetch_mssql_to_df(
        client: &mut mssql_client::Client<Ready>,
        sql: &str,
        expected_cols: &[&str],
        expected_types: &[&str],
    )->anyhow::Result<DataFrame>{
        enum ColBuilder {
            I16(Vec<Option<i16>>),
            I32(Vec<Option<i32>>),
            I64(Vec<Option<i64>>),
            F64(Vec<Option<f64>>),
            Bool(Vec<Option<bool>>),
            String(Vec<Option<String>>),
        }
        let mut builders:Vec<ColBuilder> = expected_types.iter().map(|&t|{
            match t{
                "Option<i16>" => ColBuilder::I16(Vec::new()),
                "Option<i32>" => ColBuilder::I32(Vec::new()),
                "Option<i64>" => ColBuilder::I64(Vec::new()),
                "Option<f64>" | "Option<f32>" | "Option<data_goes::rust_decimal::Decimal>" => ColBuilder::F64(Vec::new()),
                "Option<bool>" => ColBuilder::Bool(Vec::new()),
                _ => ColBuilder::String(Vec::new()),
            }
        }).collect();
        let rows = client.query(sql, &[]).await?;
        for result in rows{
            let row = result?;
            for (i, builder) in builders.iter_mut().enumerate(){
                match builder {
                    ColBuilder::I16(vec) => vec.push(row.get::<i16>(i).ok()),
                    ColBuilder::I32(vec) => vec.push(row.get::<i32>(i).ok()),
                    ColBuilder::I64(vec) => vec.push(row.get::<i64>(i).ok()),
                    ColBuilder::F64(vec) => vec.push(row.get::<f64>(i).ok()),
                    ColBuilder::Bool(vec) => vec.push(row.get::<bool>(i).ok()),
                    ColBuilder::String(vec) => vec.push(row.get::<String>(i).ok().map(|s| s.to_string())),
                }
            }
        }
        let mut columns = Vec::new();
        for (i, builder) in builders.into_iter().enumerate(){
            let col_name = expected_cols.get(i).unwrap_or(&"").split('.').last().unwrap_or("col");
            let s = match builder {
                ColBuilder::I16(vec) => Series::new(col_name.into(), vec),
                ColBuilder::I32(vec) => Series::new(col_name.into(), vec),
                ColBuilder::I64(vec) => Series::new(col_name.into(), vec),
                ColBuilder::F64(vec) => Series::new(col_name.into(), vec),
                ColBuilder::Bool(vec) => Series::new(col_name.into(), vec),
                ColBuilder::String(vec) => Series::new(col_name.into(), vec),
            };
            columns.push(s.into());
        }
        Ok(DataFrame::new(columns)?)
    }
    async fn fetch_mysql_to_df(
        client: &mut mysql_async::Conn,
        sql: &str,
        expected_cols: &[&str],
        expected_types: &[&str],
    )->anyhow::Result<DataFrame>{
        enum ColBuilder{
            I16(Vec<Option<i16>>),
            I32(Vec<Option<i32>>),
            I64(Vec<Option<i64>>),
            F64(Vec<Option<f64>>),
            Bool(Vec<Option<bool>>),
            String(Vec<Option<String>>),
        }
        let mut builders: Vec<ColBuilder>= expected_types.iter().map(|&t|{
        match t {
            "Option<i16>" => ColBuilder::I16(Vec::new()),
                "Option<i32>" => ColBuilder::I32(Vec::new()),
                "Option<i64>" => ColBuilder::I64(Vec::new()),
                "Option<f64>" | "Option<f32>" | "Option<data_goes::rust_decimal::Decimal>" => ColBuilder::F64(Vec::new()),
                "Option<bool>" => ColBuilder::Bool(Vec::new()),
                _ => ColBuilder::String(Vec::new()),
        }
    }).collect();
    let mut result  = client.query_iter(sql).await?;
    while let Some(row)= result.next().await? {
        for (i, builder) in builders.iter_mut().enumerate(){ 
        match builder{
            ColBuilder::I16(vec) => vec.push(row.get::<Option<i16>, _>(i).flatten()),
            ColBuilder::I32(vec) => vec.push(row.get::<Option<i32>, _>(i).flatten()),
            ColBuilder::I64(vec) => vec.push(row.get::<Option<i64>, _>(i).flatten()),
            ColBuilder::F64(vec) => vec.push(row.get::<Option<f64>, _>(i).flatten()),
            ColBuilder::Bool(vec) => vec.push(row.get::<Option<bool>, _>(i).flatten()),
            ColBuilder::String(vec) => vec.push(row.get::<Option<String>, _>(i).flatten()),
            }
        }
    }
    let mut columns = Vec::new();
    for (i,builder) in builders.into_iter().enumerate(){
        let col_name = expected_cols.get(i).unwrap_or(&"").split('.').last().unwrap_or("col");
        let s = match builder {
            ColBuilder::I16(vec) => Series::new(col_name.into(), vec),
            ColBuilder::I32(vec) => Series::new(col_name.into(), vec),
            ColBuilder::I64(vec) => Series::new(col_name.into(), vec),
            ColBuilder::F64(vec) => Series::new(col_name.into(), vec),
            ColBuilder::Bool(vec) => Series::new(col_name.into(), vec),
            ColBuilder::String(vec) => Series::new(col_name.into(), vec),
        };
        columns.push(s.into());
    }
    Ok(DataFrame::new(columns)?)
 }
}