From bb77165f704b413a531e6cd4c29ce66b33675e09 Mon Sep 17 00:00:00 2001
From: Krystian Chmielewski <krystian.chmielewski@intel.com>
Date: Fri, 29 Oct 2021 11:18:09 +0000
Subject: [PATCH] remove regex from yaml parser

Signed-off-by: Krystian Chmielewski <krystian.chmielewski@intel.com>
---
 .../device_binary_format/yaml/yaml_parser.cpp | 39 ++++++++++++++++++-
 .../device_binary_format/yaml/yaml_parser.h   |  3 +-
 .../yaml/yaml_parser_tests.cpp                | 14 ++++++-
 3 files changed, 51 insertions(+), 5 deletions(-)

diff --git a/shared/source/device_binary_format/yaml/yaml_parser.cpp b/shared/source/device_binary_format/yaml/yaml_parser.cpp
index 69eba7f34d..3c389fd0dd 100644
--- a/shared/source/device_binary_format/yaml/yaml_parser.cpp
+++ b/shared/source/device_binary_format/yaml/yaml_parser.cpp
@@ -108,6 +108,43 @@ bool tokenizeEndLine(ConstStringRef text, LinesCache &outLines, TokensCache &out
     return true;
 }
 
+bool isValidInlineCollectionFormat(const char *context, const char *contextEnd) {
+    auto consumeAlphaNum = [](const char *&text) {
+        while (isAlphaNumeric(*text)) {
+            text++;
+        }
+    };
+
+    bool endNum = false;
+    bool endCollection = false;
+    context++; // skip '['
+    while (context < contextEnd && *context != '\n') {
+        if (isWhitespace(*context)) {
+            context++;
+        } else if (false == endNum) {
+            if (isAlphaNumeric(*context)) {
+                consumeAlphaNum(context);
+                endNum = true;
+            } else {
+                return false;
+            }
+        } else if (false == endCollection) {
+            if (*context == ',') {
+                context++;
+                endNum = false;
+            } else if (*context == ']') {
+                context++;
+                endCollection = true;
+            } else {
+                return false;
+            }
+        } else {
+            return false;
+        }
+    }
+    return endCollection;
+}
+
 bool tokenize(ConstStringRef text, LinesCache &outLines, TokensCache &outTokens, std::string &outErrReason, std::string &outWarning) {
     if (text.empty()) {
         outWarning.append("NEO::Yaml : input text is empty\n");
@@ -200,7 +237,7 @@ bool tokenize(ConstStringRef text, LinesCache &outLines, TokensCache &outTokens,
             break;
         }
         case '[':
-            if (false == std::regex_search(context.pos, inlineCollectionRegex)) {
+            if (false == isValidInlineCollectionFormat(context.pos, text.end())) {
                 outErrReason = constructYamlError(outLines.size(), context.lineBeginPos, context.pos, inlineCollectionYamlErrorMsg.data());
                 return false;
             }
diff --git a/shared/source/device_binary_format/yaml/yaml_parser.h b/shared/source/device_binary_format/yaml/yaml_parser.h
index d4856f5386..315540a00d 100644
--- a/shared/source/device_binary_format/yaml/yaml_parser.h
+++ b/shared/source/device_binary_format/yaml/yaml_parser.h
@@ -12,7 +12,6 @@
 #include "shared/source/utilities/stackvec.h"
 
 #include <iterator>
-#include <regex>
 #include <string>
 
 namespace NEO {
@@ -256,7 +255,7 @@ using LinesCache = StackVec<Line, 512>;
 
 std::string constructYamlError(size_t lineNumber, const char *lineBeg, const char *parsePos, const char *reason = nullptr);
 
-static std::regex inlineCollectionRegex(R"regex(^\[(\s*(\d|\w)+,?)+\s*\]\s*\n)regex");
+bool isValidInlineCollectionFormat(const char *context, const char *contextEnd);
 constexpr ConstStringRef inlineCollectionYamlErrorMsg = "NEO::Yaml : Inline collection is not in valid regex format - ^\\[(\\s*(\\d|\\w)+,?)+\\s*\\]\\s*\\n";
 
 bool tokenize(ConstStringRef text, LinesCache &outLines, TokensCache &outTokens, std::string &outErrReason, std::string &outWarning);
diff --git a/shared/test/unit_test/device_binary_format/yaml/yaml_parser_tests.cpp b/shared/test/unit_test/device_binary_format/yaml/yaml_parser_tests.cpp
index ad4f7463c5..8633290587 100644
--- a/shared/test/unit_test/device_binary_format/yaml/yaml_parser_tests.cpp
+++ b/shared/test/unit_test/device_binary_format/yaml/yaml_parser_tests.cpp
@@ -166,6 +166,11 @@ TEST(YamlIsMatched, WhenTextIsTooShortThenReturnFalse) {
     EXPECT_FALSE(NEO::Yaml::isMatched(text, text.begin(), "bcd"));
 }
 
+TEST(YamlIsValidInlineCollectionFormat, WhenEndIsReachedThenReturnFalse) {
+    const char coll[8] = "[ 1, 2 ";
+    EXPECT_FALSE(NEO::Yaml::isValidInlineCollectionFormat(coll, coll + 7));
+}
+
 TEST(YamlConsumeNumberOrSign, GivenValidNumberOrSignThenReturnProperEndingPosition) {
     ConstStringRef plus5 = "a+5";
     ConstStringRef minus7 = "b  -7 [";
@@ -479,7 +484,12 @@ TEST(YamlTokenize, GivenInvalidInlineCollectionThenEmitsError) {
     warnings.clear();
     errors.clear();
 
-    success = NEO::Yaml::tokenize("[[1,2,3,4]]\n", lines, tokens, errors, warnings);
+    success = NEO::Yaml::tokenize("[[1,2,3,4]\n", lines, tokens, errors, warnings);
+    EXPECT_FALSE(success);
+    EXPECT_STREQ("NEO::Yaml : Could not parse line : [0] : [[] <-- parser position on error. Reason : NEO::Yaml : Inline collection is not in valid regex format - ^\\[(\\s*(\\d|\\w)+,?)+\\s*\\]\\s*\\n\n", errors.c_str());
+    EXPECT_TRUE(warnings.empty()) << warnings;
+
+    success = NEO::Yaml::tokenize("[1 2 3 4]\n", lines, tokens, errors, warnings);
     EXPECT_FALSE(success);
     EXPECT_STREQ("NEO::Yaml : Could not parse line : [0] : [[] <-- parser position on error. Reason : NEO::Yaml : Inline collection is not in valid regex format - ^\\[(\\s*(\\d|\\w)+,?)+\\s*\\]\\s*\\n\n", errors.c_str());
     EXPECT_TRUE(warnings.empty()) << warnings;
@@ -1317,7 +1327,7 @@ TEST(YamlBuildTree, GivenInlineCollectionThenProperlyCreatesChildNodes) {
         R"===(
       banana : yellow
       kiwi : green
-      apple : [red, green, blue]
+      apple : [ red, green, blue ]
       pear : pearish
 )===";