use super::*; use anyhow::{Context, Result}; use std::collections::HashMap; pub const EXTENSION_METADATA: &str = "__extension__"; pub const PATH_METADATA: &str = "__path__"; pub async fn load_recursive_url( loaders: &HashMap, path: &str, ) -> Result> { let extension = RECURSIVE_URL_LOADER; let pages: Vec = match loaders.get(extension) { Some(loader_command) => { let contents = run_loader_command(path, extension, loader_command)?; serde_json::from_str(&contents).context(r#"The crawler response is invalid. It should follow the JSON format: `[{"path":"...", "text":"..."}]`."#)? } None => { let options = CrawlOptions::preset(path); crawl_website(path, options).await? } }; let output = pages .into_iter() .map(|v| { let Page { path, text } = v; let mut metadata: RagMetadata = Default::default(); metadata.insert(PATH_METADATA.into(), path); metadata.insert(EXTENSION_METADATA.into(), "md".into()); (text, metadata) }) .collect(); Ok(output) } pub async fn load_file( loaders: &HashMap, path: &str, ) -> Result<(String, RagMetadata)> { let extension = get_patch_extension(path).unwrap_or_else(|| DEFAULT_EXTENSION.into()); match loaders.get(&extension) { Some(loader_command) => load_with_command(path, &extension, loader_command), None => load_plain(path, &extension).await, } } pub async fn load_url( loaders: &HashMap, path: &str, ) -> Result<(String, RagMetadata)> { let (contents, extension) = fetch(loaders, path, false).await?; let mut metadata: RagMetadata = Default::default(); metadata.insert(PATH_METADATA.into(), path.into()); metadata.insert(EXTENSION_METADATA.into(), extension); Ok((contents, metadata)) } async fn load_plain(path: &str, extension: &str) -> Result<(String, RagMetadata)> { let contents = tokio::fs::read_to_string(path).await?; let mut metadata: RagMetadata = Default::default(); metadata.insert(PATH_METADATA.into(), path.to_string()); metadata.insert(EXTENSION_METADATA.into(), extension.to_string()); Ok((contents, metadata)) } fn load_with_command( path: &str, extension: &str, loader_command: &str, ) -> Result<(String, RagMetadata)> { let contents = run_loader_command(path, extension, loader_command)?; let mut metadata: RagMetadata = Default::default(); metadata.insert(PATH_METADATA.into(), path.to_string()); metadata.insert(EXTENSION_METADATA.into(), DEFAULT_EXTENSION.to_string()); Ok((contents, metadata)) }