copyscan.hpp

/*
 SPDX-FileCopyrightText: © 2015 Siemens AG
 Author: Florian Krügel
 
 SPDX-License-Identifier: GPL-2.0-only
*/
#ifndef COPYSCAN_HPP_
#define COPYSCAN_HPP_
 
#include "scanners.hpp"
#include "regex.hpp"
 
using std::vector;
using std::pair;
 
struct CleanupResult {
  std::string content;  
  enum class Disposition { KEEP, DEACTIVATE, DISCARD } disposition;
};
 
class hCopyrightScanner : public scanner
{
public:
  void ScanString(const string& s, list<match>& results) const;
  hCopyrightScanner();
  CleanupResult Cleanup(const string &raw) const;
  void TrimPunctuation(string &text) const;
  void RemoveNoisePatterns(string& text) const;
  void NormalizeCopyright(string& text) const;
  void StripSuffixes(string& text) const;
private:
  rx::regex regCopyright, regException, regExceptionCopy, regNonBlank, regSimpleCopyright,
  regSpdxCopyright, regRemoveFileStmt, regStripLicenseTrail, regStripTrademarkTrail, regStripAllRightReserveTrail, regExceptionVerbFollow, regExceptionAdjectivePrefix, regExceptionTemplate, regExceptionPassive, regStripCopySymNonYear, regExceptionBinaryNoise, regExceptionMeta, regExceptionCharNameRun;
} ;
 
#endif