feat(datasets): Add upload id functionality to datasets create command

Author: pthurlow

src/command.rs

@@ -179,7 +179,7 @@ pub enum DatasetsCommands {
         format: String,
     },
 
-    /// Create a new dataset from a file or piped stdin
+    /// Create a new dataset from a file, piped stdin, or a pre-existing upload ID
     Create {
         /// Dataset label (derived from filename if omitted)
         #[arg(long)]
@@ -190,8 +190,16 @@ pub enum DatasetsCommands {
         table_name: Option<String>,
 
         /// Path to a file to upload (omit to read from stdin)
-        #[arg(long)]
+        #[arg(long, conflicts_with = "upload_id")]
         file: Option<String>,
+
+        /// Skip upload and use a pre-existing upload ID directly
+        #[arg(long, conflicts_with = "file")]
+        upload_id: Option<String>,
+
+        /// Source format when using --upload-id (csv, json, parquet)
+        #[arg(long, default_value = "csv", value_parser = ["csv", "json", "parquet"], requires = "upload_id")]
+        format: String,
     },
 
 }

src/datasets.rs

@@ -231,6 +231,8 @@ pub fn create(
     label: Option<&str>,
     table_name: Option<&str>,
     file: Option<&str>,
+    upload_id: Option<&str>,
+    source_format: &str,
 ) {
     let profile_config = match config::load("default") {
         Ok(c) => c,
@@ -260,31 +262,42 @@ pub fn create(
                     .to_string();
                 &label_derived
             }
-            None => match stdin_redirect_filename() {
-                Some(name) => {
-                    label_derived = name;
-                    &label_derived
-                }
-                None => {
+            None => {
+                if upload_id.is_some() {
                     eprintln!("error: no label provided. Use --label to name the dataset.");
                     std::process::exit(1);
                 }
-            },
+                match stdin_redirect_filename() {
+                    Some(name) => {
+                        label_derived = name;
+                        &label_derived
+                    }
+                    None => {
+                        eprintln!("error: no label provided. Use --label to name the dataset.");
+                        std::process::exit(1);
+                    }
+                }
+            }
         },
     };
 
     let client = reqwest::blocking::Client::new();
 
-    let (upload_id, format) = match file {
-        Some(path) => upload_from_file(&client, &api_key, workspace_id, &profile_config.api_url, path),
-        None => {
-            use std::io::IsTerminal;
-            if std::io::stdin().is_terminal() {
-                eprintln!("error: no input data. Use --file <path> or pipe data via stdin.");
-                std::process::exit(1);
+    let (upload_id, format, upload_id_was_uploaded): (String, &str, bool) = if let Some(id) = upload_id {
+        (id.to_string(), source_format, false)
+    } else {
+        let (id, fmt) = match file {
+            Some(path) => upload_from_file(&client, &api_key, workspace_id, &profile_config.api_url, path),
+            None => {
+                use std::io::IsTerminal;
+                if std::io::stdin().is_terminal() {
+                    eprintln!("error: no input data. Use --file <path>, --upload-id <id>, or pipe data via stdin.");
+                    std::process::exit(1);
+                }
+                upload_from_stdin(&client, &api_key, workspace_id, &profile_config.api_url)
             }
-            upload_from_stdin(&client, &api_key, workspace_id, &profile_config.api_url)
-        }
+        };
+        (id, fmt, true)
     };
 
     let source = json!({ "upload_id": upload_id, "format": format });
@@ -312,6 +325,16 @@ pub fn create(
     if !resp.status().is_success() {
         use crossterm::style::Stylize;
         eprintln!("{}", api_error(resp.text().unwrap_or_default()).red());
+        // Only show the resume hint when the upload_id came from a fresh upload
+        if upload_id_was_uploaded {
+            eprintln!(
+                "{}",
+                format!(
+                    "Resume dataset creation without re-uploading by passing --upload-id {upload_id}"
+                )
+                .yellow()
+            );
+        }
         std::process::exit(1);
     }
 

src/main.rs

@@ -70,8 +70,8 @@ fn main() {
                         Some(DatasetsCommands::List { limit, offset, format }) => {
                             datasets::list(&workspace_id, limit, offset, &format)
                         }
-                        Some(DatasetsCommands::Create { label, table_name, file }) => {
-                            datasets::create(&workspace_id, label.as_deref(), table_name.as_deref(), file.as_deref())
+                        Some(DatasetsCommands::Create { label, table_name, file, upload_id, format }) => {
+                            datasets::create(&workspace_id, label.as_deref(), table_name.as_deref(), file.as_deref(), upload_id.as_deref(), &format)
                         }
                         None => {
                             use clap::CommandFactory;