work off content-type header
This commit is contained in:
		
							
								
								
									
										33
									
								
								src/main.rs
									
									
									
									
									
								
							
							
						
						
									
										33
									
								
								src/main.rs
									
									
									
									
									
								
							@@ -19,7 +19,7 @@ use metrics::{counter, gauge};
 | 
				
			|||||||
use metrics_exporter_prometheus::PrometheusBuilder;
 | 
					use metrics_exporter_prometheus::PrometheusBuilder;
 | 
				
			||||||
use serde::Deserialize;
 | 
					use serde::Deserialize;
 | 
				
			||||||
use surrealdb::{engine::remote::ws::Client, Surreal};
 | 
					use surrealdb::{engine::remote::ws::Client, Surreal};
 | 
				
			||||||
use tokio::{io::AsyncWriteExt, task::JoinSet};
 | 
					use tokio::{io::{AsyncWriteExt, BufWriter}, task::JoinSet};
 | 
				
			||||||
use tracing::{debug, error, info, instrument, level_filters::LevelFilter, trace, trace_span, warn};
 | 
					use tracing::{debug, error, info, instrument, level_filters::LevelFilter, trace, trace_span, warn};
 | 
				
			||||||
use tracing_subscriber::{fmt, layer::SubscriberExt, EnvFilter, Layer, Registry};
 | 
					use tracing_subscriber::{fmt, layer::SubscriberExt, EnvFilter, Layer, Registry};
 | 
				
			||||||
 | 
					
 | 
				
			||||||
@@ -175,23 +175,40 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
 | 
				
			|||||||
    if let Ok(response) = request_builder.send().await {
 | 
					    if let Ok(response) = request_builder.send().await {
 | 
				
			||||||
        // Get body from response
 | 
					        // Get body from response
 | 
				
			||||||
        let headers = response.headers();
 | 
					        let headers = response.headers();
 | 
				
			||||||
        let ct = headers.get("Content-Type");
 | 
					 | 
				
			||||||
 | 
					
 | 
				
			||||||
        if let Some(ct) = ct {
 | 
					        #[allow(non_snake_case)]
 | 
				
			||||||
 | 
					        let CT = headers.get("Content-Type");
 | 
				
			||||||
 | 
					        let ct = headers.get("content-type");
 | 
				
			||||||
 | 
					
 | 
				
			||||||
 | 
					        if CT.is_none() && ct.is_none() {
 | 
				
			||||||
 | 
					        }
 | 
				
			||||||
 | 
					        let ct = match (CT,ct) {
 | 
				
			||||||
 | 
					            (None, None) => {
 | 
				
			||||||
 | 
					                warn!("Server did not respond with Content-Type header. Url: {} Headers: ({:?})", site.site.to_string(), headers);
 | 
				
			||||||
 | 
					                return
 | 
				
			||||||
 | 
					            },
 | 
				
			||||||
 | 
					            (None, Some(a)) => a,
 | 
				
			||||||
 | 
					            (Some(a), None) => a,
 | 
				
			||||||
 | 
					            (Some(a), Some(_)) => a,
 | 
				
			||||||
 | 
					        };
 | 
				
			||||||
 | 
					
 | 
				
			||||||
        let path = filesystem::as_path(&site.site, ct);
 | 
					        let path = filesystem::as_path(&site.site, ct);
 | 
				
			||||||
 | 
					
 | 
				
			||||||
        // make sure that the file is good to go
 | 
					        // make sure that the file is good to go
 | 
				
			||||||
            if let Some(mut file) = filesystem::init(&path).await {
 | 
					        if let Some(file) = filesystem::init(&path).await {
 | 
				
			||||||
            let should_parse = path.to_string_lossy().ends_with(".html");
 | 
					            let should_parse = path.to_string_lossy().ends_with(".html");
 | 
				
			||||||
            let mut buf: Vec<u8> = Vec::new();
 | 
					            let mut buf: Vec<u8> = Vec::new();
 | 
				
			||||||
 | 
					
 | 
				
			||||||
 | 
					            let mut writer = BufWriter::new(file);
 | 
				
			||||||
 | 
					
 | 
				
			||||||
            // stream the response onto the disk
 | 
					            // stream the response onto the disk
 | 
				
			||||||
            let mut stream = response.bytes_stream();
 | 
					            let mut stream = response.bytes_stream();
 | 
				
			||||||
 | 
					
 | 
				
			||||||
 | 
					            info!("Writing at: {:?}", path);
 | 
				
			||||||
            while let Some(data) = stream.next().await {
 | 
					            while let Some(data) = stream.next().await {
 | 
				
			||||||
                match data {
 | 
					                match data {
 | 
				
			||||||
                    Ok(data) => {
 | 
					                    Ok(data) => {
 | 
				
			||||||
                            debug!("Writing at: {:?}", path);
 | 
					                        let _ = writer.write_all(&data).await;
 | 
				
			||||||
                            let _ = file.write_all(&data).await;
 | 
					 | 
				
			||||||
                        // If we are going to parse this file later, we will save it
 | 
					                        // If we are going to parse this file later, we will save it
 | 
				
			||||||
                        // into memory as well as the disk.
 | 
					                        // into memory as well as the disk.
 | 
				
			||||||
                        if should_parse {
 | 
					                        if should_parse {
 | 
				
			||||||
@@ -203,6 +220,7 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
 | 
				
			|||||||
                    },
 | 
					                    },
 | 
				
			||||||
                }
 | 
					                }
 | 
				
			||||||
            }
 | 
					            }
 | 
				
			||||||
 | 
					            let _ = writer.flush();
 | 
				
			||||||
 | 
					
 | 
				
			||||||
            if should_parse {
 | 
					            if should_parse {
 | 
				
			||||||
                // Parse document and get relationships
 | 
					                // Parse document and get relationships
 | 
				
			||||||
@@ -230,9 +248,6 @@ async fn process(mut site: Website, db: Surreal<Client>, reqwest: reqwest::Clien
 | 
				
			|||||||
        } else {
 | 
					        } else {
 | 
				
			||||||
            error!("File failed to cooperate: {:?}", path);
 | 
					            error!("File failed to cooperate: {:?}", path);
 | 
				
			||||||
        }
 | 
					        }
 | 
				
			||||||
        } else {
 | 
					 | 
				
			||||||
            warn!("Server did not respond with Content-Type header: {}", site.site.to_string());
 | 
					 | 
				
			||||||
        }
 | 
					 | 
				
			||||||
    } else {
 | 
					    } else {
 | 
				
			||||||
        error!("Failed to get: {}", &site.site);
 | 
					        error!("Failed to get: {}", &site.site);
 | 
				
			||||||
    }
 | 
					    }
 | 
				
			||||||
 
 | 
				
			|||||||
		Reference in New Issue
	
	Block a user