/** * Substitutes vector elements if a pattern occurs in a string * * @param str character vector * @param pattern character vector * @param value character vector * @return character vector * * @version 1.0-3 (Marek Gagolewski, 2016-02-03) * FR#124 * * @version 1.0-3 (Marek Gagolewski, 2016-02-03) * FR #216: `negate` arg added */ SEXP stri_subset_charclass_replacement(SEXP str, SEXP pattern, SEXP negate, SEXP value) { bool negate_1 = stri__prepare_arg_logical_1_notNA(negate, "negate"); PROTECT(str = stri_prepare_arg_string(str, "str")); PROTECT(pattern = stri_prepare_arg_string_1(pattern, "pattern")); PROTECT(value = stri_prepare_arg_string(value, "value")); int vectorize_length = LENGTH(str); int value_length = LENGTH(value); if (value_length == 0) Rf_error(MSG__REPLACEMENT_ZERO); STRI__ERROR_HANDLER_BEGIN(3) StriContainerUTF8 str_cont(str, vectorize_length); StriContainerUTF8 value_cont(value, value_length); StriContainerCharClass pattern_cont(pattern, vectorize_length); SEXP ret; STRI__PROTECT(ret = Rf_allocVector(STRSXP, vectorize_length)); R_len_t k = 0; for (R_len_t i = str_cont.vectorize_init(); i != str_cont.vectorize_end(); i = str_cont.vectorize_next(i)) { if (str_cont.isNA(i) || pattern_cont.isNA(i)) { SET_STRING_ELT(ret, i, NA_STRING); continue; } const UnicodeSet* pattern_cur = &pattern_cont.get(i); R_len_t str_cur_n = str_cont.get(i).length(); const char* str_cur_s = str_cont.get(i).c_str(); UChar32 chr = 0; bool found = false; for (R_len_t j=0; j<str_cur_n; ) { U8_NEXT(str_cur_s, j, str_cur_n, chr); if (chr < 0) // invalid utf-8 sequence throw StriException(MSG__INVALID_UTF8); if (pattern_cur->contains(chr)) { found = true; break; } } if ((found && !negate_1) || (!found && negate_1)) SET_STRING_ELT(ret, i, value_cont.toR((k++)%value_length)); else SET_STRING_ELT(ret, i, str_cont.toR(i)); } STRI__UNPROTECT_ALL return ret; STRI__ERROR_HANDLER_END(;/* nothing special to be done on error */) }
/** * Split a single string into text lines * * @param str character vector * * @return character vector * * @version 0.1 (Marek Gagolewski, 2013-08-04) */ SEXP stri_split_lines1(SEXP str) { str = stri_prepare_arg_string_1(str, "str"); R_len_t vectorize_length = LENGTH(str); STRI__ERROR_HANDLER_BEGIN StriContainerUTF8 str_cont(str, vectorize_length); if (str_cont.isNA(0)) return str; const char* str_cur_s = str_cont.get(0).c_str(); R_len_t str_cur_n = str_cont.get(0).length(); UChar32 c; R_len_t jlast; deque<R_len_t_x2> occurences; occurences.push_back(R_len_t_x2(0, 0)); for (R_len_t j=0; j < str_cur_n; /* null */) { jlast = j; U8_NEXT(str_cur_s, j, str_cur_n, c); switch (c) { case ASCII_CR: /* CR */ /* check if next is LF */ if (str_cur_s[j] == ASCII_LF) { // look ahead one byte j++; // just one byte } break; case ASCII_LF: /* LF */ break; case UCHAR_NEL: /* NEL */ break; case ASCII_VT: /* VT */ break; case ASCII_FF: /* FF */ break; case UCHAR_LS: /* LS */ break; case UCHAR_PS: /* PS */ break; default: /* not a newline character */ occurences.back().v2 = j; continue; } occurences.back().v2 = jlast; if (j < str_cur_n) occurences.push_back(R_len_t_x2(j, j)); } SEXP ans; PROTECT(ans = Rf_allocVector(STRSXP, (R_len_t)occurences.size())); deque<R_len_t_x2>::iterator iter = occurences.begin(); for (R_len_t k = 0; iter != occurences.end(); ++iter, ++k) { R_len_t_x2 curoccur = *iter; SET_STRING_ELT(ans, k, Rf_mkCharLenCE(str_cur_s+curoccur.v1, curoccur.v2-curoccur.v1, CE_UTF8)); } UNPROTECT(1); return ans; STRI__ERROR_HANDLER_END(;/* nothing special to be done on error */) }
/** List available time zone IDs * * @param offset single numeric * @param region single string * @return character vector * * @version 0.5-1 (Marek Gagolewski, 2014-12-24) */ SEXP stri_timezone_list(SEXP region, SEXP offset) { StringEnumeration* tz_enum = NULL; PROTECT(region = stri_prepare_arg_string_1(region, "region")); PROTECT(offset = stri_prepare_arg_double_1(offset, "offset")); STRI__ERROR_HANDLER_BEGIN(2) StriContainerUTF8 region_cont(region, 1); UErrorCode status = U_ZERO_ERROR; int32_t offset_hours = 0; const int32_t* o = NULL; const char* r = NULL; if (!ISNA(REAL(offset)[0])) { // 0.5 and 0.75 are represented exactly within the double type offset_hours = (int32_t)(REAL(offset)[0]*1000.0*3600.0); o = &offset_hours; } if (!region_cont.isNA(0)) r = region_cont.get(0).c_str(); tz_enum = TimeZone::createTimeZoneIDEnumeration(UCAL_ZONE_TYPE_ANY, r, o, status); STRI__CHECKICUSTATUS_RFERROR(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; tz_enum->reset(status); STRI__CHECKICUSTATUS_RFERROR(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; R_len_t n = (R_len_t)tz_enum->count(status); STRI__CHECKICUSTATUS_RFERROR(status, {/* do nothing special on err */}) SEXP ret; STRI__PROTECT(ret = Rf_allocVector(STRSXP, n)); // SEXP nam; // STRI__PROTECT(nam = Rf_allocVector(STRSXP, n)); // MG: I reckon that IDs are more readable than DisplayNames (which are moreover localized) for (R_len_t i=0; i<n; ++i) { int len; status = U_ZERO_ERROR; const char* cur = tz_enum->next(&len, status); STRI__CHECKICUSTATUS_RFERROR(status, {/* do nothing special on err */}) SET_STRING_ELT(ret, i, Rf_mkCharLenCE(cur, len, CE_UTF8)); // TimeZone* curtz = TimeZone::createTimeZone(UnicodeString::fromUTF8(cur)); // UnicodeString curdn; // curtz->getDisplayName(locale, curdn); // delete curtz; // string out; // curdn.toUTF8String(out); // SET_STRING_ELT(nam, i, Rf_mkCharCE(out.c_str(), CE_UTF8)); } // Rf_setAttrib(ret, R_NamesSymbol, nam); if (tz_enum) { delete tz_enum; tz_enum = NULL; } STRI__UNPROTECT_ALL return ret; STRI__ERROR_HANDLER_END( if (tz_enum) { delete tz_enum; tz_enum = NULL; } ) }
/** Date-time artithmetic * * @param time * @param value * @param units * @param tz * @param locale * * @return POSIXst * * @version 0.5-1 (Marek Gagolewski, 2014-12-30) * @version 0.5-1 (Marek Gagolewski, 2015-03-06) tz arg added */ SEXP stri_datetime_add(SEXP time, SEXP value, SEXP units, SEXP tz, SEXP locale) { PROTECT(time = stri_prepare_arg_POSIXct(time, "time")); PROTECT(value = stri_prepare_arg_integer(value, "value")); if (!isNull(tz)) PROTECT(tz = stri_prepare_arg_string_1(tz, "tz")); else PROTECT(tz); /* needed to set tzone attrib */ R_len_t vectorize_length = stri__recycling_rule(true, 2, LENGTH(time), LENGTH(value)); const char* units_val = stri__prepare_arg_string_1_notNA(units, "units"); const char* units_opts[] = {"years", "months", "weeks", "days", "hours", "minutes", "seconds", "milliseconds", NULL}; int units_cur = stri__match_arg(units_val, units_opts); const char* locale_val = stri__prepare_arg_locale(locale, "locale", true); TimeZone* tz_val = stri__prepare_arg_timezone(tz, "tz", true/*allowdefault*/); Calendar* cal = NULL; STRI__ERROR_HANDLER_BEGIN(3) StriContainerDouble time_cont(time, vectorize_length); StriContainerInteger value_cont(value, vectorize_length); UCalendarDateFields units_field; switch (units_cur) { case 0: units_field = UCAL_YEAR; break; case 1: units_field = UCAL_MONTH; break; case 2: units_field = UCAL_WEEK_OF_YEAR; break; case 3: units_field = UCAL_DAY_OF_MONTH; break; case 4: units_field = UCAL_HOUR_OF_DAY; break; case 5: units_field = UCAL_MINUTE; break; case 6: units_field = UCAL_SECOND; break; case 7: units_field = UCAL_MILLISECOND; break; default: throw StriException(MSG__INCORRECT_MATCH_OPTION, "units"); } UErrorCode status = U_ZERO_ERROR; cal = Calendar::createInstance(locale_val, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) cal->adoptTimeZone(tz_val); tz_val = NULL; /* The Calendar takes ownership of the TimeZone. */ SEXP ret; STRI__PROTECT(ret = Rf_allocVector(REALSXP, vectorize_length)); double* ret_val = REAL(ret); for (R_len_t i=0; i<vectorize_length; ++i) { if (time_cont.isNA(i) || value_cont.isNA(i)) { ret_val[i] = NA_REAL; continue; } status = U_ZERO_ERROR; cal->setTime((UDate)(time_cont.get(i)*1000.0), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; cal->add(units_field, value_cont.get(i), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; ret_val[i] = ((double)cal->getTime(status))/1000.0; STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) } if (!isNull(tz)) Rf_setAttrib(ret, Rf_ScalarString(Rf_mkChar("tzone")), tz); stri__set_class_POSIXct(ret); if (tz_val) { delete tz_val; tz_val = NULL; } if (cal) { delete cal; cal = NULL; } STRI__UNPROTECT_ALL return ret; STRI__ERROR_HANDLER_END({ if (tz_val) { delete tz_val; tz_val = NULL; } if (cal) { delete cal; cal = NULL; } }) }
/** * Get values of date-time fields * * @param time * @param locale * @param tz * * @return list * * @version 0.5-1 (Marek Gagolewski, 2015-01-01) * @version 0.5-1 (Marek Gagolewski, 2015-03-03) tz arg added */ SEXP stri_datetime_fields(SEXP time, SEXP tz, SEXP locale) { PROTECT(time = stri_prepare_arg_POSIXct(time, "time")); const char* locale_val = stri__prepare_arg_locale(locale, "locale", true); if (!isNull(tz)) PROTECT(tz = stri_prepare_arg_string_1(tz, "tz")); else PROTECT(tz); /* needed to set tzone attrib */ TimeZone* tz_val = stri__prepare_arg_timezone(tz, "tz", true/*allowdefault*/); Calendar* cal = NULL; STRI__ERROR_HANDLER_BEGIN(2) R_len_t vectorize_length = LENGTH(time); StriContainerDouble time_cont(time, vectorize_length); UErrorCode status = U_ZERO_ERROR; cal = Calendar::createInstance(locale_val, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) cal->adoptTimeZone(tz_val); tz_val = NULL; /* The Calendar takes ownership of the TimeZone. */ SEXP ret; #define STRI__FIELDS_NUM 14 STRI__PROTECT(ret = Rf_allocVector(VECSXP, STRI__FIELDS_NUM)); for (R_len_t j=0; j<STRI__FIELDS_NUM; ++j) SET_VECTOR_ELT(ret, j, Rf_allocVector(INTSXP, vectorize_length)); for (R_len_t i=0; i<vectorize_length; ++i) { if (time_cont.isNA(i)) { for (R_len_t j=0; j<STRI__FIELDS_NUM; ++j) INTEGER(VECTOR_ELT(ret, j))[i] = NA_INTEGER; continue; } status = U_ZERO_ERROR; cal->setTime((UDate)(time_cont.get(i)*1000.0), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) for (R_len_t j=0; j<STRI__FIELDS_NUM; ++j) { UCalendarDateFields units_field; switch (j) { case 0: units_field = UCAL_EXTENDED_YEAR; break; case 1: units_field = UCAL_MONTH; break; case 2: units_field = UCAL_DAY_OF_MONTH; break; case 3: units_field = UCAL_HOUR_OF_DAY; break; case 4: units_field = UCAL_MINUTE; break; case 5: units_field = UCAL_SECOND; break; case 6: units_field = UCAL_MILLISECOND; break; case 7: units_field = UCAL_WEEK_OF_YEAR; break; case 8: units_field = UCAL_WEEK_OF_MONTH; break; case 9: units_field = UCAL_DAY_OF_YEAR; break; case 10: units_field = UCAL_DAY_OF_WEEK; break; case 11: units_field = UCAL_HOUR; break; case 12: units_field = UCAL_AM_PM; break; case 13: units_field = UCAL_ERA; break; default: throw StriException(MSG__INCORRECT_MATCH_OPTION, "units"); } //UCAL_IS_LEAP_MONTH //UCAL_MILLISECONDS_IN_DAY -> SecondsInDay // UCAL_AM_PM -> "AM" or "PM" (localized? or factor?+index in stri_datetime_symbols) add arg use_symbols???? // UCAL_DAY_OF_WEEK -> (localized? or factor?) SUNDAY, MONDAY // UCAL_DAY_OF_YEAR ' // isWekend status = U_ZERO_ERROR; INTEGER(VECTOR_ELT(ret, j))[i] = cal->get(units_field, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) if (units_field == UCAL_MONTH) ++INTEGER(VECTOR_ELT(ret, j))[i]; // month + 1 else if (units_field == UCAL_AM_PM) ++INTEGER(VECTOR_ELT(ret, j))[i]; // ampm + 1 else if (units_field == UCAL_ERA) ++INTEGER(VECTOR_ELT(ret, j))[i]; // era + 1 }
/** Word wrap text * * @param str character vector * @param width single integer * @param cost_exponent single double * @param indent single integer * @param exdent single integer * @param prefix single string * @param initial single string * @param locale locale identifier or NULL for default locale * @param use_length single logical value * * @return list * * @version 0.1-?? (Bartek Tartanus) * * @version 0.2-2 (Marek Gagolewski, 2014-04-27) * single function for wrap_greedy and wrap_dynamic * (dispatch inside); * use BreakIterator * * @version 0.3-1 (Marek Gagolewski, 2014-11-04) * Issue #112: str_prepare_arg* retvals were not PROTECTed from gc * * @version 0.4-1 (Marek Gagolewski, 2014-12-06) * new args: indent, exdent, prefix, initial * * @version 0.5-1 (Marek Gagolewski, 2014-12-19) * #133 allow width <= 0 * * @version 0.5-1 (Marek Gagolewski, 2015-02-28) * don't trim so many white spaces at the end of each word (normalize arg does that) * #139: allow a "whitespace" break iterator * * @version 0.5-1 (Marek Gagolewski, 2015-04-23) * `use_length` arg added * * * @version 0.5-1 (Marek Gagolewski, 2015-06-09) * BIGSKIP: no more CHARSXP on out on "" input */ SEXP stri_wrap(SEXP str, SEXP width, SEXP cost_exponent, SEXP indent, SEXP exdent, SEXP prefix, SEXP initial, SEXP whitespace_only, SEXP use_length, SEXP locale) { bool use_length_val = stri__prepare_arg_logical_1_notNA(use_length, "use_length"); double exponent_val = stri__prepare_arg_double_1_notNA(cost_exponent, "cost_exponent"); bool whitespace_only_val = stri__prepare_arg_logical_1_notNA(whitespace_only, "whitespace_only"); int width_val = stri__prepare_arg_integer_1_notNA(width, "width"); if (width_val <= 0) width_val = 0; int indent_val = stri__prepare_arg_integer_1_notNA(indent, "indent"); if (indent_val < 0) Rf_error(MSG__EXPECTED_POSITIVE, "indent"); int exdent_val = stri__prepare_arg_integer_1_notNA(exdent, "exdent"); if (exdent_val < 0) Rf_error(MSG__EXPECTED_POSITIVE, "exdent"); const char* qloc = stri__prepare_arg_locale(locale, "locale", true); /* this is R_alloc'ed */ Locale loc = Locale::createFromName(qloc); PROTECT(str = stri_prepare_arg_string(str, "str")); PROTECT(prefix = stri_prepare_arg_string_1(prefix, "prefix")); PROTECT(initial = stri_prepare_arg_string_1(initial, "initial")); BreakIterator* briter = NULL; UText* str_text = NULL; STRI__ERROR_HANDLER_BEGIN(3) UErrorCode status = U_ZERO_ERROR; briter = BreakIterator::createLineInstance(loc, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) R_len_t str_length = LENGTH(str); StriContainerUTF8_indexable str_cont(str, str_length); StriContainerUTF8 prefix_cont(prefix, 1); StriContainerUTF8 initial_cont(initial, 1); // prepare indent/exdent/prefix/initial stuff: // 1st line, 1st para (i==0, u==0): initial+indent // nth line, 1st para (i==0, u> 0): prefix +exdent // 1st line, nth para (i> 0, u==0): prefix +indent // nth line, nth para (i> 0, u> 0): prefix +exdent StriWrapLineStart ii(initial_cont.get(0), indent_val); StriWrapLineStart pi(prefix_cont.get(0), indent_val); StriWrapLineStart pe(prefix_cont.get(0), exdent_val); status = U_ZERO_ERROR; //Unicode Newline Guidelines - Unicode Technical Report #13 UnicodeSet uset_linebreaks(UnicodeString::fromUTF8("[\\u000A-\\u000D\\u0085\\u2028\\u2029]"), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) uset_linebreaks.freeze(); status = U_ZERO_ERROR; UnicodeSet uset_whitespaces(UnicodeString::fromUTF8("\\p{White_space}"), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) uset_whitespaces.freeze(); SEXP ret; STRI__PROTECT(ret = Rf_allocVector(VECSXP, str_length)); for (R_len_t i = 0; i < str_length; ++i) { if (str_cont.isNA(i) || prefix_cont.isNA(0) || initial_cont.isNA(0)) { SET_VECTOR_ELT(ret, i, stri__vector_NA_strings(1)); continue; } status = U_ZERO_ERROR; const char* str_cur_s = str_cont.get(i).c_str(); R_len_t str_cur_n = str_cont.get(i).length(); str_text = utext_openUTF8(str_text, str_cur_s, str_cont.get(i).length(), &status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; briter->setText(str_text, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) // all right, first let's generate a list of places at which we may do line breaks deque< R_len_t > occurrences_list; // this could be an R_len_t queue R_len_t match = briter->first(); while (match != BreakIterator::DONE) { if (!whitespace_only_val) occurrences_list.push_back(match); else { if (match > 0 && match < str_cur_n) { UChar32 c; U8_GET((const uint8_t*)str_cur_s, 0, match-1, str_cur_n, c); if (uset_whitespaces.contains(c)) occurrences_list.push_back(match); } else occurrences_list.push_back(match); } match = briter->next(); } R_len_t noccurrences = (R_len_t)occurrences_list.size(); // number of boundaries if (noccurrences <= 1) { // no match (1 boundary == 0) SET_VECTOR_ELT(ret, i, Rf_ScalarString(str_cont.toR(i))); continue; } // the number of "words" is: R_len_t nwords = noccurrences - 1; // convert occurrences_list to a vector // in order to obtain end positions (in a string) of each "words", // noting that occurrences_list.at(0) == 0 #ifndef NDEBUG if (occurrences_list.at(0) != 0) throw StriException("NDEBUG: stri_wrap: (occurrences_list.at(0) != 0)"); #endif std::vector<R_len_t> end_pos_orig(nwords); deque<R_len_t>::iterator iter = ++(occurrences_list.begin()); for (R_len_t j = 0; iter != occurrences_list.end(); ++iter, ++j) { end_pos_orig[j] = (*iter); // this is a UTF-8 index } // now: // we'll get the total widths/number of code points in each "word" std::vector<R_len_t> widths_orig(nwords); // we'll get the total widths/number of code points without trailing whitespaces std::vector<R_len_t> widths_trim(nwords); // we'll get the end positions without trailing whitespaces std::vector<R_len_t> end_pos_trim(nwords); // detect line endings (fail on a match) UChar32 c = 0; R_len_t j = 0; R_len_t cur_block = 0; R_len_t cur_width_orig = 0; R_len_t cur_width_trim = 0; R_len_t cur_count_orig = 0; R_len_t cur_count_trim = 0; R_len_t cur_end_pos_trim = 0; while (j < str_cur_n) { R_len_t jlast = j; U8_NEXT(str_cur_s, j, str_cur_n, c); if (c < 0) // invalid utf-8 sequence throw StriException(MSG__INVALID_UTF8); if (uset_linebreaks.contains(c)) throw StriException(MSG__NEWLINE_FOUND); cur_width_orig += stri__width_char(c); ++cur_count_orig; if (uset_whitespaces.contains(c)) { // OLD: trim all white spaces from the end: // ++cur_count_trim; // [we have the normalize arg for that] // NEW: trim just one white space at the end: cur_width_trim = stri__width_char(c); cur_count_trim = 1; cur_end_pos_trim = jlast; } else { cur_width_trim = 0; cur_count_trim = 0; cur_end_pos_trim = j; } if (j >= str_cur_n || end_pos_orig[cur_block] <= j) { // we'll start a new block in a moment if (use_length_val) { widths_orig[cur_block] = cur_count_orig; widths_trim[cur_block] = cur_count_orig-cur_count_trim; } else { widths_orig[cur_block] = cur_width_orig; widths_trim[cur_block] = cur_width_orig-cur_width_trim; } end_pos_trim[cur_block] = cur_end_pos_trim; cur_block++; cur_width_orig = 0; cur_width_trim = 0; cur_count_orig = 0; cur_count_trim = 0; cur_end_pos_trim = j; } } // do wrap std::deque<R_len_t> wrap_after; // wrap line after which word in {0..nwords-1}? if (exponent_val <= 0.0) { stri__wrap_greedy(wrap_after, nwords, width_val, widths_orig, widths_trim, (use_length_val)?((i==0)?ii.count:pi.count):((i==0)?ii.width:pi.width), (use_length_val)?pe.count:pe.width); } else { stri__wrap_dynamic(wrap_after, nwords, width_val, exponent_val, widths_orig, widths_trim, (use_length_val)?((i==0)?ii.count:pi.count):((i==0)?ii.width:pi.width), (use_length_val)?pe.count:pe.width); } // wrap_after.size() line breaks => wrap_after.size()+1 lines R_len_t nlines = (R_len_t)wrap_after.size()+1; R_len_t last_pos = 0; SEXP ans; STRI__PROTECT(ans = Rf_allocVector(STRSXP, nlines)); deque<R_len_t>::iterator iter_wrap = wrap_after.begin(); for (R_len_t u = 0; iter_wrap != wrap_after.end(); ++iter_wrap, ++u) { R_len_t wrap_after_cur = *iter_wrap; R_len_t cur_pos = end_pos_trim[wrap_after_cur]; std::string cs; if (i == 0 && u == 0) cs = ii.str; else if (i > 0 && u == 0) cs = pi.str; else cs = pe.str; cs.append(str_cur_s+last_pos, cur_pos-last_pos); SET_STRING_ELT(ans, u, Rf_mkCharLenCE(cs.c_str(), cs.size(), CE_UTF8)); last_pos = end_pos_orig[wrap_after_cur]; } // last line goes here: std::string cs; if (i == 0 && nlines-1 == 0) cs = ii.str; else if (i > 0 && nlines-1 == 0) cs = pi.str; else cs = pe.str; cs.append(str_cur_s+last_pos, end_pos_trim[nwords-1]-last_pos); SET_STRING_ELT(ans, nlines-1, Rf_mkCharLenCE(cs.c_str(), cs.size(), CE_UTF8)); SET_VECTOR_ELT(ret, i, ans); STRI__UNPROTECT(1); } if (briter) { delete briter; briter = NULL; } if (str_text) { utext_close(str_text); str_text = NULL; } STRI__UNPROTECT_ALL return ret; STRI__ERROR_HANDLER_END({ if (briter) { delete briter; briter = NULL; } if (str_text) { utext_close(str_text); str_text = NULL; } }) }
/** * Parse date-time objects * * @param str * @param format * @param tz * @param lenient * @param locale * * @return character vector * * @version 0.5-1 (Marek Gagolewski, 2015-01-08) * @version 0.5-1 (Marek Gagolewski, 2015-01-11) lenient arg added * @version 0.5-1 (Marek Gagolewski, 2015-02-22) use tz * @version 0.5-1 (Marek Gagolewski, 2015-03-01) set tzone attrib on retval */ SEXP stri_datetime_parse(SEXP str, SEXP format, SEXP lenient, SEXP tz, SEXP locale) { PROTECT(str = stri_prepare_arg_string(str, "str")); const char* locale_val = stri__prepare_arg_locale(locale, "locale", true); const char* format_val = stri__prepare_arg_string_1_notNA(format, "format"); bool lenient_val = stri__prepare_arg_logical_1_notNA(lenient, "lenient"); if (!isNull(tz)) PROTECT(tz = stri_prepare_arg_string_1(tz, "tz")); else PROTECT(tz); /* needed to set tzone attrib */ // "format" may be one of: const char* format_opts[] = { "date_full", "date_long", "date_medium", "date_short", "date_relative_full", "date_relative_long", "date_relative_medium", "date_relative_short", "time_full", "time_long", "time_medium", "time_short", "time_relative_full", "time_relative_long", "time_relative_medium", "time_relative_short", "datetime_full", "datetime_long", "datetime_medium", "datetime_short", "datetime_relative_full", "datetime_relative_long", "datetime_relative_medium", "datetime_relative_short", NULL}; int format_cur = stri__match_arg(format_val, format_opts); TimeZone* tz_val = stri__prepare_arg_timezone(tz, "tz", true/*allowdefault*/); Calendar* cal = NULL; DateFormat* fmt = NULL; STRI__ERROR_HANDLER_BEGIN(2) R_len_t vectorize_length = LENGTH(str); StriContainerUTF16 str_cont(str, vectorize_length); UnicodeString format_str(format_val); UErrorCode status = U_ZERO_ERROR; if (format_cur >= 0) { DateFormat::EStyle style = DateFormat::kNone; switch (format_cur % 8) { case 0: style = DateFormat::kFull; break; case 1: style = DateFormat::kLong; break; case 2: style = DateFormat::kMedium; break; case 3: style = DateFormat::kShort; break; case 4: style = DateFormat::kFullRelative; break; case 5: style = DateFormat::kLongRelative; break; case 6: style = DateFormat::kMediumRelative; break; case 7: style = DateFormat::kShortRelative; break; default: style = DateFormat::kNone; break; } /* ICU 54.1: Relative time styles are not currently supported. */ switch (format_cur / 8) { case 0: fmt = DateFormat::createDateInstance(style, Locale::createFromName(locale_val)); break; case 1: fmt = DateFormat::createTimeInstance((DateFormat::EStyle)(style & ~DateFormat::kRelative), Locale::createFromName(locale_val)); break; case 2: fmt = DateFormat::createDateTimeInstance(style, (DateFormat::EStyle)(style & ~DateFormat::kRelative), Locale::createFromName(locale_val)); break; default: fmt = NULL; break; } } else fmt = new SimpleDateFormat(format_str, Locale::createFromName(locale_val), status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) status = U_ZERO_ERROR; cal = Calendar::createInstance(locale_val, status); STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) cal->adoptTimeZone(tz_val); tz_val = NULL; /* The Calendar takes ownership of the TimeZone. */ cal->setLenient(lenient_val); SEXP ret; STRI__PROTECT(ret = Rf_allocVector(REALSXP, vectorize_length)); for (R_len_t i=0; i<vectorize_length; ++i) { if (str_cont.isNA(i)) { REAL(ret)[i] = NA_REAL; continue; } status = U_ZERO_ERROR; ParsePosition pos; fmt->parse(str_cont.get(i), *cal, pos); if (pos.getErrorIndex() >= 0) REAL(ret)[i] = NA_REAL; else { status = U_ZERO_ERROR; REAL(ret)[i] = ((double)cal->getTime(status))/1000.0; if (U_FAILURE(status)) REAL(ret)[i] = NA_REAL; } } if (!isNull(tz)) Rf_setAttrib(ret, Rf_ScalarString(Rf_mkChar("tzone")), tz); stri__set_class_POSIXct(ret); if (tz_val) { delete tz_val; tz_val = NULL; } if (fmt) { delete fmt; fmt = NULL; } if (cal) { delete cal; cal = NULL; } STRI__UNPROTECT_ALL return ret; STRI__ERROR_HANDLER_END({ if (tz_val) { delete tz_val; tz_val = NULL; } if (fmt) { delete fmt; fmt = NULL; } if (cal) { delete cal; cal = NULL; } }) }