1+ #include < iostream>
2+ #include < fstream>
3+ #include " json.hpp"
4+ #include < regex>
5+ #include < memory>
6+
7+ using json = nlohmann::ordered_json;
8+
9+
10+ std::string generate_uuid () {
11+ static std::random_device rd;
12+ static std::mt19937 generator (rd ());
13+ static std::uniform_int_distribution<int > distribution (0 , 15 );
14+
15+ const char *v = " 0123456789abcdef"
16+ std::stringstream uuid;
17+
18+ for (int i = 0 ; i < 8 ; ++i) {
19+ uuid << v[distribution (generator)];
20+ }
21+ return uuid.str ();
22+ }
23+
24+
25+ std::string jsonrepair (const std::string value) {
26+ std::array<char , 128 > buffer;
27+ std::string result;
28+ // Ensure the command passed to popen() is null-terminated
29+ std::string tmpfile_name = " ." generate_uuid () + " .json"
30+ std::ofstream outfile (tmpfile_name);
31+ outfile << value; // Assuming jsonStr contains your JSON string
32+ outfile.close ();
33+ std::string command = " node jsonrepair.ts "
34+ std::unique_ptr<FILE, decltype (&pclose)> pipe (popen (command.c_str (), " r"
35+ if (!pipe) {
36+ throw std::runtime_error (" popen() failed!"
37+ }
38+ while (fgets (buffer.data (), buffer.size (), pipe.get ()) != nullptr ) {
39+ result += buffer.data ();
40+ }
41+ return result;
42+ }
43+
44+
45+ json parse_if_json (const std::string& value) {
46+ try {
47+ // json repair here
48+ return json::parse (jsonrepair (value));
49+ } catch (const json::parse_error&) {
50+ return value; // Return the original string if parsing fails
51+ }
52+ }
53+
54+
55+ std::string clean_command_string (const std::string& command_str) {
56+ std::string cleaned_command = std::regex_replace (command_str, std::regex (R"( \\(?!["\\/bfnrt]|u[a-fA-F0-9]{4}))" " "
57+ cleaned_command = std::regex_replace (cleaned_command, std::regex (R"( \\")" " \" "
58+
59+ if (cleaned_command.front () == ' "' back () == ' "'
60+ cleaned_command = cleaned_command.substr (1 , cleaned_command.size () - 2 );
61+ }
62+ return cleaned_command;
63+ }
64+
65+
66+ json clean_json_strings (const std::string& input_str) {
67+ try {
68+ // json repair here
69+ std::string fixed_str = jsonrepair (input_str);
70+ json data = json::parse (fixed_str);
71+
72+ for (auto & [key, value] : data.items ()) {
73+ if (value.is_string ()) {
74+ std::string val = value.get <std::string>();
75+ if (val.front () == ' {' front () == ' ['
76+ data[key] = parse_if_json (val);
77+ } else {
78+ data[key] = clean_command_string (val);
79+ }
80+ } else if (value.is_object ()) {
81+ for (auto & [k, v] : value.items ()) {
82+ if (v.is_string ()) {
83+ v = clean_command_string (v.get <std::string>());
84+ }
85+ }
86+ }
87+ }
88+ return data;
89+ } catch (const json::parse_error& e) {
90+ std::cout << " Error decoding JSON: " what () << std::endl;
91+ return nullptr ;
92+ }
93+ }
94+
95+
96+
97+
98+ std::vector<json> rubra_fc_json_tool_extractor (const std::string& output_str) {
99+ std::vector<json> result;
100+ printf (" OUTPUT STR TO BE PARSED : %s" c_str ());
101+ if (output_str.find (" endtoolcall"
102+ return result;
103+ }
104+
105+ std::vector<std::string> listOfStrToParse;
106+ size_t start = 0 , end = 0 ;
107+
108+ // Iterate until all instances of "endtoolcall" are processed
109+ while ((end = output_str.find (" endtoolcall"
110+ std::string segment = output_str.substr (start, end - start);
111+ size_t pos = segment.find (" starttoolcall"
112+ if (pos != std::string::npos) {
113+ // Extract substring after "toolcall"
114+ listOfStrToParse.push_back (segment.substr (pos + std::string (" starttoolcall" length ()));
115+ }
116+ start = end + std::string (" endtoolcall" length (); // Move past the "endtoolcall"
117+ }
118+
119+ std::vector<json> function_call_json;
120+
121+ try {
122+ for (const auto & line : listOfStrToParse) {
123+ // json fc = json::parse(line);
124+ json fc = clean_json_strings (line);
125+ if (fc[" arguments" is_string ()) {
126+ fc[" arguments" json::parse (fc[" arguments" get <std::string>());
127+ }
128+ if (!fc.is_null ()) {
129+ function_call_json.push_back (fc);
130+ }
131+
132+ }
133+ } catch (const std::exception& e) {
134+ std::cerr << " Error: " what () << std::endl;
135+ }
136+
137+ for (const auto & fc : function_call_json) {
138+ json func_call;
139+ func_call[" id" generate_uuid ();
140+ func_call[" name" " name"
141+ func_call[" kwargs" " arguments"
142+ func_call[" type" " function"
143+ result.push_back (func_call);
144+ }
145+
146+ return result;
147+ }
0 commit comments