feat: refactoring complete?

This commit is contained in:
Per Stark
2024-11-21 21:23:49 +01:00
parent 94f328e542
commit 1e789e1153
27 changed files with 428 additions and 338 deletions

View File

@@ -0,0 +1,107 @@
use super::ingress_object::IngressObject;
use crate::storage::{db::SurrealDbClient, types::file_info::FileInfo};
use serde::{Deserialize, Serialize};
use thiserror::Error;
use tracing::info;
use url::Url;
use uuid::Uuid;
/// Struct defining the expected body when ingressing content.
#[derive(Serialize, Deserialize, Debug)]
pub struct IngressInput {
pub content: Option<String>,
pub instructions: String,
pub category: String,
pub files: Option<Vec<String>>,
}
/// Error types for processing ingress content.
#[derive(Error, Debug)]
pub enum IngressContentError {
#[error("IO error occurred: {0}")]
Io(#[from] std::io::Error),
#[error("UTF-8 conversion error: {0}")]
Utf8(#[from] std::string::FromUtf8Error),
#[error("MIME type detection failed for input: {0}")]
MimeDetection(String),
#[error("Unsupported MIME type: {0}")]
UnsupportedMime(String),
#[error("URL parse error: {0}")]
UrlParse(#[from] url::ParseError),
#[error("UUID parse error: {0}")]
UuidParse(#[from] uuid::Error),
#[error("Redis error: {0}")]
RedisError(String),
}
/// Function to create ingress objects from input.
///
/// # Arguments
/// * `input` - IngressInput containing information needed to ingress content.
/// * `redis_client` - Initialized redis client needed to retrieve file information
///
/// # Returns
/// * `Vec<IngressObject>` - An array containing the ingressed objects, one file/contenttype per object.
pub async fn create_ingress_objects(
input: IngressInput,
db_client: &SurrealDbClient,
) -> Result<Vec<IngressObject>, IngressContentError> {
// Initialize list
let mut object_list = Vec::new();
// Create a IngressObject from input.content if it exists, checking for URL or text
if let Some(input_content) = input.content {
match Url::parse(&input_content) {
Ok(url) => {
info!("Detected URL: {}", url);
object_list.push(IngressObject::Url {
url: url.to_string(),
instructions: input.instructions.clone(),
category: input.category.clone(),
});
}
Err(_) => {
info!("Treating input as plain text");
object_list.push(IngressObject::Text {
text: input_content.to_string(),
instructions: input.instructions.clone(),
category: input.category.clone(),
});
}
}
}
// Look up FileInfo objects using the db and the submitted uuids in input.files
if let Some(file_uuids) = input.files {
for uuid_str in file_uuids {
let uuid = Uuid::parse_str(&uuid_str)?;
match FileInfo::get_by_uuid(uuid, db_client).await {
Ok(file_info) => {
object_list.push(IngressObject::File {
file_info,
instructions: input.instructions.clone(),
category: input.category.clone(),
});
}
_ => {
info!("No file with UUID: {}", uuid);
}
}
}
}
// If no objects are constructed, we return Err
if object_list.is_empty() {
return Err(IngressContentError::MimeDetection(
"No valid content or files provided".into(),
));
}
Ok(object_list)
}

View File

@@ -0,0 +1,119 @@
use crate::storage::types::{file_info::FileInfo, text_content::TextContent};
use serde::{Deserialize, Serialize};
use super::ingress_input::IngressContentError;
/// Knowledge object type, containing the content or reference to it, as well as metadata
#[derive(Debug, Serialize, Deserialize, Clone)]
pub enum IngressObject {
Url {
url: String,
instructions: String,
category: String,
},
Text {
text: String,
instructions: String,
category: String,
},
File {
file_info: FileInfo,
instructions: String,
category: String,
},
}
impl IngressObject {
/// Creates a new `TextContent` instance from a `IngressObject`.
///
/// # Arguments
/// `&self` - A reference to the `IngressObject`.
///
/// # Returns
/// `TextContent` - An object containing a text representation of the object, could be a scraped URL, parsed PDF, etc.
pub async fn to_text_content(&self) -> Result<TextContent, IngressContentError> {
match self {
IngressObject::Url {
url,
instructions,
category,
} => {
let text = Self::fetch_text_from_url(url).await?;
Ok(TextContent::new(
text,
instructions.into(),
category.into(),
None,
))
}
IngressObject::Text {
text,
instructions,
category,
} => Ok(TextContent::new(
text.into(),
instructions.into(),
category.into(),
None,
)),
IngressObject::File {
file_info,
instructions,
category,
} => {
let text = Self::extract_text_from_file(file_info).await?;
Ok(TextContent::new(
text,
instructions.into(),
category.into(),
Some(file_info.to_owned()),
))
}
}
}
/// Fetches and extracts text from a URL.
async fn fetch_text_from_url(_url: &str) -> Result<String, IngressContentError> {
unimplemented!()
}
/// Extracts text from a file based on its MIME type.
async fn extract_text_from_file(file_info: &FileInfo) -> Result<String, IngressContentError> {
match file_info.mime_type.as_str() {
"text/plain" => {
// Read the file and return its content
let content = tokio::fs::read_to_string(&file_info.path).await?;
Ok(content)
}
"text/markdown" => {
// Read the file and return its content
let content = tokio::fs::read_to_string(&file_info.path).await?;
Ok(content)
}
"application/pdf" => {
// TODO: Implement PDF text extraction using a crate like `pdf-extract` or `lopdf`
Err(IngressContentError::UnsupportedMime(
file_info.mime_type.clone(),
))
}
"image/png" | "image/jpeg" => {
// TODO: Implement OCR on image using a crate like `tesseract`
Err(IngressContentError::UnsupportedMime(
file_info.mime_type.clone(),
))
}
"application/octet-stream" => {
let content = tokio::fs::read_to_string(&file_info.path).await?;
Ok(content)
}
"text/x-rust" => {
let content = tokio::fs::read_to_string(&file_info.path).await?;
Ok(content)
}
// Handle other MIME types as needed
_ => Err(IngressContentError::UnsupportedMime(
file_info.mime_type.clone(),
)),
}
}
}

2
src/ingress/types/mod.rs Normal file
View File

@@ -0,0 +1,2 @@
pub mod ingress_input;
pub mod ingress_object;