Store tags in the cache (they are very useful :/).

[apt.git] / apt-pkg / contrib / strutl.cc

diff --git a/apt-pkg/contrib/strutl.cc b/apt-pkg/contrib/strutl.cc
index 0db4c57b249f9e20fea0d0f3a7b40b4d811736f8..cf8feb970a102883908d3f83cc9aaf0983d9bc97 100644 (file)
--- a/apt-pkg/contrib/strutl.cc
+++ b/apt-pkg/contrib/strutl.cc
@@ -21,16 +21,20 @@
 #include <apt-pkg/fileutl.h>
 #include <apt-pkg/error.h>
 
 #include <apt-pkg/fileutl.h>
 #include <apt-pkg/error.h>
 

+#include <array>
+#include <algorithm>
+#include <iomanip>
+#include <locale>
+#include <sstream>
+#include <string>
+#include <vector>
+

 #include <stddef.h>
 #include <stdlib.h>
 #include <time.h>
 #include <stddef.h>
 #include <stdlib.h>
 #include <time.h>

-#include <string>
-#include <vector>

 #include <ctype.h>
 #include <string.h>
 #include <ctype.h>
 #include <string.h>

-#include <sstream>

 #include <stdio.h>
 #include <stdio.h>

-#include <algorithm>

 #include <unistd.h>
 #include <regex.h>
 #include <errno.h>
 #include <unistd.h>
 #include <regex.h>
 #include <errno.h>


@@ -71,14 +75,14 @@ bool Endswith(const std::string &s, const std::string &end)
 {
    if (end.size() > s.size())
       return false;
 {
    if (end.size() > s.size())
       return false;

-   return (s.substr(s.size() - end.size(), s.size()) == end);
+   return (s.compare(s.size() - end.size(), end.size(), end) == 0);

 }
 
 bool Startswith(const std::string &s, const std::string &start)
 {
    if (start.size() > s.size())
       return false;
 }
 
 bool Startswith(const std::string &s, const std::string &start)
 {
    if (start.size() > s.size())
       return false;

-   return (s.substr(0, start.size()) == start);
+   return (s.compare(0, start.size(), start) == 0);

 }
 
 }
 }
 
 }


@@ -331,7 +335,7 @@ string QuoteString(const string &Str, const char *Bad)
          *I == 0x25 || // percent '%' char
          *I <= 0x20 || *I >= 0x7F) // control chars
       {
          *I == 0x25 || // percent '%' char
          *I <= 0x20 || *I >= 0x7F) // control chars
       {

-        ioprintf(Res,"%%%02x",(int)*I);
+        ioprintf(Res, "%%%02hhx", *I);

       }
       else
         Res << *I;
       }
       else
         Res << *I;


@@ -464,7 +468,9 @@ string SubstVar(const string &Str,const string &Subst,const string &Contents)
 
    if (OldPos >= Str.length())
       return Temp;
 
    if (OldPos >= Str.length())
       return Temp;

-   return Temp + string(Str,OldPos);
+
+   Temp.append(Str, OldPos, string::npos);
+   return Temp;

 }
 string SubstVar(string Str,const struct SubstVar *Vars)
 {
 }
 string SubstVar(string Str,const struct SubstVar *Vars)
 {


@@ -690,9 +696,9 @@ string LookupTag(const string &Message,const char *Tag,const char *Default)
         // Find the end of line and strip the leading/trailing spaces
         string::const_iterator J;
         I += Length + 1;
         // Find the end of line and strip the leading/trailing spaces
         string::const_iterator J;
         I += Length + 1;

-        for (; isspace(*I) != 0 && I < Message.end(); ++I);
+        for (; isspace_ascii(*I) != 0 && I < Message.end(); ++I);

         for (J = I; *J != '\n' && J < Message.end(); ++J);
         for (J = I; *J != '\n' && J < Message.end(); ++J);

-        for (; J > I && isspace(J[-1]) != 0; --J);
+        for (; J > I && isspace_ascii(J[-1]) != 0; --J);

         
         return string(I,J);
       }
         
         return string(I,J);
       }


@@ -744,20 +750,27 @@ int StringToBool(const string &Text,int Default)
 /* This converts a time_t into a string time representation that is
    year 2000 complient and timezone neutral */
 string TimeRFC1123(time_t Date)
 /* This converts a time_t into a string time representation that is
    year 2000 complient and timezone neutral */
 string TimeRFC1123(time_t Date)

+{
+   return TimeRFC1123(Date, false);
+}
+string TimeRFC1123(time_t Date, bool const NumericTimezone)

 {
    struct tm Conv;
    if (gmtime_r(&Date, &Conv) == NULL)
       return "";
 
 {
    struct tm Conv;
    if (gmtime_r(&Date, &Conv) == NULL)
       return "";
 

-   char Buf[300];
-   const char *Day[] = {"Sun","Mon","Tue","Wed","Thu","Fri","Sat"};
-   const char *Month[] = {"Jan","Feb","Mar","Apr","May","Jun","Jul",
-                          "Aug","Sep","Oct","Nov","Dec"};
-
-   snprintf(Buf, sizeof(Buf), "%s, %02i %s %i %02i:%02i:%02i GMT",Day[Conv.tm_wday],
-          Conv.tm_mday,Month[Conv.tm_mon],Conv.tm_year+1900,Conv.tm_hour,
-          Conv.tm_min,Conv.tm_sec);
-   return Buf;
+   auto const posix = std::locale::classic();
+   std::ostringstream datestr;
+   datestr.imbue(posix);
+   APT::StringView const fmt("%a, %d %b %Y %H:%M:%S");
+   std::use_facet<std::time_put<char>>(posix).put(
+                    std::ostreambuf_iterator<char>(datestr),
+                    datestr, ' ', &Conv, fmt.data(), fmt.data() + fmt.size());
+   if (NumericTimezone)
+      datestr << " +0000";
+   else
+      datestr << " GMT";
+   return datestr.str();

 }
                                                                        /*}}}*/
 // ReadMessages - Read messages from the FD                            /*{{{*/
 }
                                                                        /*}}}*/
 // ReadMessages - Read messages from the FD                            /*{{{*/


@@ -788,7 +801,11 @@ bool ReadMessages(int Fd, vector<string> &List)
         return false;
 
       // No data
         return false;
 
       // No data

+#if EAGAIN != EWOULDBLOCK

       if (Res < 0 && (errno == EAGAIN || errno == EWOULDBLOCK))
       if (Res < 0 && (errno == EAGAIN || errno == EWOULDBLOCK))

+#else
+      if (Res < 0 && errno == EAGAIN)
+#endif

         return true;
       if (Res < 0)
         return false;
         return true;
       if (Res < 0)
         return false;


@@ -862,7 +879,7 @@ bool ReadMessages(int Fd, vector<string> &List)
 // ---------------------------------------------------------------------
 /* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
    Made it a bit more robust with a few tolower_ascii though. */
 // ---------------------------------------------------------------------
 /* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
    Made it a bit more robust with a few tolower_ascii though. */

-static int MonthConv(char *Month)
+static int MonthConv(char const * const Month)

 {
    switch (tolower_ascii(*Month)) 
    {
 {
    switch (tolower_ascii(*Month)) 
    {


@@ -915,28 +932,98 @@ static time_t timegm(struct tm *t)
 }
 #endif
                                                                        /*}}}*/
 }
 #endif
                                                                        /*}}}*/

-// FullDateToTime - Converts a HTTP1.1 full date strings into a time_t /*{{{*/
+// RFC1123StrToTime - Converts a HTTP1.1 full date strings into a time_t       /*{{{*/

 // ---------------------------------------------------------------------
 // ---------------------------------------------------------------------

-/* tries to parses a full date as specified in RFC2616 Section 3.3.1
-   with one exception: All timezones (%Z) are accepted but the protocol
-   says that it MUST be GMT, but this one is equal to UTC which we will
-   encounter from time to time (e.g. in Release files) so we accept all
-   here and just assume it is GMT (or UTC) later on */
+/* tries to parses a full date as specified in RFC7231 §7.1.1.1
+   with one exception: HTTP/1.1 valid dates need to have GMT as timezone.
+   As we encounter dates from UTC or with a numeric timezone in other places,
+   we allow them here to to be able to reuse the method. Either way, a date
+   must be in UTC or parsing will fail. Previous implementations of this
+   method used to ignore the timezone and assume always UTC. */

 bool RFC1123StrToTime(const char* const str,time_t &time)
 {
 bool RFC1123StrToTime(const char* const str,time_t &time)
 {

-   struct tm Tm;
-   setlocale (LC_ALL,"C");
-   bool const invalid =
-   // Sun, 06 Nov 1994 08:49:37 GMT  ; RFC 822, updated by RFC 1123
-      (strptime(str, "%a, %d %b %Y %H:%M:%S %Z", &Tm) == NULL &&
-   // Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
-       strptime(str, "%A, %d-%b-%y %H:%M:%S %Z", &Tm) == NULL &&
-   // Sun Nov  6 08:49:37 1994       ; ANSI C's asctime() format
-       strptime(str, "%a %b %d %H:%M:%S %Y", &Tm) == NULL);
-   setlocale (LC_ALL,"");
-   if (invalid == true)
+   unsigned short day = 0;
+   signed int year = 0; // yes, Y23K problem – we gonna worry then…
+   std::string weekday, month, datespec, timespec, zone;
+   std::istringstream ss(str);
+   auto const &posix = std::locale::classic();
+   ss.imbue(posix);
+   ss >> weekday;
+   // we only superficially check weekday, mostly to avoid accepting localized
+   // weekdays here and take only its length to decide which datetime format we
+   // encounter here. The date isn't stored.
+   std::transform(weekday.begin(), weekday.end(), weekday.begin(), ::tolower);
+   std::array<char const * const, 7> c_weekdays = {{ "sun", "mon", "tue", "wed", "thu", "fri", "sat" }};
+   if (std::find(c_weekdays.begin(), c_weekdays.end(), weekday.substr(0,3)) == c_weekdays.end())
+      return false;
+
+   switch (weekday.length())
+   {
+   case 4:
+      // Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
+      if (weekday[3] != ',')
+        return false;
+      ss >> day >> month >> year >> timespec >> zone;
+      break;
+   case 3:
+      // Sun Nov  6 08:49:37 1994 ; ANSI C's asctime() format
+      ss >> month >> day >> timespec >> year;
+      zone = "UTC";
+      break;
+   case 0:
+   case 1:
+   case 2:

       return false;
       return false;

+   default:
+      // Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
+      if (weekday[weekday.length() - 1] != ',')
+        return false;
+      ss >> datespec >> timespec >> zone;
+      auto const expldate = VectorizeString(datespec, '-');
+      if (expldate.size() != 3)
+        return false;
+      try {
+        size_t pos;
+        day = std::stoi(expldate[0], &pos);
+        if (pos != expldate[0].length())
+           return false;
+        year = 1900 + std::stoi(expldate[2], &pos);
+        if (pos != expldate[2].length())
+           return false;
+        strprintf(datespec, "%.4d-%.2d-%.2d", year, MonthConv(expldate[1].c_str()) + 1, day);
+      } catch (...) {
+         return false;
+      }
+      break;
+   }
+
+   if (ss.fail() || ss.bad() || !ss.eof())
+      return false;
+
+   if (zone != "GMT" && zone != "UTC" && zone != "Z") // RFC 822
+   {
+      // numeric timezones as a should of RFC 1123 and generally preferred
+      try {
+        size_t pos;
+        auto const z = std::stoi(zone, &pos);
+        if (z != 0 || pos != zone.length())
+           return false;
+      } catch (...) {
+        return false;
+      }
+   }

 
 

+   if (datespec.empty())
+   {
+      if (month.empty())
+        return false;
+      strprintf(datespec, "%.4d-%.2d-%.2d", year, MonthConv(month.c_str()) + 1, day);
+   }
+
+   std::string const datetime = datespec + ' ' + timespec;
+   struct tm Tm;
+   if (strptime(datetime.c_str(), "%Y-%m-%d %H:%M:%S", &Tm) == nullptr)
+      return false;

    time = timegm(&Tm);
    return true;
 }
    time = timegm(&Tm);
    return true;
 }


@@ -1084,10 +1171,11 @@ bool Base256ToNum(const char *Str,unsigned long long &Res,unsigned int Len)
    tar files */
 bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len)
 {
    tar files */
 bool Base256ToNum(const char *Str,unsigned long &Res,unsigned int Len)
 {

-   unsigned long long Num;
+   unsigned long long Num = 0;

    bool rc;
 
    rc = Base256ToNum(Str, Num, Len);
    bool rc;
 
    rc = Base256ToNum(Str, Num, Len);

+   // rudimentary check for overflow (Res = ulong, Num = ulonglong)

    Res = Num;
    if (Res != Num)
       return false;
    Res = Num;
    if (Res != Num)
       return false;


@@ -1106,26 +1194,36 @@ static int HexDigit(int c)
       return c - 'a' + 10;
    if (c >= 'A' && c <= 'F')
       return c - 'A' + 10;
       return c - 'a' + 10;
    if (c >= 'A' && c <= 'F')
       return c - 'A' + 10;

-   return 0;
+   return -1;

 }
                                                                        /*}}}*/
 // Hex2Num - Convert a long hex number into a buffer                   /*{{{*/
 // ---------------------------------------------------------------------
 /* The length of the buffer must be exactly 1/2 the length of the string. */
 bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length)
 }
                                                                        /*}}}*/
 // Hex2Num - Convert a long hex number into a buffer                   /*{{{*/
 // ---------------------------------------------------------------------
 /* The length of the buffer must be exactly 1/2 the length of the string. */
 bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length)

+{
+   return Hex2Num(APT::StringView(Str), Num, Length);
+}
+
+bool Hex2Num(const APT::StringView Str,unsigned char *Num,unsigned int Length)

 {
    if (Str.length() != Length*2)
       return false;
    
    // Convert each digit. We store it in the same order as the string
    int J = 0;
 {
    if (Str.length() != Length*2)
       return false;
    
    // Convert each digit. We store it in the same order as the string
    int J = 0;

-   for (string::const_iterator I = Str.begin(); I != Str.end();J++, I += 2)
+   for (auto I = Str.begin(); I != Str.end();J++, I += 2)

    {
    {

-      if (isxdigit(*I) == 0 || isxdigit(I[1]) == 0)
+      int first_half = HexDigit(I[0]);
+      int second_half;
+      if (first_half < 0)

         return false;
       
         return false;
       

-      Num[J] = HexDigit(I[0]) << 4;
-      Num[J] += HexDigit(I[1]);
+      second_half = HexDigit(I[1]);
+      if (second_half < 0)
+        return false;
+      Num[J] = first_half << 4;
+      Num[J] += second_half;

    }
    
    return true;
    }
    
    return true;


@@ -1201,7 +1299,7 @@ vector<string> StringSplit(std::string const &s, std::string const &sep,
    vector<string> split;
    size_t start, pos;
 
    vector<string> split;
    size_t start, pos;
 

-   // no seperator given, this is bogus
+   // no separator given, this is bogus

    if(sep.size() == 0)
       return split;
 
    if(sep.size() == 0)
       return split;
 


@@ -1308,7 +1406,7 @@ void ioprintf(ostream &out,const char *format,...)
    va_list args;
    ssize_t size = 400;
    while (true) {
    va_list args;
    ssize_t size = 400;
    while (true) {

-      bool ret = false;
+      bool ret;

       va_start(args,format);
       ret = iovprintf(out, format, args, size);
       va_end(args);
       va_start(args,format);
       ret = iovprintf(out, format, args, size);
       va_end(args);


@@ -1322,7 +1420,7 @@ void strprintf(string &out,const char *format,...)
    ssize_t size = 400;
    std::ostringstream outstr;
    while (true) {
    ssize_t size = 400;
    std::ostringstream outstr;
    while (true) {

-      bool ret = false;
+      bool ret;

       va_start(args,format);
       ret = iovprintf(outstr, format, args, size);
       va_end(args);
       va_start(args,format);
       ret = iovprintf(outstr, format, args, size);
       va_end(args);


@@ -1364,17 +1462,32 @@ string StripEpoch(const string &VerStr)
    return VerStr.substr(i+1);
 }
                                                                        /*}}}*/
    return VerStr.substr(i+1);
 }
                                                                        /*}}}*/

+

 // tolower_ascii - tolower() function that ignores the locale          /*{{{*/
 // ---------------------------------------------------------------------
 /* This little function is the most called method we have and tries
    therefore to do the absolut minimum - and is notable faster than
    standard tolower/toupper and as a bonus avoids problems with different
    locales - we only operate on ascii chars anyway. */
 // tolower_ascii - tolower() function that ignores the locale          /*{{{*/
 // ---------------------------------------------------------------------
 /* This little function is the most called method we have and tries
    therefore to do the absolut minimum - and is notable faster than
    standard tolower/toupper and as a bonus avoids problems with different
    locales - we only operate on ascii chars anyway. */

+#undef tolower_ascii
+int tolower_ascii(int const c) APT_CONST APT_COLD;

 int tolower_ascii(int const c)
 {
 int tolower_ascii(int const c)
 {

-   if (c >= 'A' && c <= 'Z')
-      return c + 32;
-   return c;
+   return tolower_ascii_inline(c);
+}
+                                                                       /*}}}*/
+
+// isspace_ascii - isspace() function that ignores the locale          /*{{{*/
+// ---------------------------------------------------------------------
+/* This little function is one of the most called methods we have and tries
+   therefore to do the absolut minimum - and is notable faster than
+   standard isspace() and as a bonus avoids problems with different
+   locales - we only operate on ascii chars anyway. */
+#undef isspace_ascii
+int isspace_ascii(int const c) APT_CONST APT_COLD;
+int isspace_ascii(int const c)
+{
+   return isspace_ascii_inline(c);

 }
                                                                        /*}}}*/
 
 }
                                                                        /*}}}*/
 


@@ -1528,13 +1641,15 @@ void URI::CopyFrom(const string &U)
    I = FirstColon + 1;
    if (I > SingleSlash)
       I = SingleSlash;
    I = FirstColon + 1;
    if (I > SingleSlash)
       I = SingleSlash;

-   for (; I < SingleSlash && *I != ':'; ++I);
-   string::const_iterator SecondColon = I;
-   
-   // Search for the @ after the colon
-   for (; I < SingleSlash && *I != '@'; ++I);
-   string::const_iterator At = I;
-   
+
+   // Search for the @ separating user:pass from host
+   auto const RevAt = std::find(
+        std::string::const_reverse_iterator(SingleSlash),
+        std::string::const_reverse_iterator(I), '@');
+   string::const_iterator const At = RevAt.base() == I ? SingleSlash : std::prev(RevAt.base());
+   // and then look for the colon between user and pass
+   string::const_iterator const SecondColon = std::find(I, At, ':');
+

    // Now write the host and user/pass
    if (At == SingleSlash)
    {
    // Now write the host and user/pass
    if (At == SingleSlash)
    {


@@ -1622,7 +1737,7 @@ URI::operator string()
         Res << Host;
 
       if (Port != 0)
         Res << Host;
 
       if (Port != 0)

-        Res << ':' << Port;
+        Res << ':' << std::to_string(Port);

    }
 
    if (Path.empty() == false)
    }
 
    if (Path.empty() == false)


@@ -1637,8 +1752,6 @@ URI::operator string()
 }
                                                                        /*}}}*/
 // URI::SiteOnly - Return the schema and site for the URI              /*{{{*/
 }
                                                                        /*}}}*/
 // URI::SiteOnly - Return the schema and site for the URI              /*{{{*/

-// ---------------------------------------------------------------------
-/* */

 string URI::SiteOnly(const string &URI)
 {
    ::URI U(URI);
 string URI::SiteOnly(const string &URI)
 {
    ::URI U(URI);


@@ -1648,9 +1761,18 @@ string URI::SiteOnly(const string &URI)
    return U;
 }
                                                                        /*}}}*/
    return U;
 }
                                                                        /*}}}*/

+// URI::ArchiveOnly - Return the schema, site and cleaned path for the URI /*{{{*/
+string URI::ArchiveOnly(const string &URI)
+{
+   ::URI U(URI);
+   U.User.clear();
+   U.Password.clear();
+   if (U.Path.empty() == false && U.Path[U.Path.length() - 1] == '/')
+      U.Path.erase(U.Path.length() - 1);
+   return U;
+}
+                                                                       /*}}}*/

 // URI::NoUserPassword - Return the schema, site and path for the URI  /*{{{*/
 // URI::NoUserPassword - Return the schema, site and path for the URI  /*{{{*/

-// ---------------------------------------------------------------------
-/* */

 string URI::NoUserPassword(const string &URI)
 {
    ::URI U(URI);
 string URI::NoUserPassword(const string &URI)
 {
    ::URI U(URI);