Use exiftool to parse more metadata when needed

While the EXIF crate is pretty good when it comes to parse JPEG and
PNG metadata, it does not support non-exif metadata such as MOV
metadata.

I'd like to organise my videos like my pictures. I did not find a
Rust-native swiss army knife library. So I sadly decided to shell out
metadata parsing to the exiftool program if the exif crate does not
manage to parse the picture metadata.

Despite its name, this program can parse a wide range of picture/video
shooting metadata outside of EXIF. That's really good.

If for some reason exiftool fails to parse the file creation time, we
fallback to the unix file datetime.

Cargo.lock

@@ -325,6 +325,12 @@ dependencies = [
  "windows-sys 0.48.0",
 ]
 
+[[package]]
+name = "itoa"
+version = "1.0.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "af150ab688ff2122fcef229be89cb50dd66af9e01a4ff320cc137eecc9bacc38"
+
 [[package]]
 name = "js-sys"
 version = "0.3.64"
@@ -422,6 +428,8 @@ dependencies = [
  "indicatif",
  "kamadak-exif",
  "rayon",
+ "serde",
+ "serde_json",
 ]
 
 [[package]]
@@ -483,12 +491,49 @@ dependencies = [
  "windows-sys 0.48.0",
 ]
 
+[[package]]
+name = "ryu"
+version = "1.0.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1ad4cc8da4ef723ed60bced201181d83791ad433213d8c24efffda1eec85d741"
+
 [[package]]
 name = "scopeguard"
 version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
 
+[[package]]
+name = "serde"
+version = "1.0.188"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf9e0fcba69a370eed61bcf2b728575f726b50b55cba78064753d708ddc7549e"
+dependencies = [
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_derive"
+version = "1.0.188"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4eca7ac642d82aa35b60049a6eccb4be6be75e599bd2e9adb5f875a737654af2"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "serde_json"
+version = "1.0.107"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6b420ce6e3d8bd882e9b243c6eed35dbc9a6110c9769e74b584e0d68d1f20c65"
+dependencies = [
+ "itoa",
+ "ryu",
+ "serde",
+]
+
 [[package]]
 name = "strsim"
 version = "0.10.0"

Cargo.toml

@@ -12,6 +12,8 @@ clap = { version = "4.3.24", features = ["derive"] }
 indicatif = { version = "0.17.6", features = ["rayon"] }
 kamadak-exif = "0.5.5"
 rayon = "1.7.0"
+serde = { version = "1.0.188", features = ["derive"] }
+serde_json = "1.0.107"
 
 [profile.release]
 strip = true

default.nix

@@ -4,8 +4,14 @@ pkgs.rustPlatform.buildRustPackage {
   pname = "picobak";
   version = "0.1.0";
   src = lib.cleanSource ./.;
-  cargoHash = "sha256-gytrsYdL9WuxJDZBaK/w+1KLmAKKBD711efHTzQqs4o=";
+  cargoHash = "sha256-W0SLjlrqONMdTXoOlMilEvza2WEIVaKUJRraGR//qsw=";
   meta = {
     description = "Backup and organize your pictures library";
   };
+  nativeBuildInputs = [ pkgs.makeWrapper ];
+  # Inject exiftool
+  postInstall = ''
+    wrapProgram $out/bin/picobak \
+      --prefix PATH : "${lib.makeBinPath [pkgs.exiftool]}"
+  '';
 }

src/main.rs

@@ -1,4 +1,5 @@
 use std::fs::{create_dir_all, copy};
+use std::process::{Command};
 use std::{fs::File, path::PathBuf};
 use std::path::Path;
 use std::fmt;
@@ -9,6 +10,7 @@ use exif::{Tag, In, Value};
 use chrono::{Utc, DateTime, Datelike, NaiveDateTime};
 use indicatif::ParallelProgressIterator;
 use rayon::prelude::*;
+use serde::Deserialize;
 
 #[derive(Parser, Debug)]
 #[command(author, version, about, long_about = None)]
@@ -20,6 +22,13 @@ struct CliArgs {
     file_path: Option<String>,
 }
 
+/// Structure used to parse the JSON output of the exiftool program.
+#[derive(Debug, Deserialize)]
+struct ExifToolEntry {
+    #[serde(rename(deserialize = "CreateDate"))]
+    create_date: Option<String>
+}
+
 enum BackupSuccess {
     AlreadyBackup(String),
     Backup(String, PictureDatetimeOrigin)
@@ -33,6 +42,7 @@ enum BackupFailure {
 
 enum PictureDatetimeOrigin {
     Exif,
+    ExifTool,
     FilesystemMetadata
 }
 
@@ -76,6 +86,7 @@ fn main() {
 
 fn display_backup_result(results: Vec<Result<BackupSuccess, BackupFailure>>) {
     let mut nb_copy_exif: u32 = 0;
+    let mut nb_copy_exiftool: u32 = 0;
     let mut nb_copy_filesystem: u32 = 0;
     let mut nb_duplicates: u32 = 0;
     let mut failures: Vec<BackupFailure> = Vec::new();
@@ -85,6 +96,7 @@ fn display_backup_result(results: Vec<Result<BackupSuccess, BackupFailure>>) {
                 BackupSuccess::AlreadyBackup(_) => nb_duplicates +=1,
                 BackupSuccess::Backup(_, origin) => match origin {
                     PictureDatetimeOrigin::Exif => nb_copy_exif +=1,
+                    PictureDatetimeOrigin::ExifTool => nb_copy_exiftool +=1,
                     PictureDatetimeOrigin::FilesystemMetadata => nb_copy_filesystem +=1
                 }
             }
@@ -97,6 +109,7 @@ fn display_backup_result(results: Vec<Result<BackupSuccess, BackupFailure>>) {
     eprintln!("Copied: {}", nb_copy_exif + nb_copy_filesystem);
     eprintln!("To classify these newly copied files, we used:");
     eprintln!("   {}: EXIF metadata", nb_copy_exif);
+    eprintln!("   {}: the exiftool program", nb_copy_exiftool);
     eprintln!("   {}: filesystem metadata", nb_copy_filesystem);
     eprintln!("Failures: {}", failures.len());
     if failures.len() != 0 {
@@ -153,11 +166,14 @@ fn upsert_picture_directory(picture_dir: &PathBuf) {
 /// If no datetime EXIF data is attached to the file, use the file
 /// last modification date.
 fn get_picture_datetime(file_path: &str, file: &File) -> (DateTime<Utc>, PictureDatetimeOrigin) {
-    let exif_datetime = get_picture_exif_datetime(file);
-    match exif_datetime {
-        Some(dt) => (dt, PictureDatetimeOrigin::Exif),
-        None => (get_file_modified_time(file_path, file), PictureDatetimeOrigin::FilesystemMetadata)
-    }
+    // Try exif crate.
+    get_picture_exif_datetime(file).map(|dt| (dt, PictureDatetimeOrigin::Exif))
+        // Exif failed, shell out to exiftool.
+        .or_else(|| {
+                 get_picture_exiftool_datetime(file_path)
+                .map(|dt| (dt, PictureDatetimeOrigin::ExifTool))})
+        // Exiftool failed as well. Fallback to Unix datetime.
+        .unwrap_or((get_file_modified_time(file_path, file), PictureDatetimeOrigin::FilesystemMetadata))
 }
 
 /// Retrieves the picture EXIF datetime.
@@ -178,6 +194,29 @@ fn get_picture_exif_datetime(file: &File) -> Option<DateTime<Utc>> {
     }
 }
 
+/// Shells out to the exiftool CLI. Despite its name, exiftool parses
+/// much more metadata than exif. Such as MOV metadata.
+fn get_picture_exiftool_datetime(file_path: &str) -> Option<DateTime<Utc>> {
+    let output = Command::new("exiftool")
+        .args(["-j", "-P", "-CreateDate", file_path])
+        .output()
+        .ok()?;
+    if !output.status.success() {
+        return None
+    }
+    let stdout_str = String::from_utf8(output.stdout).ok()?;
+    let parsed_output: Vec<ExifToolEntry> = serde_json::from_str(&stdout_str).ok()?;
+    if parsed_output.len() != 1 {
+        None
+    } else {
+        let entry = parsed_output.get(0)?;
+        let date: &str = &entry.create_date.as_ref()?;
+        NaiveDateTime::parse_from_str(date, "%Y:%m:%d %H:%M:%S")
+            .map(|naive_datetime| DateTime::from_utc(naive_datetime, Utc))
+            .ok()
+    }
+}
+
 /// If we cannot load the EXIF creation datetime, we end up using the
 /// last modified time of the file.
 fn get_file_modified_time(file_path: &str, file: &File) -> DateTime<Utc> {
@@ -212,6 +251,16 @@ fn validate_args(args: &CliArgs) {
     if Path::new(&args.backup_root).is_file() {
         panic!("ERROR: {} is a file, not a valid backup dir", &args.backup_root);
     };
+
+    let exif_tool_in_path = Command::new("bash")
+        .args(["-c", "command exiftool"])
+        .output()
+        .ok()
+        .map(|e| e.status.success())
+        .unwrap();
+    if !exif_tool_in_path {
+        eprintln!("Exiftool doesn't seem to be present in $PATH. Install it if you want to be able to extract more pictures metadata");
+    }
 }
 
 /// Compare two files and check if they're the same. We're not really