dingtalk-workspace-cli/scripts/dev/catalog_codegen_probe.py

#!/usr/bin/env python3
"""One-shot feasibility probe: render one product's catalog shard as Go literals.

Answers whether the shortcut storage model (compiled-in struct literals, zero
runtime parse) is viable for the 845-tool Schema catalog, by measuring compile
time, binary size and access cost on a single product first.

Deliberately drops field_provenance: it is 50% of the bytes and no runtime
consumer reads it, so a codegen design would keep it in JSON for audit only.
Keeping it here would make the size numbers pessimistic and the design wrong.
"""
import json
import sys


def q(s: str) -> str:
    return json.dumps(s if s is not None else "", ensure_ascii=False)


def strs(v) -> str:
    if not v:
        return "nil"
    return "[]string{" + ", ".join(q(x) for x in v) + "}"


def render_param(name, p) -> str:
    fields = [
        f"Name: {q(p.get('name') or name)}",
        f"Type: {q(p.get('type'))}",
        f"Description: {q(p.get('description'))}",
        f"Property: {q(p.get('property'))}",
        f"Required: {str(bool(p.get('required'))).lower()}",
        f"CLIRequired: {str(bool(p.get('cli_required'))).lower()}",
        f"RequiredWhen: {q(p.get('required_when'))}",
        f"Format: {q(p.get('format'))}",
        f"Enum: {strs(p.get('enum'))}",
    ]
    return "{" + ", ".join(fields) + "}"


def render_tool(canonical, t) -> str:
    params = t.get("parameters") or {}
    lines = [
        f"\t{q(canonical)}: {{",
        f"\t\tName: {q(t.get('name'))},",
        f"\t\tCanonicalPath: {q(t.get('canonical_path'))},",
        f"\t\tCLIPath: {q(t.get('cli_path'))},",
        f"\t\tPrimaryCLIPath: {q(t.get('primary_cli_path'))},",
        f"\t\tProductID: {q(t.get('product_id'))},",
        f"\t\tGroup: {q(t.get('group'))},",
        f"\t\tTitle: {q(t.get('title'))},",
        f"\t\tDescription: {q(t.get('description'))},",
        f"\t\tAliases: {strs(t.get('aliases'))},",
        f"\t\tIsAlias: {str(bool(t.get('is_alias'))).lower()},",
        f"\t\tEffect: {q(t.get('effect'))},",
        f"\t\tRisk: {q(t.get('risk'))},",
        f"\t\tConfirmation: {q(t.get('confirmation'))},",
        f"\t\tIdempotency: {q(t.get('idempotency'))},",
        f"\t\tInterfaceMode: {q(t.get('interface_mode'))},",
        f"\t\tAvailability: {q(t.get('availability'))},",
        f"\t\tAgentSummary: {q(t.get('agent_summary'))},",
        f"\t\tUseWhen: {strs(t.get('use_when'))},",
        f"\t\tAvoidWhen: {strs(t.get('avoid_when'))},",
        f"\t\tExamples: {strs(t.get('examples'))},",
        f"\t\tHasParameters: {str(bool(t.get('has_parameters'))).lower()},",
        f"\t\tParameterCount: {int(t.get('parameter_count') or 0)},",
    ]
    if params:
        lines.append("\t\tParameters: map[string]Param{")
        for pname in sorted(params):
            lines.append(f"\t\t\t{q(pname)}: {render_param(pname, params[pname])},")
        lines.append("\t\t},")
    lines.append("\t},")
    return "\n".join(lines)


def main() -> int:
    src, product, out = sys.argv[1], sys.argv[2], sys.argv[3]
    tools = json.load(open(src, encoding="utf-8"))["tools"]
    body = "\n".join(render_tool(c, tools[c]) for c in sorted(tools))
    with open(out, "w", encoding="utf-8") as f:
        f.write(f"""// Copyright 2026 Alibaba Group
// Code generated by scripts/dev/catalog_codegen_probe.py; DO NOT EDIT.
//
// Feasibility probe for storing the Schema catalog as compiled-in Go literals
// instead of embedded JSON, mirroring how internal/shortcut stores its 376
// definitions. field_provenance is omitted: no runtime consumer reads it.

package catalogcodegen

// Tools holds the {product} product's leaf ToolSpecs as static data, so there is
// no decode step on the first lookup.
var Tools = map[string]Tool{{
{body}
}}
""")
    print(f"generated {out}: {len(tools)} tools")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())