Remove unnecessary copying in transformations

In C++11, string data is always null-terminated[1], and can be
freely modified[2].

[1] https://stackoverflow.com/questions/6077189/will-stdstring-always-be-null-terminated-in-c11
[2] https://stackoverflow.com/questions/38875623/is-it-permitted-to-modify-the-internal-stdstring-buffer-returned-by-operator

CHANGES

@@ -1,6 +1,8 @@
 v3.x.y - YYYY-MMM-DD (to be released)
 -------------------------------------
-   
+ 
+  - Removed unnecessary while processing the transformations.
+    [#2368 - @WGH-, @zimmerle]  
   - auditlog: Computes whether or not to save while loading the rules.
     [@zimmerle]
   - actions: Computes Rule association while loading the rules given a

src/actions/transformations/css_decode.cc

@@ -32,19 +32,10 @@ namespace transformations {
 void CssDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    size_t s = in.size();
-
-    char *tmp = reinterpret_cast<char *>(
-        malloc(sizeof(char) * s + 1));
-    memcpy(tmp, in.c_str(), s + 1);
-    tmp[s] = '\0';
-
-    size_t r = CssDecode::css_decode_inplace(
-        reinterpret_cast<unsigned char *>(tmp),
-        s);
-
-    out.assign(tmp, r);
-    free(tmp);
+    out.assign(in);
+    auto size = CssDecode::css_decode_inplace(
+        reinterpret_cast<unsigned char *>(&out[0]), out.size());
+    out.resize(size);
 }
 
 

src/actions/transformations/escape_seq_decode.cc

@@ -135,14 +135,11 @@ int EscapeSeqDecode::ansi_c_sequences_decode_inplace(unsigned char *input,
 void EscapeSeqDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *tmp = (unsigned char *) malloc(sizeof(char)
-        * in.size() + 1);
-    memcpy(tmp, in.c_str(), in.size() + 1);
-    tmp[in.size()] = '\0';
 
-    int size = ansi_c_sequences_decode_inplace(tmp, in.size());
-    out.assign(reinterpret_cast<char *>(tmp), size);
-    free(tmp);
+    out.assign(in);
+    auto size = ansi_c_sequences_decode_inplace(
+        reinterpret_cast<unsigned char *>(&out[0]), out.size());
+    out.resize(size);
 }
 
 }  // namespace transformations

src/actions/transformations/hex_decode.cc

@@ -34,22 +34,10 @@ namespace transformations {
 void HexDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-    int size = 0;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    size = inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), size);
-    free(input);
+    out.assign(in);
+    auto size = inplace(reinterpret_cast<unsigned char *>(
+        &out[0]), out.length());
+    out.resize(size);
 }
 
 

src/actions/transformations/html_entity_decode.cc

@@ -32,21 +32,10 @@ namespace transformations {
 void HtmlEntityDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    size_t i = inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), i);
-    free(input);
+    out.assign(in);
+    auto i = inplace(reinterpret_cast<unsigned char *>(
+        &out[0]), out.length());
+    out.resize(i);
 }
 
 

src/actions/transformations/js_decode.cc

@@ -32,21 +32,11 @@ namespace transformations {
 void JsDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
 
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    size_t i = inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), i);
-    free(input);
+    out.assign(in);
+    auto i = inplace(reinterpret_cast<unsigned char *>(
+        &out[0]), out.length());
+    out.resize(i);
 }
 
 

src/actions/transformations/normalise_path.cc

@@ -31,18 +31,9 @@ void NormalisePath::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
     int changed = 0;
-
-    char *tmp = reinterpret_cast<char *>(
-        malloc(sizeof(char) * in.size() + 1));
-    memcpy(tmp, in.c_str(), in.size() + 1);
-    tmp[in.size()] = '\0';
-
-    int i = normalize_path_inplace((unsigned char *)tmp,
-        in.size(), 0, &changed);
-
-    std::string ret("");
-    out.assign(tmp, i);
-    free(tmp);
+    out.assign(in);
+    auto size = normalize_path_inplace(reinterpret_cast<unsigned char *>(&out[0]), out.length(), 0, &changed);
+    out.resize(size);
 }
 
 

src/actions/transformations/normalise_path_win.cc

@@ -36,18 +36,10 @@ void NormalisePathWin::execute(const Transaction *t,
     ModSecString &out) noexcept {
     int changed;
 
-    char *tmp = reinterpret_cast<char *>(
-        malloc(sizeof(char) * in.size() + 1));
-    memcpy(tmp, in.c_str(), in.size() + 1);
-    tmp[in.size()] = '\0';
-
-    int i = NormalisePath::normalize_path_inplace(
-        reinterpret_cast<unsigned char *>(tmp),
-        in.size(), 1, &changed);
-
-    std::string ret("");
-    out.assign(tmp, i);
-    free(tmp);
+    out.assign(in);
+    auto size = NormalisePath::normalize_path_inplace(
+        reinterpret_cast<unsigned char *>(&out[0]), out.length(), 1, &changed);
+    out.resize(size);
 }
 
 

src/actions/transformations/parity_even_7bit.cc

@@ -30,21 +30,8 @@ namespace transformations {
 void ParityEven7bit::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    std::memcpy(input, in.c_str(), in.length()+1);
-
-    inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), in.length());
-    free(input);
+    out.assign(in);
+    inplace(reinterpret_cast<unsigned char*>(&out[0]), out.size());
 }
 
 

src/actions/transformations/parity_odd_7bit.cc

@@ -30,21 +30,8 @@ namespace transformations {
 void ParityOdd7bit::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), in.length());
-    free(input);
+    out.assign(in);
+    inplace(reinterpret_cast<unsigned char *>(&out[0]), out.length());
 }
 
 

src/actions/transformations/parity_zero_7bit.cc

@@ -30,21 +30,8 @@ namespace transformations {
 void ParityZero7bit::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), in.length());
-    free(input);
+    out.assign(in);
+    inplace(reinterpret_cast<unsigned char *>(&out[0]), out.length());
 }
 
 

src/actions/transformations/sql_hex_decode.cc

@@ -32,22 +32,9 @@ namespace transformations {
 void SqlHexDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-    int size = 0;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    size = inplace(input, in.length());
-
-    out.assign(reinterpret_cast<char *>(input), size);
-    free(input);
+    out.assign(in);
+    auto size = inplace(reinterpret_cast<unsigned char*>(&out[0]), out.size());
+    out.resize(size);
 }
 
 

src/actions/transformations/url_decode.cc

@@ -32,19 +32,14 @@ namespace transformations {
 void UrlDecode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *val(NULL);
     int invalid_count = 0;
     int changed;
 
-    val = (unsigned char *) malloc(sizeof(char) * in.size() + 1);
-    memcpy(val, in.c_str(), in.size() + 1);
-    val[in.size()] = '\0';
-
-    int size = utils::urldecode_nonstrict_inplace(val, in.size(),
+    out.assign(in);
+    int size = utils::urldecode_nonstrict_inplace(
+        reinterpret_cast<unsigned char *>(&out[0]), out.size(),
         &invalid_count, &changed);
-    out.append((const char *)val, size);
-
-    free(val);
+    out.resize(size);
 }
 
 

src/actions/transformations/url_decode_uni.cc

@@ -33,21 +33,9 @@ namespace transformations {
 void UrlDecodeUni::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
-
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memcpy(input, in.c_str(), in.length()+1);
-
-    size_t i = inplace(input, in.length(), t);
-
-    out.assign(reinterpret_cast<char *>(input), i);
-    free(input);
+    out.assign(in);
+    size_t i = inplace(reinterpret_cast<unsigned char *>(&out[0]), out.length(), t);
+    out.resize(i);
 }
 
 

src/actions/transformations/utf8_to_unicode.cc

@@ -32,31 +32,19 @@ namespace transformations {
 void Utf8ToUnicode::execute(const Transaction *t,
     const ModSecString &in,
     ModSecString &out) noexcept {
-    unsigned char *input;
     int changed = 0;
     char *out2;
 
-    input = reinterpret_cast<unsigned char *>
-        (malloc(sizeof(char) * in.length()+1));
-
-    if (input == NULL) {
-        return;
-    }
-
-    memset(input, '\0', in.length()+1);
-    memcpy(input, in.c_str(), in.length()+1);
-
-    out2 = inplace(input, in.size() + 1, &changed);
+    out2 = inplace(reinterpret_cast<const unsigned char *>(&in[0]), in.size() + 1, &changed);
     if (out2 != NULL) {
         out.assign(reinterpret_cast<char *>(out2),
             strlen(reinterpret_cast<char *>(out2)));
         free(out2);
     }
-    free(input);
 }
 
 
-char *Utf8ToUnicode::inplace(unsigned char *input,
+char *Utf8ToUnicode::inplace(const unsigned char *input,
     uint64_t input_len, int *changed) {
     unsigned int count = 0;
     char *data;
@@ -84,7 +72,7 @@ char *Utf8ToUnicode::inplace(unsigned char *input,
         int unicode_len = 0;
         unsigned int d = 0;
         unsigned char c;
-        unsigned char *utf = (unsigned char *)&input[i];
+        const unsigned char *utf = &input[i];
 
         c = *utf;
 

src/actions/transformations/utf8_to_unicode.h

@@ -50,7 +50,7 @@ class Utf8ToUnicode : public Transformation {
         UNICODE_ERROR_DECODING_ERROR = -5
     };
 
-    static char *inplace(unsigned char *input, uint64_t input_len,
+    static char *inplace(const unsigned char *input, uint64_t input_len,
         int *changed);
 };