parse_primitives.cc File Reference

#include <iostream>
#include <strings.h>
#include <stdexcept>
#include <sstream>
#include "parse_primitives.h"
#include "hash.h"
#include "vw_exception.h"

Go to the source code of this file.

Functions
bool	substring_equal (const substring &a, const substring &b)
bool	substring_equal (const substring &ss, const char *str)
size_t	substring_len (substring &s)
hash_func_t	getHasher (const std::string &s)
bool	operator== (const substring &ss, const char *str)
bool	operator== (const char *str, const substring &ss)
bool	operator== (const substring &ss1, const substring &ss2)
bool	operator!= (const substring &ss, const char *str)
bool	operator!= (const char *str, const substring &ss)
bool	operator!= (const substring &ss1, const substring &ss2)
std::vector< substring >	escaped_tokenize (char delim, substring s, bool allow_empty)
std::ostream &	operator<< (std::ostream &os, const substring &ss)
std::ostream &	operator<< (std::ostream &os, const v_array< substring > &ss)

Function Documentation

escaped_tokenize()

std::vector<substring> escaped_tokenize	(	char	delim,
		substring	s,
		bool	allow_empty
	)

Definition at line 58 of file parse_primitives.cc.

References substring::begin, and substring::end.

Referenced by safe_index(), and VW::to_argv_escaped().

{
  std::vector<substring> tokens;
  substring current;
  current.begin = s.begin;
  bool in_escape = false;
  char* reading_head = s.begin;
  char* writing_head = s.begin;

  while (reading_head < s.end)
  {
    char current_character = *reading_head++;

    if (in_escape)
    {
      *writing_head++ = current_character;
      in_escape = false;
    }
    else
    {
      if (current_character == delim)
      {
        current.end = writing_head++;
        *current.end = '\0';
        if (current.begin != current.end || allow_empty)
        {
          tokens.push_back(current);
        }

        // Regardless of whether the token was saved, we need to reset the current token.
        current.begin = writing_head;
        current.end = writing_head;
      }
      else if (current_character == '\\')
      {
        in_escape = !in_escape;
      }
      else
      {
        *writing_head++ = current_character;
      }
    }
  }

  current.end = writing_head;
  *current.end = '\0';
  if (current.begin != current.end || allow_empty)
  {
    tokens.push_back(current);
  }

  return tokens;
}

getHasher()

hash_func_t getHasher

(

const std::string &

s

)

Definition at line 36 of file parse_primitives.cc.

References hashall(), hashstring(), and THROW.

Referenced by VW::hash_feature_static(), VW::hash_space_static(), and parse_feature_tweaks().

{
  if (s == "strings")
    return hashstring;
  else if (s == "all")
    return hashall;
  else
    THROW("Unknown hash function: " << s);
}

operator!=() [1/3]

bool operator!=	(	const substring &	ss,
		const char *	str
	)

Definition at line 52 of file parse_primitives.cc.

Referenced by tokenize().

{ return !(ss == str); }

operator!=() [2/3]

bool operator!=	(	const char *	str,
		const substring &	ss
	)

Definition at line 54 of file parse_primitives.cc.

{ return !(ss == str); }

operator!=() [3/3]

bool operator!=	(	const substring &	ss1,
		const substring &	ss2
	)

Definition at line 56 of file parse_primitives.cc.

{ return !(ss1 == ss2); }

operator<<() [1/2]

std::ostream& operator<<	(	std::ostream &	os,
		const substring &	ss
	)

Definition at line 112 of file parse_primitives.cc.

References substring::begin, and substring::end.

{
  std::string s(ss.begin, ss.end - ss.begin);
  return os << s;
}

operator<<() [2/2]

std::ostream& operator<<	(	std::ostream &	os,
		const v_array< substring > &	ss
	)

Definition at line 118 of file parse_primitives.cc.

{
  substring* it = ss.cbegin();

  if (it == ss.cend())
  {
    return os;
  }

  os << *it;

  for (it++; it != ss.cend(); it++)
  {
    os << ",";
    os << *it;
  }

  return os;
}

operator==() [1/3]

bool operator==	(	const substring &	ss,
		const char *	str
	)

Definition at line 46 of file parse_primitives.cc.

References substring_equal().

Referenced by tokenize().

{ return substring_equal(ss, str); }

operator==() [2/3]

bool operator==	(	const char *	str,
		const substring &	ss
	)

Definition at line 48 of file parse_primitives.cc.

References substring_equal().

{ return substring_equal(ss, str); }

operator==() [3/3]

bool operator==	(	const substring &	ss1,
		const substring &	ss2
	)

Definition at line 50 of file parse_primitives.cc.

References substring_equal().

{ return substring_equal(ss1, ss2); }

substring_equal() [1/2]

bool substring_equal	(	const substring &	a,
		const substring &	b
	)

Definition at line 19 of file parse_primitives.cc.

References substring::begin, and substring::end.

Referenced by namedlabels::namedlabels(), operator==(), parse_dictionary_argument(), CB::parse_label(), COST_SENSITIVE::parse_label(), CCB::parse_label(), and tokenize().

{
  return (a.end - a.begin == b.end - b.begin)  // same length
      && (strncmp(a.begin, b.begin, a.end - a.begin) == 0);
}

substring_equal() [2/2]

bool substring_equal	(	const substring &	ss,
		const char *	str
	)

Definition at line 25 of file parse_primitives.cc.

References substring::begin, and substring::end.

{
  size_t len_ss = ss.end - ss.begin;
  size_t len_str = strlen(str);
  if (len_ss != len_str)
    return false;
  return (strncmp(ss.begin, str, len_ss) == 0);
}

substring_len()

size_t substring_len

(

substring &

s

)

Definition at line 34 of file parse_primitives.cc.

References substring::begin, and substring::end.

Referenced by VW::cb_sample_data::learn_or_predict(), and tokenize().

{ return s.end - s.begin; }