[write] skip blanks (#1111)

* [write] skip entirely blank cells

* [tests] update tests

* update NEWS

NEWS.md

@@ -1,5 +1,9 @@
 # openxlsx2 (development version)
 
+## New features
+
+* When writing a file with `na.strings = NULL`, the file will not contain any reference to the blank cell. Depending on the number of missings in a data set, this can reduce the file size significantly. [1111](https://github.com/JanMarvin/openxlsx2/pull/1111)
+
 ## Fixes
 
 * The integration of the shared formula feature in the previous release broke the silent extension of dims, if a single cell `dims` was provided for an `x` that was larger than a single cell in `wb_add_formula()`. [1131](https://github.com/JanMarvin/openxlsx2/pull/1131)

src/write_file.cpp

@@ -96,6 +96,26 @@ pugi::xml_document xml_sheet_data(Rcpp::DataFrame row_attr, Rcpp::DataFrame cc)
  }
  }
 
+ // update lastrow
+ lastrow = thisrow;
+
+ if ( // skip blank cells entirely
+ cc_c_s[i] == "" &&
+ cc_c_t[i] == "" &&
+ cc_c_cm[i] == "" &&
+ cc_c_ph[i] == "" &&
+ cc_c_vm[i] == "" &&
+ cc_v[i] == "" &&
+ cc_f[i] == "" &&
+ cc_f_t[i] == "" &&
+ cc_f_ref[i] == "" &&
+ cc_f_ca[i] == "" &&
+ cc_f_si[i] == "" &&
+ cc_is[i] == ""
+ ) {
+ continue;
+ }
+
  // create node <c>
  pugi::xml_node cell = row.append_child("c");
 
@@ -170,9 +190,6 @@ pugi::xml_document xml_sheet_data(Rcpp::DataFrame row_attr, Rcpp::DataFrame cc)
  cell.append_copy(is_node.first_child());
  }
  }
-
- // update lastrow
- lastrow = thisrow;
  }
 
  return doc;

tests/testthat/test-save.R

@@ -281,7 +281,7 @@ test_that("write cells without data", {
  expect_equal(exp, got)
 
  sheet <- paste0(tmp_dir, "/xl/worksheets/sheet1.xml")
- exp <- "<sheetData><row r=\"2\"><c r=\"B2\"/><c r=\"C2\"/></row><row r=\"3\"><c r=\"B3\"/><c r=\"C3\"/></row></sheetData>"
+ exp <- "<sheetData><row r=\"2\"/><row r=\"3\"/></sheetData>"
  got <- xml_node(sheet, "worksheet", "sheetData")
  expect_equal(exp, got)
 

tests/testthat/test-write.R

@@ -809,7 +809,7 @@ test_that("writing na.strings = NULL works", {
  write_xlsx(matrix(NA, 2, 2), tmp, na.strings = NULL)
  wb <- wb_load(tmp)
 
- exp <- ""
+ exp <- NA_character_
  got <- unique(wb$worksheets[[1]]$sheet_data$cc$v[3:6])
  expect_equal(exp, got)
 
@@ -1239,3 +1239,23 @@ test_that("sheet is a valid argument in write_xlsx", {
  wb2 <- write_xlsx(x = mtcars, sheet = "data")
  expect_equal(wb1$get_sheet_names(), wb2$get_sheet_names())
 })
+
+test_that("skipping entirely blank cells works", {
+
+ tp <- temp_xlsx()
+
+ mm <- matrix(1:9, 3, 3)
+ mm[diag(mm)] <- NA
+
+ wb1 <- write_xlsx(x = mm, file = tp, col_names = FALSE, na.strings = NULL)
+ cc1 <- wb1$worksheets[[1]]$sheet_data$cc
+ got1 <- cc1[cc1$r %in% c("A1", "B2", "C3"), ]
+
+ wb2 <- wb_load(tp)
+ cc2 <- wb2$worksheets[[1]]$sheet_data$cc
+ got2 <- cc2[cc2$r %in% c("A1", "B2", "C3"), ]
+
+ expect_equal(3, nrow(got1))
+ expect_equal(0, nrow(got2))
+
+})