refactor: improve error handling and logging for input files

Author: nabil-Tounarti

pre-compute/src/compute/errors.rs

@@ -28,7 +28,7 @@ pub enum ReplicateStatusCause {
     #[error("IS_DATASET_REQUIRED environment variable is missing")]
     PreComputeIsDatasetRequiredMissing,
     #[error("Input file download failed for input {0}")]
-    PreComputeInputFileDownloadFailed(usize),
+    PreComputeInputFileDownloadFailed(String),
     #[error("Input files number related environment variable is missing")]
     PreComputeInputFilesNumberMissing,
     #[error("Invalid dataset checksum for dataset {0}")]

pre-compute/src/compute/pre_compute_app.rs

@@ -114,7 +114,7 @@ impl PreComputeAppTrait for PreComputeApp {
     /// # Returns
     ///
     /// - `Ok(())` if all files are downloaded successfully.
-    /// - `Err(ReplicateStatusCause::PreComputeInputFileDownloadFailed)` if any file fails to download.
+    /// - `Err(ReplicateStatusCause::PreComputeInputFileDownloadFailed(url))` if any file fails to download.
     ///
     /// # Panics
     ///
@@ -126,13 +126,13 @@ impl PreComputeAppTrait for PreComputeApp {
         let args = &self.pre_compute_args;
         let chain_task_id: &str = &self.chain_task_id;
 
-        for (index, url) in args.input_files.iter().enumerate() {
+        for url in args.input_files.iter() {
             info!("Downloading input file [chainTaskId:{chain_task_id}, url:{url}]");
 
             let filename = sha256(url.to_string());
             if download_file(url, &args.output_dir, &filename).is_none() {
                 exit_causes.push(ReplicateStatusCause::PreComputeInputFileDownloadFailed(
-                    index,
+                    url.to_string(),
                 ));
             }
         }
@@ -317,7 +317,9 @@ mod tests {
         let result = app.download_input_files();
         assert_eq!(
             result.unwrap_err(),
-            vec![ReplicateStatusCause::PreComputeInputFileDownloadFailed(0)]
+            vec![ReplicateStatusCause::PreComputeInputFileDownloadFailed(
+                "https://invalid-url-that-should-fail.com/file.txt".to_string()
+            )]
         );
     }
 
@@ -339,7 +341,9 @@ mod tests {
         let result = app.download_input_files();
         assert_eq!(
             result.unwrap_err(),
-            vec![ReplicateStatusCause::PreComputeInputFileDownloadFailed(1)]
+            vec![ReplicateStatusCause::PreComputeInputFileDownloadFailed(
+                "https://invalid-url-that-should-fail.com/file.txt".to_string()
+            )]
         );
 
         // First file should be downloaded with SHA256 filename

pre-compute/src/compute/pre_compute_args.rs

@@ -69,6 +69,7 @@ impl PreComputeArgs {
         ) {
             Ok(output_dir) => output_dir,
             Err(e) => {
+                error!("Failed to read output directory: {e:?}");
                 return (PreComputeArgs::default(), vec![e]);
             }
         };
@@ -80,11 +81,13 @@ impl PreComputeArgs {
             Ok(s) => match s.to_lowercase().parse::<bool>() {
                 Ok(value) => value,
                 Err(_) => {
+                    error!("Invalid boolean format for IS_DATASET_REQUIRED: {s}");
                     exit_causes.push(ReplicateStatusCause::PreComputeIsDatasetRequiredMissing);
                     false
                 }
             },
             Err(e) => {
+                error!("Failed to read IS_DATASET_REQUIRED: {e:?}");
                 exit_causes.push(e);
                 false
             }
@@ -97,11 +100,13 @@ impl PreComputeArgs {
             Ok(s) => match s.parse::<usize>() {
                 Ok(value) => value,
                 Err(_) => {
+                    error!("Invalid numeric format for IEXEC_BULK_SLICE_SIZE: {s}");
                     exit_causes.push(ReplicateStatusCause::PreComputeFailedUnknownIssue);
                     0
                 }
             },
             Err(e) => {
+                error!("Failed to read IEXEC_BULK_SLICE_SIZE: {e:?}");
                 exit_causes.push(e);
                 0
             }
@@ -118,6 +123,7 @@ impl PreComputeArgs {
             ) {
                 Ok(filename) => filename,
                 Err(e) => {
+                    error!("Failed to read dataset {i} filename: {e:?}");
                     exit_causes.push(e);
                     continue;
                 }
@@ -129,6 +135,7 @@ impl PreComputeArgs {
             ) {
                 Ok(url) => url,
                 Err(e) => {
+                    error!("Failed to read dataset {i} URL: {e:?}");
                     exit_causes.push(e);
                     continue;
                 }
@@ -140,6 +147,7 @@ impl PreComputeArgs {
             ) {
                 Ok(checksum) => checksum,
                 Err(e) => {
+                    error!("Failed to read dataset {i} checksum: {e:?}");
                     exit_causes.push(e);
                     continue;
                 }
@@ -151,6 +159,7 @@ impl PreComputeArgs {
             ) {
                 Ok(key) => key,
                 Err(e) => {
+                    error!("Failed to read dataset {i} key: {e:?}");
                     exit_causes.push(e);
                     continue;
                 }
@@ -166,30 +175,31 @@ impl PreComputeArgs {
             Ok(s) => match s.parse::<usize>() {
                 Ok(value) => value,
                 Err(_) => {
+                    error!("Invalid numeric format for IEXEC_INPUT_FILES_NUMBER: {s}");
                     exit_causes.push(ReplicateStatusCause::PreComputeInputFilesNumberMissing);
                     0
                 }
             },
             Err(e) => {
+                error!("Failed to read IEXEC_INPUT_FILES_NUMBER: {e:?}");
                 exit_causes.push(e);
                 0
             }
         };
 
-        let input_files: Vec<String> = (1..=input_files_nb)
-            .filter_map(|i| {
-                match get_env_var_or_error(
-                    TeeSessionEnvironmentVariable::IexecInputFileUrlPrefix(i),
-                    ReplicateStatusCause::PreComputeAtLeastOneInputFileUrlMissing(i),
-                ) {
-                    Ok(url) => Some(url),
-                    Err(e) => {
-                        exit_causes.push(e);
-                        None
-                    }
+        let mut input_files: Vec<String> = Vec::new();
+        for i in 1..=input_files_nb {
+            match get_env_var_or_error(
+                TeeSessionEnvironmentVariable::IexecInputFileUrlPrefix(i),
+                ReplicateStatusCause::PreComputeAtLeastOneInputFileUrlMissing(i),
+            ) {
+                Ok(url) => input_files.push(url),
+                Err(e) => {
+                    error!("Failed to read input file {i} URL: {e:?}");
+                    exit_causes.push(e)
                 }
-            })
-            .collect();
+            }
+        }
 
         if !exit_causes.is_empty() {
             error!(