2
1
mirror of https://github.com/qpdf/qpdf.git synced 2024-11-14 00:34:03 +00:00
qpdf/libqpdf/JSON.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1386 lines
39 KiB
C++
Raw Normal View History

2018-12-17 16:55:11 +00:00
#include <qpdf/JSON.hh>
#include <qpdf/BufferInputSource.hh>
#include <qpdf/Pl_Base64.hh>
#include <qpdf/Pl_Concatenate.hh>
#include <qpdf/Pl_String.hh>
2018-12-17 16:55:11 +00:00
#include <qpdf/QTC.hh>
#include <qpdf/QUtil.hh>
2022-01-17 23:40:38 +00:00
#include <cstring>
#include <stdexcept>
2018-12-17 16:55:11 +00:00
JSON::Members::Members(std::unique_ptr<JSON_value> value) :
value(std::move(value))
2018-12-17 16:55:11 +00:00
{
}
JSON::JSON(std::unique_ptr<JSON_value> value) :
m(new Members(std::move(value)))
2018-12-17 16:55:11 +00:00
{
}
void
JSON::writeClose(Pipeline* p, bool first, size_t depth, char const* delimiter)
{
2023-01-28 11:09:46 +00:00
if (first) {
*p << delimiter;
} else {
std::string s{"\n"};
s.append(2 * depth, ' ');
*p << s + delimiter;
}
}
void
JSON::writeNext(Pipeline* p, bool& first, size_t depth)
{
if (first) {
first = false;
2023-01-28 11:15:27 +00:00
std::string s{"\n"};
s.append(2 * depth, ' ');
*p << s;
} else {
2023-01-28 11:15:27 +00:00
std::string s{",\n"};
s.append(2 * depth, ' ');
*p << s;
}
}
void
JSON::writeDictionaryOpen(Pipeline* p, bool& first, size_t depth)
{
*p << "{";
first = true;
}
void
JSON::writeArrayOpen(Pipeline* p, bool& first, size_t depth)
{
*p << "[";
first = true;
}
void
JSON::writeDictionaryClose(Pipeline* p, bool first, size_t depth)
{
writeClose(p, first, depth, "}");
}
void
JSON::writeArrayClose(Pipeline* p, bool first, size_t depth)
{
writeClose(p, first, depth, "]");
}
void
JSON::writeDictionaryKey(Pipeline* p, bool& first, std::string const& key, size_t depth)
{
writeNext(p, first, depth);
2023-01-28 12:54:49 +00:00
*p << std::string("\"") + key + "\": ";
}
void
JSON::writeDictionaryItem(
Pipeline* p, bool& first, std::string const& key, JSON const& value, size_t depth)
{
writeDictionaryKey(p, first, key, depth);
value.write(p, depth);
}
void
JSON::writeArrayItem(Pipeline* p, bool& first, JSON const& element, size_t depth)
{
writeNext(p, first, depth);
element.write(p, depth);
}
void
JSON::JSON_dictionary::write(Pipeline* p, size_t depth) const
2018-12-17 16:55:11 +00:00
{
bool first = true;
writeDictionaryOpen(p, first, depth);
2022-02-04 15:52:37 +00:00
for (auto const& iter: members) {
writeDictionaryItem(p, first, iter.first, iter.second, 1 + depth);
2018-12-17 16:55:11 +00:00
}
writeDictionaryClose(p, first, depth);
2018-12-17 16:55:11 +00:00
}
void
JSON::JSON_array::write(Pipeline* p, size_t depth) const
2018-12-17 16:55:11 +00:00
{
bool first = true;
writeArrayOpen(p, first, depth);
2022-02-04 15:52:37 +00:00
for (auto const& element: elements) {
writeArrayItem(p, first, element, 1 + depth);
2018-12-17 16:55:11 +00:00
}
writeArrayClose(p, first, depth);
2018-12-17 16:55:11 +00:00
}
JSON::JSON_string::JSON_string(std::string const& utf8) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_string),
utf8(utf8),
2018-12-17 16:55:11 +00:00
encoded(encode_string(utf8))
{
}
void
JSON::JSON_string::write(Pipeline* p, size_t) const
2018-12-17 16:55:11 +00:00
{
2023-01-28 12:59:26 +00:00
*p << std::string("\"") + encoded + "\"";
2018-12-17 16:55:11 +00:00
}
JSON::JSON_number::JSON_number(long long value) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_number),
encoded(std::to_string(value))
2018-12-17 16:55:11 +00:00
{
}
JSON::JSON_number::JSON_number(double value) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_number),
2018-12-17 16:55:11 +00:00
encoded(QUtil::double_to_string(value, 6))
{
}
JSON::JSON_number::JSON_number(std::string const& value) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_number),
2018-12-17 16:55:11 +00:00
encoded(value)
{
}
void
JSON::JSON_number::write(Pipeline* p, size_t) const
2018-12-17 16:55:11 +00:00
{
*p << encoded;
2018-12-17 16:55:11 +00:00
}
JSON::JSON_bool::JSON_bool(bool val) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_bool),
2018-12-17 16:55:11 +00:00
value(val)
{
}
void
JSON::JSON_bool::write(Pipeline* p, size_t) const
2018-12-17 16:55:11 +00:00
{
*p << (value ? "true" : "false");
2018-12-17 16:55:11 +00:00
}
void
JSON::JSON_null::write(Pipeline* p, size_t) const
2018-12-17 16:55:11 +00:00
{
*p << "null";
2018-12-17 16:55:11 +00:00
}
JSON::JSON_blob::JSON_blob(std::function<void(Pipeline*)> fn) :
2023-02-11 17:46:30 +00:00
JSON_value(vt_blob),
fn(fn)
{
}
void
JSON::JSON_blob::write(Pipeline* p, size_t) const
{
*p << "\"";
Pl_Concatenate cat("blob concatenate", p);
Pl_Base64 base64("blob base64", &cat, Pl_Base64::a_encode);
fn(&base64);
base64.finish();
*p << "\"";
}
void
JSON::write(Pipeline* p, size_t depth) const
2018-12-17 16:55:11 +00:00
{
if (nullptr == m->value) {
*p << "null";
2018-12-17 16:55:11 +00:00
} else {
m->value->write(p, depth);
2018-12-17 16:55:11 +00:00
}
}
std::string
JSON::unparse() const
{
std::string s;
2022-05-06 21:37:25 +00:00
Pl_String p("unparse", nullptr, s);
write(&p, 0);
return s;
}
2018-12-17 16:55:11 +00:00
std::string
JSON::encode_string(std::string const& str)
{
2023-02-03 14:28:45 +00:00
static auto constexpr hexchars = "0123456789abcdef";
auto begin = str.cbegin();
auto end = str.cend();
auto iter = begin;
while (iter != end) {
auto c = static_cast<unsigned char>(*iter);
if ((c > 34 && c != '\\') || c == ' ' || c == 33) {
// Optimistically check that no char in str requires escaping. Hopefully we can just
// return the input str.
++iter;
} else {
// We found a char that requires escaping. Initialize result to the chars scanned so
// far, append/replace the rest of str one char at a time, and return the result.
std::string result{begin, iter};
for (; iter != end; ++iter) {
auto ch = static_cast<unsigned char>(*iter);
if ((ch > 34 && ch != '\\') || ch == ' ' || ch == 33) {
// Check for most common case first.
result += *iter;
} else {
switch (ch) {
case '\\':
result += "\\\\";
break;
case '\"':
result += "\\\"";
break;
case '\b':
result += "\\b";
break;
case '\f':
result += "\\f";
break;
case '\n':
result += "\\n";
break;
case '\r':
result += "\\r";
break;
case '\t':
result += "\\t";
break;
default:
result += ch < 16 ? "\\u000" : "\\u001";
result += hexchars[ch % 16];
}
}
2018-12-17 16:55:11 +00:00
}
2023-02-03 14:28:45 +00:00
return result;
2018-12-17 16:55:11 +00:00
}
}
2023-02-03 14:28:45 +00:00
return str;
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeDictionary()
{
return {std::make_unique<JSON_dictionary>()};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::addDictionaryMember(std::string const& key, JSON const& val)
{
2023-01-28 14:32:01 +00:00
if (auto* obj = dynamic_cast<JSON_dictionary*>(m->value.get())) {
return obj->members[encode_string(key)] = val.m->value ? val : makeNull();
} else {
2018-12-17 16:55:11 +00:00
throw std::runtime_error("JSON::addDictionaryMember called on non-dictionary");
}
}
bool
JSON::checkDictionaryKeySeen(std::string const& key)
{
2023-05-20 12:34:53 +00:00
auto* obj = dynamic_cast<JSON_dictionary*>(m->value.get());
if (nullptr == obj) {
throw std::logic_error("JSON::checkDictionaryKey called on non-dictionary");
}
if (obj->parsed_keys.count(key)) {
return true;
}
obj->parsed_keys.insert(key);
return false;
}
2018-12-17 16:55:11 +00:00
JSON
JSON::makeArray()
{
return {std::make_unique<JSON_array>()};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::addArrayElement(JSON const& val)
{
2023-05-20 12:34:53 +00:00
auto* arr = dynamic_cast<JSON_array*>(m->value.get());
if (nullptr == arr) {
2018-12-17 16:55:11 +00:00
throw std::runtime_error("JSON::addArrayElement called on non-array");
}
if (val.m->value.get()) {
arr->elements.push_back(val);
2018-12-17 16:55:11 +00:00
} else {
arr->elements.push_back(makeNull());
2018-12-17 16:55:11 +00:00
}
return arr->elements.back();
}
JSON
JSON::makeString(std::string const& utf8)
{
return {std::make_unique<JSON_string>(utf8)};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeInt(long long int value)
{
return {std::make_unique<JSON_number>(value)};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeReal(double value)
{
return {std::make_unique<JSON_number>(value)};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeNumber(std::string const& encoded)
{
return {std::make_unique<JSON_number>(encoded)};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeBool(bool value)
{
return {std::make_unique<JSON_bool>(value)};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeNull()
{
return {std::make_unique<JSON_null>()};
2018-12-17 16:55:11 +00:00
}
JSON
JSON::makeBlob(std::function<void(Pipeline*)> fn)
{
return {std::make_unique<JSON_blob>(fn)};
}
bool
JSON::isArray() const
{
2023-02-11 17:46:30 +00:00
return m->value->type_code == vt_array;
}
bool
JSON::isDictionary() const
{
2023-02-11 17:46:30 +00:00
return m->value->type_code == vt_dictionary;
}
bool
JSON::getString(std::string& utf8) const
{
2023-02-11 17:46:30 +00:00
if (m->value->type_code == vt_string) {
auto v = dynamic_cast<JSON_string const*>(m->value.get());
utf8 = v->utf8;
return true;
}
2023-02-11 17:46:30 +00:00
return false;
}
bool
JSON::getNumber(std::string& value) const
{
2023-02-11 17:46:30 +00:00
if (m->value->type_code == vt_number) {
auto v = dynamic_cast<JSON_number const*>(m->value.get());
value = v->encoded;
return true;
}
2023-02-11 17:46:30 +00:00
return false;
}
bool
JSON::getBool(bool& value) const
{
2023-02-11 17:46:30 +00:00
if (m->value->type_code == vt_bool) {
auto v = dynamic_cast<JSON_bool const*>(m->value.get());
value = v->value;
return true;
}
2023-02-11 17:46:30 +00:00
return false;
}
bool
JSON::isNull() const
{
2023-02-11 17:46:30 +00:00
return m->value->type_code == vt_null;
}
bool
JSON::forEachDictItem(std::function<void(std::string const& key, JSON value)> fn) const
{
auto v = dynamic_cast<JSON_dictionary const*>(m->value.get());
if (v == nullptr) {
return false;
}
for (auto const& k: v->members) {
fn(k.first, JSON(k.second));
}
return true;
}
bool
JSON::forEachArrayItem(std::function<void(JSON value)> fn) const
{
auto v = dynamic_cast<JSON_array const*>(m->value.get());
if (v == nullptr) {
return false;
}
for (auto const& i: v->elements) {
fn(JSON(i));
}
return true;
}
2018-12-17 16:55:11 +00:00
bool
JSON::checkSchema(JSON schema, std::list<std::string>& errors)
{
return checkSchemaInternal(m->value.get(), schema.m->value.get(), 0, errors, "");
2018-12-17 16:55:11 +00:00
}
bool
JSON::checkSchema(JSON schema, unsigned long flags, std::list<std::string>& errors)
{
return checkSchemaInternal(m->value.get(), schema.m->value.get(), flags, errors, "");
}
2018-12-17 16:55:11 +00:00
bool
JSON::checkSchemaInternal(
JSON_value* this_v,
JSON_value* sch_v,
unsigned long flags,
2018-12-17 16:55:11 +00:00
std::list<std::string>& errors,
std::string prefix)
{
2023-05-20 12:34:53 +00:00
auto* this_arr = dynamic_cast<JSON_array*>(this_v);
auto* this_dict = dynamic_cast<JSON_dictionary*>(this_v);
2018-12-17 16:55:11 +00:00
2023-05-20 12:34:53 +00:00
auto* sch_arr = dynamic_cast<JSON_array*>(sch_v);
auto* sch_dict = dynamic_cast<JSON_dictionary*>(sch_v);
2018-12-17 16:55:11 +00:00
2023-05-20 12:34:53 +00:00
auto* sch_str = dynamic_cast<JSON_string*>(sch_v);
2018-12-17 16:55:11 +00:00
std::string err_prefix;
if (prefix.empty()) {
err_prefix = "top-level object";
} else {
err_prefix = "json key \"" + prefix + "\"";
}
2020-04-04 21:31:07 +00:00
std::string pattern_key;
2018-12-17 16:55:11 +00:00
if (sch_dict) {
if (!this_dict) {
QTC::TC("libtests", "JSON wanted dictionary");
errors.push_back(err_prefix + " is supposed to be a dictionary");
return false;
}
2020-04-04 21:31:07 +00:00
auto members = sch_dict->members;
std::string key;
if ((members.size() == 1) &&
((key = members.begin()->first, key.length() > 2) && (key.at(0) == '<') &&
(key.at(key.length() - 1) == '>'))) {
pattern_key = key;
}
}
if (sch_dict && (!pattern_key.empty())) {
auto pattern_schema = sch_dict->members[pattern_key].m->value.get();
for (auto const& iter: this_dict->members) {
2020-04-04 21:31:07 +00:00
std::string const& key = iter.first;
checkSchemaInternal(
this_dict->members[key].m->value.get(),
pattern_schema,
flags,
errors,
prefix + "." + key);
2020-04-04 21:31:07 +00:00
}
} else if (sch_dict) {
for (auto& iter: sch_dict->members) {
std::string const& key = iter.first;
2018-12-17 16:55:11 +00:00
if (this_dict->members.count(key)) {
checkSchemaInternal(
this_dict->members[key].m->value.get(),
iter.second.m->value.get(),
flags,
errors,
prefix + "." + key);
2018-12-17 16:55:11 +00:00
} else {
if (flags & f_optional) {
QTC::TC("libtests", "JSON optional key");
} else {
QTC::TC("libtests", "JSON key missing in object");
errors.push_back(
err_prefix + ": key \"" + key +
"\" is present in schema but missing in object");
}
2018-12-17 16:55:11 +00:00
}
}
2022-02-04 15:52:37 +00:00
for (auto const& iter: this_dict->members) {
std::string const& key = iter.first;
2018-12-17 16:55:11 +00:00
if (sch_dict->members.count(key) == 0) {
QTC::TC("libtests", "JSON key extra in object");
errors.push_back(
err_prefix + ": key \"" + key +
"\" is not present in schema but appears in object");
}
}
} else if (sch_arr) {
auto n_elements = sch_arr->elements.size();
if (n_elements == 1) {
// A single-element array in the schema allows a single element in the object or a
// variable-length array, each of whose items must conform to the single element of the
// schema array. This doesn't apply to arrays of arrays -- we fall back to the behavior
// of allowing a single item only when the object is not an array.
if (this_arr) {
int i = 0;
for (auto const& element: this_arr->elements) {
checkSchemaInternal(
element.m->value.get(),
sch_arr->elements.at(0).m->value.get(),
flags,
errors,
prefix + "." + std::to_string(i));
++i;
}
} else {
QTC::TC("libtests", "JSON schema array for single item");
checkSchemaInternal(
this_v, sch_arr->elements.at(0).m->value.get(), flags, errors, prefix);
}
} else if (!this_arr || (this_arr->elements.size() != n_elements)) {
QTC::TC("libtests", "JSON schema array length mismatch");
2018-12-17 16:55:11 +00:00
errors.push_back(
err_prefix + " is supposed to be an array of length " + std::to_string(n_elements));
2018-12-17 16:55:11 +00:00
return false;
} else {
// A multi-element array in the schema must correspond to an element of the same length
// in the object. Each element in the object is validated against the corresponding
// element in the schema.
size_t i = 0;
for (auto const& element: this_arr->elements) {
checkSchemaInternal(
element.m->value.get(),
sch_arr->elements.at(i).m->value.get(),
flags,
errors,
prefix + "." + std::to_string(i));
++i;
}
2018-12-17 16:55:11 +00:00
}
} else if (!sch_str) {
QTC::TC("libtests", "JSON schema other type");
errors.push_back(err_prefix + " schema value is not dictionary, array, or string");
return false;
}
2018-12-17 16:55:11 +00:00
return errors.empty();
}
2022-01-17 23:40:38 +00:00
namespace
{
class JSONParser
{
public:
JSONParser(InputSource& is, JSON::Reactor* reactor) :
is(is),
2022-05-01 18:06:31 +00:00
reactor(reactor),
p(buf)
2022-01-17 23:40:38 +00:00
{
}
JSON parse();
2022-01-17 23:40:38 +00:00
private:
enum parser_state_e {
ps_top,
ps_dict_begin,
ps_dict_after_key,
ps_dict_after_colon,
ps_dict_after_item,
ps_dict_after_comma,
ps_array_begin,
ps_array_after_item,
ps_array_after_comma,
ps_done,
};
enum lex_state_e {
ls_top,
ls_number,
ls_number_minus,
ls_number_leading_zero,
ls_number_before_point,
ls_number_point,
ls_number_after_point,
ls_number_e,
ls_number_e_sign,
2022-01-17 23:40:38 +00:00
ls_alpha,
ls_string,
ls_backslash,
ls_u4,
ls_begin_array,
ls_end_array,
ls_begin_dict,
ls_end_dict,
ls_colon,
ls_comma,
2022-01-17 23:40:38 +00:00
};
2023-01-27 14:37:25 +00:00
struct StackFrame
{
StackFrame(parser_state_e state, JSON& item) :
2023-01-27 14:37:25 +00:00
state(state),
item(item)
{
}
parser_state_e state;
JSON item;
2023-01-27 14:37:25 +00:00
};
void getToken();
void handleToken();
void tokenError();
static void handle_u_code(
unsigned long codepoint,
qpdf_offset_t offset,
unsigned long& high_surrogate,
qpdf_offset_t& high_offset,
std::string& result);
inline void append();
inline void append(lex_state_e);
inline void ignore();
inline void ignore(lex_state_e);
InputSource& is;
2022-05-01 18:06:31 +00:00
JSON::Reactor* reactor;
lex_state_e lex_state{ls_top};
char buf[16384];
size_t bytes{0};
2022-01-17 23:40:38 +00:00
char const* p;
qpdf_offset_t u_count{0};
unsigned long u_value{0};
qpdf_offset_t offset{0};
bool done{false};
std::string token;
qpdf_offset_t token_start{0};
parser_state_e parser_state{ps_top};
2023-01-27 14:37:25 +00:00
std::vector<StackFrame> stack;
2022-01-17 23:40:38 +00:00
std::string dict_key;
qpdf_offset_t dict_key_offset{0};
2022-01-17 23:40:38 +00:00
};
} // namespace
void
JSONParser::handle_u_code(
unsigned long codepoint,
2022-05-21 13:05:54 +00:00
qpdf_offset_t offset,
unsigned long& high_surrogate,
2022-05-21 13:05:54 +00:00
qpdf_offset_t& high_offset,
std::string& result)
{
if ((codepoint & 0xFC00) == 0xD800) {
// high surrogate
qpdf_offset_t new_high_offset = offset;
if (high_offset) {
QTC::TC("libtests", "JSON 16 high high");
throw std::runtime_error(
"JSON: offset " + std::to_string(new_high_offset) +
": UTF-16 high surrogate found after previous high surrogate at offset " +
std::to_string(high_offset));
}
high_offset = new_high_offset;
high_surrogate = codepoint;
} else if ((codepoint & 0xFC00) == 0xDC00) {
// low surrogate
if (offset != (high_offset + 6)) {
QTC::TC("libtests", "JSON 16 low not after high");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": UTF-16 low surrogate found not immediately after high surrogate");
}
high_offset = 0;
codepoint = 0x10000U + ((high_surrogate & 0x3FFU) << 10U) + (codepoint & 0x3FF);
result += QUtil::toUTF8(codepoint);
} else {
result += QUtil::toUTF8(codepoint);
}
}
2023-01-24 00:44:16 +00:00
void
JSONParser::tokenError()
2022-01-17 23:40:38 +00:00
{
if (done) {
QTC::TC("libtests", "JSON parse ls premature end of input");
throw std::runtime_error("JSON: premature end of input");
2022-01-17 23:40:38 +00:00
}
if (lex_state == ls_u4) {
QTC::TC("libtests", "JSON parse bad hex after u");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset - u_count - 1) +
": \\u must be followed by four hex digits");
} else if (lex_state == ls_alpha) {
QTC::TC("libtests", "JSON parse keyword bad character");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": keyword: unexpected character " +
std::string(p, 1));
} else if (lex_state == ls_string) {
QTC::TC("libtests", "JSON parse control char in string");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": control character in string (missing \"?)");
} else if (lex_state == ls_backslash) {
QTC::TC("libtests", "JSON parse backslash bad character");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": invalid character after backslash: " + std::string(p, 1));
}
2023-01-24 00:44:16 +00:00
if (*p == '.') {
if (lex_state == ls_number || lex_state == ls_number_e || lex_state == ls_number_e_sign) {
QTC::TC("libtests", "JSON parse point after e");
2023-01-24 00:44:16 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": numeric literal: decimal point after e");
2022-01-17 23:40:38 +00:00
} else {
QTC::TC("libtests", "JSON parse duplicate point");
2023-01-24 00:44:16 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": numeric literal: decimal point already seen");
2022-01-17 23:40:38 +00:00
}
} else if (*p == 'e' || *p == 'E') {
QTC::TC("libtests", "JSON parse duplicate e");
2023-01-24 00:44:16 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": numeric literal: e already seen");
} else if ((*p == '+') || (*p == '-')) {
QTC::TC("libtests", "JSON parse unexpected sign");
2023-01-24 00:44:16 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": numeric literal: unexpected sign");
} else if (QUtil::is_space(*p) || strchr("{}[]:,", *p)) {
QTC::TC("libtests", "JSON parse incomplete number");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": numeric literal: incomplete number");
2023-01-24 00:44:16 +00:00
} else {
QTC::TC("libtests", "JSON parse numeric bad character");
throw std::runtime_error(
2023-01-24 00:44:16 +00:00
"JSON: offset " + std::to_string(offset) + ": numeric literal: unexpected character " +
std::string(p, 1));
}
throw std::logic_error("JSON::tokenError : unhandled error");
2023-01-24 00:44:16 +00:00
}
// Append current character to token and advance to next input character.
inline void
JSONParser::append()
{
token += *p;
++p;
++offset;
}
// Append current character to token, advance to next input character and transition to 'next' lexer
// state.
inline void
JSONParser::append(lex_state_e next)
{
lex_state = next;
token += *p;
++p;
++offset;
}
// Advance to next input character without appending the current character to token.
inline void
JSONParser::ignore()
{
++p;
++offset;
}
// Advance to next input character without appending the current character to token and transition
// to 'next' lexer state.
inline void
JSONParser::ignore(lex_state_e next)
{
lex_state = next;
++p;
++offset;
2022-01-17 23:40:38 +00:00
}
void
JSONParser::getToken()
{
token.clear();
// Keep track of UTF-16 surrogate pairs.
unsigned long high_surrogate = 0;
qpdf_offset_t high_offset = 0;
while (true) {
if (p == (buf + bytes)) {
p = buf;
bytes = is.read(buf, sizeof(buf));
if (bytes == 0) {
done = true;
break;
}
}
if ((*p < 32 && *p >= 0)) {
if (*p == '\t' || *p == '\n' || *p == '\r') {
// Legal white space not permitted in strings. This will always end the current
// token (unless we are still before the start of the token).
if (lex_state == ls_top) {
ignore();
} else {
break;
}
2022-01-17 23:40:38 +00:00
} else {
QTC::TC("libtests", "JSON parse null character");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: control or null character at offset " + std::to_string(offset));
2022-01-17 23:40:38 +00:00
}
} else if (*p == ',') {
if (lex_state == ls_top) {
ignore(ls_comma);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == ':') {
if (lex_state == ls_top) {
ignore(ls_colon);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == ' ') {
if (lex_state == ls_top) {
ignore();
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == '{') {
if (lex_state == ls_top) {
token_start = offset;
ignore(ls_begin_dict);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == '}') {
if (lex_state == ls_top) {
ignore(ls_end_dict);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == '[') {
if (lex_state == ls_top) {
token_start = offset;
ignore(ls_begin_array);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else if (*p == ']') {
if (lex_state == ls_top) {
ignore(ls_end_array);
return;
} else if (lex_state == ls_string) {
append();
} else {
break;
}
} else {
switch (lex_state) {
case ls_top:
token_start = offset;
if (*p == '"') {
ignore(ls_string);
} else if ((*p >= 'a') && (*p <= 'z')) {
append(ls_alpha);
} else if (*p == '-') {
append(ls_number_minus);
} else if ((*p >= '1') && (*p <= '9')) {
append(ls_number_before_point);
} else if (*p == '0') {
append(ls_number_leading_zero);
2022-01-17 23:40:38 +00:00
} else {
QTC::TC("libtests", "JSON parse bad character");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": unexpected character " +
std::string(p, 1));
2022-01-17 23:40:38 +00:00
}
break;
case ls_number_minus:
if ((*p >= '1') && (*p <= '9')) {
append(ls_number_before_point);
} else if (*p == '0') {
append(ls_number_leading_zero);
} else {
QTC::TC("libtests", "JSON parse number minus no digits");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": numeric literal: no digit after minus sign");
2022-01-17 23:40:38 +00:00
}
break;
case ls_number_leading_zero:
if (*p == '.') {
append(ls_number_point);
} else if (*p == 'e' || *p == 'E') {
append(ls_number_e);
2022-01-17 23:40:38 +00:00
} else {
QTC::TC("libtests", "JSON parse leading zero");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": number with leading zero");
2022-01-17 23:40:38 +00:00
}
break;
2022-01-17 23:40:38 +00:00
case ls_number_before_point:
if ((*p >= '0') && (*p <= '9')) {
append();
} else if (*p == '.') {
append(ls_number_point);
} else if (*p == 'e' || *p == 'E') {
append(ls_number_e);
} else {
tokenError();
}
break;
2022-01-17 23:40:38 +00:00
case ls_number_point:
if ((*p >= '0') && (*p <= '9')) {
append(ls_number_after_point);
} else {
tokenError();
}
break;
2022-01-17 23:40:38 +00:00
case ls_number_after_point:
if ((*p >= '0') && (*p <= '9')) {
append();
} else if (*p == 'e' || *p == 'E') {
append(ls_number_e);
} else {
tokenError();
}
break;
case ls_number_e:
if ((*p >= '0') && (*p <= '9')) {
append(ls_number);
} else if ((*p == '+') || (*p == '-')) {
append(ls_number_e_sign);
} else {
tokenError();
}
break;
case ls_number_e_sign:
if ((*p >= '0') && (*p <= '9')) {
append(ls_number);
} else {
tokenError();
}
2022-01-17 23:40:38 +00:00
break;
case ls_number:
// We only get here after we have seen an exponent.
if ((*p >= '0') && (*p <= '9')) {
append();
} else {
tokenError();
}
2022-01-17 23:40:38 +00:00
break;
case ls_alpha:
if ((*p >= 'a') && (*p <= 'z')) {
append();
} else {
tokenError();
}
break;
2022-01-17 23:40:38 +00:00
case ls_string:
if (*p == '"') {
if (high_offset) {
QTC::TC("libtests", "JSON 16 dangling high");
throw std::runtime_error(
"JSON: offset " + std::to_string(high_offset) +
": UTF-16 high surrogate not followed by low surrogate");
}
ignore();
return;
} else if (*p == '\\') {
ignore(ls_backslash);
} else {
append();
}
break;
2022-01-17 23:40:38 +00:00
case ls_backslash:
lex_state = ls_string;
switch (*p) {
case '\\':
case '\"':
case '/':
// \/ is allowed in json input, but so is /, so we don't map / to \/ in output.
token += *p;
break;
case 'b':
token += '\b';
break;
case 'f':
token += '\f';
break;
case 'n':
token += '\n';
break;
case 'r':
token += '\r';
break;
case 't':
token += '\t';
break;
case 'u':
lex_state = ls_u4;
u_count = 0;
u_value = 0;
break;
default:
lex_state = ls_backslash;
tokenError();
}
ignore();
2022-01-17 23:40:38 +00:00
break;
case ls_u4:
using ui = unsigned int;
if (ui val = ui(QUtil::hex_decode_char(*p)); val < 16) {
u_value = 16 * u_value + val;
} else {
tokenError();
}
if (++u_count == 4) {
handle_u_code(u_value, offset - 5, high_surrogate, high_offset, token);
lex_state = ls_string;
}
ignore();
break;
default:
throw std::logic_error("JSONParser::getToken : trying to handle delimiter state");
2022-01-17 23:40:38 +00:00
}
}
}
// We only get here if on end of input or if the last character was a control character or other
// delimiter.
2022-01-17 23:40:38 +00:00
if (!token.empty()) {
switch (lex_state) {
case ls_top:
// Can't happen
throw std::logic_error("tok_start set in ls_top while parsing");
break;
case ls_number_leading_zero:
case ls_number_before_point:
case ls_number_after_point:
lex_state = ls_number;
break;
case ls_number:
case ls_alpha:
// terminal state
break;
default:
tokenError();
2022-01-17 23:40:38 +00:00
}
}
}
void
JSONParser::handleToken()
{
if (lex_state == ls_top) {
2022-01-17 23:40:38 +00:00
return;
}
if (parser_state == ps_done) {
QTC::TC("libtests", "JSON parse junk after object");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) +
": material follows end of object: " + token);
2022-01-17 23:40:38 +00:00
}
const static JSON null_item = JSON::makeNull();
JSON item;
auto tos = stack.empty() ? null_item : stack.back().item;
auto ls = lex_state;
lex_state = ls_top;
2022-01-17 23:40:38 +00:00
switch (ls) {
case ls_begin_dict:
item = JSON::makeDictionary();
break;
2022-01-17 23:40:38 +00:00
case ls_begin_array:
item = JSON::makeArray();
2022-01-17 23:40:38 +00:00
break;
case ls_colon:
if (parser_state != ps_dict_after_key) {
QTC::TC("libtests", "JSON parse unexpected :");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": unexpected colon");
2022-01-17 23:40:38 +00:00
}
parser_state = ps_dict_after_colon;
return;
case ls_comma:
if (!((parser_state == ps_dict_after_item) || (parser_state == ps_array_after_item))) {
QTC::TC("libtests", "JSON parse unexpected ,");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": unexpected comma");
2022-01-17 23:40:38 +00:00
}
if (parser_state == ps_dict_after_item) {
parser_state = ps_dict_after_comma;
} else if (parser_state == ps_array_after_item) {
parser_state = ps_array_after_comma;
} else {
throw std::logic_error("JSONParser::handleToken: unexpected parser state for comma");
}
return;
case ls_end_array:
if (!(parser_state == ps_array_begin || parser_state == ps_array_after_item)) {
QTC::TC("libtests", "JSON parse unexpected ]");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": unexpected array end delimiter");
2022-01-17 23:40:38 +00:00
}
2023-01-27 14:37:25 +00:00
parser_state = stack.back().state;
tos.setEnd(offset);
if (reactor) {
reactor->containerEnd(tos);
}
if (parser_state != ps_done) {
stack.pop_back();
}
return;
case ls_end_dict:
if (!((parser_state == ps_dict_begin) || (parser_state == ps_dict_after_item))) {
QTC::TC("libtests", "JSON parse unexpected }");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": unexpected dictionary end delimiter");
}
2023-01-27 14:37:25 +00:00
parser_state = stack.back().state;
tos.setEnd(offset);
if (reactor) {
reactor->containerEnd(tos);
}
if (parser_state != ps_done) {
stack.pop_back();
}
return;
2022-01-17 23:40:38 +00:00
case ls_number:
item = JSON::makeNumber(token);
2022-01-17 23:40:38 +00:00
break;
case ls_alpha:
if (token == "true") {
item = JSON::makeBool(true);
} else if (token == "false") {
item = JSON::makeBool(false);
} else if (token == "null") {
item = JSON::makeNull();
2022-01-17 23:40:38 +00:00
} else {
QTC::TC("libtests", "JSON parse invalid keyword");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": invalid keyword " + token);
2022-01-17 23:40:38 +00:00
}
break;
case ls_string:
if (parser_state == ps_dict_begin || parser_state == ps_dict_after_comma) {
dict_key = token;
dict_key_offset = token_start;
parser_state = ps_dict_after_key;
return;
} else {
item = JSON::makeString(token);
}
2022-01-17 23:40:38 +00:00
break;
default:
throw std::logic_error("JSONParser::handleToken : non-terminal lexer state encountered");
2022-01-17 23:40:38 +00:00
break;
}
item.setStart(token_start);
item.setEnd(offset);
2022-01-17 23:40:38 +00:00
switch (parser_state) {
case ps_dict_begin:
case ps_dict_after_comma:
QTC::TC("libtests", "JSON parse string as dict key");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": expect string as dictionary key");
break;
2022-01-17 23:40:38 +00:00
case ps_dict_after_colon:
if (tos.checkDictionaryKeySeen(dict_key)) {
QTC::TC("libtests", "JSON parse duplicate key");
2022-01-17 23:40:38 +00:00
throw std::runtime_error(
"JSON: offset " + std::to_string(dict_key_offset) + ": duplicated dictionary key");
2022-01-17 23:40:38 +00:00
}
if (!reactor || !reactor->dictionaryItem(dict_key, item)) {
tos.addDictionaryMember(dict_key, item);
2022-01-17 23:40:38 +00:00
}
parser_state = ps_dict_after_item;
break;
2022-01-17 23:40:38 +00:00
case ps_array_begin:
case ps_array_after_comma:
if (!reactor || !reactor->arrayItem(item)) {
tos.addArrayElement(item);
2022-05-01 18:06:31 +00:00
}
parser_state = ps_array_after_item;
break;
2022-05-01 18:06:31 +00:00
case ps_top:
if (!(item.isDictionary() || item.isArray())) {
stack.emplace_back(ps_done, item);
parser_state = ps_done;
return;
2022-01-17 23:40:38 +00:00
}
parser_state = ps_done;
break;
2022-01-17 23:40:38 +00:00
case ps_dict_after_key:
QTC::TC("libtests", "JSON parse expected colon");
throw std::runtime_error("JSON: offset " + std::to_string(offset) + ": expected ':'");
break;
2022-01-17 23:40:38 +00:00
case ps_dict_after_item:
QTC::TC("libtests", "JSON parse expected , or }");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": expected ',' or '}'");
break;
2022-01-17 23:40:38 +00:00
case ps_array_after_item:
QTC::TC("libtests", "JSON parse expected, or ]");
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": expected ',' or ']'");
break;
2022-01-17 23:40:38 +00:00
case ps_done:
throw std::logic_error("JSONParser::handleToken: unexpected parser state");
2022-01-17 23:40:38 +00:00
}
if (item.isDictionary() || item.isArray()) {
stack.emplace_back(parser_state, item);
// Calling container start method is postponed until after adding the containers to their
// parent containers, if any. This makes it much easier to keep track of the current nesting
// level.
if (item.isDictionary()) {
if (reactor) {
reactor->dictionaryStart();
}
parser_state = ps_dict_begin;
} else if (item.isArray()) {
if (reactor) {
reactor->arrayStart();
}
parser_state = ps_array_begin;
}
2023-01-27 14:37:25 +00:00
if (stack.size() > 500) {
throw std::runtime_error(
"JSON: offset " + std::to_string(offset) + ": maximum object depth exceeded");
2022-01-17 23:40:38 +00:00
}
}
}
JSON
JSONParser::parse()
2022-01-17 23:40:38 +00:00
{
while (!done) {
2022-01-17 23:40:38 +00:00
getToken();
handleToken();
}
if (parser_state != ps_done) {
QTC::TC("libtests", "JSON parse premature EOF");
2022-01-17 23:40:38 +00:00
throw std::runtime_error("JSON: premature end of input");
}
2023-01-27 14:37:25 +00:00
auto const& tos = stack.back().item;
if (reactor && !(tos.isArray() || tos.isDictionary())) {
2022-05-01 18:06:31 +00:00
reactor->topLevelScalar();
}
return tos;
2022-01-17 23:40:38 +00:00
}
JSON
JSON::parse(InputSource& is, Reactor* reactor)
{
JSONParser jp(is, reactor);
return jp.parse();
}
JSON
JSON::parse(std::string const& s)
2022-01-17 23:40:38 +00:00
{
BufferInputSource bis("json input", s);
JSONParser jp(bis, nullptr);
return jp.parse();
2022-01-17 23:40:38 +00:00
}
2022-05-01 18:06:31 +00:00
void
2022-05-21 13:05:54 +00:00
JSON::setStart(qpdf_offset_t start)
2022-05-01 18:06:31 +00:00
{
m->start = start;
}
void
2022-05-21 13:05:54 +00:00
JSON::setEnd(qpdf_offset_t end)
2022-05-01 18:06:31 +00:00
{
m->end = end;
}
2022-05-21 13:05:54 +00:00
qpdf_offset_t
2022-05-01 18:06:31 +00:00
JSON::getStart() const
{
return m->start;
}
2022-05-21 13:05:54 +00:00
qpdf_offset_t
2022-05-01 18:06:31 +00:00
JSON::getEnd() const
{
return m->end;
}