// Copyright 2019-present, the HuggingFace Inc. team, The Google AI Language Team and Facebook, Inc.
// Copyright 2019 Guillaume Becquin
// 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.

//! # Named Entity Recognition pipeline
//! Extracts entities (Person, Location, Organization, Miscellaneous) from text.
//! Pretrained models are available for the following languages:
//! - English
//! - German
//! - Spanish
//! - Dutch
//!
//! The default NER mode is an English BERT cased large model finetuned on CoNNL03, contributed by the [MDZ Digital Library team at the Bavarian State Library](https://github.com/dbmdz)
//! All resources for this model can be downloaded using the Python utility script included in this repository.
//! 1. Set-up a Python virtual environment and install dependencies (in ./requirements.txt)
//! 2. Run the conversion script python /utils/download-dependencies_bert_ner.py.
//! The dependencies will be downloaded to the user's home directory, under ~/rustbert/bert-ner
//!
//! The example below illustrate how to run the model for the default English NER model
//! ```no_run
//! use rust_bert::pipelines::ner::NERModel;
//! # fn main() -> anyhow::Result<()> {
//! let ner_model = NERModel::new(Default::default())?;
//!
//! let input = [
//!     "My name is Amy. I live in Paris.",
//!     "Paris is a city in France.",
//! ];
//! let output = ner_model.predict(&input);
//! # Ok(())
//! # }
//! ```
//! Output: \
//! ```no_run
//! # use rust_bert::pipelines::question_answering::Answer;
//! # use rust_bert::pipelines::ner::Entity;
//! # let output =
//! [
//!     Entity {
//!         word: String::from("Amy"),
//!         score: 0.9986,
//!         label: String::from("I-PER"),
//!     },
//!     Entity {
//!         word: String::from("Paris"),
//!         score: 0.9985,
//!         label: String::from("I-LOC"),
//!     },
//!     Entity {
//!         word: String::from("Paris"),
//!         score: 0.9988,
//!         label: String::from("I-LOC"),
//!     },
//!     Entity {
//!         word: String::from("France"),
//!         score: 0.9993,
//!         label: String::from("I-LOC"),
//!     },
//! ]
//! # ;
//! ```
//!
//! To run the pipeline for another language, change the NERModel configuration from its default:
//!
//! ```no_run
//! use rust_bert::pipelines::common::ModelType;
//! use rust_bert::pipelines::ner::NERModel;
//! use rust_bert::pipelines::token_classification::TokenClassificationConfig;
//! use rust_bert::resources::{RemoteResource, Resource};
//! use rust_bert::roberta::{
//!     RobertaConfigResources, RobertaModelResources, RobertaVocabResources,
//! };
//! use tch::Device;
//!
//! # fn main() -> anyhow::Result<()> {
//! let ner_config = TokenClassificationConfig {
//!     model_type: ModelType::XLMRoberta,
//!     model_resource: Resource::Remote(RemoteResource::from_pretrained(
//!         RobertaModelResources::XLM_ROBERTA_NER_DE,
//!     )),
//!     config_resource: Resource::Remote(RemoteResource::from_pretrained(
//!         RobertaConfigResources::XLM_ROBERTA_NER_DE,
//!     )),
//!     vocab_resource: Resource::Remote(RemoteResource::from_pretrained(
//!         RobertaVocabResources::XLM_ROBERTA_NER_DE,
//!     )),
//!     lower_case: false,
//!     device: Device::cuda_if_available(),
//!     ..Default::default()
//! };
//!
//! let ner_model = NERModel::new(ner_config)?;
//!
//! //    Define input
//! let input = [
//!     "Mein Name ist Amélie. Ich lebe in Paris.",
//!     "Paris ist eine Stadt in Frankreich.",
//! ];
//! let output = ner_model.predict(&input);
//! # Ok(())
//! # }
//! ```
//! The XLMRoberta models for the languages are defined as follows:
//!
//! | **Language** |**Model name**|
//! :-----:|:----:
//! English| XLM_ROBERTA_NER_EN |
//! German| XLM_ROBERTA_NER_DE |
//! Spanish| XLM_ROBERTA_NER_ES |
//! Dutch| XLM_ROBERTA_NER_NL |

use crate::common::error::RustBertError;
use crate::pipelines::token_classification::{TokenClassificationConfig, TokenClassificationModel};

#[derive(Debug)]
/// # Entity generated by a `NERModel`
pub struct Entity {
    /// String representation of the Entity
    pub word: String,
    /// Confidence score
    pub score: f64,
    /// Entity label (e.g. ORG, LOC...)
    pub label: String,
}

//type alias for some backward compatibility
type NERConfig = TokenClassificationConfig;

/// # NERModel to extract named entities
pub struct NERModel {
    token_classification_model: TokenClassificationModel,
}

impl NERModel {
    /// Build a new `NERModel`
    ///
    /// # Arguments
    ///
    /// * `ner_config` - `NERConfig` object containing the resource references (model, vocabulary, configuration) and device placement (CPU/GPU)
    ///
    /// # Example
    ///
    /// ```no_run
    /// # fn main() -> anyhow::Result<()> {
    /// use rust_bert::pipelines::ner::NERModel;
    ///
    /// let ner_model = NERModel::new(Default::default())?;
    /// # Ok(())
    /// # }
    /// ```
    pub fn new(ner_config: NERConfig) -> Result<NERModel, RustBertError> {
        let model = TokenClassificationModel::new(ner_config)?;
        Ok(NERModel {
            token_classification_model: model,
        })
    }

    /// Extract entities from a text
    ///
    /// # Arguments
    ///
    /// * `input` - `&[&str]` Array of texts to extract entities from.
    ///
    /// # Returns
    ///
    /// * `Vec<Entity>` containing extracted entities
    ///
    /// # Example
    ///
    /// ```no_run
    /// # fn main() -> anyhow::Result<()> {
    /// # use rust_bert::pipelines::ner::NERModel;
    ///
    /// let ner_model = NERModel::new(Default::default())?;
    /// let input = [
    ///     "My name is Amy. I live in Paris.",
    ///     "Paris is a city in France.",
    /// ];
    /// let output = ner_model.predict(&input);
    /// # Ok(())
    /// # }
    /// ```
    pub fn predict<'a, S>(&self, input: S) -> Vec<Entity>
    where
        S: AsRef<[&'a str]>,
    {
        self.token_classification_model
            .predict(input, true, false)
            .into_iter()
            .filter(|token| token.label != "O")
            .map(|token| Entity {
                word: token.text,
                score: token.score,
                label: token.label,
            })
            .collect()
    }
}
#[cfg(test)]
mod test {
    use super::*;

    #[test]
    #[ignore] // no need to run, compilation is enough to verify it is Send
    fn test() {
        let config = NERConfig::default();
        let _: Box<dyn Send> = Box::new(NERModel::new(config));
    }
}