Files
poimen-memory/crates/mem-core/src/query.rs
T

250 lines
7.7 KiB
Rust
Raw Normal View History

use crate::domain::{ProjectId, QueryId};
use anyhow::{anyhow, Result};
use serde::{Deserialize, Serialize};
use std::collections::HashMap;
use std::path::Path;
/// A single standing query.
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
pub struct Query {
pub id: String,
pub question: String,
#[serde(default)]
pub exit_gate: bool,
}
/// Synthesis query (optional, for L2).
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
pub struct SynthesisQuery {
pub question: String,
#[serde(default)]
pub exit_gate: bool,
}
/// Defaults applied to queries.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct Defaults {
#[serde(default = "default_memory_budget")]
pub memory_budget: u32,
#[serde(default = "default_chunk_tokens")]
pub chunk_tokens: u32,
#[serde(default)]
pub exit_gate: bool,
}
fn default_memory_budget() -> u32 {
1024
}
fn default_chunk_tokens() -> u32 {
5000
}
impl Default for Defaults {
fn default() -> Self {
Self {
memory_budget: default_memory_budget(),
chunk_tokens: default_chunk_tokens(),
exit_gate: false,
}
}
}
/// Complete set of queries for a project.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct QuerySet {
pub project: String,
pub roots: Vec<String>,
pub sources: Vec<String>,
pub queries: Vec<Query>,
#[serde(default)]
pub synthesis: Option<SynthesisQuery>,
#[serde(default)]
pub defaults: Defaults,
}
/// Load error with context.
#[derive(Debug, Clone)]
pub struct QueryLoadError {
pub file: String,
pub query_id: Option<String>,
pub field: Option<String>,
pub message: String,
}
impl std::fmt::Display for QueryLoadError {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match (&self.query_id, &self.field) {
(Some(id), Some(field)) => {
write!(f, "{}: query '{}', field '{}': {}", self.file, id, field, self.message)
}
(Some(id), None) => {
write!(f, "{}: query '{}': {}", self.file, id, self.message)
}
(None, Some(field)) => {
write!(f, "{}: field '{}': {}", self.file, field, self.message)
}
(None, None) => {
write!(f, "{}: {}", self.file, self.message)
}
}
}
}
impl std::error::Error for QueryLoadError {}
/// Valid charset for query ids: lowercase, digits, hyphens only.
fn is_valid_query_id(id: &str) -> bool {
!id.is_empty() && id.chars().all(|c| c.is_ascii_lowercase() || c.is_ascii_digit() || c == '-')
}
impl QuerySet {
/// Load and validate a query set from a YAML file.
pub fn load<P: AsRef<Path>>(path: P) -> Result<Self> {
let path = path.as_ref();
let filename = path.to_string_lossy().to_string();
let contents = std::fs::read_to_string(path)?;
// Parse YAML
let mut set: QuerySet = serde_yaml::from_str(&contents)
.map_err(|e| anyhow!("Failed to parse {}: {}", filename, e))?;
// Validate project
if set.project.trim().is_empty() {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("project".to_string()),
message: "project field is required and cannot be empty".to_string(),
}));
}
// Validate at least one query
if set.queries.is_empty() {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("queries".to_string()),
message: "at least one query is required".to_string(),
}));
}
// Validate each query
let mut seen_ids = std::collections::HashSet::new();
for query in &mut set.queries {
// Check ID is not empty
if query.id.trim().is_empty() {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("id".to_string()),
message: "query id cannot be empty".to_string(),
}));
}
// Check ID charset
if !is_valid_query_id(&query.id) {
return Err(anyhow!(QueryLoadError {
file: filename.clone(),
query_id: Some(query.id.clone()),
field: Some("id".to_string()),
message: format!(
"query id '{}' must match [a-z0-9-]+ (it becomes a filename)",
query.id
),
}));
}
// Check for duplicate IDs
if seen_ids.contains(&query.id) {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: Some(query.id.clone()),
field: Some("id".to_string()),
message: format!("duplicate query id '{}'", query.id),
}));
}
seen_ids.insert(query.id.clone());
// Check question is not empty
if query.question.trim().is_empty() {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: Some(query.id.clone()),
field: Some("question".to_string()),
message: "question cannot be empty".to_string(),
}));
}
// Apply defaults if exit_gate not set
// (defaults already applied via serde default)
}
// Validate synthesis if present
if let Some(ref synthesis) = set.synthesis {
if synthesis.question.trim().is_empty() {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("synthesis.question".to_string()),
message: "synthesis question cannot be empty".to_string(),
}));
}
}
// Validate defaults
if set.defaults.memory_budget == 0 {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("defaults.memory_budget".to_string()),
message: "memory_budget must be greater than 0".to_string(),
}));
}
if set.defaults.chunk_tokens == 0 {
return Err(anyhow!(QueryLoadError {
file: filename,
query_id: None,
field: Some("defaults.chunk_tokens".to_string()),
message: "chunk_tokens must be greater than 0".to_string(),
}));
}
Ok(set)
}
/// Get a query by ID.
pub fn query(&self, id: &str) -> Option<&Query> {
self.queries.iter().find(|q| q.id == id)
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_valid_query_id() {
assert!(is_valid_query_id("architecture-decisions"));
assert!(is_valid_query_id("infra-root-causes"));
assert!(is_valid_query_id("id123"));
assert!(is_valid_query_id("a"));
assert!(is_valid_query_id("a-b-c-123"));
assert!(!is_valid_query_id(""));
assert!(!is_valid_query_id("infra/root-causes"));
assert!(!is_valid_query_id("UPPERCASE"));
assert!(!is_valid_query_id("with space"));
assert!(!is_valid_query_id("with_underscore"));
}
#[test]
fn test_defaults() {
let defaults = Defaults::default();
assert_eq!(defaults.memory_budget, 1024);
assert_eq!(defaults.chunk_tokens, 5000);
assert!(!defaults.exit_gate);
}
}