mrcad.py

# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# TODO: Address all TODOs and remove all explanatory comments
"""TODO: Add a description here."""


import csv
import json
import os
from pathlib import Path

import datasets
import pandas as pd
import ast
import jsonlines
from mrcad import Design
from mrcad.editing_actions import EditExecution
from .data_conversion_utils import (
    get_design_from_record,
    get_strokes_from_record,
    get_edit_actions_from_record,
)


# TODO: Add BibTeX citation
# Find for instance the citation on arxiv or on the dataset repo/website
_CITATION = """\
@InProceedings{huggingface:dataset,
title = {A great new dataset},
author={huggingface, Inc.
},
year={2020}
}
"""

# TODO: Add a description of the dataset here
_DESCRIPTION = """
"""

# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = ""

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = ""

_URLS = ["mrcad_data_feb_25.zip"]

DISTANCE_THRESHOLD = 0.2

MRCAD_PRECISION = 3


class mrCADDataset(datasets.GeneratorBasedBuilder):
    """Dataset of interaractions for multimodal refinement of computer-aided designs."""

    VERSION = datasets.Version("1.0.0")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig("full", description="Full dataset"),
        datasets.BuilderConfig(
            "drawing_only", description="Full dataset with text instructions removed"
        ),
        datasets.BuilderConfig(
            "text_only", description="Full dataset with drawing instructions removed"
        ),
    ]

    DEFAULT_CONFIG_NAME = "full"

    def _info(self):
        instruction_features = {
            "text": datasets.Value("string"),
            "drawing": datasets.Features(
                {
                    "splines": datasets.Sequence(
                        datasets.Sequence(datasets.Sequence(datasets.Value("float32")))
                    )
                },
            ),
        }

        design_features = datasets.Features(
            {
                # We use LargeList here instead of Sequence because using Sequence
                # would automatically convert the list of dictionaries to a dictionary
                # of lists. LargeList allows us to keep the list of dictionaries as is.
                "curves": datasets.LargeList(
                    feature=datasets.Features(
                        {
                            "type": datasets.Value("string"),
                            "control_points": datasets.Sequence(
                                datasets.Sequence(datasets.Value("float32"))
                            ),
                        }
                    )
                )
            }
        )

        features = datasets.Features(
            {
                "trial_id": datasets.Value("string"),
                "target_id": datasets.Value("string"),
                "target": design_features,
                "dyad_id": datasets.Value("string"),
                "trial_num": datasets.Value("int32"),
                "rounds": datasets.LargeList(
                    datasets.Features(
                        {
                            "round_num": datasets.Value("int32"),
                            "context": design_features,
                            "instruction": instruction_features,
                            "execution": {
                                "design": design_features,
                            },
                            "edit_execution": datasets.Features(
                                {
                                    "edits": datasets.LargeList(
                                        datasets.Features(
                                            {
                                                "edit_type": datasets.Value("string"),
                                                "point": datasets.Sequence(
                                                    datasets.Value("float32")
                                                ),
                                                "new_point": datasets.Sequence(
                                                    datasets.Value("float32")
                                                ),
                                                "control_points": datasets.Sequence(
                                                    datasets.Sequence(
                                                        datasets.Value("float32")
                                                    )
                                                ),
                                                "type": datasets.Value("string"),
                                                "offset": datasets.Sequence(
                                                    datasets.Value("float32")
                                                ),
                                            }
                                        )
                                    ),
                                    "design": design_features,
                                }
                            ),
                        }
                    )
                ),
            }
        )

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def prepare_trial(self, trial):
        return {
            "trial_id": trial["trialId"],
            "target_id": trial["targetId"],
            "target": ast.literal_eval(trial["target"])["design"],
            "dyad_id": trial["dyadId"],
            "trial_num": trial["trialNum"],
            "rounds": [
                {
                    "round_num": round_num,
                    "context": (
                        get_design_from_record(ast.literal_eval(context))
                        .round(MRCAD_PRECISION)
                        .model_dump(mode="json")
                        if context != "[]"
                        else Design(curves=[]).model_dump(mode="json")
                    ),
                    "instruction": {
                        "text": text if isinstance(text, str) else "",
                        "drawing": (
                            {
                                "splines": get_strokes_from_record(
                                    ast.literal_eval(strokes)
                                )
                            }
                            if strokes
                            else None
                        ),
                    },
                    "execution": {
                        "design": get_design_from_record(ast.literal_eval(execution))
                        .round(MRCAD_PRECISION)
                        .model_dump(mode="json"),
                    },
                    "edit_execution": EditExecution.execute(
                        (
                            get_design_from_record(ast.literal_eval(context)).round(
                                MRCAD_PRECISION
                            )
                            if context != "[]"
                            else Design(curves=[])
                        ),
                        get_edit_actions_from_record(
                            ast.literal_eval(actions)[
                                len(ast.literal_eval(prevActions)) :
                            ],
                            (
                                get_design_from_record(ast.literal_eval(context)).round(
                                    MRCAD_PRECISION
                                )
                                if context != "[]"
                                else Design(curves=[])
                            ),
                            get_design_from_record(ast.literal_eval(execution)).round(
                                MRCAD_PRECISION
                            ),
                        ),
                    ).model_dump(mode="json"),
                }
                for (
                    round_num,
                    text,
                    strokes,
                    context,
                    execution,
                    actions,
                    prevActions,
                ) in zip(
                    trial["roundNum"],
                    trial["text"],
                    trial["strokes"],
                    trial["prevJsGeometries_li"],
                    trial["jsGeometries"],
                    trial["actions"],
                    trial["prevActions"],
                )
            ],
        }

    def _split_generators(self, dl_manager):
        rounds_file_path = dl_manager.download_and_extract(_URLS)

        df = pd.read_csv(Path(rounds_file_path[0]) / "df_rounds_all_up_to_feb_25.csv")
        df.rename(columns={"trialId_sp": "trialId"}, inplace=True)

        if self.config.name == "drawing_only":
            df["text"] = df["text"].apply(lambda x: None)
        elif self.config.name == "text_only":
            df["strokes"] = df["strokes"].apply(lambda x: None)

        consolidated_df = (
            df[~df.practice_sp]  # exclude practice trials
            .sort_values("roundNum")
            .groupby("trialId")
            .agg(
                {
                    "trialId": "first",
                    "text": list,
                    "targetId": "first",
                    "target": "first",
                    "dyadId": "first",
                    "trialNum": "first",
                    "roundNum": list,
                    "prevJsGeometries_li": list,
                    "jsGeometries": list,
                    "strokes": list,
                    "distance": list,
                    "experiment_subset": "first",
                    "actions": list,
                    "prevActions": list,
                }
            )
        )

        consolidated_df["verified"] = consolidated_df.apply(
            lambda x: x.distance[-1] < DISTANCE_THRESHOLD
            and x.roundNum == [i + 1 for i, _ in enumerate(x.roundNum)],
            axis=1,
        )

        # mark trials based on what split they belong to by generating a mask
        # and then assigning the "split" field based on the mask
        mask = (consolidated_df.experiment_subset == "coverage") & (
            consolidated_df.verified
        )
        consolidated_df.loc[mask, "split"] = "coverage_verified"

        mask = (consolidated_df.experiment_subset == "coverage") & (
            ~consolidated_df.verified
        )
        consolidated_df.loc[mask, "split"] = "coverage_unverified"

        mask = (consolidated_df.experiment_subset == "eval") & (
            ~consolidated_df.verified
        )
        consolidated_df.loc[mask, "split"] = "eval_unverified"

        eval_verified_mask = (consolidated_df.experiment_subset == "eval") & (
            consolidated_df.verified
        )
        eval_verified_df = consolidated_df[eval_verified_mask]
        trial_counts = eval_verified_df.groupby("targetId")["trialId"].nunique()
        completed_targets = trial_counts[trial_counts >= 3].index.tolist()

        complete_mask = eval_verified_mask & consolidated_df["targetId"].isin(
            completed_targets
        )
        consolidated_df.loc[complete_mask, "split"] = "eval_verified_complete"

        incomplete_mask = eval_verified_mask & ~consolidated_df["targetId"].isin(
            completed_targets
        )
        consolidated_df.loc[incomplete_mask, "split"] = "eval_verified_incomplete"

        self.trials = consolidated_df

        return [
            datasets.SplitGenerator(
                name=datasets.Split("coverage_verified"),
                gen_kwargs={
                    "split": "coverage_verified",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split("coverage_unverified"),
                gen_kwargs={
                    "split": "coverage_unverified",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split("eval_unverified"),
                gen_kwargs={
                    "split": "eval_unverified",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split("eval_verified_complete"),
                gen_kwargs={
                    "split": "eval_verified_complete",
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split("eval_verified_incomplete"),
                gen_kwargs={
                    "split": "eval_verified_incomplete",
                },
            ),
        ]

    def _generate_examples(self, split):
        for row in self.trials[self.trials.split == split].itertuples():
            trial = self.prepare_trial(row._asdict())
            yield row.trialId, trial