revised approach

split file operation and ingress functionality
This commit is contained in:
Per Stark
2024-09-24 16:21:27 +02:00
parent 1bd2eee8e1
commit 18c84f5c88
18 changed files with 264 additions and 36464 deletions
+109
View File
@@ -0,0 +1,109 @@
use axum_typed_multipart::{FieldData, TryFromMultipart};
use mime_guess::from_path;
use serde::{Deserialize, Serialize};
use sha2::{Digest, Sha256};
use std::{io::{BufReader, Read}, path::Path};
use tempfile::NamedTempFile;
use thiserror::Error;
use tracing::info;
use url::Url;
use uuid::Uuid;
/// Error types for file and content handling.
#[derive(Error, Debug)]
pub enum FileError{
#[error("IO error occurred: {0}")]
Io(#[from] std::io::Error),
#[error("MIME type detection failed for input: {0}")]
MimeDetection(String),
#[error("Unsupported MIME type: {0}")]
UnsupportedMime(String),
}
#[derive(Debug, TryFromMultipart)]
pub struct FileUploadRequest {
#[form_data(limit = "unlimited")]
pub file: FieldData<NamedTempFile>,
}
#[derive(Debug, Clone, Deserialize, Serialize)]
pub struct FileInfo {
pub uuid: Uuid,
pub sha256: String,
pub path: String,
pub mime_type: String,
}
impl FileInfo {
pub async fn new(file: NamedTempFile) -> Result<FileInfo, FileError> {
// Calculate SHA based on file
let sha = Self::get_sha(&file).await?;
// Check if SHA exists in redis db
// If so, return existing FileInfo
// Generate UUID
// Persist file with uuid as path
// Guess the mime_type
// Construct the object
// file.persist("./data/{id}");
Ok(FileInfo { uuid: Uuid::new_v4(), sha256: sha, path:String::new(), mime_type:String::new() })
}
pub async fn get(id: String) -> Result<FileInfo, FileError> {
// Get the SHA based on file in uuid path
// Check if SHA exists in redis
// If so, return FileInfo
// Else return error
Ok(FileInfo { uuid: Uuid::new_v4(), sha256: String::new() , path:String::new(), mime_type:String::new() })
}
pub async fn update(id: String, file: NamedTempFile) -> Result<FileInfo, FileError> {
// Calculate SHA based on file
// Check if SHA exists in redis
// If so, return existing FileInfo
// Use the submitted UUID
// Replace the old file with uuid as path
// Guess the mime_type
// Construct the object
Ok(FileInfo { uuid: Uuid::new_v4(), sha256: String::new() , path:String::new(), mime_type:String::new() })
}
pub async fn delete(id: String) -> Result<(), FileError> {
// Get the SHA based on file in uuid path
// Remove the entry from redis db
Ok(())
}
async fn get_sha(file: &NamedTempFile) -> Result<String, FileError> {
let input = file.as_file();
let mut reader = BufReader::new(input);
let digest = {
let mut hasher = Sha256::new();
let mut buffer = [0; 1024];
loop {
let count = reader.read(&mut buffer)?;
if count == 0 { break }
hasher.update(&buffer[..count]);
}
hasher.finalize()
};
Ok(format!("{:X}", digest))
}
}
// let input = File::open(path)?;
// let mut reader = BufReader::new(input);
// let digest = {
// let mut hasher = Sha256::new();
// let mut buffer = [0; 1024];
// loop {
// let count = reader.read(&mut buffer)?;
// if count == 0 { break }
// hasher.update(&buffer[..count]);
// }
// hasher.finalize()
// };
// Ok(HEXLOWER.encode(digest.as_ref()))
+21 -37
View File
@@ -1,35 +1,8 @@
use axum::extract::Multipart;
use serde::{Deserialize, Serialize};
use thiserror::Error;
use tokio::fs;
use tracing::info;
use url::Url;
use uuid::Uuid;
use sha2::{Digest, Sha256};
use std::path::Path;
use mime_guess::from_path;
use axum_typed_multipart::{FieldData, TryFromMultipart };
use tempfile::NamedTempFile;
#[derive(Debug, TryFromMultipart)]
pub struct IngressMultipart {
/// JSON content field
pub content: Option<String>,
pub instructions: String,
pub category: String,
/// Optional file
#[form_data(limit = "unlimited")]
pub file: Option<FieldData<NamedTempFile>>,
}
#[derive(Debug, Clone, Deserialize, Serialize)]
pub struct FileInfo {
pub uuid: Uuid,
pub sha256: String,
pub path: String,
pub mime_type: String,
}
use super::files::FileInfo;
#[derive(Debug, Deserialize, Serialize)]
pub enum Content {
@@ -37,6 +10,14 @@ pub enum Content {
Text(String),
}
#[derive(Serialize, Deserialize, Debug)]
pub struct IngressInput {
pub content: Option<String>,
pub instructions: String,
pub category: String,
pub files: Option<Vec<String>>,
}
#[derive(Serialize, Deserialize, Debug)]
pub struct IngressContent {
pub content: Option<Content>,
@@ -65,29 +46,32 @@ pub enum IngressContentError {
}
impl IngressContent {
/// Create a new `IngressContent` from `IngressMultipart`.
/// Create a new `IngressContent` from `IngressInput`.
pub async fn new(
content: Option<String>, instructions: String, category: String,
file: Option<FileInfo>
input: IngressInput
) -> Result<IngressContent, IngressContentError> {
let content = if let Some(content_str) = content {
let content = if let Some(input_content) = input.content {
// Check if the content is a URL
if let Ok(url) = Url::parse(&content_str) {
if let Ok(url) = Url::parse(&input_content) {
info!("Detected URL: {}", url);
Some(Content::Url(url.to_string()))
} else {
info!("Treating input as plain text");
Some(Content::Text(content_str))
Some(Content::Text(input_content))
}
} else {
None
};
// Check if there is files
// Use FileInfo.get(id) with all the ids in files vec
// Return a vec<FileInfo> as file_info_list
Ok(IngressContent {
content,
instructions,
category,
files: file.map(|f| vec![f]), // Single file wrapped in a Vec
instructions: input.instructions,
category: input.category,
files: None,
})
}
}
+1
View File
@@ -1 +1,2 @@
pub mod files;
pub mod ingress;