scancode_wrapper.cc

/*
 SPDX-FileCopyrightText: © 2021 Sarita Singh <saritasingh.0425@gmail.com>
 
 SPDX-License-Identifier: GPL-2.0-only
*/
 
#include "scancode_wrapper.hpp"
 
#define MINSCORE 50
 
unsigned getFilePointer(const string &filename, size_t start_line,
                        const string &match_text) {
  ifstream checkfile(filename);
  string str;
  if (checkfile.is_open()) {
    for (size_t i = 0; i < start_line - 1; i++) {
      getline(checkfile, str);
    }
    unsigned int file_p = checkfile.tellg();
    getline(checkfile, str);
    unsigned int pos = str.find(match_text);
    if (pos != string::npos) {
      return file_p + pos;
    }else{
      LOG_NOTICE("Failed to find startbyte for %s\n", filename.c_str());
    }
  }
  return 0;
}
 
 
void scanFileWithScancode(const State &state, string fileLocation, string outputFile) {
  string projectUser = fo_config_get(sysconfig, "DIRECTORIES", "PROJECTUSER",
  NULL);
  string cacheDir = fo_config_get(sysconfig, "DIRECTORIES", "CACHEDIR",
  NULL);
 
  string command =
    "PYTHONPATH='/home/" + projectUser + "/pythondeps/' " +
    "python3 runscanonfiles.py -" + state.getCliOptions() + " " +
    ((state.getCliOptions().find('l') != string::npos) ? "-m " +
    to_string(MINSCORE): "") + " " + fileLocation + " " + outputFile;
 
  int returnvalue = system(command.c_str());
 
  if (returnvalue != 0) {
    LOG_FATAL("could not execute scancode command: %s \n", command.c_str());
    bail(1);
  }
 
  if (unlink(fileLocation.c_str()) != 0)
  {
    LOG_FATAL("Unable to delete file %s \n", fileLocation.c_str());
  }
}
 
map<string, vector<Match>> extractDataFromScancodeResult(const string& scancodeResult, const string& filename) {
  Json::CharReaderBuilder json_reader_builder;
  auto scanner = unique_ptr<Json::CharReader>(json_reader_builder.newCharReader());
  Json::Value scancodevalue;
  string errors;
  const bool isSuccessful = scanner->parse(scancodeResult.c_str(),
      scancodeResult.c_str() + scancodeResult.length(), &scancodevalue, &errors);
  map<string, vector<Match>> result;
  vector<Match> licenses;
  if (isSuccessful) {
    Json::Value licensearrays = scancodevalue["licenses"];
    if(licensearrays.empty())
    {
      result["scancode_license"].push_back(Match("No_license_found"));
    }
    else
    {
      for (auto oneresult : licensearrays)
      {
          string licensename = oneresult["spdx_license_key"].asString();
          int percentage = (int)oneresult["score"].asFloat();
          string full_name=oneresult["name"].asString();
          string text_url=oneresult["text_url"].asString();
          string match_text = oneresult["matched_text"].asString();
          unsigned long start_line=oneresult["start_line"].asUInt();
          string temp_text= match_text.substr(0,match_text.find("\n"));
          unsigned start_pointer = getFilePointer(filename, start_line, temp_text);
          unsigned length = match_text.length();
          result["scancode_license"].push_back(Match(licensename,percentage,full_name,text_url,start_pointer,length));
      }
    }
 
    Json::Value copyarrays = scancodevalue["copyrights"];
    for (auto oneresult : copyarrays) {
        string copyrightname = oneresult["value"].asString();
        unsigned long start_line=oneresult["start"].asUInt();
        string temp_text= copyrightname.substr(0,copyrightname.find("[\n\t]"));
        unsigned start_pointer = getFilePointer(filename, start_line, temp_text);
        unsigned length = copyrightname.length();
        string type="scancode_statement";
        result["scancode_statement"].push_back(Match(copyrightname,type,start_pointer,length));
    }
 
    Json::Value holderarrays = scancodevalue["holders"];
    for (auto oneresult : holderarrays) {
        string holdername = oneresult["value"].asString();
        unsigned long start_line=oneresult["start"].asUInt();
        string temp_text= holdername.substr(0,holdername.find("\n"));
        unsigned start_pointer = getFilePointer(filename, start_line, temp_text);
        unsigned length = holdername.length();
        string type="scancode_author";
        result["scancode_author"].push_back(Match(holdername,type,start_pointer,length));
    }
 
    Json::Value emailarrays = scancodevalue["emails"];
    for (auto oneresult : emailarrays) {
        string emailname = oneresult["value"].asString();
        unsigned long start_line=oneresult["start"].asUInt();
        string temp_text= emailname.substr(0,emailname.find("\n"));
        unsigned start_pointer = getFilePointer(filename, start_line, temp_text);
        unsigned length = emailname.length();
        string type="scancode_email";
        result["scancode_email"].push_back(Match(emailname,type,start_pointer,length));
    }
 
    Json::Value urlarrays = scancodevalue["urls"];
    for (auto oneresult : urlarrays) {
        string urlname = oneresult["value"].asString();
        unsigned long start_line=oneresult["start"].asUInt();
        string temp_text= urlname.substr(0,urlname.find("\n"));
        unsigned start_pointer = getFilePointer(filename, start_line, temp_text);
        unsigned length = urlname.length();
        string type="scancode_url";
        result["scancode_url"].push_back(Match(urlname,type,start_pointer,length));
    }
  } else {
    LOG_FATAL("JSON parsing failed %s \n", errors.c_str());
  }
  return result;
}