encoding should probably be the same as the SAS encoding

JanMarvin · Dec 2, 2023 · 32beda1 · 32beda1
1 parent 5dbc8bc
commit 32beda1
Show file tree

Hide file tree

Showing 5 changed files with 33 additions and 25 deletions.
diff --git a/R/RcppExports.R b/R/RcppExports.R
@@ -27,7 +27,7 @@ readsas <- function(filePath, debug, selectrows_, selectcols_, empty_to_na, temp
 #' @param dateval timestamp
 #' @keywords internal
 #' @noRd
-writesas <- function(filePath, dat, compress, debug, bit32, headersize, pagesize, dateval) {
-    invisible(.Call(`_readsas_writesas`, filePath, dat, compress, debug, bit32, headersize, pagesize, dateval))
+writesas <- function(filePath, dat, compress, debug, bit32, headersize, pagesize, dateval, encoding32) {
+    invisible(.Call(`_readsas_writesas`, filePath, dat, compress, debug, bit32, headersize, pagesize, dateval, encoding32))
 }
 
diff --git a/R/writesas.R b/R/writesas.R
@@ -9,12 +9,13 @@
 #'@param bit32 write 32bit file
 #'@param varlabels optional variable labels
 #'@param size optional header/pagesize
+#'@param encoding encoding 62 = windows, 20 = utf
 #'
 #'@useDynLib readsas, .registration=TRUE
 #'
 #'@export
 write.sas <- function(dat, filepath, compress = 0, debug = FALSE, bit32 = FALSE,
-                      varlabels, size) {
+                      varlabels, size, encoding = 20) {
 
   filepath <- path.expand(filepath)
 
@@ -96,16 +97,15 @@ write.sas <- function(dat, filepath, compress = 0, debug = FALSE, bit32 = FALSE,
   # for numerics
   # formats <- rep("BEST", ncol(dat))
 
-  attr(dat, "vartypes") <- as.integer(vartypes)
-  attr(dat, "colwidth") <- as.integer(colwidth)
-  attr(dat, "formats")  <- formats
-  attr(dat, "width")    <- width
-  attr(dat, "decim")    <- decim
-  attr(dat, "labels")   <- labels
+  attr(dat, "vartypes")  <- as.integer(vartypes)
+  attr(dat, "colwidth")  <- as.integer(colwidth)
+  attr(dat, "formats")   <- formats
+  attr(dat, "width")     <- width
+  attr(dat, "decim")     <- decim
+  attr(dat, "labels")    <- labels
   attr(dat, "varlabels") <- varlabels
 
-
   writesas(filepath, dat, compress = 0, debug = debug, bit32 = bit32,
-           headersize = size[1], pagesize = size[2], dateval = as_datetime(Sys.time()))
-
+           headersize = size[1], pagesize = size[2],
+           dateval = as_datetime(Sys.time()), encoding32 = encoding)
 }
diff --git a/man/write.sas.Rd b/man/write.sas.Rd
diff --git a/src/RcppExports.cpp b/src/RcppExports.cpp
@@ -27,8 +27,8 @@ BEGIN_RCPP
 END_RCPP
 }
 // writesas
-void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress, bool debug, bool bit32, int32_t headersize, int32_t pagesize, double dateval);
-RcppExport SEXP _readsas_writesas(SEXP filePathSEXP, SEXP datSEXP, SEXP compressSEXP, SEXP debugSEXP, SEXP bit32SEXP, SEXP headersizeSEXP, SEXP pagesizeSEXP, SEXP datevalSEXP) {
+void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress, bool debug, bool bit32, int32_t headersize, int32_t pagesize, double dateval, int32_t encoding32);
+RcppExport SEXP _readsas_writesas(SEXP filePathSEXP, SEXP datSEXP, SEXP compressSEXP, SEXP debugSEXP, SEXP bit32SEXP, SEXP headersizeSEXP, SEXP pagesizeSEXP, SEXP datevalSEXP, SEXP encoding32SEXP) {
 BEGIN_RCPP
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< const char * >::type filePath(filePathSEXP);
@@ -39,14 +39,15 @@ BEGIN_RCPP
     Rcpp::traits::input_parameter< int32_t >::type headersize(headersizeSEXP);
     Rcpp::traits::input_parameter< int32_t >::type pagesize(pagesizeSEXP);
     Rcpp::traits::input_parameter< double >::type dateval(datevalSEXP);
-    writesas(filePath, dat, compress, debug, bit32, headersize, pagesize, dateval);
+    Rcpp::traits::input_parameter< int32_t >::type encoding32(encoding32SEXP);
+    writesas(filePath, dat, compress, debug, bit32, headersize, pagesize, dateval, encoding32);
     return R_NilValue;
 END_RCPP
 }
 
 static const R_CallMethodDef CallEntries[] = {
     {"_readsas_readsas", (DL_FUNC) &_readsas_readsas, 6},
-    {"_readsas_writesas", (DL_FUNC) &_readsas_writesas, 8},
+    {"_readsas_writesas", (DL_FUNC) &_readsas_writesas, 9},
     {NULL, NULL, 0}
 };
 

diff --git a/src/writesas.cpp b/src/writesas.cpp
@@ -41,7 +41,9 @@ using namespace Rcpp;
 // [[Rcpp::export]]
 void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
               bool debug, bool bit32, int32_t headersize, int32_t pagesize,
-              double dateval) {
+              double dateval, int32_t encoding32) {
+
+  int8_t encoding = (int8_t)encoding32;
 
   uint32_t k = dat.size();
   uint64_t n = dat.nrows();
@@ -149,7 +151,7 @@ void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
 
     double created = dateval, created2 = 0;   // 8
     double modified = dateval, modified2 = 0; // 16
-    double thrdts = 0;
+    double thrdts = dateval;
 
     // possibly make headersize and pagesize variable
     // int32_t headersize = 65536;
@@ -183,7 +185,7 @@ void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
     if (bit32 == 1) U64_BYTE_CHECKER_VALUE = 50;
     if (U64_BYTE_CHECKER_VALUE == 51) ALIGN_2_VALUE = 4;
 
-    pkt2 = 20, pkt3 = 0;
+    pkt2 = 34, pkt3 = 0;
 
     writebin(ALIGN_1_CHECKER_VALUE, sas, swapit);
     writebin(pkt2, sas, swapit); // 34
@@ -194,8 +196,8 @@ void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
     int8_t ENDIANNESS = 1; // 0 is swapit = 1
     uint8_t PLATFORM = 49; // (1) 49 Unix (2) 50 Win
 
-    pkt1 = 0, pkt3 = 0;
-    // if (bit32 == 1) pkt1 = 50;
+    pkt1 = 51, pkt3 = 2;
+    if (bit32 == 1) pkt1 = 50;
     writebin(pkt1, sas, swapit); // 51
     writebin(ENDIANNESS, sas, swapit);
     writebin(pkt3, sas, swapit); // 2
@@ -258,7 +260,7 @@ void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
     writebin(pkt4, sas, swapit);
 
     // packet 10 -------------------------------------- //
-    int8_t encoding = 20; // utf8
+    // int8_t encoding = 20; // utf8
     pkt1 = 51, pkt2 = 0, pkt3 = 0, pkt4 = 20;
     if (bit32 == 1) pkt1 = 50;
     writebin(pkt1, sas, swapit); // 51
@@ -341,10 +343,12 @@ void writesas(const char * filePath, Rcpp::DataFrame dat, uint8_t compress,
 
     // large unk
     // something related to file password?
-    uint32_t pktu32 = 1157289805;
+    // 0000 and 4 byte from date
+    uint32_t pktu32 = 0;
+    std::memcpy(&pktu32, &created, sizeof(int32_t));
     writebin(pktu32, sas, swapit);
 
-    pktu32 = 563452161;
+    pktu32 = 545930268;
     // three identical smaller unks
     // required so that the file is identified as SAS file
     writebin(pktu32, sas, swapit);