Improve performance of VariableOrigin instances

eduar-hte · eduar-hte · commit dc0a06fc70a1 · 2024-07-17T00:49:27.000Z
- The previous approach would create a std::unique_ptr and store it in
  a std::list in VariableValue (Origins)
- The new approach now stores Origins in a std::vector and constructs
  VariableOrigin elements in-place on insertion.
- Instead of having two heap-allocations for every added VariableOrigin
  instance, this performs only one.
- If multiple origins are added, std::vector's growth strategy may even
  prevent a heap-allocation. There's a cost on growing the size of the
  vector, because a copy of current elements will be necessary.
  - Introduced reserveOrigin method to notify that multiple insertions
    will be made, so that we can use std::vector's reserve and do a
    single allocation (and copy of previous elements), and then just
    initialize the new elements in-place.
diff --git a/headers/modsecurity/anchored_set_variable_translation_proxy.h b/headers/modsecurity/anchored_set_variable_translation_proxy.h
@@ -47,11 +47,12 @@ class AnchoredSetVariableTranslationProxy {
                 VariableValue *newVariableValue = new VariableValue(name, &l->at(i)->getKey(), &l->at(i)->getKey());
                 const VariableValue *oldVariableValue = l->at(i);
                 l->at(i) = newVariableValue;
+                newVariableValue->reserveOrigin(oldVariableValue->getOrigin().size());
                 for (const auto &oldOrigin : oldVariableValue->getOrigin()) {
-                    std::unique_ptr<VariableOrigin> newOrigin(new VariableOrigin);
-                    newOrigin->m_length = oldVariableValue->getKey().size();
-                    newOrigin->m_offset = oldOrigin->m_offset - oldVariableValue->getKey().size() - 1;
-                    newVariableValue->addOrigin(std::move(newOrigin));
+                    newVariableValue->addOrigin(
+                        oldVariableValue->getKey().size(),
+                        oldOrigin.m_offset - oldVariableValue->getKey().size() - 1
+                    );
                 }
                 delete oldVariableValue;
             }
diff --git a/headers/modsecurity/anchored_variable.h b/headers/modsecurity/anchored_variable.h
@@ -63,7 +63,7 @@ class AnchoredVariable {
     void append(const std::string &a, size_t offset,
         bool spaceSeparator = false);
     void append(const std::string &a, size_t offset,
-        bool spaceSeparator, int size);
+        bool spaceSeparator, size_t size);
 
     void evaluate(std::vector<const VariableValue *> *l);
     std::string *  evaluate();
diff --git a/headers/modsecurity/variable_origin.h b/headers/modsecurity/variable_origin.h
@@ -15,6 +15,7 @@
 
 #ifdef __cplusplus
 #include <string>
+#include <memory>
 #endif
 
 #ifndef HEADERS_MODSECURITY_VARIABLE_ORIGIN_H_
@@ -36,14 +37,17 @@ class VariableOrigin {
     VariableOrigin()
         : m_length(0),
         m_offset(0) { }
+    VariableOrigin(size_t length, size_t offset)
+        : m_length(length),
+        m_offset(offset) { } 
 
-    std::string toText() {
-        std::string offset = std::to_string(m_offset);
-        std::string len = std::to_string(m_length);
+    std::string toText() const {
+        const auto offset = std::to_string(m_offset);
+        const auto len = std::to_string(m_length);
         return "v" + offset + "," + len;
     }
 
-    int m_length;
+    size_t m_length;
     size_t m_offset;
 };
 
diff --git a/headers/modsecurity/variable_value.h b/headers/modsecurity/variable_value.h
@@ -18,7 +18,7 @@
 #include <string>
 #include <iostream>
 #include <memory>
-#include <list>
+#include <vector>
 #include <utility>
 #endif
 
@@ -37,7 +37,7 @@ namespace modsecurity {
 class Collection;
 class VariableValue {
  public:
-    using Origins = std::list<std::unique_ptr<VariableOrigin>>;
+    using Origins = std::vector<VariableOrigin>;
 
     explicit VariableValue(const std::string *key,
         const std::string *value = nullptr)
@@ -62,11 +62,9 @@ class VariableValue {
         m_keyWithCollection(o->m_keyWithCollection),
         m_value(o->m_value)
     {
+        reserveOrigin(o->m_orign.size());
         for (const auto &i : o->m_orign) {
-            std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
-            origin->m_offset = i->m_offset;
-            origin->m_length = i->m_length;
-            m_orign.push_back(std::move(origin));
+            addOrigin(i);
         }
     }
 
@@ -98,15 +96,27 @@ class VariableValue {
     }
 
 
-    void addOrigin(std::unique_ptr<VariableOrigin> origin) {
-        m_orign.push_back(std::move(origin));
+    void addOrigin(const VariableOrigin &origin) {
+        m_orign.emplace_back(origin);
+    }
+
+
+    template<typename... Args>
+    void addOrigin(Args&&... args) {
+        m_orign.emplace_back(args...);
     }
 
 
     const Origins& getOrigin() const {
         return m_orign;
     }
 
+
+    void reserveOrigin(Origins::size_type additionalSize) {
+        m_orign.reserve(m_orign.size() + additionalSize);
+    }
+
+
  private:
     Origins m_orign;
     std::string m_collection;
diff --git a/src/anchored_set_variable.cc b/src/anchored_set_variable.cc
@@ -52,26 +52,16 @@ void AnchoredSetVariable::unset() {
 
 void AnchoredSetVariable::set(const std::string &key,
     const std::string &value, size_t offset, size_t len) {
-    std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
     VariableValue *var = new VariableValue(&m_name, &key, &value);
-
-    origin->m_offset = offset;
-    origin->m_length = len;
-
-    var->addOrigin(std::move(origin));
+    var->addOrigin(len, offset);
     emplace(key, var);
 }
 
 
 void AnchoredSetVariable::set(const std::string &key,
     const std::string &value, size_t offset) {
-    std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
     VariableValue *var = new VariableValue(&m_name, &key, &value);
-
-    origin->m_offset = offset;
-    origin->m_length = value.size();
-
-    var->addOrigin(std::move(origin));
+    var->addOrigin(value.size(), offset);
     emplace(key, var);
 }
 
diff --git a/src/anchored_variable.cc b/src/anchored_variable.cc
@@ -54,58 +54,40 @@ void AnchoredVariable::unset() {
 
 void AnchoredVariable::set(const std::string &a, size_t offset,
     size_t offsetLen) {
-    std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
-
     m_offset = offset;
     m_value.assign(a.c_str(), a.size());
-    origin->m_offset = offset;
-    origin->m_length = offsetLen;
-    m_var->addOrigin(std::move(origin));
+    m_var->addOrigin(offsetLen, offset);
 }
 
 
 void AnchoredVariable::set(const std::string &a, size_t offset) {
-    std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
-
     m_offset = offset;
     m_value.assign(a.c_str(), a.size());
-    origin->m_offset = offset;
-    origin->m_length = m_value.size();
-    m_var->addOrigin(std::move(origin));
+    m_var->addOrigin(m_value.size(), offset);
 }
 
 
 void AnchoredVariable::append(const std::string &a, size_t offset,
     bool spaceSeparator) {
-    std::unique_ptr<VariableOrigin> origin(
-        new VariableOrigin());
-
     if (spaceSeparator && !m_value.empty()) {
         m_value.append(" " + a);
     } else {
         m_value.append(a);
     }
     m_offset = offset;
-    origin->m_offset = offset;
-    origin->m_length = a.size();
-    m_var->addOrigin(std::move(origin));
+    m_var->addOrigin(a.size(), offset);
 }
 
 
 void AnchoredVariable::append(const std::string &a, size_t offset,
-    bool spaceSeparator, int size) {
-    std::unique_ptr<VariableOrigin> origin(
-        new VariableOrigin());
-
+    bool spaceSeparator, size_t size) {
     if (spaceSeparator && !m_value.empty()) {
         m_value.append(" " + a);
     } else {
         m_value.append(a);
     }
     m_offset = offset;
-    origin->m_offset = offset;
-    origin->m_length = size;
-    m_var->addOrigin(std::move(origin));
+    m_var->addOrigin({size, offset});
 }
 
 
diff --git a/src/rule_with_operator.cc b/src/rule_with_operator.cc
@@ -317,8 +317,8 @@ bool RuleWithOperator::evaluate(Transaction *trans,
                 if (ret == true) {
                     ruleMessage->m_match = m_operator->resolveMatchMessage(trans,
                         key, value);
-                    for (auto &i : v->getOrigin()) {
-                        ruleMessage->m_reference.append(i->toText());
+                    for (const auto &i : v->getOrigin()) {
+                        ruleMessage->m_reference.append(i.toText());
                     }
 
                     ruleMessage->m_reference.append(*valueTemp.second);
diff --git a/src/variables/remote_user.cc b/src/variables/remote_user.cc
@@ -39,50 +39,41 @@ namespace variables {
 void RemoteUser::evaluate(Transaction *transaction,
     RuleWithActions *rule,
     std::vector<const VariableValue *> *l) {
-    size_t pos;
-    std::string base64;
-    VariableValue *var;
-    std::string header;
+    std::vector<const VariableValue *> l2;
 
-    std::vector<const VariableValue *> *l2 = \
-	new std::vector<const VariableValue *>();
-    transaction->m_variableRequestHeaders.resolve("authorization", l2);
+    transaction->m_variableRequestHeaders.resolve("authorization", &l2);
 
-    if (l2->size() < 1) {
-        goto clear;
-    }
+    if (!l2.empty()) {
+        const auto *v = l2[0];
 
-    header = std::string(l2->at(0)->getValue());
+        const auto &header = v->getValue();
 
-    if (header.compare(0, 6, "Basic ") == 0) {
-        base64 = std::string(header, 6, header.length());
-    }
+        std::string base64;
 
-    base64 = Utils::Base64::decode(base64);
+        if (header.compare(0, 6, "Basic ") == 0) {
+            base64 = std::string(header, 6, header.length());
+        }
 
-    pos = base64.find(":");
-    if (pos == std::string::npos) {
-        goto clear;
-    }
-    transaction->m_variableRemoteUser.assign(std::string(base64, 0, pos));
+        base64 = Utils::Base64::decode(base64);
 
-    var = new VariableValue(&l2->at(0)->getKeyWithCollection(),
-        &transaction->m_variableRemoteUser);
+        const auto pos = base64.find(":");
+        if (pos != std::string::npos) {
+            transaction->m_variableRemoteUser.assign(std::string(base64, 0, pos));
 
-    for (const auto &i : l2->at(0)->getOrigin()) {
-        std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
-        origin->m_offset = i->m_offset;
-        origin->m_length = i->m_length;
-        var->addOrigin(std::move(origin));
-    }
-    l->push_back(var);
+            auto var = std::make_unique<VariableValue>(&v->getKeyWithCollection(),
+                &transaction->m_variableRemoteUser);
+
+            var->reserveOrigin(v->getOrigin().size());
+            for (const auto &i : v->getOrigin()) {
+                var->addOrigin(i);
+            }
+            l->push_back(var.release());
+        }
 
-clear:
-    for (auto &a : *l2) {
-        delete a;
+        for (auto &a : l2) {
+            delete a;
+        }
     }
-    l2->clear();
-    delete l2;
 }
 
 
diff --git a/src/variables/rule.h b/src/variables/rule.h
@@ -49,15 +49,12 @@ class Rule_DictElement : public VariableDictElement { \
         if (!r || r->m_ruleId == 0) {
             return;
         }
-        std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
         std::string *a = new std::string(std::to_string(r->m_ruleId));
         VariableValue *var = new VariableValue(&m_rule, &m_rule_id,
             a
         );
         delete a;
-        origin->m_offset = 0;
-        origin->m_length = 0;
-        var->addOrigin(std::move(origin));
+        var->addOrigin();
         l->push_back(var);
     }
 
@@ -75,15 +72,12 @@ class Rule_DictElement : public VariableDictElement { \
             return;
         }
 
-        std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
         std::string *a = new std::string(r->m_rev);
         VariableValue *var = new VariableValue(&m_rule, &m_rule_rev,
             a
         );
         delete a;
-        origin->m_offset = 0;
-        origin->m_length = 0;
-        var->addOrigin(std::move(origin));
+        var->addOrigin();
         l->push_back(var);
     }
 
@@ -98,15 +92,12 @@ class Rule_DictElement : public VariableDictElement { \
         }
 
         if (r && r->hasSeverity()) {
-            std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
             std::string *a = new std::string(std::to_string(r->severity()));
             VariableValue *var = new VariableValue(&m_rule, &m_rule_severity,
                 a
             );
             delete a;
-            origin->m_offset = 0;
-            origin->m_length = 0;
-            var->addOrigin(std::move(origin));
+            var->addOrigin();
             l->push_back(var);
         }
     }
@@ -122,15 +113,12 @@ class Rule_DictElement : public VariableDictElement { \
         }
 
         if (r && r->hasLogData()) {
-            std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
             std::string *a = new std::string(r->logData(t));
             VariableValue *var = new VariableValue(&m_rule, &m_rule_logdata,
                 a
             );
             delete a;
-            origin->m_offset = 0;
-            origin->m_length = 0;
-            var->addOrigin(std::move(origin));
+            var->addOrigin();
             l->push_back(var);
         }
     }
@@ -145,15 +133,12 @@ class Rule_DictElement : public VariableDictElement { \
         }
 
         if (r && r->hasMsg()) {
-            std::unique_ptr<VariableOrigin> origin(new VariableOrigin());
             std::string *a = new std::string(r->msg(t));
             VariableValue *var = new VariableValue(&m_rule, &m_rule_msg,
                 a
             );
             delete a;
-            origin->m_offset = 0;
-            origin->m_length = 0;
-            var->addOrigin(std::move(origin));
+            var->addOrigin();
             l->push_back(var);
         }
     }

Original file line number	Diff line number	Diff line change
`@@ -49,15 +49,12 @@ class Rule_DictElement : public VariableDictElement { \`
`49`	`49`	`if (!r \|\| r->m_ruleId == 0) {`
`50`	`50`	`return;`
`51`	`51`	`}`
`52`		`- std::unique_ptr<VariableOrigin> origin(new VariableOrigin());`
`53`	`52`	`std::string *a = new std::string(std::to_string(r->m_ruleId));`
`54`	`53`	`VariableValue *var = new VariableValue(&m_rule, &m_rule_id,`
`55`	`54`	`a`
`56`	`55`	`);`
`57`	`56`	`delete a;`
`58`		`- origin->m_offset = 0;`
`59`		`- origin->m_length = 0;`
`60`		`- var->addOrigin(std::move(origin));`
	`57`	`+ var->addOrigin();`
`61`	`58`	`l->push_back(var);`
`62`	`59`	`}`
`63`	`60`
`@@ -75,15 +72,12 @@ class Rule_DictElement : public VariableDictElement { \`
`75`	`72`	`return;`
`76`	`73`	`}`
`77`	`74`
`78`		`- std::unique_ptr<VariableOrigin> origin(new VariableOrigin());`
`79`	`75`	`std::string *a = new std::string(r->m_rev);`
`80`	`76`	`VariableValue *var = new VariableValue(&m_rule, &m_rule_rev,`
`81`	`77`	`a`
`82`	`78`	`);`
`83`	`79`	`delete a;`
`84`		`- origin->m_offset = 0;`
`85`		`- origin->m_length = 0;`
`86`		`- var->addOrigin(std::move(origin));`
	`80`	`+ var->addOrigin();`
`87`	`81`	`l->push_back(var);`
`88`	`82`	`}`
`89`	`83`
`@@ -98,15 +92,12 @@ class Rule_DictElement : public VariableDictElement { \`
`98`	`92`	`}`
`99`	`93`
`100`	`94`	`if (r && r->hasSeverity()) {`
`101`		`- std::unique_ptr<VariableOrigin> origin(new VariableOrigin());`
`102`	`95`	`std::string *a = new std::string(std::to_string(r->severity()));`
`103`	`96`	`VariableValue *var = new VariableValue(&m_rule, &m_rule_severity,`
`104`	`97`	`a`
`105`	`98`	`);`
`106`	`99`	`delete a;`
`107`		`- origin->m_offset = 0;`
`108`		`- origin->m_length = 0;`
`109`		`- var->addOrigin(std::move(origin));`
	`100`	`+ var->addOrigin();`
`110`	`101`	`l->push_back(var);`
`111`	`102`	`}`
`112`	`103`	`}`
`@@ -122,15 +113,12 @@ class Rule_DictElement : public VariableDictElement { \`
`122`	`113`	`}`
`123`	`114`
`124`	`115`	`if (r && r->hasLogData()) {`
`125`		`- std::unique_ptr<VariableOrigin> origin(new VariableOrigin());`
`126`	`116`	`std::string *a = new std::string(r->logData(t));`
`127`	`117`	`VariableValue *var = new VariableValue(&m_rule, &m_rule_logdata,`
`128`	`118`	`a`
`129`	`119`	`);`
`130`	`120`	`delete a;`
`131`		`- origin->m_offset = 0;`
`132`		`- origin->m_length = 0;`
`133`		`- var->addOrigin(std::move(origin));`
	`121`	`+ var->addOrigin();`
`134`	`122`	`l->push_back(var);`
`135`	`123`	`}`
`136`	`124`	`}`
`@@ -145,15 +133,12 @@ class Rule_DictElement : public VariableDictElement { \`
`145`	`133`	`}`
`146`	`134`
`147`	`135`	`if (r && r->hasMsg()) {`
`148`		`- std::unique_ptr<VariableOrigin> origin(new VariableOrigin());`
`149`	`136`	`std::string *a = new std::string(r->msg(t));`
`150`	`137`	`VariableValue *var = new VariableValue(&m_rule, &m_rule_msg,`
`151`	`138`	`a`
`152`	`139`	`);`
`153`	`140`	`delete a;`
`154`		`- origin->m_offset = 0;`
`155`		`- origin->m_length = 0;`
`156`		`- var->addOrigin(std::move(origin));`
	`141`	`+ var->addOrigin();`
`157`	`142`	`l->push_back(var);`
`158`	`143`	`}`
`159`	`144`	`}`