From c6fa1602ded1b92ee0a685d8836730d0622d4f73 Mon Sep 17 00:00:00 2001
From: Paul Hauner <paul@paulhauner.com>
Date: Wed, 15 May 2019 17:14:28 +1000
Subject: [PATCH] Improve efficiency with manual YAML operations

---
 tests/ef_tests/src/cases.rs             | 34 ++++++++++++++++---------
 tests/ef_tests/src/doc.rs               | 29 +++++++++++++--------
 tests/ef_tests/src/yaml_decode/utils.rs | 11 ++++++++
 3 files changed, 51 insertions(+), 23 deletions(-)

diff --git a/tests/ef_tests/src/cases.rs b/tests/ef_tests/src/cases.rs
index 5bb1bdc2fc..90872707d9 100644
--- a/tests/ef_tests/src/cases.rs
+++ b/tests/ef_tests/src/cases.rs
@@ -16,21 +16,31 @@ pub struct Cases<T> {
 impl<T: YamlDecode> YamlDecode for Cases<T> {
     /// Decodes a YAML list of test cases
     fn yaml_decode(yaml: &String) -> Result<Self, Error> {
-        let doc = &YamlLoader::load_from_str(yaml).unwrap()[0];
+        let mut p = 0;
+        let mut elems: Vec<&str> = yaml
+            .match_indices("\n- ")
+            // Skip the `\n` used for matching a new line
+            .map(|(i, _)| i + 1)
+            .map(|i| {
+                let yaml_element = &yaml[p..i];
+                p = i;
 
-        let mut test_cases: Vec<T> = vec![];
+                yaml_element
+            })
+            .collect();
 
-        let mut i = 0;
-        loop {
-            // `is_badvalue` indicates when we have reached the end of the YAML list.
-            if doc[i].is_badvalue() {
-                break;
-            } else {
-                test_cases.push(T::yaml_decode(&yaml_to_string(&doc[i])).unwrap())
-            }
+        elems.push(&yaml[p..]);
 
-            i += 1;
-        }
+        let test_cases = elems
+            .iter()
+            .map(|s| {
+                // Remove the `- ` prefix.
+                let s = &s[2..];
+                // Remove a single level of indenting.
+                s.replace("\n  ", "\n")
+            })
+            .map(|s| T::yaml_decode(&s.to_string()).unwrap())
+            .collect();
 
         Ok(Self { test_cases })
     }
diff --git a/tests/ef_tests/src/doc.rs b/tests/ef_tests/src/doc.rs
index 1a93c738a7..2898ed3f86 100644
--- a/tests/ef_tests/src/doc.rs
+++ b/tests/ef_tests/src/doc.rs
@@ -2,7 +2,7 @@ use crate::case_result::CaseResult;
 use crate::cases::*;
 use crate::doc_header::DocHeader;
 use crate::eth_specs::{MainnetEthSpec, MinimalEthSpec};
-use crate::yaml_decode::{extract_yaml_by_key, YamlDecode};
+use crate::yaml_decode::{extract_yaml_by_key, yaml_split_header_and_cases, YamlDecode};
 use crate::EfTest;
 use serde_derive::Deserialize;
 use std::{fs::File, io::prelude::*, path::PathBuf};
@@ -10,7 +10,8 @@ use types::EthSpec;
 
 #[derive(Debug, Deserialize)]
 pub struct Doc {
-    pub yaml: String,
+    pub header_yaml: String,
+    pub cases_yaml: String,
     pub path: PathBuf,
 }
 
@@ -21,20 +22,26 @@ impl Doc {
         let mut yaml = String::new();
         file.read_to_string(&mut yaml).unwrap();
 
-        Self { yaml, path }
+        let (header_yaml, cases_yaml) = yaml_split_header_and_cases(yaml.clone());
+
+        Self {
+            header_yaml,
+            cases_yaml,
+            path,
+        }
     }
 
     pub fn test_results(&self) -> Vec<CaseResult> {
-        let header: DocHeader = serde_yaml::from_str(&self.yaml.as_str()).unwrap();
+        let header: DocHeader = serde_yaml::from_str(&self.header_yaml.as_str()).unwrap();
 
         match (
             header.runner.as_ref(),
             header.handler.as_ref(),
             header.config.as_ref(),
         ) {
-            ("ssz", "uint", _) => run_test::<SszGeneric, MainnetEthSpec>(&self.yaml),
-            ("ssz", "static", "minimal") => run_test::<SszStatic, MinimalEthSpec>(&self.yaml),
-            ("ssz", "static", "mainnet") => run_test::<SszStatic, MainnetEthSpec>(&self.yaml),
+            ("ssz", "uint", _) => run_test::<SszGeneric, MainnetEthSpec>(&self),
+            ("ssz", "static", "minimal") => run_test::<SszStatic, MinimalEthSpec>(&self),
+            ("ssz", "static", "mainnet") => run_test::<SszStatic, MainnetEthSpec>(&self),
             (runner, handler, config) => panic!(
                 "No implementation for runner: \"{}\", handler: \"{}\", config: \"{}\"",
                 runner, handler, config
@@ -55,21 +62,21 @@ impl Doc {
     }
 }
 
-pub fn run_test<T, E: EthSpec>(test_doc_yaml: &String) -> Vec<CaseResult>
+pub fn run_test<T, E: EthSpec>(doc: &Doc) -> Vec<CaseResult>
 where
     Cases<T>: EfTest + YamlDecode,
 {
     // Extract only the "test_cases" YAML as a stand-alone string.
-    let test_cases_yaml = extract_yaml_by_key(test_doc_yaml, "test_cases");
+    //let test_cases_yaml = extract_yaml_by_key(self., "test_cases");
 
     // Pass only the "test_cases" YAML string to `yaml_decode`.
-    let test_cases: Cases<T> = Cases::yaml_decode(&test_cases_yaml.to_string()).unwrap();
+    let test_cases: Cases<T> = Cases::yaml_decode(&doc.cases_yaml).unwrap();
 
     test_cases.test_results::<E>()
 }
 
 pub fn print_failures(doc: &Doc, results: &[CaseResult]) {
-    let header: DocHeader = serde_yaml::from_str(&doc.yaml).unwrap();
+    let header: DocHeader = serde_yaml::from_str(&doc.header_yaml).unwrap();
     let failures: Vec<&CaseResult> = results.iter().filter(|r| r.result.is_err()).collect();
 
     println!("--------------------------------------------------");
diff --git a/tests/ef_tests/src/yaml_decode/utils.rs b/tests/ef_tests/src/yaml_decode/utils.rs
index ae13ba3b45..9b7d7ef53e 100644
--- a/tests/ef_tests/src/yaml_decode/utils.rs
+++ b/tests/ef_tests/src/yaml_decode/utils.rs
@@ -22,3 +22,14 @@ pub fn yaml_to_string(yaml: &Yaml) -> String {
 
     out_str
 }
+
+pub fn yaml_split_header_and_cases(mut yaml: String) -> (String, String) {
+    let test_cases_start = yaml.find("\ntest_cases:\n").unwrap();
+    // + 1 to skip the \n we used for matching.
+    let mut test_cases = yaml.split_off(test_cases_start + 1);
+
+    let end_of_first_line = test_cases.find("\n").unwrap();
+    let test_cases = test_cases.split_off(end_of_first_line + 1);
+
+    (yaml, test_cases)
+}