#include "analyze.hpp"
#include "lolly/data/numeral.hpp"
#include "ntuple.hpp"

Include dependency graph for analyze.cpp:

Functions
bool	is_iso_alpha (char c)
	Checks if a character is an ISO alphabetic character.

bool	is_iso_locase (char c)
	Checks if a character is an ISO lowercase alphabetic character.

bool	is_iso_upcase (char c)
	Checks if a character is an ISO uppercase alphabetic character.

bool	is_alpha (string s)
	Checks if a string contains only alphabetic characters.

bool	is_alphanum (string s)

bool	is_locase_alpha (string s)
	Checks if a string contains only lowercase alphabetic characters.

bool	is_iso_alpha (string s)
	Checks if a string contains only ISO alphabetic characters.

bool	is_numeric (string s)
	Checks if a string contains only numeric characters.

char	upcase (char c)
	Converts a lowercase character to uppercase.

char	locase (char c)
	Converts an uppercase character to lowercase.

char	closing_delimiter (char c)
	Finds the closing delimiter corresponding to the given opening delimiter.

string	upcase_first (string s)
	Converts the first character of a string to uppercase.

string	locase_first (string s)
	Converts the first character of a string to lowercase.

string	upcase_all (string s)
	Converts all lowercase characters in a string to uppercase.

string	locase_all (string s)
	Converts all uppercase characters in a string to lowercase.

string	string_union (string s1, string s2)
	Union of two strings.

string	string_minus (string s1, string s2)
	Remove characters from one string that are in another string.

string	remove_prefix (string s, string prefix)
	Remove the prefix from s if matches.

string	remove_suffix (string s, string suffix)
	Remove the suffix from s if matches.

string	ispanish_to_spanish (string s)
	Convert ispanish string to Spanish string.

string	spanish_to_ispanish (string s)
	Convert Spanish string to ispanish string.

string	igerman_to_german (string s)
	Convert igerman string to german string.

string	german_to_igerman (string s)
	Convert german string to igerman string.

static char	il2_to_cork (char c)

static char	cork_to_il2 (char c)

string	il2_to_cork (string s)

string	cork_to_il2 (string s)

string	alpha_nr (int nr)
	Generates an alphabetic string for an integer.

string	Alpha_nr (int nr)
	Generates an uppercase alphabetic string for an integer.

string	fnsymbol_nr (int nr)
	Generates footnote symbols for a given integer.

string	raw_quote (string s)
	Add quotes around a string to indicate it's a string, not a symbol.

string	raw_unquote (string s)
	Remove quotes from a string label.

string	escape_sh (string s)
	Escape a string for use in shell scripts.

string	escape_generic (string s)
	Escape a string with generic escape sequences.

string	escape_verbatim (string s)
	Escape a string to be displayed verbatim.

string	escape_spaces (string s)
	Escape spaces in a string with a backslash.

string	dos_to_better (string s)
	Convert DOS line endings to more standard line endings.

string	unescape_guile (string s)
	Unescape a Guile-syntax string.

bool	test (string s, int i, const char *test)

bool	test (string s, int i, string test)

bool	starts (string s, const char *what)

bool	starts (string s, const string what)

bool	ends (string s, const char *what)

bool	ends (string s, const string r)

bool	read (string s, int &i, const char *test)

bool	read (string s, string test)

bool	read (string s, int &i, string test)

bool	read_line (string s, int &i, string &result)

bool	read_int (string s, int &i, int &result)

bool	read_double (string s, int &i, double &result)

bool	read_word (string s, int &i, string &result)

bool	is_whitespace (string s)

void	skip_spaces (string s, int &i)

void	skip_whitespace (string s, int &i)

void	skip_line (string s, int &i)

void	skip_symbol (string s, int &i)

string	convert_tabs_to_spaces (string s, int tw)
	Converts tabs in a string to spaces.

void	parse (string s, int &pos, QI &ret)

void	parse (string s, int &pos, QN &ret)

void	parse (string s, int &pos, HI &ret)

void	parse (string s, int &pos, HN &ret)

void	parse (string s, int &pos, SI &ret)

void	parse (string s, int &pos, SI *&a, int len)

int	index_of (string s, char c)

int	search_forwards (array< string > a, int pos, string in)

int	search_forwards (string s, int pos, string in)

int	search_forwards (string s, string in)

bool	occurs (string what, string in)

bool	contains (string s, string what)

bool	contains (string s, char c)

int	search_backwards (string s, int pos, string in)

int	search_backwards (string s, string in)

int	count_occurrences (string s, string in)

int	overlapping (string s1, string s2)

string	replace (string s, string what, string by)

static bool	match_wildcard (string s, int spos, string w, int wpos)

bool	match_wildcard (string s, string w)

int	find_non_alpha (string s, int pos, bool forward)

array< string >	tokenize (string s, string sep)

string	recompose (array< string > a, string sep)

string	trim_spaces_left (string s)

string	trim_spaces_right (string s)

string	trim_spaces (string s)

array< string >	trim_spaces (array< string > a)

static int	find_longest (string s1, string s2, int &c1, int &c2)

static void	find_common (string s1, string s2, int &c1, int &c2)

array< int >	differences (string s1, string s2)
	the differences between two strings by identifying the common substrings and returning the different sections' indices.

int	distance (string s1, string s2)
	a measure of difference (distance) between two strings.

void	parse_length (string s, double &len, string &unit)
	Parses a string containing a length value and its unit.

Variables
static string	il2_to_cork_string

static string	cork_to_il2_string

Function Documentation

◆ is_iso_alpha() [1/2]

bool is_iso_alpha ( char c )

Checks if a character is an ISO alphabetic character.

Parameters

c	The character to be checked.

Returns: True if the character is an ISO alphabetic character; otherwise, returns False.

Definition at line 21 of file analyze.cpp.

                      {
  int i= ((int) ((unsigned char) c));
  return ((c >= 'a') && (c <= 'z')) || ((c >= 'A') && (c <= 'Z')) ||
         ((i >= 128) && (i != 159) && (i != 189) && (i != 190) && (i != 191));
}

◆ is_iso_locase()

bool is_iso_locase ( char c )

Checks if a character is an ISO lowercase alphabetic character.

Parameters

c	The character to check.

Returns: True if the character is ISO lowercase alphabetic, otherwise false.

Definition at line 28 of file analyze.cpp.

                       {
  int code= (int) ((unsigned char) c);
  return ((c >= 'a') && (c <= 'z')) || ((code >= 160) && (code < 189)) ||
         (code >= 224);
}

◆ is_iso_upcase()

bool is_iso_upcase ( char c )

Checks if a character is an ISO uppercase alphabetic character.

Parameters

c	The character to check.

Returns: True if the character is ISO uppercase alphabetic, otherwise false.

Definition at line 35 of file analyze.cpp.

                       {
  int code= (int) ((unsigned char) c);
  return ((c >= 'A') && (c <= 'Z')) || ((code >= 128) && (code < 159)) ||
         ((code >= 192) && (code < 224));
}

◆ is_alpha()

bool is_alpha ( string s )

Checks if a string contains only alphabetic characters.

Parameters

s	The string to check.

Returns: True if all characters in the string are alphabetic, otherwise false.

Definition at line 46 of file analyze.cpp.

                    {
  int i;
  if (N (s) == 0) return false;
  for (i= 0; i < N (s); i++)
    if (!is_alpha (s[i])) return false;
  return true;
}

◆ is_alphanum()

bool is_alphanum ( string s )

Definition at line 55 of file analyze.cpp.

                       {
  int i;
  if (N (s) == 0) return false;
  for (i= 0; i < N (s); i++)
    if (!(is_alpha (s[i]) || is_digit (s[i]))) return false;
  return true;
}

◆ is_locase_alpha()

bool is_locase_alpha ( string s )

Checks if a string contains only lowercase alphabetic characters.

Parameters

s	The string to check.

Returns: True if all characters in the string are lowercase alphabetic, otherwise false.

Definition at line 64 of file analyze.cpp.

                           {
  int i;
  if (N (s) == 0) return false;
  for (i= 0; i < N (s); i++)
    if (s[i] < 'a' || s[i] > 'z') return false;
  return true;
}

◆ is_iso_alpha() [2/2]

bool is_iso_alpha ( string s )

Checks if a string contains only ISO alphabetic characters.

Parameters

s	The string to check.

Returns: True if all characters in the string are ISO alphabetic, otherwise false.

Definition at line 73 of file analyze.cpp.

                        {
  int i;
  if (N (s) == 0) return false;
  for (i= 0; i < N (s); i++)
    if (!is_iso_alpha (s[i])) return false;
  return true;
}

◆ is_numeric()

bool is_numeric ( string s )

Checks if a string contains only numeric characters.

Parameters

s	The string to check.

Returns: True if all characters in the string are numeric, otherwise false.

Definition at line 82 of file analyze.cpp.

                      {
  int i;
  if (N (s) == 0) return false;
  for (i= 0; i < N (s); i++)
    if (!is_numeric (s[i])) return false;
  return true;
}

◆ upcase()

char upcase ( char s )

Converts a lowercase character to uppercase.

Parameters

c	The character to convert.

Returns: The uppercase version of the character if it is an ISO lowercase alphabetic character; otherwise, returns the original character.

Definition at line 95 of file analyze.cpp.

                {
  if (is_iso_locase (c)) return (char) (((int) ((unsigned char) c)) - 32);
  else return c;
}

◆ locase()

char locase ( char s )

Converts an uppercase character to lowercase.

Parameters

c	The character to convert.

Returns: The lowercase version of the character if it is an ISO uppercase alphabetic character; otherwise, returns the original character.

Definition at line 101 of file analyze.cpp.

                {
  if (is_iso_upcase (c)) return (char) (((int) ((unsigned char) c)) + 32);
  else return c;
}

◆ closing_delimiter()

char closing_delimiter ( char c )

Finds the closing delimiter corresponding to the given opening delimiter.

Parameters

c	The opening delimiter character.

Returns: The corresponding closing delimiter if known, otherwise returns the original character.

Definition at line 107 of file analyze.cpp.

                           {
  if (c == '{') return '}';
  if (c == '(') return ')';
  if (c == '[') return ']';
  return c;
}

◆ upcase_first()

string upcase_first ( string s )

Converts the first character of a string to uppercase.

Parameters

s	The string to convert.

Returns: A new string where the first character is converted to uppercase, if it is an ISO lowercase alphabetic character.

Definition at line 115 of file analyze.cpp.

                        {
  if ((N (s) == 0) || (!is_iso_locase (s[0]))) return s;
  return string ((char) (((int) ((unsigned char) s[0])) - 32)) * s (1, N (s));
}

◆ locase_first()

string locase_first ( string s )

Converts the first character of a string to lowercase.

Parameters

s	The string to convert.

Returns: A new string where the first character is converted to lowercase, if it is an ISO uppercase alphabetic character.

Definition at line 121 of file analyze.cpp.

                        {
  if ((N (s) == 0) || (!is_iso_upcase (s[0]))) return s;
  return string ((char) (((int) ((unsigned char) s[0])) + 32)) * s (1, N (s));
}

◆ upcase_all()

string upcase_all ( string s )

Converts all lowercase characters in a string to uppercase.

Parameters

s	The string to convert.

Returns: A new string where all ISO lowercase alphabetic characters are converted to uppercase.

Definition at line 127 of file analyze.cpp.

                      {
  int    i;
  string r (N (s));
  for (i= 0; i < N (s); i++)
    if (!is_iso_locase (s[i])) r[i]= s[i];
    else r[i]= (char) (((int) ((unsigned char) s[i])) - 32);
  return r;
}

◆ locase_all()

string locase_all ( string s )

Converts all uppercase characters in a string to lowercase.

Parameters

s	The string to convert.

Returns: A new string where all ISO uppercase alphabetic characters are converted to lowercase.

Definition at line 137 of file analyze.cpp.

                      {
  int    i;
  string r (N (s));
  for (i= 0; i < N (s); i++)
    if (!is_iso_upcase (s[i])) r[i]= s[i];
    else r[i]= (char) (((int) ((unsigned char) s[i])) + 32);
  return r;
}

◆ string_union()

string string_union	(	string	s1,
		string	s2 )

Union of two strings.

This function performs the union operation between two strings.

Parameters

s1	The first string.
s2	The second string.

Returns: A new string containing the union of s1 and s2.

Definition at line 151 of file analyze.cpp.

                                    {
  return string_minus (s1, s2) * s2;
}

◆ string_minus()

string string_minus	(	string	s1,
		string	s2 )

Remove characters from one string that are in another string.

Parameters

s1	The first string.
s2	The second string.

Returns: A new string containing s1 - s2.

Definition at line 156 of file analyze.cpp.

                                    {
  string r;
  int    i1, n1= N (s1), i2, n2= N (s2);
  for (i1= 0; i1 < n1; i1++) {
    for (i2= 0; i2 < n2; i2++)
      if (s1[i1] == s2[i2]) break;
    if (i2 == n2) r << s1[i1];
  }
  return r;
}

◆ remove_prefix()

string remove_prefix	(	string	s,
		string	prefix )

Remove the prefix from s if matches.

Parameters

s	the string
prefix	the prefix

Returns: If the prefix matches, return s with prefix removed, otherwise, return s

Definition at line 168 of file analyze.cpp.

                                        {
  if (is_empty (s) || is_empty (prefix)) return s;
  if (starts (s, prefix)) return s (N (prefix), N (s));
  return s;
}

◆ remove_suffix()

string remove_suffix	(	string	s,
		string	suffix )

Remove the suffix from s if matches.

Parameters

s	the string
prefix	the suffix

Returns: If the suffix matches, return s with suffix removed, otherwise, return s

Definition at line 175 of file analyze.cpp.

                                        {
  if (is_empty (s) || is_empty (suffix)) return s;
  if (ends (s, suffix)) return s (0, N (s) - N (suffix));
  return s;
}

◆ ispanish_to_spanish()

string ispanish_to_spanish ( string s )

Convert ispanish string to Spanish string.

Parameters

s	The ispanish string.

Returns: The converted Spanish string.

Definition at line 186 of file analyze.cpp.

                               {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    if ((s[i] == '\'') && ((i + 1) < n)) {
      switch (s[i + 1]) {
      case 'A':
        r << '\301';
        break;
      case 'E':
        r << '\311';
        break;
      case 'I':
        r << '\315';
        break;
      case 'N':
        r << '\321';
        break;
      case 'O':
        r << '\323';
        break;
      case 'U':
        r << '\332';
        break;
      case 'Y':
        r << '\335';
        break;
      case 'a':
        r << '\341';
        break;
      case 'e':
        r << '\351';
        break;
      case 'i':
        r << '\355';
        break;
      case 'n':
        r << '\361';
        break;
      case 'o':
        r << '\363';
        break;
      case 'u':
        r << '\372';
        break;
      case 'y':
        r << '\375';
        break;
      default:
        r << '\'' << s[i + 1];
      }
      i++;
    }
    else r << s[i];
  return r;
}

◆ spanish_to_ispanish()

string spanish_to_ispanish ( string s )

Convert Spanish string to ispanish string.

Parameters

s	The Spanish string.

Returns: The converted ispanish string.

Definition at line 244 of file analyze.cpp.

                               {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    switch (s[i]) {
    case '\301':
      r << "'A";
      break;
    case '\311':
      r << "'E";
      break;
    case '\315':
      r << "'I";
      break;
    case '\321':
      r << "'N";
      break;
    case '\323':
      r << "'O";
      break;
    case '\332':
      r << "'U";
      break;
    case '\335':
      r << "'Y";
      break;
    case '\341':
      r << "'a";
      break;
    case '\351':
      r << "'e";
      break;
    case '\355':
      r << "'i";
      break;
    case '\361':
      r << "'n";
      break;
    case '\363':
      r << "'o";
      break;
    case '\372':
      r << "'u";
      break;
    case '\375':
      r << "'y";
      break;
    default:
      r << s[i];
    }
  return r;
}

◆ igerman_to_german()

string igerman_to_german ( string s )

Convert igerman string to german string.

Parameters

s	The igerman string.

Returns: The converted german string.

Definition at line 298 of file analyze.cpp.

                             {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    if (s[i] == '\337') r << '\377';
    else r << s[i];
  return r;
}

◆ german_to_igerman()

string german_to_igerman ( string s )

Convert german string to igerman string.

Parameters

s	The german string.

Returns: The converted igerman string.

Definition at line 308 of file analyze.cpp.

                             {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    if (s[i] == '\377') r << '\337';
    else r << s[i];
  return r;
}

◆ il2_to_cork() [1/2]

static char il2_to_cork ( char c )

static

Definition at line 341 of file analyze.cpp.

                     {
  int i= (int) ((unsigned char) c);
  if (i < 128) return c;
  return il2_to_cork_string[i - 128];
}

◆ cork_to_il2() [1/2]

static char cork_to_il2 ( char c )

static

Definition at line 348 of file analyze.cpp.

                     {
  int i= (int) ((unsigned char) c);
  if (i < 128) return c;
  return cork_to_il2_string[i - 128];
}

◆ il2_to_cork() [2/2]

string il2_to_cork ( string s )

Definition at line 355 of file analyze.cpp.

                       {
  int    i, n= N (s);
  string r (n);
  for (i= 0; i < n; i++)
    r[i]= il2_to_cork (s[i]);
  return r;
}

◆ cork_to_il2() [2/2]

string cork_to_il2 ( string s )

Definition at line 364 of file analyze.cpp.

                       {
  int    i, n= N (s);
  string r (n);
  for (i= 0; i < n; i++)
    r[i]= cork_to_il2 (s[i]);
  return r;
}

◆ alpha_nr()

string alpha_nr ( int nr )

Generates an alphabetic string for an integer.

Parameters

nr	The integer to be converted to an alphabetic string.

Returns: A string representing the alphabetic character.

Definition at line 377 of file analyze.cpp.

                  {
  if (nr < 0) return "-" * alpha_nr (-nr);
  if (nr == 0) return "0";
  if (nr <= 26) return string ((char) (((int) 'a') + nr - 1));
  return alpha_nr ((nr - 1) / 26) * alpha_nr (((nr - 1) % 26) + 1);
}

◆ Alpha_nr()

string Alpha_nr ( int nr )

Generates an uppercase alphabetic string for an integer.

Parameters

nr	The integer to be converted to an alphabetic string.

Returns: A string representing the uppercase alphabetic character.

Definition at line 385 of file analyze.cpp.

                  {
  return upcase_all (alpha_nr (nr));
}

◆ fnsymbol_nr()

string fnsymbol_nr ( int nr )

Generates footnote symbols for a given integer.

Parameters

nr	The integer to be converted to a footnote symbol.

Returns: A string representing the footnote symbol.

Definition at line 390 of file analyze.cpp.

                     {
  if (nr < 0) nr= -nr;
  string sym, r;
  int    i, m= (nr - 1) % 6, n= ((nr - 1) / 6) + 1;
  switch (m) {
  case 0:
    sym= "<asterisk>";
    break;
  case 1:
    sym= "<dag>";
    break;
  case 2:
    sym= "<ddag>";
    break;
  case 3:
    sym= "<paragraph>";
    break;
  case 4:
    sym= "<endofline>";
    break;
  case 5:
    sym= "||";
    break;
  }
  for (i= 0; i < n; i++)
    r << sym;
  return r;
}

◆ raw_quote()

string raw_quote ( string s )

Add quotes around a string to indicate it's a string, not a symbol.

This function is used for marking the label of a STRING tree as representing a string and not a symbol.

Parameters

s	The input string.

Returns: A new string with quotes around it.

Definition at line 420 of file analyze.cpp.

                     {
  // Mark the label of a STRING tree as representing a string and not a symbol.
  return "\"" * s * "\"";
}

◆ raw_unquote()

string raw_unquote ( string s )

Remove quotes from a string label.

This function is used to get the string value of a STRING tree label representing a string.

Parameters

s	The input string.

Returns: The string without the quotes.

Definition at line 426 of file analyze.cpp.

                       {
  // Get the string value of a STRING tree label representing a string.
  if (is_quoted (s)) return s (1, N (s) - 1);
  else return s;
}

◆ escape_sh()

string escape_sh ( string s )

Escape a string for use in shell scripts.

Parameters

s	The input string.

Returns: An escaped string.

Definition at line 437 of file analyze.cpp.

                     {
#if (defined OS_MINGW || defined OS_WIN)
  return raw_quote (s);
#else
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    switch (s[i]) {
    case '(':
    case ')':
    case '<':
    case '>':
    case '?':
    case '&':
    case '$':
    case '`':
    case '\"':
    case '\\':
    case ' ':
      r << '\\' << s[i];
      break;
    case '\n':
      r << "\\n";
      break;
    default:
      r << s[i];
    }
  return r;
#endif
}

◆ escape_generic()

string escape_generic ( string s )

Escape a string with generic escape sequences.

Parameters

s	The input string.

Returns: An escaped string.

Definition at line 469 of file analyze.cpp.

                          {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++) {
    if ((s[i] == '\2') || (s[i] == '\5') || (s[i] == '\33')) r << '\33';
    r << s[i];
  }
  return r;
}

◆ escape_verbatim()

string escape_verbatim ( string s )

Escape a string to be displayed verbatim.

Parameters

s	The input string.

Returns: An escaped string.

Definition at line 480 of file analyze.cpp.

                           {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++) {
    unsigned char c= (unsigned char) s[i];
    if ((c == '\n') || (c == '\t')) r << ' ';
    else if (((int) c) >= 32) r << s[i];
  }
  return r;
}

◆ escape_spaces()

string escape_spaces ( string s )

Escape spaces in a string with a backslash.

Parameters

s	The input string.

Returns: A string with escaped spaces.

Definition at line 492 of file analyze.cpp.

                         {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++) {
    unsigned char c= (unsigned char) s[i];
    if (c == ' ') r << '\\';
    r << c;
  }
  return r;
}

◆ dos_to_better()

string dos_to_better ( string s )

Convert DOS line endings to more standard line endings.

Parameters

s	The input string.

Returns: A string with Unix-style line endings.

Definition at line 504 of file analyze.cpp.

                         {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++)
    if (s[i] == '\015')
      ;
    else r << s[i];
  return r;
}

◆ unescape_guile()

string unescape_guile ( string s )

Unescape a Guile-syntax string.

Parameters

s	The input string.

Returns: A string with special characters unescaped.

Definition at line 515 of file analyze.cpp.

                          {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n; i++) {
    if (s[i] == '\\') {
      if (i + 1 < n && s[i + 1] == '\\') {
        r << "\\\\\\\\";
        i+= 1;
      }
      else if (i + 3 < n && s[i + 1] == 'x' && is_hex_digit (s[i + 2]) &&
               is_hex_digit (s[i + 3])) {
        string e= s (i + 2, i + 4);
        r << (unsigned char) lolly::data::from_hex (e);
        i+= 3;
      }
      else r << s[i];
    }
    else r << s[i];
  }
  return r;
}

◆ test() [1/2]

bool test	(	string	s,
		int	i,
		const char *	test )

Definition at line 542 of file analyze.cpp.

                                         {
  int n= N (s), j= 0;
  while (test[j] != '\0') {
    if (i >= n) return false;
    if (s[i] != test[j]) return false;
    i++;
    j++;
  }
  return true;
}

◆ test() [2/2]

bool test	(	string	s,
		int	i,
		string	test )

Definition at line 554 of file analyze.cpp.

                                    {
  int n= N (s), m= N (test), j= 0;
  while (j < m) {
    if (i >= n) return false;
    if (s[i] != test[j]) return false;
    i++;
    j++;
  }
  return true;
}

◆ starts() [1/2]

bool starts	(	string	s,
		const char *	what )

Definition at line 566 of file analyze.cpp.

                                    {
  return test (s, 0, what);
}

◆ starts() [2/2]

bool starts	(	string	s,
		const string	what )

Definition at line 571 of file analyze.cpp.

                                     {
  return test (s, 0, what);
}

◆ ends() [1/2]

bool ends	(	string	s,
		const char *	what )

Definition at line 576 of file analyze.cpp.

                                  {
  string r (what);
  if (N (r) > N (s)) return false;
  return s (N (s) - N (r), N (s)) == r;
}

◆ ends() [2/2]

bool ends	(	string	s,
		const string	r )

Definition at line 583 of file analyze.cpp.

                                {
  if (N (r) > N (s)) return false;
  return s (N (s) - N (r), N (s)) == r;
}

◆ read() [1/3]

bool read	(	string	s,
		int &	i,
		const char *	test )

Definition at line 589 of file analyze.cpp.

                                          {
  int n= N (s), j= 0, k= i;
  while (test[j] != '\0') {
    if (k >= n) return false;
    if (s[k] != test[j]) return false;
    j++;
    k++;
  }
  i= k;
  return true;
}

◆ read() [2/3]

bool read	(	string	s,
		string	test )

Definition at line 602 of file analyze.cpp.

                             {
  int i= 0;
  return read (s, i, test);
}

◆ read() [3/3]

bool read	(	string	s,
		int &	i,
		string	test )

Definition at line 608 of file analyze.cpp.

                                     {
  int n= N (s), m= N (test), j= 0, k= i;
  while (j < m) {
    if (k >= n) return false;
    if (s[k] != test[j]) return false;
    j++;
    k++;
  }
  i= k;
  return true;
}

◆ read_line()

bool read_line	(	string	s,
		int &	i,
		string &	result )

Definition at line 621 of file analyze.cpp.

                                             {
  int start= i;
  for (; i < N (s); i++) {
    if (s[i] == '\n') {
      result= s (start, i++);
      return true;
    }
  }
  result= s (start, i);
  return false;
}

◆ read_int()

bool read_int	(	string	s,
		int &	i,
		int &	result )

Definition at line 634 of file analyze.cpp.

                                         {
  int n= N (s), start= i;
  result= 0;
  if (i == n) return false;
  if (s[i] == '-') {
    if (i + 1 == n) return false;
    if (!is_digit (s[i + 1])) return false;
    i++;
  }
  else if (!is_digit (s[i])) return false;
  while ((i < n) && is_digit (s[i]))
    i++;
  result= as_int (s (start, i));
  return true;
}

◆ read_double()

bool read_double	(	string	s,
		int &	i,
		double &	result )

Definition at line 651 of file analyze.cpp.

                                               {
  int n= N (s), start= i;
  result= 0.0;
  if (i == n) return false;
  if (s[i] == '-') {
    if (i + 1 == n) return false;
    if (!is_numeric (s[i + 1])) return false;
    i++;
  }
  else if (!is_numeric (s[i])) return false;
  while ((i < n) && is_digit (s[i]))
    i++;
  if ((i < n) && (s[i] == '.')) i++;
  while ((i < n) && is_digit (s[i]))
    i++;
  if ((i < n) && ((s[i] == 'e') || (s[i] == 'E'))) {
    i++;
    if ((i < n) && (s[i] == '-')) i++;
    if ((i == n) || (!is_digit (s[i]))) {
      i= start;
      return false;
    }
    while ((i < n) && is_digit (s[i]))
      i++;
  }
  result= as_double (s (start, i));
  return true;
}

◆ read_word()

bool read_word	(	string	s,
		int &	i,
		string &	result )

Definition at line 681 of file analyze.cpp.

                                             {
  int opos= i;
  while (i < N (s) && is_alpha (s[i])) {
    i++;
  }
  result= s (opos, i);
  return i > opos;
}

◆ is_whitespace()

bool is_whitespace ( string s )

Definition at line 691 of file analyze.cpp.

                         {
  for (int i= 0; i < N (s); i++)
    if (s[i] != ' ' && s[i] != '\t' && s[i] != '\n') return false;
  return true;
}

◆ skip_spaces()

void skip_spaces	(	string	s,
		int &	i )

Definition at line 698 of file analyze.cpp.

                               {
  int n= N (s);
  while ((i < n) && ((s[i] == ' ') || (s[i] == '\t')))
    i++;
}

◆ skip_whitespace()

void skip_whitespace	(	string	s,
		int &	i )

Definition at line 705 of file analyze.cpp.

                                   {
  int n= N (s);
  while ((i < n) && ((s[i] == ' ') || (s[i] == '\t') || (s[i] == '\n')))
    i++;
}

◆ skip_line()

void skip_line	(	string	s,
		int &	i )

Definition at line 712 of file analyze.cpp.

                             {
  int n= N (s);
  while ((i < n) && (s[i] != '\n'))
    i++;
  if (i < n) i++;
}

◆ skip_symbol()

void skip_symbol	(	string	s,
		int &	i )

Definition at line 720 of file analyze.cpp.

                               {
  int n= N (s);
  if (i < n) {
    if (s[i] == '<') {
      for (i++; i < n; i++)
        if (s[i - 1] == '>') break;
    }
    else i++;
  }
}

◆ convert_tabs_to_spaces()

string convert_tabs_to_spaces	(	string	s,
		int	w )

Converts tabs in a string to spaces.

Parameters

s	The original string with tabs.
tw	The tab width.

Returns: A new string with tabs replaced by spaces.

Definition at line 732 of file analyze.cpp.

                                          {
  int    i= 0, ts= 0, n= N (s);
  string r= "";
  while (i < n) {
    if (s[i] == '\t') {
      r << string (' ', tw - ((i - ts) % tw));
      ts= i + 1;
    }
    else if (s[i] == '\n') {
      ts= i + 1;
      r << s[i];
    }
    else r << s[i];
    i++;
  }
  return r;
}

◆ parse() [1/6]

void parse	(	string	s,
		int &	pos,
		QI &	ret )

Definition at line 755 of file analyze.cpp.

                                    {
  ret= (QI) s[pos++];
}

◆ parse() [2/6]

void parse	(	string	s,
		int &	pos,
		QN &	ret )

Definition at line 760 of file analyze.cpp.

                                    {
  ret= (QN) s[pos++];
}

◆ parse() [3/6]

void parse	(	string	s,
		int &	pos,
		HI &	ret )

Definition at line 765 of file analyze.cpp.

                                    {
  QI c1= (QI) s[pos++];
  QN c2= (QN) s[pos++];
  ret  = (((HI) c1) << 8) + c2;
}

◆ parse() [4/6]

void parse	(	string	s,
		int &	pos,
		HN &	ret )

Definition at line 772 of file analyze.cpp.

                                    {
  QN c1= (QN) s[pos++];
  QN c2= (QN) s[pos++];
  ret  = (((HN) c1) << 8) + c2;
}

◆ parse() [5/6]

void parse	(	string	s,
		int &	pos,
		SI &	ret )

Definition at line 779 of file analyze.cpp.

                                    {
  QI c1= (QI) s[pos++];
  QN c2= (QN) s[pos++];
  QN c3= (QN) s[pos++];
  QN c4= (QN) s[pos++];
  ret  = (((((((SI) c1) << 8) + ((SI) c2)) << 8) + ((SI) c3)) << 8) + c4;
}

◆ parse() [6/6]

void parse	(	string	s,
		int &	pos,
		SI *&	a,
		int	len )

Definition at line 788 of file analyze.cpp.

                                            {
  int i;
  a= tm_new_array<int> (len);
  for (i= 0; i < len; i++)
    parse (s, pos, a[i]);
}

◆ index_of()

int index_of	(	string	s,
		char	c )

Definition at line 800 of file analyze.cpp.

                            {
  int s_N= N (s);
  for (int i= 0; i < s_N; i++) {
    if (s[i] == c) {
      return i;
    }
  }
  return -1;
}

◆ search_forwards() [1/3]

int search_forwards	(	array< string >	what_list,
		int	pos,
		string	in )

Searches for a substring in a string starting from a specified position, in reverse.

Parameters

what_list	The substring to search for.
pos	The starting position in the string to search from.
in	The string to search in.

Returns: Position where the substring was found, or -1 if not found.

Definition at line 811 of file analyze.cpp.

                                                      {
  int n= N (in), na= N (a);
  while (pos <= n) {
    for (int i= 0; i < na; i++)
      if (N (a[i]) > 0 && in[pos] == a[i][0] && test (in, pos, a[i]))
        return pos;
    pos++;
  }
  return -1;
}

◆ search_forwards() [2/3]

int search_forwards	(	string	what,
		int	pos,
		string	in )

Searches for a substring in a string starting from a specified position.

Parameters

what	The substring to search for.
pos	The starting position in the string to search from.
in	The string to search in.

Returns: Position where the substring was found, or -1 if not found.

Definition at line 823 of file analyze.cpp.

                                               {
  int k= N (s), n= N (in);
  if (k == 0) return pos;
  char c= s[0];
  while (pos + k <= n) {
    if (in[pos] == c && test (in, pos, s)) return pos;
    pos++;
  }
  return -1;
}

◆ search_forwards() [3/3]

int search_forwards	(	string	what,
		string	in )

Searches for a substring in a string.

Parameters

what	The substring to search for.
in	The string to search in.

Returns: Position where the substring was found, or -1 if not found.

Definition at line 835 of file analyze.cpp.

                                      {
  return search_forwards (s, 0, in);
}

◆ occurs()

bool occurs	(	string	what,
		string	in )

Checks whether a substring occurs within another string.

Parameters

what	The string to find.
in	The string in which to search.

Returns: True if the string occurs, otherwise false.

Definition at line 840 of file analyze.cpp.

                                {
  return search_forwards (what, 0, in) >= 0;
}

◆ contains() [1/2]

bool contains	(	string	s,
		string	what )

Definition at line 845 of file analyze.cpp.

                                 {
  return search_forwards (what, 0, s) >= 0;
}

◆ contains() [2/2]

bool contains	(	string	s,
		char	c )

Definition at line 850 of file analyze.cpp.

                            {
  int s_N= N (s);
  for (int i= 0; i < s_N; i++) {
    if (s[i] == c) {
      return true;
    }
  }
  return false;
}

◆ search_backwards() [1/2]

int search_backwards	(	string	what,
		int	pos,
		string	in )

Searches for a substring in a string starting from a specified position, in reverse.

Parameters

s	The substring to search for.
pos	The starting position in the string to search from.
in	The string to search in.

Returns: Position where the substring was found, or -1 if not found.

Definition at line 861 of file analyze.cpp.

                                                {
  while (pos >= 0) {
    if (test (in, pos, s)) return pos;
    pos--;
  }
  return -1;
}

◆ search_backwards() [2/2]

int search_backwards	(	string	what,
		string	in )

Searches for a substring in a string, in reverse.

Parameters

s	The substring to search for.
in	The string to search in.

Returns: Position where the substring was found, or -1 if not found.

Definition at line 870 of file analyze.cpp.

                                       {
  return search_backwards (s, N (in) - N (s), in);
}

◆ count_occurrences()

int count_occurrences	(	string	what,
		string	in )

Counts the occurrences of a substring in a string.

Parameters

s	The substring to count.
in	The string to search in.

Returns: Number of occurrences of the substring.

Definition at line 875 of file analyze.cpp.

                                        {
  int count= 0;
  int i= 0, next, n= N (in);
  while (i < n) {
    next= search_forwards (s, i, in);
    if (next == -1) break;
    count++;
    i= next + 1;
  }
  return count;
}

◆ overlapping()

int overlapping	(	string	s1,
		string	s2 )

Finds the length of the longest string that is both a suffix of the first string and a prefix of the second string.

Parameters

s1	The first string.
s2	The second string.

Returns: The length of the longest overlapping string.

Definition at line 888 of file analyze.cpp.

                                   {
  // return the longuest string being suffix of s1 and prefix of s2
  int i= min (N (s1), N (s2)), n= N (s1);
  while (i > 0) {
    if (s1 (n - i, n) == s2 (0, i)) return i;
    i--;
  }
  return 0;
}

◆ replace()

string replace	(	string	s,
		string	what,
		string	by )

Replaces all occurrences of a specific substring within a string.

Parameters

s	The original string.
what	The substring to replace.
by	The string with which to replace the substring.

Returns: The new string with all occurrences of the substring replaced.

Definition at line 899 of file analyze.cpp.

                                           {
  int    i, n= N (s);
  string r;
  for (i= 0; i < n;)
    if (test (s, i, what)) {
      r << by;
      i+= N (what);
    }
    else {
      r << s[i];
      i++;
    }
  return r;
}

◆ match_wildcard() [1/2]

static bool match_wildcard	(	string	s,
		int	spos,
		string	w,
		int	wpos )

static

Definition at line 915 of file analyze.cpp.

                                                        {
  if (wpos == N (w)) return spos == N (s);
  if (w[wpos] != '*')
    return (spos < N (s)) && (s[spos] == w[wpos]) &&
           match_wildcard (s, spos + 1, w, wpos + 1);
  while ((wpos < N (w)) && (w[wpos] == '*'))
    wpos++;
  while (spos <= N (s)) {
    if (match_wildcard (s, spos, w, wpos)) return true;
    spos++;
  }
  return false;
}

◆ match_wildcard() [2/2]

bool match_wildcard	(	string	s,
		string	w )

Matches a string against a wildcard pattern.

Parameters

s	The string to match.
w	The wildcard pattern.

Returns: True if the string matches the wildcard pattern, otherwise false.

Definition at line 930 of file analyze.cpp.

                                    {
  return match_wildcard (s, 0, w, 0);
}

◆ find_non_alpha()

int find_non_alpha	(	string	s,
		int	pos,
		bool	forward )

Finds the position of the first non-alphabetic character in a string.

Parameters

s	The string to search.
pos	The position from which to start the search.
forward	True to search forward, false to search backward.

Returns: The position of the first non-alphabetic character; -1 if not found.

Definition at line 935 of file analyze.cpp.

                                                 {
  if (forward) {
    for (; pos < N (s); pos++)
      if (!is_alpha (s[pos])) return pos;
  }
  else {
    for (; pos > 0; pos--)
      if (!is_alpha (s[pos - 1])) return pos - 1;
  }
  return -1;
}

◆ tokenize()

array< string > tokenize	(	string	s,
		string	sep )

Splits a string into an array of strings based on a separator string.

Parameters

s	The string to split.
sep	The separator string.

Returns: An array of strings split based on the separator.

Definition at line 948 of file analyze.cpp.

                                {
  int           start= 0;
  array<string> a;
  for (int i= 0; i < N (s);)
    if (test (s, i, sep)) {
      a << s (start, i);
      i+= N (sep);
      start= i;
    }
    else i++;
  a << s (start, N (s));
  return a;
}

◆ recompose()

string recompose	(	array< string >	a,
		string	sep )

Joins an array of strings into a single string, separated by a specific string.

Parameters

a	The array of strings to join.
sep	The separator string.

Returns: A single string consisting of the array elements separated by the separator.

Definition at line 963 of file analyze.cpp.

                                        {
  string r;
  for (int i= 0; i < N (a); i++) {
    if (i != 0) r << sep;
    r << a[i];
  }
  return r;
}

◆ trim_spaces_left()

string trim_spaces_left ( string s )

Removes all leading spaces from a string.

Parameters

s	The string to trim.

Returns: The string without leading spaces.

Definition at line 973 of file analyze.cpp.

                            {
  int start;
  for (start= 0; start < N (s) && is_space (s[start]); start++)
    ;
  return s (start, N (s));
}

◆ trim_spaces_right()

string trim_spaces_right ( string s )

Removes all trailing spaces from a string.

Parameters

s	The string to trim.

Returns: The string without trailing spaces.

Definition at line 981 of file analyze.cpp.

                             {
  int end;
  for (end= N (s) - 1; end >= 0 && is_space (s[end]); end--)
    ;
  return s (0, end + 1);
}

◆ trim_spaces() [1/2]

string trim_spaces ( string s )

Removes all leading and trailing spaces from a string.

Parameters

s	The string to trim.

Returns: The string without leading and trailing spaces.

Definition at line 989 of file analyze.cpp.

                       {
  return trim_spaces_left (trim_spaces_right (s));
}

◆ trim_spaces() [2/2]

array< string > trim_spaces ( array< string > a )

Removes all leading and trailing spaces from an array of strings.

Parameters

a	The array of strings to trim.

Returns: An array of strings with all leading and trailing spaces removed.

Definition at line 994 of file analyze.cpp.

                              {
  array<string> b (N (a));
  for (int i= 0; i < N (a); i++)
    b[i]= trim_spaces (a[i]);
  return b;
}

◆ find_longest()

static int find_longest	(	string	s1,
		string	s2,
		int &	c1,
		int &	c2 )

static

Definition at line 1006 of file analyze.cpp.

                                                      {
  int n1= N (s1), n2= N (s2), bc= 0, bl= 0, br= 0;
  for (c2= 0; c2 < n2; c2++)
    if (s1[c1] == s2[c2]) {
      int l= 0, r= 0;
      while (c1 + r < n1 && c2 + r < n2 && s1[c1 + r] == s2[c2 + r])
        r++;
      while (l < c1 && l < c2 && s1[c1 - l - 1] == s2[c2 - l - 1])
        l++;
      if (l + r > bl + br) {
        bc= c2;
        bl= l;
        br= r;
      }
    }
  if (bl + br > 0) {
    c1= c1 - bl;
    c2= bc - bl;
  }
  return bl + br;
}

◆ find_common()

static void find_common	(	string	s1,
		string	s2,
		int &	c1,
		int &	c2 )

static

Definition at line 1029 of file analyze.cpp.

                                                     {
  int best_len= 0;
  c1= c2= 0;
  int n1= N (s1), n2= N (s2);
  if (n1 == 0 || n2 == 0) return;
  int t= min (min (n1, n2), 6);
  for (int k= 1; k < t; k++) {
    int a1= (k * n1) / t, a2= (k * n2) / t;
    int len= find_longest (s1, s2, a1, a2);
    if (len > best_len) {
      best_len= len;
      c1      = a1;
      c2      = a2;
    }
  }
}

◆ differences()

array< int > differences	(	string	s1,
		string	s2 )

the differences between two strings by identifying the common substrings and returning the different sections' indices.

Parameters

s1	The first string.
s2	The second string.

Returns: An array containing sets of four integers:

Definition at line 1047 of file analyze.cpp.

                                   {
  int n1= N (s1), n2= N (s2);
  int i1= 0, i2= 0, j1= n1, j2= n2;
  while (i1 < j1 && i2 < j2 && s1[i1] == s2[i2]) {
    i1++;
    i2++;
  }
  while (i1 < j1 && i2 < j2 && s1[j1 - 1] == s2[j2 - 1]) {
    j1--;
    j2--;
  }
  if (i1 == i2 && j1 == j2) return array<int> ();
  if (i1 > 0 || i2 > 0 || j1 < n1 || j2 < n2) {
    array<int> r= differences (s1 (i1, j1), s2 (i2, j2));
    for (int k= 0; k < N (r); k+= 4) {
      r[k]+= i1;
      r[k + 1]+= i1;
      r[k + 2]+= i2;
      r[k + 3]+= i2;
    }
    return r;
  }
  else {
    int c1, c2;
    find_common (s1, s2, c1, c2);
    if (c1 == 0 && c2 == 0) {
      array<int> r;
      r << i1 << j1 << i2 << j2;
      return r;
    }
    else {
      array<int> r1= differences (s1 (0, c1), s2 (0, c2));
      array<int> r2= differences (s1 (c1, n1), s2 (c2, n2));
      for (int k= 0; k < N (r2); k+= 4) {
        r2[k]+= c1;
        r2[k + 1]+= c1;
        r2[k + 2]+= c2;
        r2[k + 3]+= c2;
      }
      r1 << r2;
      return r1;
    }
  }
}

◆ distance()

int distance	(	string	s1,
		string	s2 )

a measure of difference (distance) between two strings.

Parameters

s1	The first string.
s2	The second string.

Returns: An integer representing the sum of lengths of all differing sections between the two strings.

Definition at line 1093 of file analyze.cpp.

                                {
  int        d= 0;
  array<int> r= differences (s1, s2);
  for (int k= 0; k < N (r); k+= 4)
    d+= max (r[k + 1] - r[k], r[k + 3] - r[k + 2]);
  return d;
}

◆ parse_length()

void parse_length	(	string	s,
		double &	len,
		string &	unit )

Parses a string containing a length value and its unit.

Parameters

[in]	s	The string to parse, e.g., "12.34cm".
[out]	len	The parsed numerical value of the length.
[out]	unit	The unit of the length as a string, e.g., "cm".

Definition at line 1106 of file analyze.cpp.

                                                   {
  int start= 0;
  int i, n= N (s);
  for (i= start; i < n && !is_locase (s[i]); i++) {
  }
  string s1= s (start, i);
  string s2= s (i, n);
  if (is_double (s1) && (is_locase_alpha (s2) || is_empty (s2))) {
    len = as_double (s1);
    unit= s2;
  }
  else {
    len = 0.0;
    unit= "error";
  }
}

Variable Documentation

◆ il2_to_cork_string

string il2_to_cork_string

static

Initial value:

=
    "\200\201\202\203\204\205\206\207\210\211\212\213\214\215\216\217\220\221"
    "\222\223\224\225\226\227\230\231\232\233\234\235\236\237 \20\212 "
    "\211\221\237¨\222\223\224\231‐\232\233 "
    "\241˛\252´\251\261ˇ¸\262\263\264\271˝\272\273\217\301\302\200\304\210\202"
    "\307\203\311\206\313\205\315\316\204\320\213\214\323\324\216\326."
    "\220\227\332\226\334\335\225\377\257\341\342\240\344\250\242\347\243\351"
    "\246\353\245\355\356\244\236\253\254\363\364\256\366/"
    "\260\267\372\266\374\375\265 "

Definition at line 321 of file analyze.cpp.

◆ cork_to_il2_string

string cork_to_il2_string

static

Initial value:

=
    "\303\241\306\310\317\314\312G\305\245\243\321\322 "
    "\325\300\330\246\251\252\253\336\333\331Y\254\256\257II\360\247\343\261"
    "\346\350\357\354\352g\345\265\263\361\362 "
    "\365\340\370\266\271\272\273\376\373\371y\274\276\277i!?"
    "LA\301\302A\304AA\307E\311E\313I\315\316I\320NO\323\324O\326OOU\332U\334"
    "\335 "
    "Sa\341\342a\344aa\347e\351e\353i\355\356i\360no\363\364o\366oou\372u\374"
    "\375 \337"

Definition at line 330 of file analyze.cpp.

Functions

Variables

Function Documentation

◆ is_iso_alpha() [1/2]

◆ is_iso_locase()

◆ is_iso_upcase()

◆ is_alpha()

◆ is_alphanum()

◆ is_locase_alpha()

◆ is_iso_alpha() [2/2]

◆ is_numeric()

◆ upcase()

◆ locase()

◆ closing_delimiter()

◆ upcase_first()

◆ locase_first()

◆ upcase_all()

◆ locase_all()

◆ string_union()

◆ string_minus()

◆ remove_prefix()

◆ remove_suffix()

◆ ispanish_to_spanish()

◆ spanish_to_ispanish()

◆ igerman_to_german()

◆ german_to_igerman()

◆ il2_to_cork() [1/2]

◆ cork_to_il2() [1/2]

◆ il2_to_cork() [2/2]

◆ cork_to_il2() [2/2]

◆ alpha_nr()

◆ Alpha_nr()

◆ fnsymbol_nr()

◆ raw_quote()

◆ raw_unquote()

◆ escape_sh()

◆ escape_generic()

◆ escape_verbatim()

◆ escape_spaces()

◆ dos_to_better()

◆ unescape_guile()

◆ test() [1/2]

◆ test() [2/2]

◆ starts() [1/2]

◆ starts() [2/2]

◆ ends() [1/2]

◆ ends() [2/2]

◆ read() [1/3]

◆ read() [2/3]

◆ read() [3/3]

◆ read_line()

◆ read_int()

◆ read_double()

◆ read_word()

◆ is_whitespace()

◆ skip_spaces()

◆ skip_whitespace()

◆ skip_line()

◆ skip_symbol()

◆ convert_tabs_to_spaces()

◆ parse() [1/6]

◆ parse() [2/6]

◆ parse() [3/6]

◆ parse() [4/6]

◆ parse() [5/6]

◆ parse() [6/6]

◆ index_of()

◆ search_forwards() [1/3]

◆ search_forwards() [2/3]

◆ search_forwards() [3/3]

◆ occurs()

◆ contains() [1/2]

◆ contains() [2/2]

◆ search_backwards() [1/2]

◆ search_backwards() [2/2]

◆ count_occurrences()

◆ overlapping()

◆ replace()

◆ match_wildcard() [1/2]

◆ match_wildcard() [2/2]