Lil-Ran/Pyarmor-Static-Unpack-1shot
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Simplify string object printing to reduce code duplication

Browse Source
This commit is contained in:
Michael Hansen
2023-06-09 09:09:03 -07:00
parent 4b4f9f8edb
commit 753d42d94a
5 changed files with 54 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
ASTree.cpp
View File

@@ -3224,32 +3224,14 @@ bool print_docstring(PycRef<PycObject> obj, int indent, PycModule* mod,
std::ostream& pyc_output) std::ostream& pyc_output)
{ {
// docstrings are translated from the bytecode __doc__ = 'string' to simply '''string''' // docstrings are translated from the bytecode __doc__ = 'string' to simply '''string'''
signed char prefix = -1; auto doc = obj.try_cast<PycString>();
switch (obj.type()) { if (doc != nullptr) {
case PycObject::TYPE_STRING:
prefix = mod->strIsUnicode() ? 'b' : 0;
break;
case PycObject::TYPE_UNICODE:
prefix = mod->strIsUnicode() ? 0 : 'u';
break;
case PycObject::TYPE_INTERNED:
case PycObject::TYPE_ASCII:
case PycObject::TYPE_ASCII_INTERNED:
case PycObject::TYPE_SHORT_ASCII:
case PycObject::TYPE_SHORT_ASCII_INTERNED:
if (mod->majorVer() >= 3)
prefix = 0;
else
prefix = mod->strIsUnicode() ? 'b' : 0;
break;
}
if (prefix != -1) {
start_line(indent, pyc_output); start_line(indent, pyc_output);
OutputString(pyc_output, obj.cast<PycString>(), prefix, true); doc->print(pyc_output, mod, true);
pyc_output << "\n"; pyc_output << "\n";
return true; return true;
} else }
return false; return false;
} }
void decompyle(PycRef<PycCode> code, PycModule* mod, std::ostream& pyc_output) void decompyle(PycRef<PycCode> code, PycModule* mod, std::ostream& pyc_output)

12
bytecode.cpp
View File

@@ -170,23 +170,13 @@ void print_const(std::ostream& pyc_output, PycRef<PycObject> obj, PycModule* mod
switch (obj->type()) { switch (obj->type()) {
case PycObject::TYPE_STRING: case PycObject::TYPE_STRING:
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 'b' : 0,
false, parent_f_string_quote);
break;
case PycObject::TYPE_UNICODE: case PycObject::TYPE_UNICODE:
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 0 : 'u',
false, parent_f_string_quote);
break;
case PycObject::TYPE_INTERNED: case PycObject::TYPE_INTERNED:
case PycObject::TYPE_ASCII: case PycObject::TYPE_ASCII:
case PycObject::TYPE_ASCII_INTERNED: case PycObject::TYPE_ASCII_INTERNED:
case PycObject::TYPE_SHORT_ASCII: case PycObject::TYPE_SHORT_ASCII:
case PycObject::TYPE_SHORT_ASCII_INTERNED: case PycObject::TYPE_SHORT_ASCII_INTERNED:
if (mod->majorVer() >= 3) obj.cast<PycString>()->print(pyc_output, mod, false, parent_f_string_quote);
OutputString(pyc_output, obj.cast<PycString>(), 0, false, parent_f_string_quote);
else
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 'b' : 0,
false, parent_f_string_quote);
break; break;
case PycObject::TYPE_TUPLE: case PycObject::TYPE_TUPLE:
case PycObject::TYPE_SMALL_TUPLE: case PycObject::TYPE_SMALL_TUPLE:

72
pyc_string.cpp
View File

@@ -56,15 +56,35 @@ bool PycString::isEqual(PycRef<PycObject> obj) const
return isEqual(strObj->m_value); return isEqual(strObj->m_value);
} }
void OutputString(std::ostream &pyc_output, PycRef<PycString> str, char prefix, void PycString::print(std::ostream &pyc_output, PycModule* mod, bool triple,
bool triple, const char* parent_f_string_quote) const char* parent_f_string_quote)
{ {
char prefix = 0;
switch (type()) {
case TYPE_STRING:
prefix = mod->strIsUnicode() ? 'b' : 0;
break;
case PycObject::TYPE_UNICODE:
prefix = mod->strIsUnicode() ? 0 : 'u';
break;
case PycObject::TYPE_INTERNED:
case PycObject::TYPE_ASCII:
case PycObject::TYPE_ASCII_INTERNED:
case PycObject::TYPE_SHORT_ASCII:
case PycObject::TYPE_SHORT_ASCII_INTERNED:
if (mod->majorVer() >= 3)
prefix = 0;
else
prefix = mod->strIsUnicode() ? 'b' : 0;
break;
default:
throw std::runtime_error("Invalid string type");
}
if (prefix != 0) if (prefix != 0)
pyc_output << prefix; pyc_output << prefix;
const char* ch = str->value(); if (m_value.empty()) {
int len = str->length();
if (len == 0) {
pyc_output << "''"; pyc_output << "''";
return; return;
} }
@@ -72,20 +92,17 @@ void OutputString(std::ostream &pyc_output, PycRef<PycString> str, char prefix,
// Determine preferred quote style (Emulate Python's method) // Determine preferred quote style (Emulate Python's method)
bool useQuotes = false; bool useQuotes = false;
if (!parent_f_string_quote) { if (!parent_f_string_quote) {
while (len--) { for (char ch : m_value) {
if (*ch == '\'') { if (ch == '\'') {
useQuotes = true; useQuotes = true;
} else if (*ch == '"') { } else if (ch == '"') {
useQuotes = false; useQuotes = false;
break; break;
} }
ch++;
} }
} else { } else {
useQuotes = parent_f_string_quote[0] == '"'; useQuotes = parent_f_string_quote[0] == '"';
} }
ch = str->value();
len = str->length();
// Output the string // Output the string
if (!parent_f_string_quote) { if (!parent_f_string_quote) {
@@ -94,42 +111,41 @@ void OutputString(std::ostream &pyc_output, PycRef<PycString> str, char prefix,
else else
pyc_output << (useQuotes ? '"' : '\''); pyc_output << (useQuotes ? '"' : '\'');
} }
while (len--) { for (char ch : m_value) {
if ((unsigned char)(*ch) < 0x20 || *ch == 0x7F) { if (static_cast<unsigned char>(ch) < 0x20 || ch == 0x7F) {
if (*ch == '\r') { if (ch == '\r') {
pyc_output << "\\r"; pyc_output << "\\r";
} else if (*ch == '\n') { } else if (ch == '\n') {
if (triple) if (triple)
pyc_output << '\n'; pyc_output << '\n';
else else
pyc_output << "\\n"; pyc_output << "\\n";
} else if (*ch == '\t') { } else if (ch == '\t') {
pyc_output << "\\t"; pyc_output << "\\t";
} else { } else {
formatted_print(pyc_output, "\\x%02x", (*ch & 0xFF)); formatted_print(pyc_output, "\\x%02x", (ch & 0xFF));
} }
} else if ((unsigned char)(*ch) >= 0x80) { } else if (static_cast<unsigned char>(ch) >= 0x80) {
if (str->type() == PycObject::TYPE_UNICODE) { if (type() == TYPE_UNICODE) {
// Unicode stored as UTF-8... Let the stream interpret it // Unicode stored as UTF-8... Let the stream interpret it
pyc_output << *ch; pyc_output << ch;
} else { } else {
formatted_print(pyc_output, "\\x%x", (*ch & 0xFF)); formatted_print(pyc_output, "\\x%x", (ch & 0xFF));
} }
} else { } else {
if (!useQuotes && *ch == '\'') if (!useQuotes && ch == '\'')
pyc_output << R"(\')"; pyc_output << R"(\')";
else if (useQuotes && *ch == '"') else if (useQuotes && ch == '"')
pyc_output << R"(\")"; pyc_output << R"(\")";
else if (*ch == '\\') else if (ch == '\\')
pyc_output << R"(\\)"; pyc_output << R"(\\)";
else if (parent_f_string_quote && *ch == '{') else if (parent_f_string_quote && ch == '{')
pyc_output << "{{"; pyc_output << "{{";
else if (parent_f_string_quote && *ch == '}') else if (parent_f_string_quote && ch == '}')
pyc_output << "}}"; pyc_output << "}}";
else else
pyc_output << *ch; pyc_output << ch;
} }
ch++;
} }
if (!parent_f_string_quote) { if (!parent_f_string_quote) {
if (triple) if (triple)

6
pyc_string.h
View File

@@ -27,11 +27,11 @@ public:
void setValue(std::string str) { m_value = std::move(str); } void setValue(std::string str) { m_value = std::move(str); }
void print(std::ostream& stream, class PycModule* mod, bool triple = false,
const char* parent_f_string_quote = nullptr);
private: private:
std::string m_value; std::string m_value;
}; };
void OutputString(std::ostream& stream, PycRef<PycString> str, char prefix,
bool triple = false, const char* parent_f_string_quote = nullptr);
#endif #endif

13
pycdas.cpp
View File

@@ -153,25 +153,14 @@ void output_object(PycRef<PycObject> obj, PycModule* mod, int indent,
} }
break; break;
case PycObject::TYPE_STRING: case PycObject::TYPE_STRING:
iputs(pyc_output, indent, "");
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 'b' : 0);
pyc_output << "\n";
break;
case PycObject::TYPE_UNICODE: case PycObject::TYPE_UNICODE:
iputs(pyc_output, indent, "");
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 0 : 'u');
pyc_output << "\n";
break;
case PycObject::TYPE_INTERNED: case PycObject::TYPE_INTERNED:
case PycObject::TYPE_ASCII: case PycObject::TYPE_ASCII:
case PycObject::TYPE_ASCII_INTERNED: case PycObject::TYPE_ASCII_INTERNED:
case PycObject::TYPE_SHORT_ASCII: case PycObject::TYPE_SHORT_ASCII:
case PycObject::TYPE_SHORT_ASCII_INTERNED: case PycObject::TYPE_SHORT_ASCII_INTERNED:
iputs(pyc_output, indent, ""); iputs(pyc_output, indent, "");
if (mod->majorVer() >= 3) obj.cast<PycString>()->print(pyc_output, mod);
OutputString(pyc_output, obj.cast<PycString>(), 0);
else
OutputString(pyc_output, obj.cast<PycString>(), mod->strIsUnicode() ? 'b' : 0);
pyc_output << "\n"; pyc_output << "\n";
break; break;
case PycObject::TYPE_TUPLE: case PycObject::TYPE_TUPLE: