[WIP]

Cargo.toml

@@ -8,10 +8,18 @@ authors = ["Pietro Albini <[email protected]>"]
 failure = "0.1.8"
 reqwest = { version = "0.12.8", features = ["blocking", "json", "rustls-tls"], default-features = false }
 serde = "1.0.210"
-serde_derive = "1.0.210"
-serde_json = "1.0.128"
+serde_derive = "1"
+serde_json = "1"
 log = "0.4.22"
 env_logger = "0.11.5"
 csv = "1.3.0"
 ctrlc = "3.4.5"
 crossbeam-utils = "0.8.20"
+
+tokio = { version = "1.40", features = ["full", "tracing"] }
+dotenvy = "0.15.7"
+color-eyre = "0.6.3"
+tracing = "0.1.40"
+tracing-subscriber = "0.3"
+enum-map = { version = "2.7.3", features = ["serde"] }
+thiserror = "1"

src/config.rs

@@ -20,8 +20,9 @@
 
 use std::path::PathBuf;
 
+#[derive(Debug, Clone)]
 pub struct Config {
-    pub github_token: String,
+    pub github_token: Vec<String>,
     pub data_dir: PathBuf,
     pub timeout: Option<u64>,
 }

src/data.rs

@@ -18,99 +18,162 @@
 // OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 // SOFTWARE.
 
+use enum_map::{Enum, EnumMap};
+use serde_derive::{Deserialize, Serialize};
+use tokio::sync::Mutex;
+use tokio::task::{spawn_blocking, JoinSet};
+
 use crate::config::Config;
-use crate::prelude::*;
-use std::collections::HashMap;
+use std::collections::BTreeMap;
+use std::fs::OpenOptions;
 use std::path::PathBuf;
-use std::sync::{Arc, Mutex};
+use std::sync::atomic::AtomicUsize;
+use std::sync::Arc;
 use std::{
-    fs::{self, File, OpenOptions},
+    fs::{self, File},
     io::{prelude::*, BufWriter},
 };
 
-#[derive(Default, Serialize, Deserialize)]
-struct State {
-    last_id: HashMap<String, usize>,
+#[derive(Debug, Enum, Serialize, Deserialize, Copy, Clone)]
+pub enum Forge {
+    Github,
 }
 
-#[derive(Serialize, Deserialize)]
+#[derive(Debug, Default, Serialize, Deserialize)]
+struct State(EnumMap<Forge, AtomicUsize>);
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
 pub struct Repo {
     pub id: String,
     pub name: String,
     pub has_cargo_toml: bool,
     pub has_cargo_lock: bool,
 }
 
+#[derive(Debug, Clone)]
 pub struct Data {
-    base_dir: PathBuf,
+    data_dir: PathBuf,
 
     csv_write_lock: Arc<Mutex<()>>,
+    state_lock: Arc<Mutex<()>>,
 
-    state_path: PathBuf,
-    state_cache: Arc<Mutex<Option<State>>>,
+    state_cache: Arc<State>,
+
+    repos_state: Arc<Mutex<EnumMap<Forge, BTreeMap<String, Repo>>>>,
 }
 
 impl Data {
-    pub fn new(config: &Config) -> Self {
-        Data {
-            base_dir: config.data_dir.clone(),
+    pub fn new(config: &Config) -> color_eyre::Result<Self> {
+        let mut data = Data {
+            data_dir: config.data_dir.clone(),
 
             csv_write_lock: Arc::new(Mutex::new(())),
 
-            state_path: config.data_dir.join("state.json"),
-            state_cache: Arc::new(Mutex::new(None)),
-        }
-    }
+            state_lock: Arc::new(Mutex::new(())),
+            state_cache: Arc::new(State::default()),
+            repos_state: Arc::new(Mutex::new(EnumMap::default())),
+        };
+
+        // TODO: create CSV files if not exist
+
 
-    fn edit_state<T, F: Fn(&mut State) -> Fallible<T>>(&self, f: F) -> Fallible<T> {
-        let mut state_cache = self.state_cache.lock().unwrap();
+        let state_path = data.state_path();
+        if state_path.exists() {
+            let state_cache: State = serde_json::from_slice(&fs::read(&state_path)?)?;
 
-        if state_cache.is_none() {
-            if self.state_path.exists() {
-                *state_cache = Some(serde_json::from_slice(&fs::read(&self.state_path)?)?);
-            } else {
-                *state_cache = Some(Default::default());
-            }
+            data.state_cache = Arc::new(state_cache)
         }
 
-        let state = state_cache.as_mut().unwrap();
-        let result = f(state)?;
+        Ok(data)
+    }
 
-        let mut file = BufWriter::new(File::create(&self.state_path)?);
-        serde_json::to_writer_pretty(&mut file, &state)?;
-        file.write_all(&[b'\n'])?;
+    pub fn state_path(&self) -> PathBuf {
+        self.data_dir.join("state.json")
+    }
 
-        Ok(result)
+    pub fn csv_path(&self, forge: Forge) -> PathBuf {
+        match forge {
+            Forge::Github => self.data_dir.join("github"),
+        }
     }
 
-    pub fn get_last_id(&self, platform: &str) -> Fallible<Option<usize>> {
-        self.edit_state(|state| Ok(state.last_id.get(platform).cloned()))
+    pub fn get_last_id(&self, forge: Forge) -> usize {
+        self.state_cache.0[forge].load(std::sync::atomic::Ordering::SeqCst)
     }
 
-    pub fn set_last_id(&self, platform: &str, id: usize) -> Fallible<()> {
-        self.edit_state(|state| {
-            state.last_id.insert(platform.to_string(), id);
+    /// Store the state cache to disk, i.e. last fetched ids
+    async fn store_state_cache(&self) -> color_eyre::Result<()> {
+        let state = self.state_cache.clone();
+        let lock = self.state_lock.clone();
+        let state_path = self.state_path();
+        spawn_blocking(move || -> color_eyre::Result<()> {
+            let guard = lock.blocking_lock();
+
+            let file = File::create(state_path)?;
+            let mut file = BufWriter::new(file);
+            serde_json::to_writer_pretty(&mut file, state.as_ref())?;
+            file.write_all(b"\n")?;
+
+            drop(guard);
+
             Ok(())
         })
+        .await
+        .unwrap()
     }
 
-    pub fn store_repo(&self, platform: &str, repo: Repo) -> Fallible<()> {
-        // Ensure only one thread can write to CSV files at once
-        let _lock = self.csv_write_lock.lock().unwrap();
+    /// Stores the repos found to disk in a CSV
+    async fn store_csv(&self) -> color_eyre::Result<()> {
+        let mut repos = self.repos_state.lock().await;
 
-        let file = self.base_dir.join(format!("{}.csv", platform));
+        let mut js = JoinSet::new();
 
-        // Create the new file or append to it
-        let mut csv = if file.exists() {
-            csv::WriterBuilder::new()
-                .has_headers(false)
-                .from_writer(OpenOptions::new().append(true).open(&file)?)
-        } else {
-            csv::WriterBuilder::new().from_path(&file)?
-        };
+        for (forge, repos) in repos.iter() {
+            let path = self.csv_path(forge);
+            let repos = repos.clone(); // is this necessary?
+            js.spawn_blocking(|| -> color_eyre::Result<()> {
+                let mut write_headers = false;
+                if !path.exists() {
+                    File::create(&path)?;
+                    write_headers = true;
+                }
 
-        csv.serialize(repo)?;
+                let file = OpenOptions::new()
+                    .append(true)
+                    .open(path)?;
+
+                let mut writer = csv::WriterBuilder::new()
+                    .has_headers(write_headers)
+                    .from_writer(file);
+
+                for (_, repo) in repos {
+                    writer.serialize(repo)?;
+                }
+
+                Ok(())
+            });
+        }
+
+        js.join_all().await.into_iter().collect::<Result<(), _>>()?;
+
+        // Clear the map
+        repos.iter_mut().for_each(|(_, m)| m.clear());
+
+        Ok(())
+    }
+
+    pub async fn set_last_id(&self, forge: Forge, n: usize) -> color_eyre::Result<()> {
+        self.state_cache.0[forge].store(n, std::sync::atomic::Ordering::SeqCst);
+
+        self.store_csv().await?;
+        self.store_state_cache().await?;
 
         Ok(())
     }
+
+    pub async fn store_repo(&self, forge: Forge, repo: Repo) {
+        let mut repos_state = self.repos_state.lock().await;
+        repos_state[forge].insert(repo.name.clone(), repo);
+    }
 }
+