chore: update user_queries dataset

Author: enitrat

python/optimizers/datasets/user_queries.json

@@ -19,12 +19,12 @@
 RESOURCE_DESCRIPTIONS = {
     DocumentSource.CAIRO_BOOK: "The Cairo Programming Language Book. Essential for core language syntax, semantics, types (felt252, structs, enums, Vec), traits, generics, control flow, memory management, writing tests, organizing a project, standard library usage, starknet interactions. Crucial for smart contract structure, storage, events, ABI, syscalls, contract deployment, interaction, L1<>L2 messaging, Starknet-specific attributes. Very important for interactions with the Starknet state and context (e.g. block, transaction) through syscalls.",
     DocumentSource.STARKNET_DOCS: "The Starknet Documentation. For the Starknet protocol, the STWO prover, architecture, APIs, syscalls, network interaction, deployment, ecosystem tools (Starkli, indexers, StarknetJS, wallets), general Starknet knowledge. This should not be included for Coding and Programming questions, but rather, only for questions about Starknet, Proving, ZK, STWO, SHARP itself.",
-    DocumentSource.STARKNET_FOUNDRY: "The Starknet Foundry Documentation. For using the Foundry toolchain: writing, compiling, testing (unit tests, integration tests), and debugging Starknet contracts.",
+    DocumentSource.STARKNET_FOUNDRY: "The Starknet Foundry Documentation. For using the Foundry toolchain: `snforge` for writing, compiling, testing (unit tests, integration tests), and debugging Starknet contracts. `sncast` for deploying and interacting with contracts to Starknet.",
     DocumentSource.CAIRO_BY_EXAMPLE: "Cairo by Example Documentation. Provides practical Cairo code snippets for specific language features or common patterns. Useful for how-to syntax questions. This should not be included for Smart Contract questions, but for all other Cairo programming questions.",
     DocumentSource.OPENZEPPELIN_DOCS: "OpenZeppelin Cairo Contracts Documentation. For using the OZ library: standard implementations (ERC20, ERC721), access control, security patterns, contract upgradeability. Crucial for building standard-compliant contracts.",
     DocumentSource.CORELIB_DOCS: "Cairo Core Library Documentation. For using the Cairo core library: basic types, stdlib functions, stdlib structs, macros, and other core concepts. Essential for Cairo programming questions.",
     DocumentSource.SCARB_DOCS: "Scarb Documentation. For using the Scarb package manager: building, compiling, generating compilation artifacts, managing dependencies, configuration of Scarb.toml.",
-    DocumentSource.STARKNET_JS: "StarknetJS Documentation. For using the StarknetJS library: interacting with Starknet contracts, StarknetJS APIs, StarknetJS examples, StarknetJS tutorials, StarknetJS guides, StarknetJS documentation.",
+    DocumentSource.STARKNET_JS: "StarknetJS Documentation. For using the StarknetJS library: interacting with Starknet contracts, (calls and transactions), deploying Starknet contracts, front-end APIs, javascript integration examples, guides, tutorials and general JS/TS documentation for starknet.",
 }
 
 # Ensure all DocumentSource variants are covered
@@ -61,7 +61,7 @@ class CairoQueryAnalysis(dspy.Signature):
     )
 
     resources: list[str] = dspy.OutputField(
-        desc="List of documentation sources. Available sources: "
+        desc="List of documentation sources. If unsure what to use or if the query is not clear, use all of the available sources. Available sources: "
         + ", ".join([f"{key.value}: {value}" for key, value in RESOURCE_DESCRIPTIONS.items()])
     )
 

python/optimizers/results/optimized_mcp_program.json

@@ -108,7 +108,7 @@ def _(dspy, os, retriever):
     with open(dataset_path, encoding="utf-8") as f:
         example_dataset = json.load(f)
 
-    data = [dspy.Example(**d).with_inputs("query") for d in example_dataset]
+    data = [dspy.Example({"query": d}).with_inputs("query") for d in example_dataset]
 
     # Take maximum 300 random values from the dataset
     random.seed(42)
@@ -253,7 +253,7 @@ def compute_overall_score_with_feedback(
             score=score,
             feedback=feedback_text,
         )
-    return RESOURCE_DESCRIPTIONS, compute_overall_score_with_feedback
+    return (compute_overall_score_with_feedback,)
 
 
 @app.cell
@@ -368,6 +368,7 @@ def _(dspy, example, loading_progr):
 def _():
     return
 
+
 @app.cell
 def _():
     return

python/optimizers/results/optimized_rag.json

@@ -18,37 +18,14 @@ class HelpOnInvalidCommand(TyperGroup):
     def get_command(self, ctx, cmd_name):  # type: ignore[override]
         cmd = super().get_command(ctx, cmd_name)
         if cmd is None:
+            # Show a friendly message and the group's help, then exit with code 2
             typer.secho(f"Error: Unknown command '{cmd_name}'.", fg=typer.colors.RED, err=True)
             typer.echo()
             typer.echo(ctx.get_help())
-            ctx.exit(2)
+            # Use Click's normal control flow to avoid rich tracebacks
+            raise click.exceptions.Exit(2)
         return cmd
 
-    def main(self, *args, **kwargs):  # type: ignore[override]
-        # Ensure we handle errors ourselves to show full help
-        kwargs.setdefault("standalone_mode", False)
-        try:
-            return super().main(*args, **kwargs)
-        except click.exceptions.UsageError as e:  # missing option, bad invocation, etc.
-            ctx = e.ctx or click.Context(self)
-            # Show the specific command help if available
-            typer.secho(f"Error: {e.format_message().strip()}", fg=typer.colors.RED, err=True)
-            typer.echo()
-            typer.echo(ctx.get_help())
-            ctx.exit(2)
-
-    def __call__(self, *args, **kwargs):  # type: ignore[override]
-        # Force standalone_mode=False so exceptions propagate
-        kwargs["standalone_mode"] = False
-        try:
-            return super().__call__(*args, **kwargs)
-        except click.exceptions.UsageError as e:
-            ctx = e.ctx or click.Context(self)
-            typer.secho(f"Error: {e.format_message().strip()}", fg=typer.colors.RED, err=True)
-            typer.echo()
-            typer.echo(ctx.get_help())
-            ctx.exit(2)
-
 
 app = typer.Typer(
     cls=HelpOnInvalidCommand,
@@ -87,16 +64,48 @@ def extract_starknet_agent(
         "--output",
         help="Where to write the extracted [{query, answer}] JSON array.",
     ),
+    query_only: bool = typer.Option(
+        False,
+        "--query",
+        help="Write only queries as a JSON array instead of [{query, answer}] objects.",
+    ),
 ) -> None:
     """Extract de-duplicated QA pairs from a Starknet Agent chat dataset."""
     pairs = extract_starknet_agent_pairs(str(input))
 
+    # De-duplicate pairs by (query, answer) while preserving order
+    seen_pairs: set[tuple[str, str]] = set()
+    unique_pairs: list[dict] = []
+    for p in pairs:
+        q = p.get("query")
+        a = p.get("answer")
+        key = (q, a)
+        if key not in seen_pairs:
+            seen_pairs.add(key)
+            unique_pairs.append(p)
+
+    if query_only:
+        # De-duplicate queries while preserving order
+        seen_queries: set[str] = set()
+        queries_only: list[str] = []
+        for p in unique_pairs:
+            q = p.get("query")
+            if q not in seen_queries:
+                seen_queries.add(q)
+                queries_only.append(q)
+        data_to_write = queries_only
+    else:
+        data_to_write = unique_pairs
+
     output = Path(output).expanduser()
     output.parent.mkdir(parents=True, exist_ok=True)
     with output.open("w", encoding="utf-8") as f:
-        json.dump(pairs, f, ensure_ascii=False, indent=2)
+        json.dump(data_to_write, f, ensure_ascii=False, indent=2)
 
-    typer.echo(f"Wrote {len(pairs)} pairs to {output}")
+    if query_only:
+        typer.echo(f"Wrote {len(data_to_write)} queries to {output}")
+    else:
+        typer.echo(f"Wrote {len(data_to_write)} pairs to {output}")
 
 
 @extract_app.command("cairo-coder")
@@ -133,6 +142,11 @@ def extract_cairo_coder(
             "`reasoning=`."
         ),
     ),
+    query_only: bool = typer.Option(
+        False,
+        "--query",
+        help="Write only queries as a JSON array instead of [{query, answer}] objects.",
+    ),
 ) -> None:
     """Extract QA pairs from a Cairo-Coder LangSmith export with `outputs.output`.
 
@@ -145,16 +159,42 @@ def extract_cairo_coder(
         str(input), only_mcp=only_mcp, only_generated_answers=only_generated_answers
     )
 
+    # De-duplicate pairs by (query, answer) while preserving order
+    seen_pairs: set[tuple[str, str]] = set()
+    unique_pairs: list[dict] = []
+    for p in pairs:
+        q = p.get("query")
+        a = p.get("answer")
+        key = (q, a)
+        if key not in seen_pairs:
+            seen_pairs.add(key)
+            unique_pairs.append(p)
+
+    if query_only:
+        # De-duplicate queries while preserving order
+        seen_queries: set[str] = set()
+        queries_only: list[str] = []
+        for p in unique_pairs:
+            q = p.get("query")
+            if q not in seen_queries:
+                seen_queries.add(q)
+                queries_only.append(q)
+        data_to_write = queries_only
+    else:
+        data_to_write = unique_pairs
+
     output = Path(output).expanduser()
     output.parent.mkdir(parents=True, exist_ok=True)
     with output.open("w", encoding="utf-8") as f:
-        json.dump(pairs, f, ensure_ascii=False, indent=2)
+        json.dump(data_to_write, f, ensure_ascii=False, indent=2)
 
     typer.echo(
         json.dumps(
             {
                 "input": str(Path(input).expanduser()),
                 "output": str(output),
+                "wrote": len(data_to_write),
+                "format": "queries" if query_only else "pairs",
                 **stats,
             },
             indent=2,