apt-pkg/contrib/strutl.h

   1 // -*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: strutl.h,v 1.22 2003/02/02 22:20:27 jgg Exp $
   4 /* ######################################################################
   5
   6    String Util - These are some useful string functions
   7
   8    _strstrip is a function to remove whitespace from the front and end
   9    of a string.
  10
  11    This source is placed in the Public Domain, do with it what you will
  12    It was originally written by Jason Gunthorpe <jgg@gpu.srv.ualberta.ca>
  13
  14    ##################################################################### */
  15                                                                         /*}}}*/
  16 #ifndef STRUTL_H
  17 #define STRUTL_H
  18
  19
  20 #include <limits>
  21 #include <string>
  22 #include <cstring>
  23 #include <vector>
  24 #include <iostream>
  25 #ifdef APT_PKG_EXPOSE_STRING_VIEW
  26 #include <apt-pkg/string_view.h>
  27 #endif
  28 #include <time.h>
  29 #include <stddef.h>
  30
  31 #include "macros.h"
  32
  33 #ifndef APT_10_CLEANER_HEADERS
  34 #include <stdlib.h>
  35 #endif
  36 #ifndef APT_8_CLEANER_HEADERS
  37 using std::string;
  38 using std::vector;
  39 using std::ostream;
  40 #endif
  41
  42 namespace APT {
  43    namespace String {
  44       std::string Strip(const std::string &s);
  45       bool Endswith(const std::string &s, const std::string &ending);
  46       bool Startswith(const std::string &s, const std::string &starting);
  47    }
  48 }
  49
  50
  51 bool UTF8ToCodeset(const char *codeset, const std::string &orig, std::string *dest);
  52 char *_strstrip(char *String);
  53 char *_strrstrip(char *String); // right strip only
  54 char *_strtabexpand(char *String,size_t Len);
  55 bool ParseQuoteWord(const char *&String,std::string &Res);
  56 bool ParseCWord(const char *&String,std::string &Res);
  57 std::string QuoteString(const std::string &Str,const char *Bad);
  58 std::string DeQuoteString(const std::string &Str);
  59 std::string DeQuoteString(std::string::const_iterator const &begin, std::string::const_iterator const &end);
  60
  61 // unescape (\0XX and \xXX) from a string
  62 std::string DeEscapeString(const std::string &input);
  63
  64 std::string SizeToStr(double Bytes);
  65 std::string TimeToStr(unsigned long Sec);
  66 std::string Base64Encode(const std::string &Str);
  67 std::string OutputInDepth(const unsigned long Depth, const char* Separator="  ");
  68 std::string URItoFileName(const std::string &URI);
  69 APT_DEPRECATED_MSG("Specify if GMT is required or a numeric timezone can be used") std::string TimeRFC1123(time_t Date);
  70 /** returns a datetime string as needed by HTTP/1.1 and Debian files.
  71  *
  72  * Note: The date will always be represented in a UTC timezone
  73  *
  74  * @param Date to be represented as a string
  75  * @param NumericTimezone is preferred in general, but HTTP/1.1 requires the use
  76  *    of GMT as timezone instead. \b true means that the timezone should be denoted
  77  *    as "+0000" while \b false uses "GMT".
  78  */
  79 std::string TimeRFC1123(time_t Date, bool const NumericTimezone);
  80 /** parses time as needed by HTTP/1.1 and Debian files.
  81  *
  82  * HTTP/1.1 prefers dates in RFC1123 format (but the other two obsolete date formats
  83  * are supported to) and e.g. Release files use the same format in Date & Valid-Until
  84  * fields.
  85  *
  86  * Note: datetime strings need to be in UTC timezones (GMT, UTC, Z, +/-0000) to be
  87  * parsed. Other timezones will be rejected as invalid. Previous implementations
  88  * accepted other timezones, but treated them as UTC.
  89  *
  90  * @param str is the datetime string to parse
  91  * @param[out] time will be the seconds since epoch of the given datetime if
  92  *    parsing is successful, undefined otherwise.
  93  * @return \b true if parsing was successful, otherwise \b false.
  94  */
  95 bool RFC1123StrToTime(const char* const str,time_t &time) APT_MUSTCHECK;
  96 bool FTPMDTMStrToTime(const char* const str,time_t &time) APT_MUSTCHECK;
  97 APT_DEPRECATED_MSG("Use RFC1123StrToTime or FTPMDTMStrToTime as needed instead") bool StrToTime(const std::string &Val,time_t &Result);
  98 std::string LookupTag(const std::string &Message,const char *Tag,const char *Default = 0);
  99 int StringToBool(const std::string &Text,int Default = -1);
 100 bool ReadMessages(int Fd, std::vector<std::string> &List);
 101 bool StrToNum(const char *Str,unsigned long &Res,unsigned Len,unsigned Base = 0);
 102 bool StrToNum(const char *Str,unsigned long long &Res,unsigned Len,unsigned Base = 0);
 103 bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len);
 104 bool Base256ToNum(const char *Str,unsigned long long &Res,unsigned int Len);
 105 bool Hex2Num(const std::string &Str,unsigned char *Num,unsigned int Length);
 106 #ifdef APT_PKG_EXPOSE_STRING_VIEW
 107 APT_HIDDEN bool Hex2Num(const APT::StringView Str,unsigned char *Num,unsigned int Length);
 108 #endif
 109 // input changing string split
 110 bool TokSplitString(char Tok,char *Input,char **List,
 111                     unsigned long ListMax);
 112
 113 // split a given string by a char
 114 std::vector<std::string> VectorizeString(std::string const &haystack, char const &split) APT_PURE;
 115
 116 /* \brief Return a vector of strings from string "input" where "sep"
 117  * is used as the delimiter string.
 118  *
 119  * \param input The input string.
 120  *
 121  * \param sep The separator to use.
 122  *
 123  * \param maxsplit (optional) The maximum amount of splitting that
 124  * should be done .
 125  *
 126  * The optional "maxsplit" argument can be used to limit the splitting,
 127  * if used the string is only split on maxsplit places and the last
 128  * item in the vector contains the remainder string.
 129  */
 130 std::vector<std::string> StringSplit(std::string const &input,
 131                                      std::string const &sep,
 132                                      unsigned int maxsplit=std::numeric_limits<unsigned int>::max()) APT_CONST;
 133
 134 void ioprintf(std::ostream &out,const char *format,...) APT_PRINTF(2);
 135 void strprintf(std::string &out,const char *format,...) APT_PRINTF(2);
 136 char *safe_snprintf(char *Buffer,char *End,const char *Format,...) APT_PRINTF(3);
 137 bool CheckDomainList(const std::string &Host, const std::string &List);
 138
 139 /* Do some compat mumbo jumbo */
 140 #define tolower_ascii  tolower_ascii_inline
 141 #define isspace_ascii  isspace_ascii_inline
 142
 143 APT_CONST APT_HOT
 144 static inline int tolower_ascii_unsafe(int const c)
 145 {
 146    return c | 0x20;
 147 }
 148 APT_CONST APT_HOT
 149 static inline int tolower_ascii_inline(int const c)
 150 {
 151    return (c >= 'A' && c <= 'Z') ? c + 32 : c;
 152 }
 153 APT_CONST APT_HOT
 154 static inline int isspace_ascii_inline(int const c)
 155 {
 156    // 9='\t',10='\n',11='\v',12='\f',13='\r',32=' '
 157    return (c >= 9 && c <= 13) || c == ' ';
 158 }
 159
 160 // StringViewCompareFast - awkward attempt to optimize cache generation /*{{{*/
 161 #ifdef APT_PKG_EXPOSE_STRING_VIEW
 162 /**
 163  * \brief Faster comparison for string views (compare size before data)
 164  *
 165  * Still stable, but faster than the normal ordering.
 166  *  As this is used for package comparison this *MUST* be case insensitive,
 167  * as the alternative is to lower case all dependency fields which is slow. */
 168 static inline int StringViewCompareFast(APT::StringView a, APT::StringView b) {
 169     if (a.size() != b.size())
 170         return a.size() - b.size();
 171     auto l(a.data()), r(b.data());
 172     for (auto e(a.size()), i(decltype(e)(0)); i != e; ++i)
 173         if (tolower_ascii_inline(l[i]) != tolower_ascii_inline(r[i]))
 174             return tolower_ascii(l[i]) < tolower_ascii(r[i]) ? -1 : 1;
 175     return 0;
 176 }
 177 #endif
 178                                                                         /*}}}*/
 179
 180
 181 std::string StripEpoch(const std::string &VerStr);
 182
 183 #define APT_MKSTRCMP(name,func) \
 184 inline APT_PURE int name(const char *A,const char *B) {return func(A,A+strlen(A),B,B+strlen(B));} \
 185 inline APT_PURE int name(const char *A,const char *AEnd,const char *B) {return func(A,AEnd,B,B+strlen(B));} \
 186 inline APT_PURE int name(const std::string& A,const char *B) {return func(A.c_str(),A.c_str()+A.length(),B,B+strlen(B));} \
 187 inline APT_PURE int name(const std::string& A,const std::string& B) {return func(A.c_str(),A.c_str()+A.length(),B.c_str(),B.c_str()+B.length());} \
 188 inline APT_PURE int name(const std::string& A,const char *B,const char *BEnd) {return func(A.c_str(),A.c_str()+A.length(),B,BEnd);}
 189
 190 #define APT_MKSTRCMP2(name,func) \
 191 inline APT_PURE int name(const char *A,const char *AEnd,const char *B) {return func(A,AEnd,B,B+strlen(B));} \
 192 inline APT_PURE int name(const std::string& A,const char *B) {return func(A.begin(),A.end(),B,B+strlen(B));} \
 193 inline APT_PURE int name(const std::string& A,const std::string& B) {return func(A.begin(),A.end(),B.begin(),B.end());} \
 194 inline APT_PURE int name(const std::string& A,const char *B,const char *BEnd) {return func(A.begin(),A.end(),B,BEnd);}
 195
 196 int APT_PURE stringcmp(const char *A,const char *AEnd,const char *B,const char *BEnd);
 197 int APT_PURE stringcasecmp(const char *A,const char *AEnd,const char *B,const char *BEnd);
 198
 199 /* We assume that GCC 3 indicates that libstdc++3 is in use too. In that
 200    case the definition of string::const_iterator is not the same as
 201    const char * and we need these extra functions */
 202 #if __GNUC__ >= 3
 203 int APT_PURE stringcmp(std::string::const_iterator A,std::string::const_iterator AEnd,
 204               const char *B,const char *BEnd);
 205 int APT_PURE stringcmp(std::string::const_iterator A,std::string::const_iterator AEnd,
 206               std::string::const_iterator B,std::string::const_iterator BEnd);
 207 int APT_PURE stringcasecmp(std::string::const_iterator A,std::string::const_iterator AEnd,
 208                   const char *B,const char *BEnd);
 209 int APT_PURE stringcasecmp(std::string::const_iterator A,std::string::const_iterator AEnd,
 210                   std::string::const_iterator B,std::string::const_iterator BEnd);
 211
 212 inline APT_PURE int stringcmp(std::string::const_iterator A,std::string::const_iterator Aend,const char *B) {return stringcmp(A,Aend,B,B+strlen(B));}
 213 inline APT_PURE int stringcasecmp(std::string::const_iterator A,std::string::const_iterator Aend,const char *B) {return stringcasecmp(A,Aend,B,B+strlen(B));}
 214 #endif
 215
 216 APT_MKSTRCMP2(stringcmp,stringcmp)
 217 APT_MKSTRCMP2(stringcasecmp,stringcasecmp)
 218
 219 // Return the length of a NULL-terminated string array
 220 size_t APT_PURE strv_length(const char **str_array);
 221
 222
 223 inline const char *DeNull(const char *s) {return (s == 0?"(null)":s);}
 224
 225 class URI
 226 {
 227    void CopyFrom(const std::string &From);
 228
 229    public:
 230
 231    std::string Access;
 232    std::string User;
 233    std::string Password;
 234    std::string Host;
 235    std::string Path;
 236    unsigned int Port;
 237
 238    operator std::string();
 239    inline void operator =(const std::string &From) {CopyFrom(From);}
 240    inline bool empty() {return Access.empty();};
 241    static std::string SiteOnly(const std::string &URI);
 242    static std::string ArchiveOnly(const std::string &URI);
 243    static std::string NoUserPassword(const std::string &URI);
 244
 245    URI(std::string Path) {CopyFrom(Path);}
 246    URI() : Port(0) {}
 247 };
 248
 249 struct SubstVar
 250 {
 251    const char *Subst;
 252    const std::string *Contents;
 253 };
 254 std::string SubstVar(std::string Str,const struct SubstVar *Vars);
 255 std::string SubstVar(const std::string &Str,const std::string &Subst,const std::string &Contents);
 256
 257 struct RxChoiceList
 258 {
 259    void *UserData;
 260    const char *Str;
 261    bool Hit;
 262 };
 263 unsigned long RegexChoice(RxChoiceList *Rxs,const char **ListBegin,
 264                       const char **ListEnd);
 265
 266 #endif