morethantext-web/src/morethantext/cache.rs

671 lines
22 KiB
Rust

use super::{Database, DBError, ErrorCode, FileData, SessionData, Store};
use async_std::{
fs::{read, remove_file, write},
path::{Path, PathBuf},
};
use rand::{distributions::Alphanumeric, thread_rng, Rng};
use std::{
cell::Cell,
slice, str,
time::{Duration, Instant},
};
const ENTRY: &str = "EntryPoint";
#[derive(Clone)]
enum DataType {
DBMap(Store),
TableMap(Database),
}
impl DataType {
fn new(data_type: &str) -> Result<Self, DBError> {
match data_type {
"store" => Ok(DataType::DBMap(Store::new())),
_ => Err(DBError::from_code(ErrorCode::DataTypeIncorrect(
data_type.to_string(),
))),
}
}
}
impl SessionData for DataType {
fn add(&mut self, key: &str, value: &str, data: &str) -> Result<Vec<String>, DBError> {
match self {
DataType::DBMap(dbs) => dbs.add(key, value, data),
DataType::TableMap(_) => todo!(),
}
}
fn eq(&self, key: &str, value: &str) -> Result<Vec<String>, DBError> {
match self {
DataType::DBMap(dbs) => dbs.eq(key, value),
DataType::TableMap(_) => todo!(),
}
}
fn list(&self, keys: Vec<&str>) -> Result<Vec<String>, DBError> {
match self {
DataType::DBMap(dbs) => dbs.list(keys),
DataType::TableMap(_) => todo!(),
}
}
}
impl FileData<Self> for DataType {
fn to_bytes(&self) -> Vec<u8> {
let mut output = Vec::new();
match self {
DataType::DBMap(store) => {
output.append(&mut "DBMap".as_bytes().to_vec());
output.push(0);
output.append(&mut store.to_bytes());
},
DataType::TableMap(_) => todo!(),
}
output
}
fn from_bytes(data: &mut slice::Iter<u8>) -> Result<Self, DBError> {
let mut header: Vec<u8> = Vec::new();
loop {
let letter = match data.next() {
Some(a) => a.clone(),
None => 0,
};
if letter == 0 {
break;
} else {
header.push(letter);
}
}
let header = match str::from_utf8(&header) {
Ok(item) => item,
Err(_) => return Err(DBError::from_code(ErrorCode::CorruptFile)),
};
match header {
"DBMap" => match Store::from_bytes(data) {
Ok(store) => Ok(DataType::DBMap(store)),
Err(err) => Err(err),
},
_ => Err(DBError::from_code(ErrorCode::CorruptFile)),
}
}
}
struct Entry {
data: DataType,
filename: PathBuf,
last_used: Cell<Instant>,
}
impl Entry {
async fn new<P>(filename: P, data: DataType) -> Result<Self, DBError>
where
P: Into<PathBuf>,
{
let pathbuf = filename.into();
if pathbuf.as_path().exists().await {
return Err(DBError::from_code(ErrorCode::EntryExists(pathbuf)));
} else {
match write(&pathbuf, data.to_bytes()).await {
Ok(_) => (),
Err(err) => {
let mut error = DBError::from_code(ErrorCode::EntryWriteFailure(pathbuf));
error.add_source(err);
return Err(error);
}
}
}
Ok(Self {
data: data,
filename: pathbuf,
last_used: Cell::new(Instant::now()),
})
}
async fn get<P>(filename: P) -> Result<Self, DBError>
where
P: Into<PathBuf>,
{
let pathbuf = filename.into();
let content = match read(&pathbuf).await {
Ok(text) => text,
Err(err) => {
let mut error = DBError::from_code(ErrorCode::EntryReadFailure(pathbuf));
error.add_source(err);
return Err(error);
}
};
let data = match DataType::from_bytes(&mut content.iter()) {
Ok(raw) => raw,
Err(err) => {
let mut error = DBError::from_code(ErrorCode::EntryReadFailure(pathbuf));
error.add_source(err);
return Err(error);
}
};
Ok(Self {
data: data,
filename: pathbuf,
last_used: Cell::new(Instant::now()),
})
}
fn elapsed(&self) -> Duration {
self.last_used.get().elapsed()
}
fn data(&self) -> DataType {
self.last_used.set(Instant::now());
self.data.clone()
}
async fn update(&mut self, data: DataType) -> Result<(), DBError> {
self.last_used.set(Instant::now());
match write(&self.filename, data.to_bytes()).await {
Ok(_) => (),
Err(err) => {
let mut error = DBError::from_code(ErrorCode::EntryWriteFailure(self.filename.clone()));
error.add_source(err);
return Err(error);
}
};
self.data = data;
Ok(())
}
async fn remove(&self) -> Result<(), DBError> {
match remove_file(&self.filename).await {
Ok(_) => Ok(()),
Err(err) => {
let mut error = DBError::from_code(ErrorCode::EntryDeleteFailure(self.filename.clone()));
error.add_source(err);
Err(error)
}
}
}
}
struct Cache;
impl Cache {
async fn new<P>(dir: P) -> Result<Self, DBError>
where
P: Into<PathBuf>,
{
let pathbuf = dir.into();
let entry = pathbuf.as_path().join(ENTRY);
match Entry::get(entry.clone()).await {
Ok(_) => Ok(Self {}),
Err(_) => {
let store = DataType::new("store").unwrap();
match Entry::new(entry, store).await {
Ok(_) => Ok(Self {}),
Err(err) => {
let mut error = DBError::from_code(ErrorCode::CacheReadWrite);
error.add_source(err);
Err(error)
},
}
},
}
}
}
#[cfg(test)]
mod datatype_sesssion {
use super::*;
#[test]
fn invalid_cache_type() -> Result<(), DBError> {
match DataType::new("dkhgdl") {
Ok(_) => Err(DBError::new("invalid data type should raise an error")),
Err(err) => match err.code {
ErrorCode::DataTypeIncorrect(_) => Ok(()),
_ => Err(DBError::new("Invalid error code")),
},
}
}
#[test]
fn create_storage() {
let dbs = DataType::new("store").unwrap();
let expected: Vec<String> = Vec::new();
assert_eq!(dbs.list(["database"].to_vec()).unwrap(), expected);
}
#[test]
fn update_storage() {
let mut dbs = DataType::new("store").unwrap();
let name = "new_database";
let id = "someid";
dbs.add("database", name, id).unwrap();
assert_eq!(dbs.eq("database", name).unwrap(), [id].to_vec());
assert_eq!(dbs.list(["database"].to_vec()).unwrap(), [name].to_vec());
}
}
#[cfg(test)]
mod datatype_file {
use super::*;
#[test]
fn new_store_bytes() {
let dbs = DataType::new("store").unwrap();
let mut expected = "DBMap".as_bytes().to_vec();
expected.push(0);
assert_eq!(dbs.to_bytes(), expected);
}
#[test]
fn store_bytes_with_info() {
let name = "title";
let id = "king";
let mut store = Store::new();
let mut dt_store = DataType::new("store").unwrap();
let mut expected = dt_store.to_bytes();
store.add("database", name, id).unwrap();
expected.append(&mut store.to_bytes());
dt_store.add("database", name, id).unwrap();
assert_eq!(dt_store.to_bytes(), expected);
}
#[test]
fn read_empty_store() {
let dt_store = DataType::new("store").unwrap();
let data = dt_store.to_bytes();
let mut feed = data.iter();
let output = DataType::from_bytes(&mut feed).unwrap();
assert_eq!(
dt_store.list(["database"].to_vec()).unwrap(),
output.list(["database"].to_vec()).unwrap()
);
}
#[test]
fn read_store_info() {
let mut dt_store = DataType::new("store").unwrap();
dt_store.add("database", "raven", "beastboy").unwrap();
let data = dt_store.to_bytes();
let mut feed = data.iter();
let output = DataType::from_bytes(&mut feed).unwrap();
assert_eq!(
dt_store.list(["database"].to_vec()).unwrap(),
output.list(["database"].to_vec()).unwrap()
);
}
#[test]
fn read_bad_header() -> Result<(), DBError> {
let data = "sdghsdl".as_bytes().to_vec();
let mut feed = data.iter();
match DataType::from_bytes(&mut feed) {
Ok(_) => Err(DBError::new("should have raised an error")),
Err(err) => match err.code {
ErrorCode::CorruptFile => Ok(()),
_ => Err(DBError::new("incorrect error")),
},
}
}
#[test]
fn read_bad_store() -> Result<(), DBError> {
let mut data = "DBMap".as_bytes().to_vec();
data.push(0);
data.append(&mut "sdfgs".as_bytes().to_vec());
let mut feed = data.iter();
match DataType::from_bytes(&mut feed) {
Ok(_) => Err(DBError::new("should have raised an error")),
Err(err) => match err.code {
ErrorCode::CorruptFile => Ok(()),
_ => Err(DBError::new("incorrect error code")),
}
}
}
}
#[cfg(test)]
mod entry {
use super::*;
use std::error::Error;
use tempfile::tempdir;
#[async_std::test]
async fn get_elapsed_time() {
let dir = tempdir().unwrap();
let data = DataType::new("store").unwrap();
let filepath = dir.path().join("count");
let filename = filepath.to_str().unwrap();
let item = Entry::new(filename.to_string(), data).await.unwrap();
assert!(
Duration::from_secs(1) > item.elapsed(),
"last_used should have been now."
);
item.last_used
.set(Instant::now() - Duration::from_secs(500));
assert!(
Duration::from_secs(499) < item.elapsed(),
"The duration should have increased."
);
}
#[async_std::test]
async fn create() {
let dir = tempdir().unwrap();
let mut data = DataType::new("store").unwrap();
data.add("database", "roger", "moore").unwrap();
let filepath = dir.path().join("wiliam");
let filename = filepath.to_str().unwrap();
let item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
assert!(
Duration::from_secs(1) > item.elapsed(),
"last_used should have been now."
);
let output = item.data();
assert_eq!(
data.list(["database"].to_vec()).unwrap(),
output.list(["database"].to_vec()).unwrap()
);
assert!(filepath.is_file(), "Should have created the entry file.");
let content = read(&filepath).await.unwrap();
assert_eq!(content, data.to_bytes());
}
#[async_std::test]
async fn create_errors_on_bad_files() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let data = DataType::new("store").unwrap();
let filepath = dir.path().join("bad").join("path");
let filename = filepath.to_str().unwrap();
match Entry::new(filename.to_string(), data).await {
Ok(_) => Err(DBError::new("bad file names should raise an error")),
Err(err) => match err.code {
ErrorCode::EntryWriteFailure(_) => {
assert!(err.source().is_some(), "Must include the source error.");
assert!(err
.source()
.unwrap()
.to_string()
.contains("could not write to file"));
Ok(())
}
_ => Err(DBError::new("incorrect error code")),
},
}
}
#[async_std::test]
async fn create_does_not_over_writes() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let id = "wicked";
let file = dir.path().join(id);
let filename = file.to_str().unwrap();
write(&file, b"previous").await.unwrap();
let data = DataType::new("store").unwrap();
match Entry::new(filename.to_string(), data).await {
Ok(_) => {
return Err(DBError::new(
"Should produce an error for an existing Entry",
))
}
Err(err) => match err.code {
ErrorCode::EntryExists(_) => Ok(()),
_ => Err(DBError::new("incorrect error code")),
},
}
}
#[async_std::test]
async fn get_updates_last_used() {
let dir = tempdir().unwrap();
let data = DataType::new("store").unwrap();
let filepath = dir.path().join("holder");
let filename = filepath.to_str().unwrap();
let item = Entry::new(filename.to_string(), data).await.unwrap();
item.last_used
.set(Instant::now() - Duration::from_secs(300));
item.data();
assert!(
Duration::from_secs(1) > item.elapsed(),
"last_used should have been reset."
);
}
#[async_std::test]
async fn update_entry() {
let dir = tempdir().unwrap();
let mut data = DataType::new("store").unwrap();
let filepath = dir.path().join("changing");
let filename = filepath.to_str().unwrap();
let mut item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
item.last_used
.set(Instant::now() - Duration::from_secs(500));
data.add("database", "new", "stuff").unwrap();
item.update(data.clone()).await.unwrap();
assert!(
Duration::from_secs(1) > item.elapsed(),
"last_used should have been reset."
);
let output = item.data();
assert_eq!(
data.list(["database"].to_vec()).unwrap(),
output.list(["database"].to_vec()).unwrap()
);
let content = read(&filepath).await.unwrap();
assert_eq!(content, data.to_bytes());
}
#[async_std::test]
async fn update_write_errors() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let data = DataType::new("store").unwrap();
let filepath = dir.path().join("changing");
let filename = filepath.to_str().unwrap();
let mut item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
drop(dir);
match item.update(data).await {
Ok(_) => Err(DBError::new("file writes should return an error")),
Err(err) => match err.code {
ErrorCode::EntryWriteFailure(_) => {
assert!(err.source().is_some(), "Must include the source error.");
assert!(err
.source()
.unwrap()
.to_string()
.contains("could not write to file"));
Ok(())
},
_ => Err(DBError::new("incorrect error code")),
}
}
}
#[async_std::test]
async fn retrieve() {
let dir = tempdir().unwrap();
let mut data = DataType::new("store").unwrap();
data.add("database", "something_old", "3.14159").unwrap();
let filepath = dir.path().join("existing");
let filename = filepath.to_str().unwrap();
let item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
let output = Entry::get(filename).await.unwrap();
assert_eq!(
output.data().list(["database"].to_vec()).unwrap(),
data.list(["database"].to_vec()).unwrap()
);
assert_eq!(output.filename.to_str().unwrap(), filename);
assert!(
Duration::from_secs(1) > item.elapsed(),
"last_used should have been reset."
);
}
#[async_std::test]
async fn retrieve_file_missing() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let filepath = dir.path().join("justnotthere");
let filename = filepath.to_str().unwrap();
match Entry::get(filename).await {
Ok(_) => Err(DBError::new("should have returned an error")),
Err(err) => match err.code {
ErrorCode::EntryReadFailure(_) => {
assert!(err.source().is_some(), "Error should have a source.");
assert!(
err.source()
.unwrap()
.to_string()
.contains("could not read file"),
"Source Error Message: {}",
err.source().unwrap().to_string()
);
Ok(())
}
_ => Err(DBError::new("incorrect error code")),
}
}
}
#[async_std::test]
async fn retrieve_corrupt_file() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let filepath = dir.path().join("garbage");
let filename = filepath.to_str().unwrap();
write(&filepath, b"jhsdfghlsdf").await.unwrap();
match Entry::get(filename).await {
Ok(_) => Err(DBError::new("should have returned an error")),
Err(err) => match err.code {
ErrorCode::EntryReadFailure(_) => {
assert!(err.source().is_some(), "Error should have a source.");
assert!(
err.source().unwrap().to_string().contains("corrupt file"),
"Source Error Message: {}",
err.source().unwrap().to_string()
);
Ok(())
},
_ => Err(DBError::new("incorrect error code")),
}
}
}
#[async_std::test]
async fn delete() {
let dir = tempdir().unwrap();
let filepath = dir.path().join("byebye");
let filename = filepath.to_str().unwrap();
let data = DataType::new("store").unwrap();
let item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
item.remove().await.unwrap();
assert!(!filepath.exists(), "Entry file should be removed.");
}
#[async_std::test]
async fn delete_bad_file() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let filepath = dir.path().join("itsnotthere");
let filename = filepath.to_str().unwrap();
let data = DataType::new("store").unwrap();
let item = Entry::new(filename.to_string(), data.clone())
.await
.unwrap();
remove_file(filename).await.unwrap();
match item.remove().await {
Ok(_) => Err(DBError::new("should have produced an error")),
Err(err) => match err.code {
ErrorCode::EntryDeleteFailure(_) => {
assert!(err.source().is_some(), "Error should have a source.");
assert!(
err.source()
.unwrap()
.to_string()
.contains("could not remove file"),
"Source Error Message: {}",
err.source().unwrap().to_string()
);
Ok(())
},
_ => Err(DBError::new("incorrect error code")),
}
}
}
}
#[cfg(test)]
mod cache {
use super::*;
use std::error::Error;
use tempfile::tempdir;
#[async_std::test]
async fn create() {
let dir = tempdir().unwrap();
Cache::new(dir.path().to_str().unwrap()).await.unwrap();
let epoint = dir.path().join(ENTRY);
assert!(
epoint.is_file(),
"{} did not get created.",
epoint.display()
);
let entry = Entry::get(epoint.to_str().unwrap()).await.unwrap();
assert_eq!(
entry.data().list(["database"].to_vec()).unwrap(),
Vec::<String>::new()
);
}
#[async_std::test]
async fn entry_failure() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let path = dir.path().join("bad").join("path");
match Cache::new(path).await {
Ok(_) => Err(DBError::new("Should have produced an error.")),
Err(err) => {
match err.code {
ErrorCode::CacheReadWrite => {
assert!(err.source().is_some(), "Error should have a source.");
assert!(
err.source().unwrap().to_string().contains("write failure"),
"Source Error Message: {}",
err.source().unwrap().to_string()
);
Ok(())
},
_ => Err(DBError::new("incorrect error code")),
}
}
}
}
#[async_std::test]
async fn existing_entry_point() {
let dir = tempdir().unwrap();
let data = DataType::new("store").unwrap();
Entry::new(dir.path().join(ENTRY), data.clone())
.await
.unwrap();
let cache = Cache::new(dir.path()).await.unwrap();
}
#[async_std::test]
async fn corrupt_enty_point() -> Result<(), DBError> {
let dir = tempdir().unwrap();
let file = dir.path().join(ENTRY);
write(file, b"Really bad data.").await.unwrap();
match Cache::new(dir.path()).await {
Ok(_) => Err(DBError::new("should have errored")),
Err(_) => Ok(()),
}
}
}