More robust and better error reporting of errors in fetch (#1438)

Potentially fixes #1437 although the cause of the failure is unknown.
This patch should at least give us better insight to what is the cause.

EDIT: with several successful runs, it seems that the fix was being
stricter on the status code. Nevertheless, the rest of the code helps
better tracing potential issues that might arise.

Author: beta-ziliani

crates/solidity/testing/perf/cargo/src/dataset.rs

@@ -3,7 +3,7 @@ use std::fs;
 use std::path::Path;
 use std::sync::OnceLock;
 
-use anyhow::{anyhow, Result};
+use anyhow::{anyhow, bail, Result};
 use semver::{BuildMetadata, Prerelease};
 use serde::Deserialize;
 use slang_solidity::compilation::{CompilationBuilder, CompilationBuilderConfig, CompilationUnit};
@@ -142,8 +142,14 @@ fn import_resolver_from<'a, T: Iterator<Item = &'a String>>(
 
 impl SolidityProject {
     pub fn build(json_file: &Path) -> Result<Self> {
-        let json_str = fs::read_to_string(json_file)?;
-        let metadata: Metadata = serde_json::from_str(&json_str)?;
+        let json_str = match fs::read_to_string(json_file) {
+            Ok(json_str) => json_str,
+            Err(e) => bail!("Error reading file {json_file:?}: {e}"),
+        };
+        let metadata = match serde_json::from_str::<Metadata>(&json_str) {
+            Ok(metadata) => metadata,
+            Err(e) => bail!("Error parsing file {json_file:?}: {e}\nWhile parsing:\n{json_str}"),
+        };
         let sself: Self = Self::try_from(metadata)?;
         Ok(sself)
     }

crates/solidity/testing/utils/src/fetch.rs

@@ -2,10 +2,12 @@ use std::path::Path;
 use std::time::Duration;
 use std::{fs, thread};
 
-use anyhow::{Ok, Result};
-use reqwest::blocking::get;
+use anyhow::{bail, Result};
+use reqwest::header::CONTENT_TYPE;
 use serde_json::Value;
 
+const MAX_RETRIES: u64 = 6;
+
 // Given an address and a path, it downloads the json file from sourcify,
 // and stores it in the path.
 pub fn fetch(address: &str, base_path: &Path) -> Result<()> {
@@ -20,21 +22,56 @@ pub fn fetch(address: &str, base_path: &Path) -> Result<()> {
     let url =
         format!("https://sourcify.dev/server/v2/contract/1/{address}/?fields=sources,compilation");
 
-    let mut body = get(&url);
-
     // Try with exponential backoff
-    let mut tries = 0;
-    while body.is_err() && tries < 6 {
-        tries += 1;
-        thread::sleep(Duration::from_secs(2 ^ tries));
-        body = get(&url);
-    }
+    let mut retries = 0;
+    let project = loop {
+        retries += 1;
 
-    let project_json: Value = body?.json()?;
+        match try_fetch_project(&url) {
+            Ok(project) => break project,
+            Err(err) => {
+                println!("Error fetching {url} (attempt {retries}/{MAX_RETRIES}): {err}");
+                if retries <= MAX_RETRIES {
+                    thread::sleep(Duration::from_secs(2 ^ retries));
+                } else {
+                    bail!("Giving up after {retries} attempts.");
+                }
+            }
+        }
+    };
 
-    let content = serde_json::to_string_pretty(&project_json)?;
+    // write project to disk
     fs::create_dir_all(base_path)?;
-    fs::write(project_file_path, content)?;
+    fs::write(project_file_path, project)?;
 
     Ok(())
 }
+
+fn try_fetch_project(url: &str) -> Result<String> {
+    let response = reqwest::blocking::get(url)?;
+
+    if response.status() != reqwest::StatusCode::OK {
+        bail!("Status code is {status}", status = response.status());
+    }
+
+    let Some(content_type) = response.headers().get(CONTENT_TYPE) else {
+        bail!("No content-type header")
+    };
+
+    let content_type_str = content_type.to_str().unwrap_or("");
+    if !content_type_str
+        .trim()
+        .to_ascii_lowercase()
+        .starts_with("application/json")
+    {
+        bail!("content-type is not json: {content_type_str:?}");
+    }
+
+    if response.content_length().unwrap_or_default() == 0 {
+        bail!("body is empty");
+    }
+
+    let project_json: Value = response.json()?;
+
+    Ok(serde_json::to_string_pretty(&project_json)?)
+}