Refresh the GTFS bundle every time the program starts.

This commit is contained in:
Nahuel Lofeudo 2026-05-25 07:45:35 +01:00
parent 2957ccf1ff
commit dc368ca811
6 changed files with 122 additions and 147 deletions

View File

@ -9,8 +9,10 @@ chrono = "0.4"
csv = "1.4"
gtfs-structures = "0.47"
log = "0.4"
reqwest = { version = "0.11", features = ["blocking"] }
sdl3 = {version = "0.17", features = ["ttf"]}
serde = { version = "1.0", features = ["derive"]}
time-format = "1.2"
yaml_serde = "0.10"
zip = "8.3"

View File

@ -1,132 +0,0 @@
use gtfs_structures::{Agency, Calendar, CalendarDate, Route, Stop};
use serde::{de::DeserializeOwned};
use std::{collections::HashMap, fs::File};
use zip::ZipArchive;
// The main GTFS struct. This is similar to (but not exactly) gtfs-structures::Gtfs because we don't need everything
#[derive(Debug)]
pub struct Gtfs {
/// All agencies. They can not be read by `agency_id`, as it is not a required field
pub agencies: Vec<Agency>,
/// All Calendar by `service_id`
pub calendar: HashMap<String, Calendar>,
/// All calendar dates grouped by service_id
pub calendar_dates: HashMap<String, Vec<CalendarDate>>,
/// All routes by `route_id`
pub routes: HashMap<String, Route>,
/// All stop by `stop_id`.
pub stops: HashMap<String, Stop>,
}
// Utility function to load all records in a dataset
fn load_all<V>(_: &V) -> bool { true }
// Loads a vector of the selected type
fn load_vector<T: serde::de::DeserializeOwned>(
destination: &mut Vec<T>,
zip_reader: &mut ZipArchive<File>,
table_name: &str,
) {
let file_reader = zip_reader.by_name(table_name).unwrap();
let mut rdr = csv::Reader::from_reader(file_reader);
for row in rdr.deserialize() {
let record: T = row.unwrap();
destination.push(record);
}
}
// Loads a HashMap of the selected type, using the provided index function as the key
fn load_map<'a, V: DeserializeOwned>(
destination: &mut HashMap<String, V>,
zip_reader: &mut ZipArchive<File>,
table_name: &str,
index: fn(&V) -> String,
accept: fn(&V) -> bool,
) {
let file_reader = zip_reader.by_name(table_name).unwrap();
let mut rdr = csv::Reader::from_reader(file_reader);
for row in rdr.deserialize() {
let record: V = row.unwrap();
if accept(&record) {
let idx: String = index(&record);
destination.insert(idx, record);
}
}
}
// Loads a HashMap of a vector of the selected type, using the provided index function as the key
// And a predicate as a filter
fn load_vector_map<'a, V: DeserializeOwned + Clone>(
destination: &mut HashMap<String, Vec<V>>,
zip_reader: &mut ZipArchive<File>,
table_name: &str,
index: fn(&V) -> String,
accept: fn(&V) -> bool,
) {
let file_reader = zip_reader.by_name(table_name).unwrap();
let mut rdr = csv::Reader::from_reader(file_reader);
for row in rdr.deserialize() {
let record: V = row.unwrap();
if accept(&record) {
let idx: String = index(&record);
destination.entry(idx).or_insert_with(Vec::new).push(record);
}
}
}
pub fn init(src_file: &str, routes: Vec<String>) -> Gtfs {
// Open zip file
let mut zip_reader = zip::ZipArchive::new(File::open(src_file).unwrap()).unwrap();
let mut gtfs: Gtfs = Gtfs {
agencies: Vec::new(),
calendar: HashMap::new(),
calendar_dates: HashMap::new(),
routes: HashMap::new(),
stops: HashMap::new(),
};
// Agencies
load_vector(&mut gtfs.agencies, &mut zip_reader, "agency.txt");
// Calendars
load_map(
&mut gtfs.calendar,
&mut zip_reader,
"calendar.txt",
|c: &Calendar| String::from(&c.id),
load_all
);
// Calendar Dates
load_vector_map(
&mut gtfs.calendar_dates,
&mut zip_reader,
"calendar_dates.txt",
|d: &CalendarDate| String::from(&d.service_id),
load_all
);
// Stops
load_map(&mut gtfs.stops,
&mut zip_reader,
"stops.txt",
|s: &Stop| {String::from(&s.id)},
load_all
);
// Routes
let accept_filter: fn(&Route) -> bool = (| rs: Vec<String>, r: &Route | {rs.contains(&r.short_name.unwrap())}).curry(routes);
load_map(
&mut gtfs.routes,
&mut zip_reader,
"routes.txt",
|r: &Route| String::from(&r.id),
accept_filter
);
return gtfs;
}

View File

@ -1,15 +1,13 @@
mod arrival;
mod loader;
mod utils;
mod refresher;
pub mod structs;
use chrono::{DateTime, Local, Timelike};
use log::{debug};
use std::{
collections::{HashMap, HashSet}, fs::File, io::Error
};
use log::{debug, trace};
use std::{collections::{HashMap, HashSet}, fs::File };
use gtfs_structures::{Exception, RawTrip};
use crate::gtfs::{loader::load_gtfs, structs::{Arrival, Gtfs, Preferences}};
use crate::gtfs::{loader::load_gtfs, structs::{Arrival, Gtfs, Preferences, Error}};
impl Gtfs {
@ -75,7 +73,7 @@ impl Gtfs {
trip: &trip,
departure_time: stop_timestamp.into()
};
debug!("{:#?}: Arrival to {:#?} for trip ID {:#?}.",
trace!("{:#?}: Arrival to {:#?} for trip ID {:#?}.",
format!("{:02}:{:02}", (arrival.departure_time/3600) as u32, ((arrival.departure_time / 60) % 60) as u32),
arrival.trip.trip_headsign.as_ref().unwrap(),
arrival.trip.id);
@ -91,9 +89,12 @@ impl Gtfs {
/// Load a GTFS structure from a zip file
pub fn load(src_file: &str, prefs: &Preferences) -> Result<Gtfs, Error> {
pub fn load(prefs: &Preferences) -> Result<Gtfs, Error> {
_ = refresher::refresh(prefs);
// Open zip file
let zip_file = File::open(src_file)?;
let zip_file = File::open(prefs.gtfs_file_path()?)?;
let mut zip_reader = zip::ZipArchive::new(zip_file)?;
let mut gtfs: Gtfs = Gtfs {

61
src/gtfs/refresher.rs Normal file
View File

@ -0,0 +1,61 @@
use log::{debug, error, info};
use time_format::format_common_utc;
use std::{fs::{self}, time::{Duration, SystemTime, UNIX_EPOCH}};
use crate::gtfs::structs::{Error, Preferences};
use reqwest::{StatusCode, blocking::Client};
pub(crate) fn refresh(prefs: &Preferences) -> Result<(), Error> {
let mut modified_timestamp: SystemTime = SystemTime::UNIX_EPOCH;
let gtfs_file_name = prefs.gtfs_file_path()?;
// Obtain the GTFS zip's creation time.
let metadata = fs::metadata(&gtfs_file_name);
if metadata.is_ok() {
modified_timestamp = metadata?.modified()?;
}
let modified_seconds = modified_timestamp.duration_since(UNIX_EPOCH).unwrap().as_secs() as i64;
let modified_header = format_common_utc(modified_seconds, time_format::DateFormat::HTTP).unwrap();
debug!("Using if-modified-since: {}", modified_header);
// request updates from the server if-modified-since last time
let client = Client::builder()
.timeout(Duration::from_secs(30))
.connect_timeout(Duration::from_secs(5))
.build()?;
let mut response = client
.get(&prefs.gtfs_url)
.header(reqwest::header::IF_MODIFIED_SINCE, modified_header)
.send()
.unwrap();
if response.status() == StatusCode::NOT_MODIFIED {
debug!("GTFS data is still up-to-date");
return Ok(());
}
if response.status() == StatusCode::OK {
info!("Refreshing GTFS data");
// Stream the response data into a temp file and swap them
let tmp_full_file_name = format!("{}.new", &gtfs_file_name);
let mut tmp_file = fs::File::create(&tmp_full_file_name)?;
_ = response.copy_to(&mut tmp_file)?;
drop(tmp_file);
if fs::exists(&gtfs_file_name)? {
_ = fs::remove_file(&gtfs_file_name);
}
_ = fs::rename(&tmp_full_file_name, &gtfs_file_name);
info!("GTFS data refreshed");
return Ok(());
}
let errmsg = format!("GET on {} returned result code {}", prefs.gtfs_url, response.status());
return Err(Error { message: errmsg });
}

View File

@ -1,6 +1,33 @@
use std::collections::{HashMap, HashSet};
use gtfs_structures::{Agency, Calendar, CalendarDate, RawStopTime, RawTrip, Route, Stop};
use log::error;
use serde::{Serialize, Deserialize};
use zip::result::ZipError;
#[derive(Debug)]
pub struct Error {
pub(crate) message: String,
}
impl From<std::io::Error> for Error {
fn from(value: std::io::Error) -> Self {
return Error { message: value.to_string() }
}
}
impl From<reqwest::Error> for Error {
fn from(value: reqwest::Error) -> Self {
return Error { message: value.to_string() }
}
}
impl From<ZipError> for Error {
fn from(value: ZipError) -> Self {
return Error { message: value.to_string() }
}
}
// This is to store the preferences for the GTFS(-R) side of the code.
#[derive(Debug, PartialEq, Serialize, Deserialize)]
@ -27,6 +54,23 @@ pub struct Preferences {
pub refresh_seconds: u64,
}
// Utility functions
impl Preferences {
pub fn gtfs_file_name(&self) -> Result<String, Error> {
let file_name_part = self.gtfs_url.split("/").last();
if file_name_part.is_none() {
error!("The config for gtfs-url is not a valid URL: {}", self.gtfs_url);
return Err(Error {message: String::from("Failed to refresh GTFS data")});
}
return Ok(String::from(file_name_part.unwrap()));
}
pub fn gtfs_file_path(&self) -> Result<String, Error> {
return Ok(format!("{}/{}", self.data_folder, self.gtfs_file_name()?));
}
}
// The main GTFS struct. This is similar to (but not exactly) gtfs-structures::Gtfs because we don't need everything
#[derive(Debug)]

View File

@ -3,11 +3,10 @@ mod renderer;
mod config;
use std::{ops::Add, process, thread::Builder, time::SystemTime};
use chrono::{DateTime, Duration, Local, NaiveTime};
use log::{Metadata, Record, error, info};
use log::{Level, Metadata, Record, error, info};
use sdl3::event::Event;
use crate::{config::load_config, gtfs::structs::{Arrival, Gtfs}, renderer::structs::{DisplayData, DisplayEntry, Screen}};
const SRC_FILE: &str = "/home/nahuel/Downloads/GTFS_Realtime.zip";
const NUM_ARRIVALS: usize = 4;
// Custom Event to signal data refresh
@ -59,8 +58,8 @@ fn main() {
static MY_LOGGER: MyLogger = MyLogger;
struct MyLogger;
impl log::Log for MyLogger {
fn enabled(&self, _metadata: &Metadata) -> bool {
true
fn enabled(&self, metadata: &Metadata) -> bool {
return metadata.level() < Level::Trace;
}
fn log(&self, record: &Record) {
@ -92,10 +91,10 @@ fn main() {
// Init GTFS static info
info!("Loading GTFS data...");
let res = Gtfs::load(SRC_FILE, &gtfs_prefs);
let res = Gtfs::load(&gtfs_prefs);
if let Err(e) = res {
error!("Error loading GTFS data: {}", e);
error!("Error loading GTFS data: {:#?}", e);
process::exit(-1);
}
let gtfs = res.unwrap();