data_viewer.py

from fasthtml.common import *
from fasthtml.components import *
import json
import string
import random
import jsonlines


def gen_random_id() -> str:
    return "".join(random.choices(string.ascii_lowercase, k=8))


def view_data(
    before,
    after,
    doc_id,
    data_source: str = None,
    data_sources=None,
    target: str = "colcontent",
):
    if data_sources is not None:
        drop_down = Select(
            *[
                Option(ds, value=ds, selected=(ds == data_source))
                for ds in data_sources
            ],
            name=f"data_source_{target}",
            hx_get=f"/curated/{target}",
            hx_target=f"#{target}",
            hx_trigger="change",
            hx_swap="innerHTML",
        )

    slider = Input(
        type="range",
        name=f"doc_id_{target}",
        min="0",
        max="9",
        value=str(doc_id),
        hx_get=f"/curated/{target}",
        hx_target=f"#{target}",
        hx_trigger="change",
        hx_swap="innerHTML",
        hx_include=f'[name="data_source_{target}"]',
    )

    form = Form(
        Div(
            Label("Data source: ", drop_down),
        )
        if (data_sources is not None)
        else None,
        Div(
            Label("Data sample: ", slider, f"{doc_id}", cls="plotly_slider"),
        ),
        cls="plotly_input_container",
    )

    col1 = Div(
        H3("Raw format"),
        Pre(
            json.dumps(before, indent=4),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="width: 48%; float: left; overflow-x: auto;",
    )

    col2 = Div(
        H3("Extracted format"),
        Pre(
            json.dumps(after, indent=4),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="width: 48%; float: right; overflow-x: auto;",
    )

    data_display = Div(
        col1,
        col2,
        style="overflow: auto; clear: both; height: 600px; border: 1px solid #ccc; padding: 20px;",
    )
    return Div(form, data_display, style="margin-top: 10px;", id=target)



def DVS(
    left,
    header,
):
    col1 = Div(
        Pre(
            json.dumps(left, indent=4, ensure_ascii=False),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="float: left; overflow-x: auto;",
    )

    data_display = Div(
        col1,
        style="overflow: auto; clear: both; height: 200px; border: 1px solid #ccc; padding: 20px;",
    )
    return Div(H3(header), data_display, style="margin-top: 10px;")


def DV(
    left_file,
    doc_id,
    header,
    target: str = None,
):
    if target is None:
        target = "".join(random.choices(string.ascii_lowercase, k=8))

    if left_file.endswith("jsonl"):
        left = [x for x in jsonlines.open(left_file)]
    else:
        left = json.load(open(left_file, encoding="utf-8"))
    max_doc_id = len(left) - 1
    slider = Input(
        type="range",
        name=f"doc_id_{target}",
        min="0",
        max=str(max_doc_id),
        value=str(doc_id),
        hx_get=f"/update/{target}",
        hx_target=f"#{target}",
        hx_trigger="change",
        hx_swap="innerHTML",
        hx_vals=json.dumps({"left_file": f"{left_file}", "header": f"{header}"}),
    )

    form = Div(
        H3(header),
        Label(
            "Data sample: ", slider, f"{doc_id} of {max_doc_id}", cls="plotly_slider"
        ),
        cls="plotly_input_container",
        style="padding: 20px;",
    )

    col1 = Div(
        Pre(
            json.dumps(left[doc_id], indent=4, ensure_ascii=False),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="float: left; overflow-x: auto;",
    )

    data_display = Div(
        col1,
        style="overflow: auto; clear: both; height: 600px; border: 1px solid #ccc; padding: 20px;",
    )
    return Div(form, data_display, style="margin-top: 10px;", id=target)


def DV2(
    left_file,
    right_file,
    doc_id,
    target: str = None,
):
    if target is None:
        target = "".join(random.choices(string.ascii_lowercase, k=8))

    left = json.load(open(left_file, encoding="utf-8"))
    right = json.load(open(right_file, encoding="utf-8"))
    max_doc_id = len(left) - 1

    slider = Input(
        type="range",
        name=f"doc_id_{target}",
        min="0",
        max=str(max_doc_id),
        value=str(doc_id),
        hx_get=f"/update/{target}",
        hx_target=f"#{target}",
        hx_trigger="change",
        hx_swap="innerHTML",
        hx_vals=json.dumps(
            {"left_file": f"{left_file}", "right_file": f"{right_file}"}
        ),
    )

    form = Div(
        Label(
            "Data sample: ", slider, f"{doc_id} of {max_doc_id}", cls="plotly_slider"
        ),
        cls="plotly_input_container",
        style="padding: 20px;",
    )

    col1 = Div(
        H3("Raw format", style="margin-top: 0px;"),
        Pre(
            json.dumps(left[doc_id], indent=4, ensure_ascii=False),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="width: 48%; float: left; overflow-x: auto;",
    )

    col2 = Div(
        H3("Extracted format", style="margin-top: 0px;"),
        Pre(
            json.dumps(right[doc_id], indent=4, ensure_ascii=False),
            style="white-space: pre-wrap; word-break: break-all;",
        ),
        style="width: 48%; float: right; overflow-x: auto;",
    )

    data_display = Div(
        col1,
        col2,
        style="overflow: auto; clear: both; height: 600px; border: 1px solid #ccc; padding: 20px;",
    )
    return Div(form, data_display, style="margin-top: 10px;", id=target)


def update(target: str, request):
    params = request.query_params
    doc_id = int(params.get(f"doc_id_{target}", 3))
    left_file = params.get("left_file")
    right_file = params.get("right_file")
    if left_file and right_file:
        return (
            DV2(
                left_file,
                right_file,
                doc_id,
                target,
            ),
        )
    else:
        return DV(
            left_file,
            doc_id,
            params.get("header"),
            target,
        )