upend/cli/src/extractors/exif.rs

use std::sync::Arc;

use super::Extractor;
use anyhow::{anyhow, Result};
use lazy_static::lazy_static;
use upend_base::entry::Attribute;
use upend_base::{
    addressing::Address,
    constants::{ATTR_IN, ATTR_KEY, ATTR_LABEL, ATTR_OF},
    entry::{Entry, EntryValue, InvariantEntry},
};
use upend_db::{
    jobs::{JobContainer, JobState},
    stores::{fs::FILE_MIME_KEY, UpStore},
    OperationContext, UpEndConnection,
};

pub struct ExifExtractor;

// TODO: EXIF metadata is oftentimes a constant/enum value. What's the proper
//       model for enum-like values in UpEnd?

lazy_static! {
    pub static ref EXIF_TYPE_INVARIANT: InvariantEntry = InvariantEntry {
        attribute: ATTR_KEY.parse().unwrap(),
        value: "TYPE_EXIF".into(),
    };
    pub static ref EXIF_TYPE_LABEL: Entry = Entry {
        entity: EXIF_TYPE_INVARIANT.entity().unwrap(),
        attribute: ATTR_LABEL.parse().unwrap(),
        value: "EXIF".into(),
        provenance: "INVARIANT".to_string(),
        timestamp: chrono::Utc::now().naive_utc(),
        user: None
    };
}

impl Extractor for ExifExtractor {
    fn get(
        &self,
        address: &Address,
        _connection: &UpEndConnection,
        store: Arc<Box<dyn UpStore + Send + Sync>>,
        mut job_container: JobContainer,
        context: OperationContext,
    ) -> Result<Vec<Entry>> {
        if let Address::Hash(hash) = address {
            let files = store.retrieve(hash)?;

            if let Some(file) = files.first() {
                let file_path = file.get_file_path();
                let mut job_handle = job_container.add_job(
                    None,
                    &format!(
                        r#"Getting EXIF info from "{:}""#,
                        file_path
                            .components()
                            .last()
                            .unwrap()
                            .as_os_str()
                            .to_string_lossy()
                    ),
                )?;

                let file = std::fs::File::open(file_path)?;
                let mut bufreader = std::io::BufReader::new(&file);
                let exifreader = exif::Reader::new();
                let exif = exifreader.read_from_container(&mut bufreader)?;

                let mut result: Vec<Entry> = vec![];

                for field in exif
                    .fields()
                    .filter(|field| !matches!(field.value, exif::Value::Undefined(..)))
                {
                    if let Some(tag_description) = field.tag.description() {
                        let attribute: Attribute = format!("EXIF_{}", field.tag.1).parse()?;

                        result.extend(vec![
                            Entry {
                                entity: address.clone(),
                                attribute: attribute.clone(),
                                value: match field.tag {
                                    exif::Tag::ExifVersion => {
                                        EntryValue::String(format!("{}", field.display_value()))
                                    }
                                    _ => {
                                        EntryValue::guess_from(format!("{}", field.display_value()))
                                    }
                                },
                                provenance: context.provenance.clone() + "EXTRACTOR",
                                user: context.user.clone(),
                                timestamp: chrono::Utc::now().naive_utc(),
                            },
                            Entry {
                                entity: Address::Attribute(attribute),
                                attribute: ATTR_LABEL.parse().unwrap(),
                                value: format!("EXIF: {}", tag_description).into(),
                                provenance: context.provenance.clone() + "EXTRACTOR",
                                user: context.user.clone(),
                                timestamp: chrono::Utc::now().naive_utc(),
                            },
                        ]);
                    }
                }

                if !result.is_empty() {
                    result.extend(
                        result
                            .iter()
                            .filter(|e| e.attribute != ATTR_LABEL)
                            .map(|e| Entry {
                                entity: Address::Attribute(e.attribute.clone()),
                                attribute: ATTR_OF.parse().unwrap(),
                                value: EntryValue::Address(EXIF_TYPE_INVARIANT.entity().unwrap()),
                                provenance: context.provenance.clone() + "EXTRACTOR",
                                user: context.user.clone(),
                                timestamp: chrono::Utc::now().naive_utc(),
                            })
                            .collect::<Vec<Entry>>(),
                    );
                    result.extend(vec![
                        (&EXIF_TYPE_INVARIANT as &InvariantEntry)
                            .try_into()
                            .unwrap(),
                        EXIF_TYPE_LABEL.clone(),
                        Entry {
                            entity: address.clone(),
                            attribute: ATTR_IN.parse().unwrap(),
                            value: EntryValue::Address(EXIF_TYPE_INVARIANT.entity().unwrap()),
                            provenance: context.provenance.clone() + "EXTRACTOR",
                            user: context.user.clone(),
                            timestamp: chrono::Utc::now().naive_utc(),
                        },
                    ]);
                }

                let _ = job_handle.update_state(JobState::Done);

                Ok(result)
            } else {
                Err(anyhow!("Couldn't find file for {hash:?}!"))
            }
        } else {
            Ok(vec![])
        }
    }

    fn is_needed(&self, address: &Address, connection: &UpEndConnection) -> Result<bool> {
        let is_exif = connection.retrieve_object(address)?.iter().any(|e| {
            if e.attribute == FILE_MIME_KEY {
                if let EntryValue::String(mime) = &e.value {
                    return mime.starts_with("image");
                }
            }
            false
        });

        if !is_exif {
            return Ok(false);
        }

        let is_extracted = !connection
            .query(format!("(matches @{} (contains \"EXIF\") ?)", address).parse()?)?
            .is_empty();

        if is_extracted {
            return Ok(false);
        }

        Ok(true)
    }
}