Struct ProjectData 

pub struct ProjectData {
    pub paired: C4PairedProjectData,
}

Legacy ProjectData — a thin wrapper around knowdit_project::C4PairedProjectData.

All LLM-driven methods (categorize / extract / merge / link / is_completed) live in [crate::learn::impl ProjectData] and project this wrapper down to the inner fields via the accessor methods. Per-call tunables (chunk-input budget, agent step caps, merge chunking) are passed in as method parameters, not stored on this struct.

Fields

paired: C4PairedProjectData

The project + optional audit-report payload. Owned so this wrapper can be cloned cheaply and threaded into async tasks.

Implementations

impl ProjectData

pub async fn categorize_and_extract( &self, llm: &LLM, agent_options: &AgentRunOptions, chunk_input_budget: Option<usize>, ) -> Result<ExtractResult>

Phase 1: Categorize the project and extract semantics. Safe to run concurrently across multiple projects.

pub async fn categorize_and_extract_semantics( &self, llm: &LLM, agent_options: &AgentRunOptions, chunk_input_budget: Option<usize>, ) -> Result<ExtractResult>

Categorize the project and extract only project semantics. This is used by consumers that need semantic context but do not need audit findings.

pub async fn merge_and_write( &self, db: &HistoricalDatabase, llm: &LLM, extract: &ExtractResult, agent_options: &AgentRunOptions, merge_chunking: MergeChunkingOptions, ) -> Result<()>

Phase 2: Merge extracted semantics with existing KB and write to DB. MUST be run serially (one project at a time) to avoid merge conflicts. Commit one project’s full merge output to the historical KG.

Begins its own transaction via HistoricalDatabase::write_project_completed under the hood and discards the new-canonical id list. Use this when the caller doesn’t need to compose additional writes (pending_semantic enqueue etc.) inside the same transaction. Bulk learn paths (learn moves / learn c4 / learn projects) go through here because they never run retro-link.

pub async fn merge_and_write_txn( &self, conn: &DatabaseTransaction, db: &HistoricalDatabase, llm: &LLM, extract: &ExtractResult, agent_options: &AgentRunOptions, merge_chunking: MergeChunkingOptions, ) -> Result<Vec<i32>>

Transaction-scoped variant of Self::merge_and_write. Performs the merge-LLM passes (against the LIVE DB — these are reads, not writes, so don’t depend on the txn) and then writes the resulting rows through HistoricalDatabase::write_project_completed_txn using the supplied transaction. Returns the canonical semantic ids this project newly introduced, so the caller can chain HistoricalDatabase::enqueue_pending_canonical_semantics_txn in the same transaction when needed (incremental workflow learn flow).

pub async fn is_completed(&self, db: &HistoricalDatabase) -> Result<bool>

Check if this project is already completed in the DB.

pub async fn extract_semantics( &self, llm: &LLM, categories: &[DeFiCategory], agent_options: &AgentRunOptions, chunk_input_budget: Option<usize>, ) -> Result<Vec<ExtractedSemantic>>

Extract semantics from the project’s source files. Splits the source text into chunks that fit the context window and runs one Agent per chunk; each chunk’s semantics are emitted via emit_semantic tool calls and the chunk is closed with finalize_semantic_extraction.

impl ProjectData

pub async fn from_path_spec(spec: &str) -> Result<Self>

Parse name:path / name:path:platform_id and load the project as Solidity (no language auto-detection).

pub async fn from_source_dir_spec(spec: &str) -> Result<Self>

Parse name:path / name:path:platform_id and load the project, auto-detecting Solidity vs Move.

pub async fn from_dir( name: &str, root_dir: &Path, platform_id: Option<&str>, ) -> Result<Self>

Solidity-only loader over a known directory.

pub async fn from_c4(dataset_dir: &Path, contest_id: u32) -> Result<Self>

Load a Code4rena contest: (audit_meta + source contracts + markdown report). The wrapper preserves the audit material inside the underlying C4PairedProjectData.audit so the learn-pipeline’s prompt builders pick it up automatically.

pub async fn from_move_snapshot( name: &str, root_dir: &Path, commit_hash: &str, audit_report: Option<MoveVulnerabilitySnippet>, ) -> Result<Self>

Load a Move snapshot, optionally paired with its vulnerability-snippet audit report.

pub fn from_project_view(view: &ProjectData) -> Self

Adapt a knowdit_project::ProjectData view (no audit) into this legacy shape. Per-call tunables like extract_chunk_input_budget are no longer carried on the project — they flow as method parameters into the learn pipeline.

pub fn name(&self) -> &str

pub fn platform_id(&self) -> Option<&str>

pub fn root_dir(&self) -> &Path

pub fn source_language(&self) -> SourceLanguage

pub fn source_files(&self) -> &[SourceFile]

pub fn audit_report(&self) -> Option<&AuditReportMaterial>

pub fn display_id(&self) -> String

Display-friendly identifier: platform_id when set, otherwise the project name. Returns an owned String to keep call-site ergonomics stable — earlier versions had this shape and a number of format!() sites still expect it.

Trait Implementations

impl Clone for ProjectData

fn clone(&self) -> ProjectData

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ProjectData

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.