src/lib/intl/model/src/lib.rs

// Copyright 2020 The Fuchsia Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

//! # The JSON message internationalization format
//!
//! This module contains the code used to turn a dictionary of localized messages into a
//! JSON-formatted file usable by Fuchsia's localization lookup system.
//!
//! The data model used for the JSON schema is defined below.  No formal JSON schema has been
//! specified yet. (But probably should be!)

use {
    anyhow::Result,
    serde::{Deserialize, Serialize},
    std::collections::BTreeMap,
    std::io,
};

/// The message catalog.  Maps unique IDs (as generated by the message_id::gen_ids) to individual
/// message.  The catalog does not know about the locale it is intended for, and additional
/// metadata is needed to assert that.  See [Model] for details on how it fits together.
pub type Messages = BTreeMap<u64, String>;

#[cfg(test)]
pub fn as_messages(pairs: &Vec<(u64, String)>) -> Messages {
    let mut result = Messages::new();
    for (k, v) in pairs {
        result.insert(*k, v.clone());
    }
    result as Messages
}

/// The data model for a set of internationalized messages.  Every
/// file has a locale ID that it applies to, as well as the number
/// of total messages analyzed when this locale was produced
///
/// Use [Model::from_json_reader] to make a new instance of the
/// Model from its JSON serialization.  Use [Model::from_dictionaries] to make a new instance of
/// the Model based on the supplied dictionaries.
///
/// In tests, you can use [Model::from_parts] to create [Model] quickly without any checks.
#[derive(Serialize, Deserialize, Debug, Clone, Default, Eq, PartialEq)]
pub struct Model {
    /// The locale for the messages described in this file.  An
    /// example value could be "en-US".
    locale_id: String,

    /// The source-of-truth locale from which this model was
    /// generated.  Source of truth locales exist so as to provide
    /// fallback messages or such.
    source_locale_id: String,

    /// The total number of messages that are expected to exist in
    /// this bundle.  This number is not the same as the number of
    /// key-value pairs in [messages] below, though.
    num_messages: usize,

    /// The message catalog, listing the mapping of message IDs to
    /// respective messages.  This mapping does not define a fallback
    /// language.  The ordering of the messages in this map is not
    /// defined: all serializations that end up with an identical
    /// map model are equally valid.
    messages: Messages,
}

impl Model {
    pub fn locale(&self) -> &str {
        &self.locale_id
    }

    pub fn messages(&self) -> &Messages {
        &self.messages
    }

    /// Deserializes the Model from the supplied reader.  Use to
    /// create a functional Model from JSON.
    pub fn from_json_reader<R: io::Read>(r: R) -> Result<Model> {
        serde_json::from_reader(r).map_err(|e| e.into())
    }

    /// Writes the Model into the supplied writer encoded as JSON.  Use to persist the Model into
    /// JSON.  The message ordering is not guaranteed.
    pub fn to_json_writer<W: io::Write>(&self, writer: W) -> Result<()> {
        serde_json::to_writer(writer, self)
            .map_err(|e| -> io::Error { e.into() })
            .map_err(|e| -> anyhow::Error { e.into() })
    }

    /// Creates a [Model] quickly from supplied parts and without any checks,
    /// for use in tests only.
    pub fn from_parts(
        locale_id: &str,
        source_locale_id: &str,
        num_messages: usize,
        messages: Messages,
    ) -> Model {
        Model {
            locale_id: locale_id.to_string(),
            source_locale_id: source_locale_id.to_string(),
            num_messages,
            messages,
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use anyhow::Context;

    #[test]
    fn read_json_test() -> Result<()> {
        let message = r#"
        {
            "locale_id": "ru-RU",
            "source_locale_id": "en-US",
            "num_messages": 1000,
            "messages": {
                "42": "Привет",
                "100": "Что это сорок-два?"
            }
        }
        "#;
        let model = Model::from_json_reader(message.as_bytes())
            .with_context(|| format!("while loading JSON: \n{}", message))?;
        assert_eq!(model.locale_id, "ru-RU");
        assert_eq!(model.num_messages, 1000);
        assert_eq!(model.messages.get(&42).unwrap(), "Привет");
        Ok(())
    }

    /// Note that we don't care about the message ordering just yet.
    #[test]
    fn round_trip_test() -> Result<()> {
        struct TestCase {
            name: &'static str,
            message: &'static str,
        }
        let tests = vec![
            TestCase {
                name: "Random text in Russian",
                message: r#"
                {
                    "locale_id": "ru-RU",
                    "source_locale_id": "en-US",
                    "num_messages": 42,
                    "messages": {
                        "42": "Привет",
                        "100": "Что это сорок-два?"
                    }
                }
                "#,
            },
            TestCase {
                name: "Serbian pangram",
                message: r#"
                {
                    "locale_id": "sr-RS",
                    "source_locale_id": "en-US",
                    "num_messages": 2,
                    "messages": {
                        "1": "Љубазни фењерџија чађавог лица хоће да ми покаже штос"
                    }
                }
                "#,
            },
        ];
        for test in tests {
            let model = Model::from_json_reader(test.message.as_bytes()).with_context(|| {
                format!("in test '{}', while loading JSON: \n{}", test.name, test.message)
            })?;
            let mut output: Vec<u8> = vec![];
            model
                .to_json_writer(&mut output)
                .with_context(|| format!("while writing JSON, in test '{}'", test.name))?;
            let model2 = Model::from_json_reader(&output[..])
                .with_context(|| format!("while reading JSON again, in test '{}'", test.name))?;
            assert_eq!(model, model2);
        }
        Ok(())
    }
}