diff options
| -rw-r--r-- | src/util/serialize.cpp | 217 |
1 files changed, 113 insertions, 104 deletions
diff --git a/src/util/serialize.cpp b/src/util/serialize.cpp index 32e0a8936..ee46fd941 100644 --- a/src/util/serialize.cpp +++ b/src/util/serialize.cpp @@ -18,15 +18,14 @@ with this program; if not, write to the Free Software Foundation, Inc., */ #include "serialize.h" -#include "pointer.h" #include "porting.h" #include "util/string.h" +#include "util/hex.h" #include "exceptions.h" #include "irrlichttypes.h" -#include <sstream> -#include <iomanip> -#include <vector> +#include <iostream> +#include <cassert> FloatType g_serialize_f32_type = FLOATTYPE_UNKNOWN; @@ -120,118 +119,148 @@ std::string deSerializeString32(std::istream &is) } //// -//// JSON +//// JSON-like strings //// std::string serializeJsonString(const std::string &plain) { - std::ostringstream os(std::ios::binary); - os << "\""; + std::string tmp; + + tmp.reserve(plain.size() + 2); + tmp.push_back('"'); for (char c : plain) { switch (c) { case '"': - os << "\\\""; + tmp.append("\\\""); break; case '\\': - os << "\\\\"; + tmp.append("\\\\"); break; case '\b': - os << "\\b"; + tmp.append("\\b"); break; case '\f': - os << "\\f"; + tmp.append("\\f"); break; case '\n': - os << "\\n"; + tmp.append("\\n"); break; case '\r': - os << "\\r"; + tmp.append("\\r"); break; case '\t': - os << "\\t"; + tmp.append("\\t"); break; default: { if (c >= 32 && c <= 126) { - os << c; + tmp.push_back(c); } else { - u32 cnum = (u8)c; - os << "\\u" << std::hex << std::setw(4) - << std::setfill('0') << cnum; + // We pretend that Unicode codepoints map to bytes (they don't) + u8 cnum = static_cast<u8>(c); + tmp.append("\\u00"); + tmp.push_back(hex_chars[cnum >> 4]); + tmp.push_back(hex_chars[cnum & 0xf]); } break; } } } - os << "\""; - return os.str(); + tmp.push_back('"'); + return tmp; +} + +static void deSerializeJsonString(std::string &s) +{ + assert(s.size() >= 2); + assert(s.front() == '"' && s.back() == '"'); + + size_t w = 0; // write index + size_t i = 1; // read index + const size_t len = s.size() - 1; // string length with trailing quote removed + + while (i < len) { + char c = s[i++]; + assert(c != '"'); + + if (c != '\\') { + s[w++] = c; + continue; + } + + if (i >= len) + throw SerializationError("JSON string ended prematurely"); + char c2 = s[i++]; + switch (c2) { + case 'b': + s[w++] = '\b'; + break; + case 'f': + s[w++] = '\f'; + break; + case 'n': + s[w++] = '\n'; + break; + case 'r': + s[w++] = '\r'; + break; + case 't': + s[w++] = '\t'; + break; + case 'u': { + if (i + 3 >= len) + throw SerializationError("JSON string ended prematurely"); + unsigned char v[4] = {}; + for (int j = 0; j < 4; j++) + hex_digit_decode(s[i+j], v[j]); + i += 4; + u32 hexnumber = (v[0] << 12) | (v[1] << 8) | (v[2] << 4) | v[3]; + // Note that this does not work for anything other than ASCII + // but these functions do not actually interact with real JSON input. + s[w++] = (int) hexnumber; + break; + } + default: + s[w++] = c2; + break; + } + } + + assert(w <= i && i <= len); + // Truncate string to current write index + s.resize(w); } std::string deSerializeJsonString(std::istream &is) { - std::ostringstream os(std::ios::binary); - char c, c2; + std::string tmp; + char c; + bool was_backslash = false; // Parse initial doublequote - is >> c; + c = is.get(); if (c != '"') throw SerializationError("JSON string must start with doublequote"); + tmp.push_back(c); - // Parse characters + // Grab the entire json string for (;;) { c = is.get(); if (is.eof()) throw SerializationError("JSON string ended prematurely"); - if (c == '"') { - return os.str(); - } - - if (c == '\\') { - c2 = is.get(); - if (is.eof()) - throw SerializationError("JSON string ended prematurely"); - switch (c2) { - case 'b': - os << '\b'; - break; - case 'f': - os << '\f'; - break; - case 'n': - os << '\n'; - break; - case 'r': - os << '\r'; - break; - case 't': - os << '\t'; - break; - case 'u': { - int hexnumber; - char hexdigits[4 + 1]; - - is.read(hexdigits, 4); - if (is.eof()) - throw SerializationError("JSON string ended prematurely"); - hexdigits[4] = 0; - - std::istringstream tmp_is(hexdigits, std::ios::binary); - tmp_is >> std::hex >> hexnumber; - os << (char)hexnumber; - break; - } - default: - os << c2; - break; - } - } else { - os << c; - } + tmp.push_back(c); + if (was_backslash) + was_backslash = false; + else if (c == '\\') + was_backslash = true; + else if (c == '"') + break; // found end of string } - return os.str(); + deSerializeJsonString(tmp); + return tmp; } std::string serializeJsonStringIfNeeded(const std::string &s) @@ -245,41 +274,21 @@ std::string serializeJsonStringIfNeeded(const std::string &s) std::string deSerializeJsonStringIfNeeded(std::istream &is) { - std::stringstream tmp_os(std::ios_base::binary | std::ios_base::in | std::ios_base::out); - bool expect_initial_quote = true; - bool is_json = false; - bool was_backslash = false; - for (;;) { - char c = is.get(); - if (is.eof()) - break; - - if (expect_initial_quote && c == '"') { - tmp_os << c; - is_json = true; - } else if(is_json) { - tmp_os << c; - if (was_backslash) - was_backslash = false; - else if (c == '\\') - was_backslash = true; - else if (c == '"') - break; // Found end of string - } else { - if (c == ' ') { - // Found end of word - is.unget(); - break; - } - - tmp_os << c; - } - expect_initial_quote = false; - } - if (is_json) { - return deSerializeJsonString(tmp_os); + // Check for initial quote + char c = is.peek(); + if (is.eof()) + return ""; + + if (c == '"') { + // json string: defer to the right implementation + return deSerializeJsonString(is); } - return tmp_os.str(); + // not a json string: + std::string tmp; + std::getline(is, tmp, ' '); + if (!is.eof()) + is.unget(); // we hit a space, put it back + return tmp; } |