Ingestable: use ingest_path for local paths

This way we don't error-out when a path is a directory. That said, we're
still only including the root in sidx.db (e.g. manifests/ but not
manifests/*.json). We should change that next.

Also renamed "blob_id" to "blake3_id" because datasette has a special
branch for ${column}_id referencing a table that contains ${column}

src/main.rs

@@ -1,16 +1,14 @@
-use std::{
-    path::{absolute, PathBuf},
-    pin::Pin,
-};
+use std::path::{absolute, PathBuf};
 
 use anyhow::anyhow;
+use anyhow::Context;
 use clap::Parser;
 use futures::{stream, StreamExt, TryStreamExt};
 use rusqlite::{params, OptionalExtension};
-use snix_castore::{blobservice, directoryservice};
+use snix_castore::{blobservice, directoryservice, import::fs::ingest_path};
 use url::Url;
 
-#[derive(Clone)]
+#[derive(Clone, Debug)]
 enum Ingestable {
     Url(Url),
     Path(PathBuf),
@@ -143,7 +141,7 @@ async fn main() {
     let samples = stream::iter(args.inputs.iter().map(|uri| {
         let client = &client;
         let blob_service = &blob_service;
-        let _dir_service = &dir_service;
+        let dir_service = &dir_service;
         let con = &con;
         let mut find_sample = con
             .prepare(include_str!("q/latest-download.sql"))
@@ -174,40 +172,47 @@ async fn main() {
                     }));
                 }
             }
-            let mut r: Pin<Box<dyn tokio::io::AsyncRead>> = {
-                match uri {
-                    Ingestable::Path(path) => match tokio::fs::File::open(path).await {
-                        Ok(f) => Box::pin(f),
-                        Err(e) => {
-                            return Err(anyhow!("Failed to read {:?}: {}", path, e));
+            let (digest, n_bytes) = match uri {
+                Ingestable::Path(path) => {
+                    match ingest_path::<_, _, _, &[u8]>(&blob_service, &dir_service, path, None)
+                        .await?
+                    {
+                        snix_castore::Node::Directory { digest, size } => (digest, size),
+                        snix_castore::Node::File {
+                            digest,
+                            size,
+                            executable: _,
+                        } => (digest, size),
+                        snix_castore::Node::Symlink { target: _ } => {
+                            return Err(anyhow!("TODO: Figure out what to do with symlink roots"))
                         }
-                    },
-                    Ingestable::Url(url) => {
-                        let res = match client.get(url.clone()).send().await {
-                            Ok(res) => res.error_for_status()?,
-                            Err(e) => {
-                                return Err(anyhow!("Failed to GET {}: {}", url, e));
-                            }
-                        };
-                        let r = tokio_util::io::StreamReader::new(
-                            res.bytes_stream().map_err(std::io::Error::other),
-                        );
-                        Box::pin(r)
                     }
                 }
-            };
-            let mut w = blob_service.open_write().await;
-            let n_bytes = match tokio::io::copy(&mut r, &mut w).await {
-                Ok(n) => n,
-                Err(e) => {
-                    return Err(anyhow!(
-                        "tokio::io::copy failed for uri={} with {}",
-                        uri_s,
-                        e
-                    ));
+                Ingestable::Url(url) => {
+                    let res = client
+                        .get(url.clone())
+                        .send()
+                        .await
+                        .context(format!("Request.send failed early for {:?}", uri))?
+                        .error_for_status()?;
+                    let mut r = tokio_util::io::StreamReader::new(
+                        res.bytes_stream().map_err(std::io::Error::other),
+                    );
+                    let mut w = blob_service.open_write().await;
+                    let n_bytes = match tokio::io::copy(&mut r, &mut w).await {
+                        Ok(n) => n,
+                        Err(e) => {
+                            return Err(anyhow!(
+                                "tokio::io::copy failed for uri={} with {}",
+                                uri_s,
+                                e
+                            ));
+                        }
+                    };
+                    let digest = w.close().await?;
+                    (digest, n_bytes)
                 }
             };
-            let digest = w.close().await?;
             let digest64 = format!("{}", digest);
             add_blob.execute(params![digest64, n_bytes,])?;
             add_uri.execute(params![uri_s])?;
@@ -229,7 +234,7 @@ async fn main() {
     for s in samples {
         match s {
             Err(e) => {
-                println!("Failed to fetch ...: {}", e);
+                println!("Failed to fetch: {}", e);
             }
             Ok(None) => {}
             Ok(Some(ingested)) => {

src/q/add-sample.sql

@@ -1,4 +1,4 @@
-INSERT INTO sidx_uri_sample(uri_id, blob_id)
+INSERT INTO sidx_uri_sample(uri_id, blake3_id)
 VALUES(
     (
         SELECT
@@ -13,7 +13,7 @@ VALUES(
         SELECT
             id
         FROM
-            sidx_blob
+            sidx_blake3
         WHERE
             blake3 = ?
     )

src/q/init.sql

@@ -3,20 +3,20 @@ CREATE TABLE IF NOT EXISTS sidx_uri(
     uri TEXT UNIQUE,
     PRIMARY KEY(id)
     );
-CREATE TABLE IF NOT EXISTS sidx_blob(
+CREATE TABLE IF NOT EXISTS sidx_blake3(
     id INTEGER,
-    blake3 TEXT UNIQUE,
+    blake3 TEXT UNIQUE, /* snix-castore node */
     n_bytes INTEGER NOT NULL,
     PRIMARY KEY(id)
     );
 CREATE TABLE IF NOT EXISTS sidx_uri_sample(
     id INTEGER,
     uri_id INTEGER NOT NULL,
-    blob_id INTEGER,
+    blake3_id INTEGER,
     epoch INTEGER NOT NULL DEFAULT (unixepoch()),
     PRIMARY KEY(id),
     FOREIGN KEY(uri_id) REFERENCES sidx_uri(id),
-    FOREIGN KEY(blob_id) REFERENCES sidx_blob(id)
+    FOREIGN KEY(blake3_id) REFERENCES sidx_blake3(id)
 );
-CREATE INDEX IF NOT EXISTS sidx_uri_blob_idx
-ON sidx_uri_sample(uri_id, blob_id, epoch);
+CREATE INDEX IF NOT EXISTS sidx_uri_blake3_idx
+ON sidx_uri_sample(uri_id, blake3_id, epoch);

src/q/latest-download.sql

@@ -5,10 +5,10 @@ SELECT
 FROM
     sidx_uri_sample AS s,
     sidx_uri AS u,
-    sidx_blob AS b
+    sidx_blake3 AS b
 ON
     s.uri_id = u.id
-    AND s.blob_id = b.id
+    AND s.blake3_id = b.id
 WHERE
     u.uri = ?
 ORDER BY

src/q/upsert-blob.sql

@@ -1,4 +1,4 @@
-INSERT INTO sidx_blob(blake3, n_bytes)
+INSERT INTO sidx_blake3(blake3, n_bytes)
 VALUES
 (?, ?)
 ON CONFLICT DO NOTHING;