nancyrs/src/main.rs

use clap::{Parser, Subcommand};
use rusqlite::{Connection, OpenFlags};

use std::path::PathBuf;
use std::process;

// Composable provenance tracking for scientific data analysis
#[derive(Parser)]
#[command(author, version, about, long_about = None, arg_required_else_help=true)]
struct Cli {
    #[command(subcommand)]
    command: Option<Commands>,
}

#[derive(Subcommand)]
enum Commands {
    /// Record changes to files/directories within a dataset (or create a new dataset)
    #[command()]
    Record {
        /// If not within an existing dataset, create one
        #[arg(short, long)]
        initialize: bool,
        /// A short descriptive message for this recording, i.e. "Re-run with lr=1e-3"
        #[arg(short, long)]
        message: String,
        #[arg()]
        record_paths: Vec<PathBuf>,
    },
    /// Check for changes in dataset and print basic statistics
    Status {},
    /// Just say hello
    Hello {
        //#[arg(short, long)]
    //fromlib: bool,
    },
}

fn main() {
    env_logger::init();
    let args = Cli::parse();

    match &args.command {
        Some(Commands::Hello {}) => {
            println!("Hello from nancy (binary)!");
            match nancy::print_uuid() {
                Ok(_) => {
                    println!("OK");
                }
                Err(e) => {
                    println!("SQLite error: {}", e);
                }
            };
        }
        Some(Commands::Record {
            initialize,
            message,
            record_paths,
        }) => {
            // If no paths are given, use ["."] for the following steps.

            // Determine dataset dir (ds_dir)
            let mut conn = match nancy::db::find_dataset_dir(record_paths) {
                Err(nancy::db::FindDatasetError::NoDataset(path)) => {
                    log::info!("No dataset at or above nearest ancestor path: {:?}", path);
                    if !initialize {
                        log::error!("Refusing to initialize a new dataset at {path:?}. \
                                    Pass the -i or --initialize flag to request initialization.");
                        process::exit(1);
                    }
                    let dbpath = &path.join("nancy.db");
                    log::info!("Initializing new database at {:?}", dbpath);
                    let mut c = match Connection::open(dbpath) {
                        Err(e) => {
                            log::error!(
                                "Could not open new SQLite database at {dbpath:?}: {:?}",
                                e
                            );
                            process::exit(1);
                        }
                        Ok(cc) => cc,
                    };
                    c.pragma_update(None, "foreign_keys", &"ON").unwrap();
                    match nancy::db::init(&mut c) {
                        Err(e) => {
                            log::error!(
                                "Encountered error in initializing schema: {:?}",
                                e
                            );
                            process::exit(1);
                        }
                        Ok(dataset_uuid) => {
                            log::trace!("Init OK");
                            log::info!("Dataset UUID is {dataset_uuid}");
                            // Run an empty program so that the dataset log reflects when it was
                            // initialized
                            nancy::program::with_program(
                                &mut c,
                                "INIT",
                                "Initialize dataset",
                                |prog| {
                                    let _ = prog.perform_task(&[], |task| {
                                        log::debug!("INIT task UUID is {}", task.uuid);
                                    });
                                    let okres: Result<(), ()> = Ok(());
                                    okres
                                },
                            )
                            .expect("Empty program should not throw error");
                        }
                    }
                    c
                },
                Err(e) => {
                    log::error!("Could not determine dataset directory: {:?}", e);
                    process::exit(1);
                },
                Ok(path) => {
                    // existing
                    log::info!("Found existing dataset at path: {:?}", path);
                    let dbpath = &path.join("nancy.db");
                    // open with flags to prevent creating when we believe the db exists
                    let mut c = match Connection::open_with_flags(
                        dbpath,
                        OpenFlags::SQLITE_OPEN_READ_WRITE | OpenFlags::SQLITE_OPEN_NO_MUTEX,
                    ) {
                        Err(e) => {
                            log::error!(
                                "Could not open existing SQLite database at {dbpath:?}: {e:?}"
                            );
                            process::exit(1);
                        }
                        Ok(cc) => cc,
                    };
                    c.pragma_update(None, "foreign_keys", &"ON").unwrap();
                    match nancy::db::ensure_schema(&mut c) {
                        Err(e) => {
                            log::error!("Error ensuring schema: {}", e);
                            process::exit(1);
                        }
                        Ok(update_result) => {
                            log::debug!("Schema ensured: {:?}", update_result);
                        }
                    }
                    c
                },
            };

            if let Err(e) = nancy::program::with_program(&mut conn, "RECORD", message, |prog| {
                prog.perform_task(&[], |_task| {
                    let dataset_path = PathBuf::from(".");
                    // Note that this may fail, in which case we should roll back only this program
                    // but keep the dataset initialized.
                    nancy::fs::record(&dataset_path, message)
                })
            }) {
                log::error!("Encountered error in RECORD program: {:?}", e);
                process::exit(1);
            };
        }
        Some(Commands::Status {}) => {
            println!("status not yet implemented");
        }
        None => {}
    }
}