diff options
author | Keuin <[email protected]> | 2024-02-24 02:46:28 +0800 |
---|---|---|
committer | Keuin <[email protected]> | 2024-02-24 03:06:16 +0800 |
commit | e670f5389171a7a124d990fe2093e1f9290eccfc (patch) | |
tree | 9c9b313db57159f019b2d4f5770ab7a2e1085fe5 /src/main.rs |
Initial commit
Diffstat (limited to 'src/main.rs')
-rw-r--r-- | src/main.rs | 301 |
1 files changed, 301 insertions, 0 deletions
diff --git a/src/main.rs b/src/main.rs new file mode 100644 index 0000000..3cb93db --- /dev/null +++ b/src/main.rs @@ -0,0 +1,301 @@ +use std::collections::{BTreeMap, HashMap, HashSet}; +use std::fmt::Debug; +use std::hash::Hash; +use std::sync::Arc; + +use anyhow::{anyhow, Error}; +use axum::{Json, Router, routing::get}; +use axum::extract::{Query, State}; +use futures_util::TryStreamExt; +use serde_derive::{Deserialize, Serialize}; +use sqlx::{Column, Row, SqlitePool}; +use tracing::{debug, info}; +use tracing_subscriber::EnvFilter; + +mod config; + +#[derive(Clone)] +struct AppState { + pub config: config::Root, + pub connections: Vec<RelationReader>, + pub field_relations: HashMap<String, Vec<RelationReader>>, + pub fields: HashMap<String, config::Field>, +} + +#[derive(Clone)] +struct RelationReader { + cfg: config::Relation, + db: SqlitePool, +} + +impl RelationReader { + pub async fn from_config(cfg: &config::Relation) -> Result<RelationReader, Error> { + if cfg.fields.is_empty() { + return Err(anyhow!("relation does not have any field")); + } + match SqlitePool::connect(&cfg.connect).await { + Ok(p) => Ok(RelationReader { + cfg: cfg.clone(), + db: p, + }), + Err(why) => { + Err(anyhow!("failed to connect to sqlite database `{}` for relation {}: {}", + &cfg.connect, &cfg.name, why)) + } + } + } +} + +impl RelationReader { + pub async fn query(&mut self, field: &String, value: &String) -> Result<HashMap<String, String>, Error> { + let fields = self.cfg.fields.iter().map(|r| { + return format!("({}) as {}", r.query, r.id); + }).reduce(|acc, s| acc + "," + &*s) + .expect("relation does not have fields"); + let sql = format!( + "select {fields} from {table_name} where {field} = ?", + table_name = &self.cfg.table_name, + ); + debug!("SQL: {}", sql); + let mut rows = sqlx::query(&*sql).bind(value).fetch(&self.db); + let mut rr: HashMap<String, String> = HashMap::new(); + while let Some(row) = rows.try_next().await? { + for c in row.columns() { + let name = String::from(c.name()); + let value: String = match row.try_get(&*name) { + Ok(v) => v, + Err(why) => { + return Err(anyhow!("failed to get field `{}` when querying relation {}: {}", + field, self.cfg.name, why)); + } + }; + rr.insert(name, value); + } + } + Ok(rr) + } +} + +#[tokio::main] +async fn main() { + tracing_subscriber::fmt() + .with_env_filter(EnvFilter::from_default_env()) + .init(); + debug!("debug log is enabled"); + + let cfg = config::read_file("config.toml") + .expect("error loading config file"); + + let field_names = get_fields(&cfg).expect("error reading fields"); + + let mut all_relation_names = HashSet::new(); + for r in &cfg.relations { + if !all_relation_names.insert(&r.name) { + panic!("duplicate relation name `{r}`", r = r.name); + } + for f in &r.fields { + if !field_names.contains(&f.id) { + panic!("undeclared field `{}` used in relation `{}`, you have to declare it in `fields`", + f.id, r.name); + } + } + } + + + let mut relations = Vec::new(); + for r in &cfg.relations { + relations.push(match RelationReader::from_config(r).await { + Ok(v) => v, + Err(why) => panic!("error loading relation `{}`: {}", r.name, why), + }); + } + + let mut field_relations: HashMap<String, Vec<RelationReader>> = HashMap::new(); + for r in &relations { + for f in &r.cfg.fields { + if !field_relations.contains_key(&f.id) { + field_relations.insert(f.id.clone(), Vec::new()); + } + field_relations.get_mut(&f.id).unwrap().push(r.clone()); + } + } + + let fields: HashMap<String, config::Field> = cfg.fields.iter().map(|f| (f.id.clone(), f.clone())).collect(); + + let endpoint = cfg.listen.clone(); + + let state = Arc::new(AppState { + config: cfg, + connections: relations, + field_relations, + fields, + }); + + // build our application with a single route + let app = Router::new() + .route("/query", get(query)) + .with_state(state); + + // run our app with hyper, listening globally on port 3000 + info!("starting HTTP server on {}", endpoint); + let listener = tokio::net::TcpListener::bind(endpoint).await + .expect("error binding tcp socket"); + axum::serve(listener, app).await + .expect("error running http server"); +} + +#[derive(Eq, PartialEq, Clone)] +#[derive(Hash)] +struct RelationFieldValue { + relation: String, + field: String, + value: String, +} + +#[derive(Default, Debug, Clone, PartialEq, Serialize, Deserialize)] +#[serde(rename_all = "snake_case")] +pub struct QueryResponse { + pub success: bool, + pub message: String, + pub data: BTreeMap<String, Vec<String>>, +} + + +async fn query( + State(state): State<Arc<AppState>>, + Query(args): Query<HashMap<String, String>>, +) -> Json<QueryResponse> { + // Input: known field values + // Output: all reachable field values + // + // Program: + // populate unvisited field value set with separate initial conditions + // for all (field, value) in unvisited field value set: + // for all relation containing this field: + // mark (relation, field, value) as visited + // get all other field values from relation with field, for each (field2, value2), + // add all unvisited combinations (relation, field2, value2) to the unvisited field value set + // stop if the unvisited set is empty, or iteration count exceeds limit + + let mut unvisited = HashSet::new(); + let mut visited = HashSet::new(); + let known_field_values = args; + // for the first query, we use even non-distinct fields as query condition + for (field, value) in known_field_values { + let relations = match state.field_relations.get(&field) { + Some(v) => v, + None => { + return Json(QueryResponse { + success: false, + message: format!("no relation has field `{}`", &field), + data: Default::default(), + }); + } + }; + for r in relations { + unvisited.insert(RelationFieldValue { + relation: r.cfg.name.clone(), + field: field.clone(), + value: value.clone(), + }); + } + } + // TODO make this configurable + const MAX_DEPTH: i32 = 10; + let mut depth = 0; + let mut depth_limit_exceeded = false; + let mut all_result: HashMap<_, HashSet<String>> = HashMap::new(); + while !unvisited.is_empty() { + if depth > MAX_DEPTH { + depth_limit_exceeded = true; + break; + } + // visit a cloned snapshot, updates will be reflected at once in the next loop round + for task in unvisited.clone() { + let (field, value) = (&task.field, &task.value); + let mut relations = match state.field_relations.get(field) { + Some(v) => (*v).clone(), + None => { + return Json(QueryResponse { + success: false, + message: format!("no relation has field `{}`", field), + data: Default::default(), + }); + } + }; + for rel in relations.iter_mut() { + info!("visit: relation {}, field {}, value {}", rel.cfg.name, field, value); + // ensure every (relation, field, value) is visited only once + if !visited.insert(RelationFieldValue { + relation: rel.cfg.name.clone(), + field: field.clone(), + value: value.clone(), + }) { + continue; + } + let result = match rel.query(field, value).await { + Ok(v) => v, + Err(why) => { + return Json(QueryResponse { + success: false, + message: format!("failed to query relation `{r}` with field `{f}`, value `{v}`: {why}", + r = &rel.cfg.name, f = field, v = value), + data: Default::default(), + }); + } + }; + for (field, value) in result { + let v = RelationFieldValue { + relation: rel.cfg.name.clone(), + field: field.clone(), + value: value.clone(), + }; + // skip non-distinct fields to prevent generating irrelevant results + if !state.fields.get(&field).expect("missing field info").distinct { + continue; + } + if visited.contains(&v) { + // don't re-add visited values + continue; + } + unvisited.insert(v); + if let Some(set) = all_result.get_mut(&field) { + set.insert(value); + } else { + let mut s = HashSet::new(); + s.insert(value); + all_result.insert(field, s); + } + } + unvisited.remove(&RelationFieldValue { + relation: rel.cfg.name.clone(), + field: field.clone(), + value: value.clone(), + }); + } + } + depth += 1; + } + // use BTree map to keep the result ordered + let all_result: BTreeMap<String, Vec<String>> = all_result.iter().map(|(k, v)| { + let mut v: Vec<String> = v.iter().map(|v| v.clone()).collect(); + v.sort(); + (k.clone(), v) + }).collect(); + Json(QueryResponse { + success: true, + message: String::from(if depth_limit_exceeded { "depth length limit exceeded" } else { "" }), + data: all_result, + }) +} + + +fn get_fields(cfg: &config::Root) -> Result<HashSet<String>, anyhow::Error> { + let mut s: HashSet<String> = HashSet::new(); + for field in &cfg.fields { + if !s.insert(field.id.clone()) { + return Err(anyhow!("duplicate field `{:?}`", &field.id)); + } + } + Ok(s) +}
\ No newline at end of file |