Medios-Macina/TUI/pipeline_runner.py

"""Pipeline execution utilities for the Textual UI.

The TUI is a frontend to the CLI, so it must use the same pipeline executor
implementation as the CLI (`CLI.PipelineExecutor`).
"""

from __future__ import annotations

import contextlib
import io
import shlex
import sys
from pathlib import Path
from dataclasses import dataclass, field
from typing import Any, Callable, Dict, List, Optional, Sequence

BASE_DIR = Path(__file__).resolve().parent
ROOT_DIR = BASE_DIR.parent
for path in (ROOT_DIR, BASE_DIR):
    str_path = str(path)
    if str_path not in sys.path:
        sys.path.insert(0, str_path)

import pipeline as ctx
from CLI import ConfigLoader, PipelineExecutor as CLIPipelineExecutor, WorkerManagerRegistry
from SYS.logger import set_debug
from rich_display import capture_rich_output
from result_table import ResultTable


@dataclass(slots=True)
class PipelineStageResult:
    """Summary for a single pipeline stage."""

    name: str
    args: Sequence[str]
    emitted: List[Any] = field(default_factory=list)
    result_table: Optional[Any] = None  # ResultTable object if available
    status: str = "pending"
    error: Optional[str] = None


@dataclass(slots=True)
class PipelineRunResult:
    """Aggregate result for a pipeline run."""

    pipeline: str
    success: bool
    stages: List[PipelineStageResult] = field(default_factory=list)
    emitted: List[Any] = field(default_factory=list)
    result_table: Optional[Any] = None  # Final ResultTable object if available
    stdout: str = ""
    stderr: str = ""
    error: Optional[str] = None

    def to_summary(self) -> Dict[str, Any]:
        """Provide a JSON-friendly representation for logging or UI."""
        return {
            "pipeline":
            self.pipeline,
            "success":
            self.success,
            "error":
            self.error,
            "stages": [
                {
                    "name": stage.name,
                    "status": stage.status,
                    "error": stage.error,
                    "emitted": len(stage.emitted),
                } for stage in self.stages
            ],
        }


class PipelineRunner:
    """TUI wrapper that delegates to the canonical CLI pipeline executor."""

    def __init__(self) -> None:
        self._config_loader = ConfigLoader(root=ROOT_DIR)
        self._executor = CLIPipelineExecutor(config_loader=self._config_loader)
        self._worker_manager = None

    @property
    def worker_manager(self):
        return self._worker_manager

    def run_pipeline(
        self,
        pipeline_text: str,
        *,
        seeds: Optional[Any] = None,
        isolate: bool = False,
        on_log: Optional[Callable[[str],
                                  None]] = None,
    ) -> PipelineRunResult:
        snapshot: Optional[Dict[str, Any]] = None
        if isolate:
            snapshot = self._snapshot_ctx_state()

        normalized = str(pipeline_text or "").strip()
        result = PipelineRunResult(pipeline=normalized, success=False)
        if not normalized:
            result.error = "Pipeline is empty"
            return result

        try:
            from SYS.cli_syntax import validate_pipeline_text

            syntax_error = validate_pipeline_text(normalized)
            if syntax_error:
                result.error = syntax_error.message
                result.stderr = syntax_error.message
                return result
        except Exception:
            pass

        try:
            tokens = shlex.split(normalized)
        except Exception as exc:
            result.error = f"Syntax error: {exc}"
            result.stderr = result.error
            return result

        if not tokens:
            result.error = "Pipeline contains no tokens"
            return result

        config = self._config_loader.load()
        try:
            set_debug(bool(config.get("debug", False)))
        except Exception:
            pass

        try:
            self._worker_manager = WorkerManagerRegistry.ensure(config)
        except Exception:
            self._worker_manager = None

        ctx.reset()
        ctx.set_current_command_text(normalized)

        if seeds is not None:
            try:
                if not isinstance(seeds, list):
                    seeds = [seeds]
                ctx.set_last_result_items_only(list(seeds))
            except Exception:
                pass

        stdout_buffer = io.StringIO()
        stderr_buffer = io.StringIO()

        try:
            with capture_rich_output(stdout=stdout_buffer, stderr=stderr_buffer):
                with (
                        contextlib.redirect_stdout(stdout_buffer),
                        contextlib.redirect_stderr(stderr_buffer),
                ):
                    if on_log:
                        on_log("Executing pipeline via CLI executor...")
                    self._executor.execute_tokens(list(tokens))
        except Exception as exc:
            result.error = f"{type(exc).__name__}: {exc}"
        finally:
            try:
                ctx.clear_current_command_text()
            except Exception:
                pass
            result.stdout = stdout_buffer.getvalue()
            result.stderr = stderr_buffer.getvalue()

        # Pull the canonical state out of pipeline context.
        table = None
        try:
            table = (
                ctx.get_display_table() or ctx.get_current_stage_table()
                or ctx.get_last_result_table()
            )
        except Exception:
            table = None

        items: List[Any] = []
        try:
            items = list(ctx.get_last_result_items() or [])
        except Exception:
            items = []

        if table is None and items:
            try:
                synth = ResultTable("Results")
                for item in items:
                    synth.add_result(item)
                table = synth
            except Exception:
                table = None

        result.emitted = items
        result.result_table = table

        combined = (result.stdout + "\n" + result.stderr).strip().lower()
        failure_markers = (
            "unknown command:",
            "pipeline order error:",
            "invalid selection:",
            "invalid pipeline syntax",
            "failed to execute pipeline",
            "[error]",
        )
        if result.error:
            result.success = False
        elif any(m in combined for m in failure_markers):
            result.success = False
            if not result.error:
                result.error = "Pipeline failed"
        else:
            result.success = True

        if isolate and snapshot is not None:
            try:
                self._restore_ctx_state(snapshot)
            except Exception:
                # Best-effort; isolation should never break normal operation.
                pass

        return result

    @staticmethod
    def _snapshot_ctx_state() -> Dict[str, Any]:
        """Best-effort snapshot of pipeline context so TUI popups don't clobber UI state."""

        def _copy(val: Any) -> Any:
            if isinstance(val, list):
                return val.copy()
            if isinstance(val, dict):
                return val.copy()
            return val

        snap: Dict[str,
                   Any] = {}
        keys = [
            "_LIVE_PROGRESS",
            "_CURRENT_CONTEXT",
            "_LAST_SEARCH_QUERY",
            "_PIPELINE_REFRESHED",
            "_PIPELINE_LAST_ITEMS",
            "_LAST_RESULT_TABLE",
            "_LAST_RESULT_ITEMS",
            "_LAST_RESULT_SUBJECT",
            "_RESULT_TABLE_HISTORY",
            "_RESULT_TABLE_FORWARD",
            "_CURRENT_STAGE_TABLE",
            "_DISPLAY_ITEMS",
            "_DISPLAY_TABLE",
            "_DISPLAY_SUBJECT",
            "_PIPELINE_LAST_SELECTION",
            "_PIPELINE_COMMAND_TEXT",
            "_CURRENT_CMDLET_NAME",
            "_CURRENT_STAGE_TEXT",
            "_PIPELINE_VALUES",
            "_PENDING_PIPELINE_TAIL",
            "_PENDING_PIPELINE_SOURCE",
            "_UI_LIBRARY_REFRESH_CALLBACK",
        ]

        for k in keys:
            snap[k] = _copy(getattr(ctx, k, None))

        # Deepen copies where nested lists are common.
        try:
            hist = list(getattr(ctx, "_RESULT_TABLE_HISTORY", []) or [])
            snap["_RESULT_TABLE_HISTORY"] = [
                (
                    t,
                    (
                        items.copy()
                        if isinstance(items,
                                      list) else list(items) if items else []
                    ),
                    subj,
                ) for (t, items, subj) in hist if isinstance((t, items, subj), tuple)
            ]
        except Exception:
            pass

        try:
            fwd = list(getattr(ctx, "_RESULT_TABLE_FORWARD", []) or [])
            snap["_RESULT_TABLE_FORWARD"] = [
                (
                    t,
                    (
                        items.copy()
                        if isinstance(items,
                                      list) else list(items) if items else []
                    ),
                    subj,
                ) for (t, items, subj) in fwd if isinstance((t, items, subj), tuple)
            ]
        except Exception:
            pass

        try:
            tail = list(getattr(ctx, "_PENDING_PIPELINE_TAIL", []) or [])
            snap["_PENDING_PIPELINE_TAIL"] = [
                list(stage) for stage in tail if isinstance(stage, list)
            ]
        except Exception:
            pass

        try:
            values = getattr(ctx, "_PIPELINE_VALUES", None)
            if isinstance(values, dict):
                snap["_PIPELINE_VALUES"] = values.copy()
        except Exception:
            pass

        return snap

    @staticmethod
    def _restore_ctx_state(snapshot: Dict[str, Any]) -> None:
        for k, v in (snapshot or {}).items():
            try:
                setattr(ctx, k, v)
            except Exception:
                pass