Compare commits

...

2 Commits

Author SHA1 Message Date
Gwen Lg
1ad5dfbe78 improve error management with thiserror use
- add thiserror dependency
- use distinct Error enum in lib and server
2024-06-14 12:05:45 +02:00
Gwen Lg
0d43729a02 update dependencies in Cargo.toml
update: zip, reqwest and env_logger
fix fmt in Cargo.toml
2024-06-14 12:05:05 +02:00
4 changed files with 1079 additions and 538 deletions

1486
Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@ -14,17 +14,18 @@ categories = ["text-processing"]
[dependencies]
serde_json = "1.0.96"
reqwest = { version = "0.11.18", features = ["json"] }
reqwest = { version = "0.12.4", features = ["json"] }
serde = { version = "1.0.163", features = ["derive"] }
log = "0.4.17"
zip = { version = "0.6.3", optional = true }
zip = { version = "2.1.3", optional = true }
mktemp = { version = "0.5.0", optional = true }
rand = { version = "0.8.5", optional = true }
port_scanner = { version = "0.1.5", optional = true }
thiserror = "1.0.61"
[features]
embedded-server = ["zip", "mktemp", "rand", "port_scanner"]
[dev-dependencies]
env_logger = "0.10.0"
env_logger = "0.11.3"
tokio = { version = "1.28.1", features = ["full"] }

View File

@ -36,10 +36,11 @@
//! println!("RESULT = {:#?}", res);
//! ```
use thiserror::Error;
#[cfg(feature = "embedded-server")]
use crate::server::EmbeddedServer;
use std::collections::HashMap;
use std::error::Error;
#[cfg(feature = "embedded-server")]
pub mod server;
@ -302,6 +303,28 @@ pub struct SuggestResult {
pub suggestions: Vec<String>,
}
#[derive(Debug, Error)]
pub enum Error {
#[cfg(feature = "embedded-server")]
#[error("Grammalecte-server failed to start")]
ServerStartFailed(#[from] server::Error),
#[error("Failed to Serialize Option in Json")]
OptionJsonSerialization(#[source] serde_json::Error),
#[error("Failed to send request `check with option`")]
RequestSendCheckWithOptions(#[source] reqwest::Error),
#[error("Failed to send request `suggest`")]
RequestSendSuggest(#[source] reqwest::Error),
#[error("Failed to Deserialize Check result")]
CheckResultDeserialize(#[source] reqwest::Error),
#[error("Failed to Deserialize Suggest result")]
SuggestDeserialize(#[source] reqwest::Error),
}
/// The Grammalecte client itself
pub struct GrammalecteClient {
base_url: String,
@ -335,7 +358,7 @@ impl GrammalecteClient {
///
/// Python 3.7 or higher must is required at runtime
#[cfg(feature = "embedded-server")]
pub fn start_server() -> Result<Self, Box<dyn Error>> {
pub fn start_server() -> Result<Self, Error> {
let server = EmbeddedServer::start()?;
Ok(Self {
base_url: server.base_url(),
@ -344,7 +367,7 @@ impl GrammalecteClient {
}
/// Run spell check on text
pub async fn spell_check(&self, text: &str) -> Result<CheckResult, Box<dyn Error>> {
pub async fn spell_check(&self, text: &str) -> Result<CheckResult, Error> {
self.spell_check_with_options(text, &HashMap::new()).await
}
@ -353,7 +376,7 @@ impl GrammalecteClient {
&self,
text: &str,
options: &HashMap<GramOpt, bool>,
) -> Result<CheckResult, Box<dyn Error>> {
) -> Result<CheckResult, Error> {
let url = format!("{}/gc_text/fr", self.base_url);
log::debug!("Will use URL {} for spell check", url);
@ -361,7 +384,7 @@ impl GrammalecteClient {
.iter()
.map(|t| (t.0.id(), t.1))
.collect::<HashMap<_, _>>();
let options = serde_json::to_string(&options)?;
let options = serde_json::to_string(&options).map_err(Error::OptionJsonSerialization)?;
let mut params = HashMap::new();
params.insert("text", text);
@ -371,28 +394,32 @@ impl GrammalecteClient {
.post(url)
.form(&params)
.send()
.await?
.await
.map_err(Error::RequestSendCheckWithOptions)?
.json::<CheckResult>()
.await?;
.await
.map_err(Error::CheckResultDeserialize)?;
Ok(result)
}
/// Ask for word suggestion
pub async fn suggest(&self, token: &str) -> Result<SuggestResult, Box<dyn Error>> {
pub async fn suggest(&self, token: &str) -> Result<SuggestResult, Error> {
let url = format!("{}/suggest/fr", self.base_url);
log::debug!("Will use URL {} for word suggestion", url);
let mut params = HashMap::new();
params.insert("token", token);
Ok(reqwest::Client::new()
reqwest::Client::new()
.post(&url)
.form(&params)
.send()
.await?
.await
.map_err(Error::RequestSendSuggest)?
.json()
.await?)
.await
.map_err(Error::SuggestDeserialize)
}
}

View File

@ -1,9 +1,39 @@
use crate::server::utils::{get_free_port, wait_for_port};
use mktemp::Temp;
use std::error::Error;
use std::io::{Cursor, Read};
use std::io::{self, Cursor, Read};
use std::process::{Child, Stdio};
use zip::ZipArchive;
use thiserror::Error;
use zip::{result::ZipError, ZipArchive};
#[derive(Debug, Error)]
pub enum Error {
#[error("Grammalecte-server failed to launch process")]
StartServerProcess(#[source] io::Error),
#[error("Get an available port failed")]
GetFreePort(#[source] io::Error),
#[error("Port {port} did not open in time!")]
WaitPortOpen { port: u16 },
#[error("Create temporary directory failed")]
CreateTempDir(#[source] io::Error),
#[error("Zip archive loading failed")]
ZipArchiveLoading(#[source] ZipError),
#[error("Access file by index failed")]
ZipFileIndex(#[source] ZipError),
#[error("Create directory for files from zip")]
CreateDirectoryForZipFile(#[source] io::Error),
#[error("Read file from zip archive")]
ZipFileReadToEnd(#[source] io::Error),
#[error("Write file from archive on disk")]
WriteFile(#[source] io::Error),
}
pub struct EmbeddedServer {
_srv_dir: Temp,
@ -13,29 +43,31 @@ pub struct EmbeddedServer {
impl EmbeddedServer {
/// Start embedded Grammalecte server on a random free port
pub fn start() -> Result<Self, Box<dyn Error>> {
Self::start_listen_on_port(get_free_port()?)
pub fn start() -> Result<Self, Error> {
Self::start_listen_on_port(get_free_port())
}
/// Start embedded Grammalecte server on a given port
pub fn start_listen_on_port(port: u16) -> Result<Self, Box<dyn Error>> {
pub fn start_listen_on_port(port: u16) -> Result<Self, Error> {
log::info!("Will start server");
// First, unpack server
let dest = mktemp::Temp::new_dir()?;
let dest = mktemp::Temp::new_dir().map_err(Error::CreateTempDir)?;
let cursor = Cursor::new(include_bytes!("GrammalecteDist.zip"));
let mut zip = ZipArchive::new(cursor)?;
let mut zip = ZipArchive::new(cursor).map_err(Error::ZipArchiveLoading)?;
for i in 0..zip.len() {
let mut file = zip.by_index(i)?;
let mut file = zip.by_index(i).map_err(Error::ZipFileIndex)?;
if file.is_dir() {
log::debug!("Create directory: {}", file.name());
std::fs::create_dir_all(dest.join(file.name()))?;
std::fs::create_dir_all(dest.join(file.name()))
.map_err(Error::CreateDirectoryForZipFile)?;
} else {
log::debug!("Decompress file: {}", file.name());
let mut buff = Vec::with_capacity(file.size() as usize);
file.read_to_end(&mut buff)?;
file.read_to_end(&mut buff)
.map_err(Error::ZipFileReadToEnd)?;
std::fs::write(dest.join(file.name()), buff)?;
std::fs::write(dest.join(file.name()), buff).map_err(Error::WriteFile)?;
}
}
@ -54,7 +86,8 @@ impl EmbeddedServer {
.arg(port.to_string())
.stdout(Stdio::null())
.stderr(Stdio::null())
.spawn()?;
.spawn()
.map_err(Error::StartServerProcess)?;
wait_for_port(port)?;
@ -78,11 +111,12 @@ impl Drop for EmbeddedServer {
}
mod utils {
use std::io::ErrorKind;
use std::time::Duration;
use super::Error;
/// Get a free port
pub fn get_free_port() -> std::io::Result<u16> {
pub fn get_free_port() -> u16 {
let mut port = 0;
while !(2000..=64000).contains(&port) {
@ -93,10 +127,10 @@ mod utils {
port += 1;
}
Ok(port)
port
}
pub fn wait_for_port(port: u16) -> std::io::Result<()> {
pub fn wait_for_port(port: u16) -> Result<(), Error> {
for _ in 0..50 {
if port_scanner::scan_port(port) {
return Ok(());
@ -104,9 +138,6 @@ mod utils {
std::thread::sleep(Duration::from_millis(100));
}
Err(std::io::Error::new(
ErrorKind::Other,
format!("Port {} did not open in time!", port),
))?
Err(Error::WaitPortOpen { port })
}
}