From c0c50662843d8b1879713af947e9065c39931899 Mon Sep 17 00:00:00 2001
From: Nulo <git@nulo.in>
Date: Sun, 23 Jun 2024 13:53:04 -0300
Subject: [PATCH] fix: print errors en best selling + refactor

---
 scraper-rs/src/best_selling.rs | 44 ++++++++++++++++++++--------------
 scraper-rs/src/sites/vtex.rs   | 31 ++++++++++++++++--------
 2 files changed, 47 insertions(+), 28 deletions(-)

diff --git a/scraper-rs/src/best_selling.rs b/scraper-rs/src/best_selling.rs
index 23f5cd9..420d087 100644
--- a/scraper-rs/src/best_selling.rs
+++ b/scraper-rs/src/best_selling.rs
@@ -3,8 +3,9 @@ use std::collections::HashMap;
 use crate::{build_client, db::Db, sites::vtex, supermercado::Supermercado};
 use chrono::{DateTime, Utc};
 use clap::ValueEnum;
-use futures::{stream, FutureExt, StreamExt, TryStreamExt};
+use futures::{stream, FutureExt, StreamExt};
 use itertools::Itertools;
+use simple_error::SimpleError;
 use tracing::warn;
 
 #[derive(ValueEnum, Clone, Debug)]
@@ -77,10 +78,6 @@ async fn try_get_best_selling_eans(
     }
 }
 
-async fn noop<T>(t: T) -> anyhow::Result<T> {
-    Ok(t)
-}
-
 fn rank_eans(eans: Vec<Vec<String>>) -> Vec<String> {
     let mut map: HashMap<String, usize> = HashMap::new();
     for eans in eans {
@@ -98,34 +95,45 @@ fn rank_eans(eans: Vec<Vec<String>>) -> Vec<String> {
 
 pub async fn get_all_best_selling(db: &Db) -> anyhow::Result<Vec<BestSellingRecord>> {
     let client = &build_client();
-
-    stream::iter(Category::value_variants())
+    let records = stream::iter(Category::value_variants())
         .map(|category| {
             stream::iter(Supermercado::value_variants())
                 .map(|supermercado| {
-                    let db = db.clone();
-                    let client = client.clone();
                     tokio::spawn(try_get_best_selling_eans(
-                        client,
-                        db,
+                        client.clone(),
+                        db.clone(),
                         supermercado,
                         category,
                     ))
                 })
                 .buffer_unordered(5)
                 .map(|f| f.unwrap())
-                .try_filter_map(noop)
-                .try_collect::<Vec<Vec<String>>>()
+                .filter_map(|r| async {
+                    match r {
+                        Err(err) => {
+                            tracing::error!("Error getting best selling: {}", err);
+                            None
+                        }
+                        Ok(v) => v,
+                    }
+                })
+                .collect::<Vec<Vec<String>>>()
                 .map(|r| {
-                    r.map(rank_eans).map(|eans| BestSellingRecord {
+                    let ranked = rank_eans(r);
+                    BestSellingRecord {
                         fetched_at: Utc::now(),
                         category: category.clone(),
-                        eans,
-                    })
+                        eans: ranked,
+                    }
                 })
         })
         .buffer_unordered(5)
         .boxed()
-        .try_collect()
-        .await
+        .collect::<Vec<BestSellingRecord>>()
+        .await;
+    if records.len() < 10 {
+        Err(SimpleError::new("Too few BestSellingRecords").into())
+    } else {
+        Ok(records)
+    }
 }
diff --git a/scraper-rs/src/sites/vtex.rs b/scraper-rs/src/sites/vtex.rs
index 7204151..edb36cc 100644
--- a/scraper-rs/src/sites/vtex.rs
+++ b/scraper-rs/src/sites/vtex.rs
@@ -225,19 +225,30 @@ pub async fn get_best_selling_by_category(
         url
     };
     let body = fetch_body(client, url.as_str()).await?;
-    let urls: Vec<String> = serde_json::from_str::<serde_json::Value>(&body)?
+    tracing::debug!("best selling body: {}", body);
+    let json = &serde_json::from_str::<serde_json::Value>(&body)?;
+    if let Some(errors_array) = json.pointer("/errors") {
+        if let Some(error_messages) = errors_array.as_array().map(|a| {
+            a.into_iter()
+                .map(|obj| obj.get("message").and_then(|v| v.as_str()))
+                .collect_vec()
+        }) {
+            bail!("Errors from API: {:?}", error_messages);
+        } else {
+            bail!("Unknown error from API")
+        }
+    }
+    let urls: Vec<String> = json
         .pointer("/data/productSearch/products")
         .and_then(|v| v.as_array())
-        .map(|a| {
-            a.iter()
-                .filter_map(|p| {
-                    p.get("link")
-                        .and_then(|v| v.as_str())
-                        .map(|s| format!("https://{}{}", domain, s))
-                })
-                .collect()
+        .ok_or(SimpleError::new("failed to get best selling product urls"))?
+        .iter()
+        .filter_map(|p| {
+            p.get("link")
+                .and_then(|v| v.as_str())
+                .map(|s| format!("https://{}{}", domain, s))
         })
-        .ok_or(SimpleError::new("failed to get best selling product urls"))?;
+        .collect();
 
     if urls.len() < 2 {
         bail!("Too few best selling");