/**
* Count pattern occurcess in a string [with collation]
*
* @param str character vector
* @param pattern character vector
* @param opts_collator passed to stri__ucol_open()
* @return integer vector
*
* @version 0.1-?? (Marek Gagolewski)
*
* @version 0.1-?? (Marek Gagolewski)
* corrected behavior on empty str/pattern
*
* @version 0.1-?? (Marek Gagolewski, 2013-06-23)
* make StriException-friendly,
* use StriContainerUStringSearch
*
* @version 0.2-3 (Marek Gagolewski, 2014-05-08)
* new fun: stri_count_coll (opts_collator == NA not allowed)
*
* @version 0.3-1 (Marek Gagolewski, 2014-11-04)
* Issue #112: str_prepare_arg* retvals were not PROTECTed from gc
*/
SEXP stri_count_coll(SEXP str, SEXP pattern, SEXP opts_collator)
{
PROTECT(str = stri_prepare_arg_string(str, "str"));
PROTECT(pattern = stri_prepare_arg_string(pattern, "pattern"));
// call stri__ucol_open after prepare_arg:
// if prepare_arg had failed, we would have a mem leak
UCollator* collator = NULL;
collator = stri__ucol_open(opts_collator);
STRI__ERROR_HANDLER_BEGIN(2)
R_len_t vectorize_length = stri__recycling_rule(true, 2, LENGTH(str), LENGTH(pattern));
StriContainerUTF16 str_cont(str, vectorize_length);
StriContainerUStringSearch pattern_cont(pattern, vectorize_length, collator); // collator is not owned by pattern_cont
SEXP ret;
STRI__PROTECT(ret = Rf_allocVector(INTSXP, vectorize_length));
int* ret_tab = INTEGER(ret);
for (R_len_t i = pattern_cont.vectorize_init();
i != pattern_cont.vectorize_end();
i = pattern_cont.vectorize_next(i))
{
STRI__CONTINUE_ON_EMPTY_OR_NA_STR_PATTERN(str_cont, pattern_cont,
ret_tab[i] = NA_INTEGER,
ret_tab[i] = 0)
UStringSearch *matcher = pattern_cont.getMatcher(i, str_cont.get(i));
usearch_reset(matcher);
UErrorCode status = U_ZERO_ERROR;
R_len_t found = 0;
while (!U_FAILURE(status) && ((int)usearch_next(matcher, &status) != USEARCH_DONE))
++found;
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
ret_tab[i] = found;
}
if (collator) { ucol_close(collator); collator=NULL; }
STRI__UNPROTECT_ALL
return ret;
STRI__ERROR_HANDLER_END(
if (collator) ucol_close(collator);
)
}
/**
* Sets current (default) ICU charset
*
* If given charset is unavailable, an error is raised
*
* @param enc new charset (single string)
* @return nothing (\code{R_NilValue})
*
* @version 0.1-?? (Marek Gagolewski)
*
* @version 0.2-1 (Marek Gagolewski)
* use StriUcnv; make StriException-friendly
*
* @version 0.3-1 (Marek Gagolewski, 2014-11-04)
* Issue #112: str_prepare_arg* retvals were not PROTECTed from gc
*/
SEXP stri_enc_set(SEXP enc)
{
// here, the default encoding may not be requested:
const char* selected_enc
= stri__prepare_arg_enc(enc, "enc", false/*no default*/); /* this is R_alloc'ed */
STRI__ERROR_HANDLER_BEGIN(0)
StriUcnv uconv_obj(selected_enc);
// this will generate an error if selected_enc is not supported:
UConverter* uconv = uconv_obj.getConverter();
UErrorCode status = U_ZERO_ERROR;
// get "official" encoding name:
const char* name = ucnv_getName(uconv, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
ucnv_setDefaultName(name); // set as default
return R_NilValue;
STRI__ERROR_HANDLER_END({/* no special action on error */})
}
/**
* Convert character vector between marked encodings and the encoding provided
*
* @param str input character vector or list of raw vectors
* @param to target encoding, \code{NULL} or \code{""} for default enc
* @param to_raw single logical, should list of raw vectors be returned?
* @return a converted character vector or list of raw vectors
*
* @version 0.1-?? (Marek Gagolewski, 2013-11-12)
*
* @version 0.2-1 (Marek Gagolewski, 2014-03-28)
* use StriUcnv
*
* @version 0.2-1 (Marek Gagolewski, 2014-04-01)
* calc required buf size a priori
*
* @version 0.3-1 (Marek Gagolewski, 2014-11-04)
* Issue #112: str_prepare_arg* retvals were not PROTECTed from gc
*/
SEXP stri_encode_from_marked(SEXP str, SEXP to, SEXP to_raw)
{
PROTECT(str = stri_prepare_arg_string(str, "str"));
const char* selected_to = stri__prepare_arg_enc(to, "to", true); /* this is R_alloc'ed */
bool to_raw_logical = stri__prepare_arg_logical_1_notNA(to_raw, "to_raw");
STRI__ERROR_HANDLER_BEGIN(1)
R_len_t str_n = LENGTH(str);
StriContainerUTF16 str_cont(str, str_n);
// get the number of strings to convert; if == 0, then you know what's the result
if (str_n <= 0) return Rf_allocVector(to_raw_logical?VECSXP:STRSXP, 0);
// Open converters
StriUcnv ucnv(selected_to);
UConverter* uconv_to = ucnv.getConverter(true /*register_callbacks*/);
// Get target encoding mark
cetype_t encmark_to = to_raw_logical?CE_BYTES:ucnv.getCE();
// Prepare out val
SEXP ret;
STRI__PROTECT(ret = Rf_allocVector(to_raw_logical?VECSXP:STRSXP, str_n));
// calculate required buf size
R_len_t bufsize = 0;
for (R_len_t i=0; i<str_n; ++i) {
if (!str_cont.isNA(i) && str_cont.get(i).length() > bufsize)
bufsize = str_cont.get(i).length();
}
bufsize = UCNV_GET_MAX_BYTES_FOR_STRING(bufsize, ucnv_getMaxCharSize(uconv_to));
// "The calculated size is guaranteed to be sufficient for this conversion."
String8buf buf(bufsize);
for (R_len_t i=0; i<str_n; ++i) {
if (str_cont.isNA(i)) {
if (to_raw_logical) SET_VECTOR_ELT(ret, i, R_NilValue);
else SET_STRING_ELT(ret, i, NA_STRING);
continue;
}
R_len_t curn_tmp = str_cont.get(i).length();
const UChar* curs_tmp = str_cont.get(i).getBuffer(); // The buffer content is (probably) not NUL-terminated.
if (!curs_tmp)
throw StriException(MSG__INTERNAL_ERROR);
UErrorCode status = U_ZERO_ERROR;
ucnv_resetFromUnicode(uconv_to);
R_len_t bufneed = ucnv_fromUChars(uconv_to, buf.data(), buf.size(),
curs_tmp, curn_tmp, &status);
if (bufneed <= buf.size()) {
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
}
else {// larger buffer needed [this shouldn't happen?]
buf.resize(bufneed, false/*destroy contents*/);
status = U_ZERO_ERROR;
bufneed = ucnv_fromUChars(uconv_to, buf.data(), buf.size(),
curs_tmp, curn_tmp, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
}
if (to_raw_logical) {
SEXP outobj;
STRI__PROTECT(outobj = Rf_allocVector(RAWSXP, bufneed));
memcpy(RAW(outobj), buf.data(), (size_t)bufneed);
SET_VECTOR_ELT(ret, i, outobj);
STRI__UNPROTECT(1);
}
else {
SET_STRING_ELT(ret, i,
Rf_mkCharLenCE(buf.data(), bufneed, encmark_to));
}
}
STRI__UNPROTECT_ALL
return ret;
STRI__ERROR_HANDLER_END({/* nothing special on error */})
}
/**
* Convert case (TitleCase)
*
*
* @param str character vector
* @param opts_brkiter list
* @return character vector
*
* @version 0.4-1 (Marek Gagolewski, 2014-12-03)
* separated from stri_trans_casemap;
* use StriUBreakIterator
*/
SEXP stri_trans_totitle(SEXP str, SEXP opts_brkiter) {
StriBrkIterOptions opts_brkiter2(opts_brkiter, "word");
PROTECT(str = stri_prepare_arg_string(str, "str")); // prepare string argument
// version 0.2-1 - Does not work with ICU 4.8 (but we require ICU >= 50)
UCaseMap* ucasemap = NULL;
STRI__ERROR_HANDLER_BEGIN(1)
StriUBreakIterator brkiter(opts_brkiter2);
UErrorCode status = U_ZERO_ERROR;
ucasemap = ucasemap_open(brkiter.getLocale(), U_FOLD_CASE_DEFAULT, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
status = U_ZERO_ERROR;
ucasemap_setBreakIterator(ucasemap, brkiter.getIterator(), &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
brkiter.free(false);
// ucasemap_setOptions(ucasemap, U_TITLECASE_NO_LOWERCASE, &status); // to do?
// now briter is owned by ucasemap.
// it will be released on ucasemap_close
// (checked with ICU man & src code)
R_len_t str_n = LENGTH(str);
StriContainerUTF8 str_cont(str, str_n);
SEXP ret;
STRI__PROTECT(ret = Rf_allocVector(STRSXP, str_n));
// STEP 1.
// Estimate the required buffer length
// Notice: The resulting number of codepoints may be larger or smaller than
// the number before casefolding
R_len_t bufsize = str_cont.getMaxNumBytes();
bufsize += 10; // a small margin
String8buf buf(bufsize);
// STEP 2.
// Do case folding
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)) {
SET_STRING_ELT(ret, i, NA_STRING);
continue;
}
R_len_t str_cur_n = str_cont.get(i).length();
const char* str_cur_s = str_cont.get(i).c_str();
status = U_ZERO_ERROR;
int buf_need = ucasemap_utf8ToTitle(ucasemap, buf.data(), buf.size(),
(const char*)str_cur_s, str_cur_n, &status);
if (U_FAILURE(status)) {
buf.resize(buf_need, false/*destroy contents*/);
status = U_ZERO_ERROR;
buf_need = ucasemap_utf8ToTitle(ucasemap, buf.data(), buf.size(),
(const char*)str_cur_s, str_cur_n, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) // this shouldn't happen
// we do have the buffer size required to complete this op
}
SET_STRING_ELT(ret, i, Rf_mkCharLenCE(buf.data(), buf_need, CE_UTF8));
}
if (ucasemap) { ucasemap_close(ucasemap); ucasemap = NULL;}
STRI__UNPROTECT_ALL
return ret;
STRI__ERROR_HANDLER_END({
if (ucasemap) { ucasemap_close(ucasemap); ucasemap = NULL; }
})
}
/**
* Convert case (upper, lowercase)
*
*
* @param str character vector
* @param locale single string identifying
* the locale ("" or NULL for default locale)
* @return character vector
*
*
* @version 0.1-?? (Marek Gagolewski)
*
* @version 0.1-?? (Marek Gagolewski)
* use StriContainerUTF16
*
* @version 0.1-?? (Marek Gagolewski, 2013-06-16)
* make StriException-friendly
*
* @version 0.1-?? (Marek Gagolewski, 2013-11-19)
* use UCaseMap + StriContainerUTF8
* **THIS DOES NOT WORK WITH ICU 4.8**, we have to revert the changes
* ** BTW, since stringi_0.1-25 we require ICU>=50 **
*
* @version 0.2-1 (Marek Gagolewski, 2014-03-18)
* use UCaseMap + StriContainerUTF8
* (this is much faster for UTF-8 and slightly faster for 8bit enc)
* Estimates minimal buffer size.
*
* @version 0.3-1 (Marek Gagolewski, 2014-10-24)
* Use a custom BreakIterator with stri_trans_totitle
*
* @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, 2014-12-03)
* use StriUBreakIterator
*
* @version 0.6-1 (Marek Gagolewski, 2015-07-11)
* now this is an internal function
*/
SEXP stri_trans_casemap(SEXP str, int _type, SEXP locale)
{
if (_type < 1 || _type > 2) Rf_error(MSG__INCORRECT_INTERNAL_ARG);
const char* qloc = stri__prepare_arg_locale(locale, "locale", true); /* this is R_alloc'ed */
PROTECT(str = stri_prepare_arg_string(str, "str")); // prepare string argument
// version 0.2-1 - Does not work with ICU 4.8 (but we require ICU >= 50)
UCaseMap* ucasemap = NULL;
STRI__ERROR_HANDLER_BEGIN(1)
UErrorCode status = U_ZERO_ERROR;
ucasemap = ucasemap_open(qloc, U_FOLD_CASE_DEFAULT, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
R_len_t str_n = LENGTH(str);
StriContainerUTF8 str_cont(str, str_n);
SEXP ret;
STRI__PROTECT(ret = Rf_allocVector(STRSXP, str_n));
// STEP 1.
// Estimate the required buffer length
// Notice: The resulting number of codepoints may be larger or smaller than
// the number before casefolding
R_len_t bufsize = str_cont.getMaxNumBytes();
bufsize += 10; // a small margin
String8buf buf(bufsize);
// STEP 2.
// Do case folding
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)) {
SET_STRING_ELT(ret, i, NA_STRING);
continue;
}
R_len_t str_cur_n = str_cont.get(i).length();
const char* str_cur_s = str_cont.get(i).c_str();
status = U_ZERO_ERROR;
int buf_need;
if (_type == 1) buf_need = ucasemap_utf8ToLower(ucasemap,
buf.data(), buf.size(), (const char*)str_cur_s, str_cur_n, &status);
else buf_need = ucasemap_utf8ToUpper(ucasemap,
buf.data(), buf.size(), (const char*)str_cur_s, str_cur_n, &status);
if (U_FAILURE(status)) { /* retry */
buf.resize(buf_need, false/*destroy contents*/);
status = U_ZERO_ERROR;
if (_type == 1) buf_need = ucasemap_utf8ToLower(ucasemap,
buf.data(), buf.size(), (const char*)str_cur_s, str_cur_n, &status);
else buf_need = ucasemap_utf8ToUpper(ucasemap,
buf.data(), buf.size(), (const char*)str_cur_s, str_cur_n, &status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */}) // this shouldn't happen
// we do have the buffer size required to complete this op
}
SET_STRING_ELT(ret, i, Rf_mkCharLenCE(buf.data(), buf_need, CE_UTF8));
}
if (ucasemap) { ucasemap_close(ucasemap); ucasemap = NULL;}
STRI__UNPROTECT_ALL
return ret;
STRI__ERROR_HANDLER_END({
if (ucasemap) { ucasemap_close(ucasemap); ucasemap = 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; }
})
}
/**
* Format date-time objects
*
* @param time
* @param format
* @param tz
* @param locale
*
* @return character vector
*
* @version 0.5-1 (Marek Gagolewski, 2015-01-05)
*
* @version 0.5-1 (Marek Gagolewski, 2015-02-22)
* use tz
*/
SEXP stri_datetime_format(SEXP time, SEXP format, SEXP tz, SEXP locale) {
PROTECT(time = stri_prepare_arg_POSIXct(time, "time"));
const char* locale_val = stri__prepare_arg_locale(locale, "locale", true);
const char* format_val = stri__prepare_arg_string_1_notNA(format, "format");
// "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(1)
R_len_t vectorize_length = LENGTH(time);
StriContainerDouble time_cont(time, 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. */
SEXP ret;
STRI__PROTECT(ret = Rf_allocVector(STRSXP, vectorize_length));
for (R_len_t i=0; i<vectorize_length; ++i) {
if (time_cont.isNA(i)) {
SET_STRING_ELT(ret, i, NA_STRING);
continue;
}
status = U_ZERO_ERROR;
cal->setTime((UDate)(time_cont.get(i)*1000.0), status);
STRI__CHECKICUSTATUS_THROW(status, {/* do nothing special on err */})
FieldPosition pos;
UnicodeString out;
fmt->format(*cal, out, pos);
std::string s;
out.toUTF8String(s);
SET_STRING_ELT(ret, i, Rf_mkCharLenCE(s.c_str(), (int)s.length(), (cetype_t)CE_UTF8));
}
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; }
})
}
/**
* 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; }
})
}
