GKaszewski/libertas
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: integrate EXIF data extraction using nom-exif and refactor related components

Browse Source
This commit is contained in:
Gabriel Kaszewski
2025-11-14 06:35:08 +01:00
parent 60860cf508
commit ea95c2255f
7 changed files with 155 additions and 105 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Cargo.lock generated
View File

@@ -1661,6 +1661,7 @@ dependencies = [
"bytes",
"chrono",
"futures",
"nom-exif",
"serde",
"thiserror 2.0.17",
"uuid",
@@ -1714,7 +1715,6 @@ dependencies = [
"image",
"libertas_core",
"libertas_infra",
"nom-exif",
"serde",
"serde_json",
"sqlx",

1
libertas_core/Cargo.toml
View File

@@ -12,3 +12,4 @@ futures = "0.3.31"
thiserror = "2.0.17"
uuid = {version = "1.18.1", features = ["v4", "serde"] }
serde = { version = "1.0.228", features = ["derive"] }
nom-exif = { version = "2.5.4", features = ["serde", "async", "tokio"] }

1
libertas_core/src/lib.rs
View File

@@ -6,3 +6,4 @@ pub mod plugins;
pub mod repositories;
pub mod schema;
pub mod services;
pub mod media_utils;

71
libertas_core/src/media_utils.rs Normal file
View File

@@ -0,0 +1,71 @@
use std::path::Path;
use chrono::{DateTime, NaiveDateTime, Utc};
use nom_exif::{AsyncMediaParser, AsyncMediaSource, Exif, ExifIter, ExifTag};
use crate::error::{CoreError, CoreResult};
#[derive(Default, Debug)]
pub struct ExtractedExif {
pub width: Option<i32>,
pub height: Option<i32>,
pub location: Option<String>,
pub date_taken: Option<DateTime<Utc>>,
}
fn parse_exif_datetime(s: &str) -> Option<DateTime<Utc>> {
NaiveDateTime::parse_from_str(s, "%Y:%m:%d %H:%M:%S")
.ok()
.map(|ndt| ndt.and_local_timezone(Utc).unwrap())
}
pub async fn extract_exif_data(file_path: &Path) -> CoreResult<ExtractedExif> {
let ms = AsyncMediaSource::file_path(file_path)
.await
.map_err(|e| CoreError::Unknown(format!("Failed to open file for EXIF: {}", e)))?;
if !ms.has_exif() {
return Ok(ExtractedExif::default());
}
let mut parser = AsyncMediaParser::new();
let iter: ExifIter = match parser.parse(ms).await {
Ok(iter) => iter,
Err(e) => {
println!("Could not parse EXIF: {}", e);
return Ok(ExtractedExif::default());
}
};
let location = iter.parse_gps_info().ok().flatten().map(|g| g.format_iso6709());
let exif: Exif = iter.into();
let width = exif
.get(ExifTag::ExifImageWidth)
.and_then(|f| f.as_u32())
.map(|v| v as i32);
let height = exif
.get(ExifTag::ExifImageHeight)
.and_then(|f| f.as_u32())
.map(|v| v as i32);
let dt_original = exif
.get(ExifTag::DateTimeOriginal)
.and_then(|f| f.as_str())
.and_then(parse_exif_datetime);
let dt_modify = exif
.get(ExifTag::ModifyDate)
.and_then(|f| f.as_str())
.and_then(parse_exif_datetime);
let date_taken = dt_original.or(dt_modify);
Ok(ExtractedExif {
width,
height,
location,
date_taken,
})
}

97
libertas_importer/src/main.rs
View File

@@ -1,15 +1,21 @@
use std::{path::{Path, PathBuf}, sync::Arc};
use anyhow::Result;
use std::{
path::{Path, PathBuf},
sync::Arc,
};
use chrono::{DateTime, Datelike, NaiveDateTime, Utc};
use chrono::Datelike;
use clap::Parser;
use libertas_core::{config::Config, error::{CoreError, CoreResult}, models::{Media, User}, repositories::{MediaRepository, UserRepository}};
use libertas_core::{
config::Config, error::{CoreError, CoreResult}, media_utils::extract_exif_data, models::{Media, User}, repositories::{MediaRepository, UserRepository}
};
use libertas_infra::factory::{build_database_pool, build_media_repository, build_user_repository};
use nom_exif::{AsyncMediaParser, AsyncMediaSource, Exif, ExifIter, ExifTag};
use nom_exif::{AsyncMediaParser, AsyncMediaSource, ExifIter};
use serde_json;
use sha2::{Digest, Sha256};
use tokio::fs;
use uuid::Uuid;
use walkdir::WalkDir;
use tokio::fs;
mod config;
@@ -34,7 +40,10 @@ struct ImporterState {
#[tokio::main]
async fn main() -> Result<()> {
let cli = Cli::parse();
println!("Starting import for user: '{}' from path '{}'...", cli.username, cli.path);
println!(
"Starting import for user: '{}' from path '{}'...",
cli.username, cli.path
);
let config = config::load_config()?;
let db_pool = build_database_pool(&config.database).await?;
@@ -70,7 +79,7 @@ async fn main() -> Result<()> {
match process_file(path, &user, &state).await {
Ok(media) => {
println!("-> Imported: '{}'", media.original_filename);
},
}
Err(e) => {
eprintln!("!! Skipped: '{}' (Reason: {})", path.display(), e);
}
@@ -83,11 +92,7 @@ async fn main() -> Result<()> {
Ok(())
}
async fn process_file(
file_path: &Path,
user: &User,
state: &ImporterState,
) -> CoreResult<Media> {
async fn process_file(file_path: &Path, user: &User, state: &ImporterState) -> CoreResult<Media> {
let file_bytes = fs::read(file_path).await?;
let file_size = file_bytes.len() as i64;
let hash = format!("{:x}", Sha256::digest(&file_bytes));
@@ -107,32 +112,56 @@ async fn process_file(
));
}
let (width, height, location, date_taken) =
let (width, height, location, date_taken) = match extract_exif_data(file_path).await {
Ok(data) => {
println!(" -> Parsed EXIF: DateTimeOriginal={:?}, GPS={:?}", data.date_taken, data.location);
(data.width, data.height, data.location, data.date_taken)
},
Err(e) => {
eprintln!(" -> EXIF parsing failed for {}: {}. Skipping.", file_path.display(), e);
(None, None, None, None)
}
};
match AsyncMediaSource::file_path(file_path).await {
Ok(ms) => {
if ms.has_exif() {
let mut parser = AsyncMediaParser::new();
if let Ok(iter) = parser.parse::<_, _, ExifIter>(ms).await {
let gps = iter.parse_gps_info().ok().flatten().map(|g| g.format_iso6709());
println!(" -> EXIF GPS Info: {:?}", gps);
let exif: Exif = iter.into();
let modified_date = exif.get(ExifTag::ModifyDate).and_then(|f| f.as_str()).and_then(parse_exif_datetime);
println!(" -> EXIF ModifyDate: {:?}", modified_date);
let w = exif.get(ExifTag::ExifImageWidth).and_then(|f| f.as_u32()).map(|v| v as i32);
println!(" -> EXIF ExifImageWidth: {:?}", w);
let h = exif.get(ExifTag::ExifImageHeight).and_then(|f| f.as_u32()).map(|v| v as i32);
println!(" -> EXIF ExifImageHeight: {:?}", h);
let dt = exif.get(ExifTag::DateTimeOriginal).and_then(|f| f.as_str()).and_then(parse_exif_datetime);
println!(" -> EXIF DateTimeOriginal: {:?}", dt);
(w, h, gps, dt)
let values = iter
.into_iter()
.filter_map(|mut x| {
let res = x.take_result();
match res {
Ok(v) => Some((
x.tag().map(|x| x.to_string()).unwrap_or_else(|| {
format!("Unknown(0x{:04x})", x.tag_code())
}),
v,
)),
Err(e) => {
println!(
" !! EXIF parsing error for tag 0x{:04x}: {}",
x.tag_code(),
e
);
None
}
}
})
.collect::<Vec<_>>();
values.iter().for_each(|x| {
println!("{:<32}=> {}", x.0, x.1);
});
} else {
(None, None, None, None)
()
}
} else {
(None, None, None, None)
()
}
}
Err(_) => (None, None, None, None),
Err(_) => (),
};
let file_date = date_taken.unwrap_or_else(|| chrono::Utc::now());
@@ -174,21 +203,17 @@ async fn process_file(
};
state.media_repo.create(&media_model).await?;
state.user_repo
state
.user_repo
.update_storage_used(user.id, file_size)
.await?;
let job_payload = serde_json::json!({ "media_id": media_model.id });
state.nats_client
state
.nats_client
.publish("media.new".to_string(), job_payload.to_string().into())
.await
.map_err(|e| CoreError::Unknown(format!("Failed to publish NATS message: {}", e)))?;
Ok(media_model)
}
fn parse_exif_datetime(s: &str) -> Option<DateTime<Utc>> {
NaiveDateTime::parse_from_str(s, "%Y:%m:%d %H:%M:%S")
.ok()
.map(|ndt| ndt.and_local_timezone(Utc).unwrap())
}

1
libertas_worker/Cargo.toml
View File

@@ -22,7 +22,6 @@ sqlx = { version = "0.8.6", features = [
futures-util = "0.3.31"
bytes = "1.10.1"
uuid = { version = "1.18.1", features = ["v4", "serde"] }
nom-exif = { version = "2.5.4", features = ["serde", "tokio", "async"] }
async-trait = "0.1.89"
xmp_toolkit = "1.11.0"
chrono = "0.4.42"

53
libertas_worker/src/plugins/exif_reader.rs
View File

@@ -1,13 +1,9 @@
use async_trait::async_trait;
use chrono::{DateTime, NaiveDateTime, Utc};
use std::path::PathBuf;
use libertas_core::{
error::{CoreError, CoreResult},
models::Media,
plugins::{MediaProcessorPlugin, PluginContext, PluginData},
error::CoreResult, media_utils::extract_exif_data, models::Media, plugins::{MediaProcessorPlugin, PluginContext, PluginData}
};
use nom_exif::{AsyncMediaParser, AsyncMediaSource, Exif, ExifIter, ExifTag};
pub struct ExifReaderPlugin;
@@ -20,51 +16,15 @@ impl MediaProcessorPlugin for ExifReaderPlugin {
async fn process(&self, media: &Media, context: &PluginContext) -> CoreResult<PluginData> {
let file_path = PathBuf::from(&context.media_library_path).join(&media.storage_path);
let ms = match AsyncMediaSource::file_path(file_path).await {
Ok(ms) => ms,
Err(e) => return Err(CoreError::Unknown(format!("Failed to open a file: {}", e))),
};
if !ms.has_exif() {
return Ok(PluginData {
message: "No EXIF data found in file.".to_string(),
});
}
let mut parser = AsyncMediaParser::new();
let iter: ExifIter = match parser.parse(ms).await {
Ok(iter) => iter,
let (width, height, location, date_taken) = match extract_exif_data(&file_path).await {
Ok(data) => (data.width, data.height, data.location, data.date_taken),
Err(e) => {
// It's not a fatal error, just means parsing failed (e.g., corrupt data)
return Ok(PluginData {
message: format!("Could not parse EXIF: {}", e),
});
}
};
let location: Option<String> = match iter.parse_gps_info() {
Ok(Some(gps_info)) => Some(gps_info.format_iso6709()),
Ok(None) => None,
Err(_) => None,
};
let exif: Exif = iter.into();
let width = exif
.get(ExifTag::ExifImageWidth)
.and_then(|f| f.as_u32())
.map(|v| v as i32);
let height = exif
.get(ExifTag::ExifImageHeight)
.and_then(|f| f.as_u32())
.map(|v| v as i32);
let date_taken = exif
.get(ExifTag::DateTimeOriginal)
.and_then(|f| f.as_str())
.and_then(parse_exif_datetime);
if width.is_some() || height.is_some() || location.is_some() || date_taken.is_some() {
context
.media_repo
@@ -83,10 +43,3 @@ impl MediaProcessorPlugin for ExifReaderPlugin {
}
}
}
fn parse_exif_datetime(s: &str) -> Option<DateTime<Utc>> {
// EXIF datetime format is 'YYYY:MM:DD HH:MM:SS'
NaiveDateTime::parse_from_str(s, "%Y:%m:%d %H:%M:%S")
.ok()
.map(|ndt| ndt.and_local_timezone(Utc).unwrap())
}