diff options
Diffstat (limited to 'libbutl/timestamp.cxx')
| -rw-r--r-- | libbutl/timestamp.cxx | 127 |
1 files changed, 83 insertions, 44 deletions
diff --git a/libbutl/timestamp.cxx b/libbutl/timestamp.cxx index 9be2a82..260fbef 100644 --- a/libbutl/timestamp.cxx +++ b/libbutl/timestamp.cxx @@ -1,9 +1,7 @@ // file : libbutl/timestamp.cxx -*- C++ -*- // license : MIT; see accompanying LICENSE file -#ifndef __cpp_modules_ts -#include <libbutl/timestamp.mxx> -#endif +#include <libbutl/timestamp.hxx> #include <time.h> // localtime_{r,s}(), gmtime_{r,s}(), strptime(), timegm() #include <errno.h> // EINVAL @@ -25,22 +23,18 @@ #ifdef __GLIBCXX__ extern "C" { -#include "strptime.c" +# include "strptime.c" } #else -#include <locale.h> // LC_ALL +# include <locale.h> // LC_ALL #endif #endif -#ifndef __cpp_lib_modules_ts -#include <string> -#include <chrono> - -#include <ctime> // tm, time_t, mktime(), strftime()[__GLIBCXX__] +#include <ctime> // tm, time_t, mktime(), strftime()[libstdc++] #include <cstdlib> // strtoull() -#include <sstream> +#include <sstream> // ostringstream, stringstream[VC] #include <iomanip> // put_time(), setw(), dec, right -#include <cstring> // strlen(), memcpy() +#include <cstring> // strlen(), memcpy(), strchr()[VC] #include <ostream> #include <utility> // pair, make_pair() #include <stdexcept> // runtime_error @@ -49,30 +43,14 @@ extern "C" // #ifdef _WIN32 #ifndef __GLIBCXX__ -#include <locale> -#include <clocale> -#include <iomanip> -#endif +# include <ios> +# include <locale> +# include <clocale> +# include <iomanip> #endif #endif -// Other includes. - -#ifdef __cpp_modules_ts -module butl.timestamp; - -// Only imports additional to interface. -#ifdef __clang__ -#ifdef __cpp_lib_modules_ts -import std.core; -import std.io; -#endif -#endif - -import butl.utility; -#else -#include <libbutl/utility.mxx> // throw_generic_error() -#endif +#include <libbutl/utility.hxx> // throw_generic_error() using namespace std; @@ -180,24 +158,85 @@ strptime (const char* input, const char* format, tm* time) { // VC std::get_time()-based implementation. // - istringstream is (input); + // Note that the major difference in semantics of strptime() and + // std::get_time() is that the former always fails if the format string is + // not fully processed, while the latter can succeed in such a case, + // specifically if the end of the stream is reached after a conversion + // specifier was successfully applied. See this post for some background: + // + // https://stackoverflow.com/questions/67060956/what-is-the-correct-behavior-of-stdget-time-for-short-input + // + // The consequence of this fact is that there is no easy way to detect if + // the format was fully processed when the end of input is reached. It seems + // that the only way to resolve this ambiguity is to append some end marker + // to both the input and format and re-parse. We can dedicate some character + // that is unlikely to be used in the time format/input (for example '\x1') + // to serve as an end marker. + // + // Alternatively, we can abandon the use of std::get_time() altogether and, + // for example, use a FreeBSD-based strptime() implementation. This feels a + // bit too radical at the moment, though. + // + const char em ('\x1'); + + if (strchr (input, em) != nullptr || strchr (format, em) != nullptr) + return nullptr; + + stringstream ss (input); // Input/output stream. // The original strptime() function behaves according to the process' C // locale (set with std::setlocale()), which can differ from the process C++ // locale (set with std::locale::global()). // - is.imbue (locale (setlocale (LC_ALL, nullptr))); + ss.imbue (locale (setlocale (LC_ALL, nullptr))); - if (!(is >> get_time (time, format))) + // Bail out on the parsing error. + // + if (!(ss >> get_time (time, format))) return nullptr; - else - // tellg() behaves as UnformattedInputFunction, so returns failure status - // if eofbit is set. - // - return const_cast<char*> ( - input + (is.eof () - ? strlen (input) - : static_cast<size_t> (is.tellg ()))); + + // If the end of input is not reached then the format string is fully + // processed. + // + if (!ss.eof ()) + return const_cast<char*> (input + static_cast<size_t> (ss.tellg ())); + + // Since eof is reached, we cannot say if the format string was fully + // processed or not. For example: + // + // %b %Y - format + // Feb 2016 - eofbit is set with a format fully processed + // Feb - eofbit is set with a format partially processed + // + // So append the end marker character to both input and format and re-parse. + // + ss.clear (); // Clear eof. + ss.seekp (0, ios_base::end); // Position to the end for writing. + ss.put (em); // Append the end marker. + ss.seekg (0); // Rewind for reading. + + string fm (format); + fm += em; // Append the end marker. + + // Fail if the input is "shorter" than the format. For example: + // + // %b %Y\x1 - format + // Feb\x1 - stream + // + // Note that we can reuse the time object for re-parsing, since on success + // its fields will be overwritten with the same values. + // + if (!(ss >> get_time (time, fm.c_str ()))) + return nullptr; + + // We would fail earlier otherwise. + // + assert (ss.eof () || ss.get () == stringstream::traits_type::eof ()); + + // tellg() behaves as UnformattedInputFunction, so returns failure status if + // eofbit is set. + // + return const_cast<char*> (input + strlen (input)); } #endif |