release: 1.0.5

fix

fix

ingestion-pipeline/src/lib.rs

@@ -7,14 +7,14 @@ use chrono::Utc;
 use common::storage::{
     db::SurrealDbClient,
     indexes::maybe_run_scheduled_index_rebuild,
-    types::ingestion_task::{IngestionTask, DEFAULT_LEASE_SECS},
+    types::ingestion_task::{DEFAULT_LEASE_SECS, IngestionTask},
 };
 pub use pipeline::{
-    persist_artifacts, EmbeddedKnowledgeEntity, EmbeddedTextChunk, IngestionConfig,
-    IngestionPipeline, IngestionTuning, PipelineArtifacts,
+    EmbeddedKnowledgeEntity, EmbeddedTextChunk, IngestionConfig, IngestionPipeline,
+    IngestionTuning, PipelineArtifacts, persist_artifacts,
 };
 use std::sync::Arc;
-use tokio::time::{sleep, Duration};
+use tokio::time::{Duration, sleep};
 use tracing::{error, info, warn};
 use uuid::Uuid;
 

ingestion-pipeline/src/pipeline/mod.rs

@@ -42,7 +42,7 @@ use tracing::{debug, info, warn};
 use self::{
     context::PipelineContext,
     stages::{enrich, persist, prepare_content, retrieve_related},
-    state::{ready, Enriched, IngestionMachine},
+    state::{Enriched, IngestionMachine, ready},
 };
 
 /// Wall-clock duration of each pre-persistence pipeline stage.
@@ -355,10 +355,10 @@ mod finalize_tests {
     use tokio::time::sleep;
 
     use super::{
+        IngestionPipeline, PipelineServices,
         config::IngestionTuning,
         test_support::setup_db,
-        tests::{pipeline_config, reserve_task, MockServices},
-        IngestionPipeline, PipelineServices,
+        tests::{MockServices, pipeline_config, reserve_task},
     };
 
     #[tokio::test]

ingestion-pipeline/src/pipeline/persistence.rs

@@ -12,14 +12,13 @@ use common::{
     storage::{
         db::SurrealDbClient,
         types::{
-            knowledge_entity::KnowledgeEntity,
+            EmbeddingRecord, StoredObject, knowledge_entity::KnowledgeEntity,
             knowledge_entity_embedding::KnowledgeEntityEmbedding, text_chunk::TextChunk,
-            text_chunk_embedding::TextChunkEmbedding, text_content::TextContent, EmbeddingRecord,
-            StoredObject,
+            text_chunk_embedding::TextChunkEmbedding, text_content::TextContent,
         },
     },
 };
-use tokio::time::{sleep, Duration};
+use tokio::time::{Duration, sleep};
 use tracing::warn;
 
 use super::{
@@ -268,8 +267,8 @@ mod tests {
 
     use super::*;
     use crate::pipeline::test_support::{
-        self, count_chunks_for_source, count_entities_for_source, count_relationships_for_source,
-        large_artifacts, persist, sample_artifacts, setup_db, TEST_EMBEDDING_DIM,
+        self, TEST_EMBEDDING_DIM, count_chunks_for_source, count_entities_for_source,
+        count_relationships_for_source, large_artifacts, persist, sample_artifacts, setup_db,
     };
 
     #[tokio::test]

ingestion-pipeline/src/pipeline/services.rs

@@ -15,14 +15,14 @@ use common::{
         db::SurrealDbClient,
         store::StorageManager,
         types::{
-            ingestion_payload::IngestionPayload, knowledge_relationship::KnowledgeRelationship,
-            system_settings::SystemSettings, text_chunk::TextChunk, text_content::TextContent,
-            StoredObject,
+            StoredObject, ingestion_payload::IngestionPayload,
+            knowledge_relationship::KnowledgeRelationship, system_settings::SystemSettings,
+            text_chunk::TextChunk, text_content::TextContent,
         },
     },
     utils::{config::AppConfig, embedding::EmbeddingProvider},
 };
-use retrieval_pipeline::{reranking::RerankerPool, retrieved_entities_to_json, RetrievedEntity};
+use retrieval_pipeline::{RetrievedEntity, reranking::RerankerPool, retrieved_entities_to_json};
 use text_splitter::{ChunkCapacity, ChunkConfig, TextSplitter};
 
 use super::{enrichment_result::LLMEnrichmentResult, preparation::to_text_content};
@@ -358,7 +358,7 @@ mod tests {
     use std::sync::Arc;
 
     use anyhow::Context;
-    use async_openai::{config::OpenAIConfig, types::chat::ChatCompletionRequestMessage, Client};
+    use async_openai::{Client, config::OpenAIConfig, types::chat::ChatCompletionRequestMessage};
     use common::{
         storage::{
             db::SurrealDbClient, store::StorageManager, types::system_settings::SystemSettingsPatch,

ingestion-pipeline/src/pipeline/tests.rs

@@ -1,6 +1,7 @@
 use std::sync::Arc;
 
 use super::{
+    IngestionPipeline,
     config::{IngestionConfig, IngestionTuning},
     enrichment_result::LLMEnrichmentResult,
     services::PipelineServices,
@@ -8,7 +9,6 @@ use super::{
         count_chunks_for_source, count_entities_for_source, count_relationships_for_source,
         persist, sample_artifacts, setup_db,
     },
-    IngestionPipeline,
 };
 use crate::pipeline::context::{EmbeddedKnowledgeEntity, EmbeddedTextChunk};
 use anyhow::{self, Context};
@@ -405,9 +405,11 @@ async fn ingestion_pipeline_happy_path_persists_artifacts() -> anyhow::Result<()
         call_log.get(0..4),
         Some(&["prepare", "retrieve", "enrich", "convert"][..])
     );
-    assert!(call_log
-        .get(4..)
-        .is_some_and(|tail| tail.iter().all(|entry| *entry == "chunk")));
+    assert!(
+        call_log
+            .get(4..)
+            .is_some_and(|tail| tail.iter().all(|entry| *entry == "chunk"))
+    );
     Ok(())
 }
 

ingestion-pipeline/src/utils/file_text_extraction.rs

@@ -38,11 +38,11 @@ async fn materialize_temp_file(
     let mut path = env::temp_dir();
     let mut file_name = format!("minne-ingest-{}", Uuid::new_v4());
 
-    if let Some(ext) = extension {
-        if !ext.is_empty() {
-            file_name.push('.');
-            file_name.push_str(ext);
-        }
+    if let Some(ext) = extension
+        && !ext.is_empty()
+    {
+        file_name.push('.');
+        file_name.push_str(ext);
     }
 
     path.push(file_name);
@@ -142,7 +142,7 @@ pub async fn extract_text_from_file(
 #[cfg(test)]
 mod tests {
     use super::*;
-    use async_openai::{config::OpenAIConfig, Client};
+    use async_openai::{Client, config::OpenAIConfig};
     use bytes::Bytes;
     use chrono::Utc;
     use common::{

ingestion-pipeline/src/utils/image_parsing.rs

@@ -3,7 +3,7 @@ use async_openai::types::chat::{
     ChatCompletionRequestMessageContentPartTextArgs, ChatCompletionRequestUserMessageArgs,
     CreateChatCompletionRequestArgs, ImageDetail, ImageUrlArgs,
 };
-use base64::{engine::general_purpose::STANDARD, Engine as _};
+use base64::{Engine as _, engine::general_purpose::STANDARD};
 use common::{
     error::AppError,
     storage::{db::SurrealDbClient, types::system_settings::SystemSettings},

ingestion-pipeline/src/utils/pdf/render.rs

@@ -148,20 +148,27 @@ async fn maybe_dump_debug_image(page_index: u32, bytes: &[u8]) -> Result<(), App
 mod tests {
     use super::*;
     use anyhow::{self};
-    use lopdf::dictionary;
     use lopdf::Object;
+    use lopdf::dictionary;
 
     #[test]
     fn test_debug_dump_directory_env_var() -> anyhow::Result<()> {
-        std::env::remove_var(DEBUG_IMAGE_ENV_VAR);
+        // SAFETY: test runs serially; env is restored before return.
+        unsafe {
+            std::env::remove_var(DEBUG_IMAGE_ENV_VAR);
+        }
         assert!(debug_dump_directory().is_none());
 
-        std::env::set_var(DEBUG_IMAGE_ENV_VAR, "/tmp/minne_pdf_debug");
+        unsafe {
+            std::env::set_var(DEBUG_IMAGE_ENV_VAR, "/tmp/minne_pdf_debug");
+        }
         let dir =
             debug_dump_directory().ok_or_else(|| anyhow::anyhow!("expected debug directory"))?;
         assert_eq!(dir, PathBuf::from("/tmp/minne_pdf_debug"));
 
-        std::env::remove_var(DEBUG_IMAGE_ENV_VAR);
+        unsafe {
+            std::env::remove_var(DEBUG_IMAGE_ENV_VAR);
+        }
         Ok(())
     }
 

ingestion-pipeline/src/utils/pdf/vision.rs

@@ -5,7 +5,7 @@ use async_openai::types::chat::{
     ChatCompletionRequestMessageContentPartTextArgs, ChatCompletionRequestUserMessageArgs,
     CreateChatCompletionRequest, CreateChatCompletionRequestArgs, ImageDetail, ImageUrlArgs,
 };
-use base64::{engine::general_purpose::STANDARD, Engine as _};
+use base64::{Engine as _, engine::general_purpose::STANDARD};
 use tracing::{debug, warn};
 
 use common::{