{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0",
   "metadata": {},
   "source": [
    "Creating custom pipelines in QAT\n",
    "========================================\n",
    "\n",
    "We can bring together all the granular components of a pipeline to create a custom pipeline.\n",
    "See below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1",
   "metadata": {
    "tags": [
     "remove-cell"
    ]
   },
   "outputs": [],
   "source": [
    "# used to disable output from logs; not shown in the docs because of the\n",
    "# remove-cell tag\n",
    "import logging\n",
    "\n",
    "logging.disable(logging.CRITICAL)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2",
   "metadata": {},
   "source": [
    "Let's recreate the echo pipeline!\n",
    "* `AutoFrontend`: dispatches the correct frontend, decided by the input type.\n",
    "* `DefaultMiddleend`: Applies default optimization and lowering of pulse-level IR.\n",
    "* `WaveformBackend`: Transforms the IR into a binary for \"Waveform\" hardware.\n",
    "* `SimpleRuntime`: Applies a hybrid of execution on the hardware, followed by a results post-processing pipeline with an EchoEngine in this example.\n",
    "* `EchoEngine`: Executes the binary against an Echo simulator."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3",
   "metadata": {},
   "outputs": [],
   "source": [
    "from qat.backend.waveform import WaveformBackend\n",
    "from qat.core.pipeline import Pipeline\n",
    "from qat.engines.waveform import EchoEngine\n",
    "from qat.frontend import AutoFrontend\n",
    "from qat.middleend.default import DefaultMiddleend\n",
    "from qat.model.loaders.lucy import LucyModelLoader\n",
    "from qat.model.target_data import DefaultTargetData\n",
    "from qat.runtime import SimpleRuntime\n",
    "from qat.runtime.results_pipeline import get_results_pipeline\n",
    "\n",
    "model = LucyModelLoader(qubit_count=8).load()\n",
    "target_data = DefaultTargetData()\n",
    "results_pipeline = get_results_pipeline(model=model, target_data=target_data)\n",
    "new_echo8 = Pipeline(\n",
    "    name=\"new_echo8\",\n",
    "    frontend=AutoFrontend.default_for_pydantic(model),\n",
    "    middleend=DefaultMiddleend(model, target_data),\n",
    "    backend=WaveformBackend(model, target_data),\n",
    "    runtime=SimpleRuntime(EchoEngine(), results_pipeline=results_pipeline),\n",
    "    model=model,\n",
    "    target_data=target_data,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4",
   "metadata": {},
   "source": [
    "If we want to make this an updateable pipeline, we can do so as shown below. The `_build_pipeline` acts a factory method with the recipe to build a pipeline provided the model and target data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5",
   "metadata": {},
   "outputs": [],
   "source": [
    "from qat.backend.waveform import WaveformBackend\n",
    "from qat.core.pipeline import Pipeline\n",
    "from qat.engines.waveform import EchoEngine\n",
    "from qat.frontend import AutoFrontend\n",
    "from qat.middleend.default import DefaultMiddleend\n",
    "from qat.model.loaders.lucy import LucyModelLoader\n",
    "from qat.model.target_data import DefaultTargetData, TargetData\n",
    "from qat.pipelines.updateable import Model, PipelineConfig, UpdateablePipeline\n",
    "from qat.runtime import SimpleRuntime\n",
    "from qat.runtime.results_pipeline import get_results_pipeline\n",
    "\n",
    "\n",
    "class MyCoolPipeline(UpdateablePipeline):\n",
    "    @staticmethod\n",
    "    def _build_pipeline(\n",
    "        config: PipelineConfig,\n",
    "        model: Model,\n",
    "        target_data: TargetData | None,\n",
    "        *args,\n",
    "        **kwargs,\n",
    "    ) -> Pipeline:\n",
    "        results_pipeline = get_results_pipeline(model=model, target_data=target_data)\n",
    "        return Pipeline(\n",
    "            name=\"new_echo8\",\n",
    "            frontend=AutoFrontend.default_for_pydantic(model),\n",
    "            middleend=DefaultMiddleend(model, target_data),\n",
    "            backend=WaveformBackend(model, target_data),\n",
    "            runtime=SimpleRuntime(EchoEngine(), results_pipeline=results_pipeline),\n",
    "            model=model,\n",
    "            target_data=target_data,\n",
    "        )\n",
    "\n",
    "\n",
    "model = LucyModelLoader(qubit_count=8).load()\n",
    "target_data = DefaultTargetData()\n",
    "pipeline_instance = MyCoolPipeline(\n",
    "    config=dict(name=\"my_pipeline\"), model=model, target_data=target_data\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "main_language": "python",
   "notebook_metadata_filter": "-kernelspec"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}