from typing import Any

import pandas as pd

import streamlit as st

from spacy import displacy

import edsnlp

import edsnlp.pipes as eds

from edsnlp.utils.filter import filter_spans

DEFAULT_TEXT = """\

Motif :

Le patient est admis le 29 août pour des difficultés respiratoires.

Antécédents familiaux :

Le père du patient n'est pas asthmatique.

HISTOIRE DE LA MALADIE

Le patient dit avoir de la toux depuis trois jours. \

Elle a empiré jusqu'à nécessiter un passage aux urgences.

A noter deux petits kystes bénins de 1 et 2cm biopsiés en 2005.

Priorité: 2 (établie par l'IAO à l'entrée)

adicaps ABCD0A12 et ABCD0A13

Conclusion

Possible infection au coronavirus. Prescription de paracétomol pour la fièvre.\

"""

REGEX = """

# RegEx and terms matcher

nlp.add_pipe(

    eds.matcher(

        regex=dict(custom=r"{custom_regex}"),

        attr="NORM",

    ),

)

"""

CODE = """

import edsnlp, edsnlp.pipes as eds

# Declare the pipeline

nlp = edsnlp.blank("eds")

# General-purpose components

nlp.add_pipe(eds.normalizer())

nlp.add_pipe(eds.sentences())

{pipes}

# Qualifier pipes

nlp.add_pipe(eds.negation())

nlp.add_pipe(eds.family())

nlp.add_pipe(eds.hypothesis())

nlp.add_pipe(eds.rspeech())

# Define the note text

text = {text}

# Apply the pipeline

doc = nlp(text)

# Explore matched elements

doc.ents

"""

PIPES = {

    "Drugs": "drugs",

    "CIM10": "cim10",

    "Dates": "dates",

    "Quantities": "quantities",

    "Charlson": "charlson",

    "SOFA": "sofa",

    "Elston & Ellis": "elston_ellis",

    "TNM": "tnm",

    "Priority": "emergency_priority",

    "CCMU": "emergency_ccmu",

    "GEMSA": "emergency_gemsa",

    "Covid": "covid",

    "Adicap": "adicap",

    "Diabetes": "diabetes",

    "Tobacco": "tobacco",

    "AIDS": "aids",

    "Lymphoma": "lymphoma",

    "Leukemia": "leukemia",

    "Solid Tumor": "solid_tumor",

    "CKD": "ckd",

    "Hemiplegia": "hemiplegia",

    "Liver Disease": "liver_disease",

    "Peptic Ulcer Disease": "peptic_ulcer_disease",

    "Connective Tissue Disease": "connective_tissue_disease",

    "COPD": "copd",

    "Dementia": "dementia",

    "Cerebrovascular Accident": "cerebrovascular_accident",

    "Peripheral Vascular Disease": "peripheral_vascular_disease",

    "Congestive Heart Failure": "congestive_heart_failure",

    "Myocardial Infarction": "myocardial_infarction",

    "Alcohol": "alcohol",

}

@st.cache_resource()

def load_model(custom_regex: str, **enabled):

    pipes = []

    # Declare the pipeline

    nlp = edsnlp.blank("eds")

    nlp.add_pipe(eds.normalizer())

    nlp.add_pipe(eds.sentences())

    for title, name in PIPES.items():

        if name == "drugs":

            if enabled["drugs"]:

                if enabled["fuzzy_drugs"]:

                    nlp.add_pipe(eds.drugs(term_matcher="simstring"))

                    pipes.append('nlp.add_pipe(eds.drugs(term_matcher="simstring"))')

                else:

                    nlp.add_pipe(eds.drugs())

                    pipes.append("nlp.add_pipe(eds.drugs())")

        else:

            if enabled[name]:

                nlp.add_pipe(f"eds.{name}")

                pipes.append(f"nlp.add_pipe(eds.{name}())")

    if pipes:

        pipes.insert(0, "# Entity extraction pipes")

    if custom_regex:

        nlp.add_pipe(

            eds.matcher(

                regex=dict(custom=custom_regex),

                attr="NORM",

            ),

        )

        regex = REGEX.format(custom_regex=custom_regex)

    else:

        regex = ""

    nlp.add_pipe(eds.negation())

    nlp.add_pipe(eds.family())

    nlp.add_pipe(eds.hypothesis())

    nlp.add_pipe(eds.rspeech())

    return nlp, pipes, regex

st.set_page_config(

    page_title="EDS-NLP Demo",

    page_icon="📄",

)

st.title("EDS-NLP")

st.warning(

    "You should **not** put sensitive data in the example, as this application "

    "**is not secure**."

)

st.sidebar.header("About")

st.sidebar.markdown(

    "EDS-NLP is a contributive effort maintained by AP-HP's Data Science team. "

    "Have a look at the "

    "[documentation](https://aphp.github.io/edsnlp/) for "

    "more information on the available components."

)

st.sidebar.header("Pipeline")

st.sidebar.markdown(

    "This example runs a simplistic pipeline detecting a few synonyms for "

    "COVID-related entities.\n\n"

    "You can add or remove pre-defined pipeline components, and see how "

    "the pipeline reacts. You can also search for your own custom RegEx."

)

st.sidebar.header("Custom RegEx")

st_custom_regex = st.sidebar.text_input(

    "Regular Expression:",

    r"asthmatique|difficult[ée]s?\srespiratoires?",

)

st.sidebar.markdown("The RegEx you defined above is detected under the `custom` label.")

st.sidebar.subheader("Pipeline Components")

st_pipes = {}

st_pipes["cim10"] = st.sidebar.checkbox("CIM10 (loading can be slow)", value=False)

st_drugs_container = st.sidebar.columns([1, 2])

st_pipes["drugs"] = st_drugs_container[0].checkbox("Drugs", value=True)

st_fuzzy_drugs = st_drugs_container[1].checkbox(

    "Fuzzy drugs search", value=True, disabled=not st_pipes["drugs"]

)

for title, name in PIPES.items():

    if name == "drugs" or name == "cim10":

        continue

    st_pipes[name] = st.sidebar.checkbox(title, value=True)

st.sidebar.markdown(

    "These are just a few of the components provided out-of-the-box by EDS-NLP. "

    "See the [documentation](https://aphp.github.io/edsnlp/latest/pipes/) "

    "for detail."

)

model_load_state = st.info("Loading model...")

nlp, pipes, regex = load_model(

    fuzzy_drugs=st_fuzzy_drugs,

    custom_regex=st_custom_regex,

    **st_pipes,

)

model_load_state.empty()

st.header("Enter a text to analyse:")

text = st.text_area(

    "Modify the following text and see the pipeline react :",

    DEFAULT_TEXT,

    height=375,

)

doc = nlp(text)

doc.ents = filter_spans(

    (*doc.ents, *doc.spans.get("dates", []), *doc.spans.get("quantities", []))

)

st.header("Visualisation")

st.markdown(

    "The pipeline extracts simple entities using a dictionnary of RegEx (see the "

    "[Export the pipeline section](#export-the-pipeline) for more information)."

)

category20 = [

    "#1f77b4",

    "#aec7e8",

    "#ff7f0e",

    "#ffbb78",

    "#2ca02c",

    "#98df8a",

    "#d62728",

    "#ff9896",

    "#9467bd",

    "#c5b0d5",

    "#8c564b",

    "#c49c94",

    "#e377c2",

    "#f7b6d2",

    "#7f7f7f",

    "#c7c7c7",

    "#bcbd22",

    "#dbdb8d",

    "#17becf",

    "#9edae5",

]

labels = [

    "date",

    "covid",

    "drug",

    "cim10",

    "emergency_priority",

    "sofa",

    "charlson",

    "size",

    "weight",

    "adicap",

]

colors = {label: cat for label, cat in zip(labels, category20)}

colors["custom"] = "linear-gradient(90deg, #aa9cfc, #fc9ce7)"

options = {

    "colors": colors,

}

html = displacy.render(doc, style="ent", options=options)

html = html.replace("line-height: 2.5;", "line-height: 2.25;")

html = (

    '<div style="padding: 10px; border: solid 2px; border-radius: 10px; '

    f'border-color: #afc6e0;">{html}</div>'

)

st.write(html, unsafe_allow_html=True)

data = []

for ent in doc.ents:

    d = dict(

        start=ent.start_char,

        end=ent.end_char,

        text=ent.text,

        label=ent.label_,

        normalized_value=str(ent._.value or ""),

        negation="YES" if ent._.negation else "NO",

        family="YES" if ent._.family else "NO",

        hypothesis="YES" if ent._.hypothesis else "NO",

        reported_speech="YES" if ent._.reported_speech else "NO",

    )

    data.append(d)

st.header("Entity qualification")

def color_qualifiers(val: Any) -> str:

    """

    Add color to qualifiers.

    Parameters

    ----------

    val : Any

        DataFrame value

    Returns

    -------

    str

        style

    """

    if val == "NO":

        return "color: #dc3545;"

    elif val == "YES":

        return "color: #198754;"

    return ""

if data:

    df = pd.DataFrame.from_records(data)

    df.normalized_value = df.normalized_value.replace({"None": ""})

    df = df.style.applymap(color_qualifiers)

    st.dataframe(df)

else:

    st.markdown("You pipeline did not match any entity...")

pipes_text = ""

if pipes:

    pipes_text += "\n" + "\n".join(pipes) + "\n"

if regex:

    pipes_text += regex

code = CODE.format(

    pipes=pipes_text,

    text=f'"""\n{text}\n"""',

)

st.header("Export the pipeline")

st.markdown(

    "The code below recreates the pipeline. Copy and paste it "

    "in a Jupyter Notebook to interact with it."

)

with st.expander("Show the runnable code"):

    st.markdown(f"```python\n{code}\n```\n\nThis code runs as is.")