Import in-cell formatting (fixes #5)

The new column `character_formatted` is a list-column of data frames, one per
cell.  Each row is a substring, with the applicable formatting.  Where formats
are `NA`, the overall cell format applies.

NEWS.md

@@ -17,6 +17,7 @@
 * Provides a utility function `is_range()` to check whether formulas are simply
     ranges of cells.
 * Returns formatting of alignment and cell protection (#20).
+* Returns in-cell formatting of strings (#5)
 
 # tidyxl 0.2.3
 

R/tidy_xlsx.R

@@ -83,6 +83,8 @@
 #' * `numeric` The numeric value of a cell.
 #' * `date` The date value of a cell.
 #' * `character` The string value of a cell.
+#' * `character_formatted` A data frame of substrings and their individual
+#'     formatting.
 #' * `formula` The formula in a cell (see 'Details').
 #' * `is_array` Whether or not the formula is an array formula.
 #' * `formula_ref` The address of a range of cells group to which an array
@@ -140,6 +142,12 @@
 #'   the RGB values in a spreadsheet program (e.g. Excel, LibreOffice,
 #'   Gnumeric), and use the [grDevices::rgb()] function to convert these to a
 #'   hexadecimal string.
+#'
+#'   Strings can be formatted within a cell, so that a single cell can contain
+#'   substrings with different formatting.  This in-cell formatting is available
+#'   in the column `character_formatted`, which is a list-column of data frames.
+#'   Each row of each data frame describes a substring and its formatting.
+#'   When a particular format is `NA`, the overall cell format applies.
 #' }
 #'
 #' @export
@@ -171,6 +179,11 @@
 #' # the relevant indices, and then filter the cells by those indices.
 #' bold_indices <- which(x$formats$local$font$bold)
 #' Sheet1[Sheet1$local_format_id %in% bold_indices, ]
+#'
+#' # In-cell formatting is available in the `character_formatted` column as a
+#' # data frame, one row per substring.  When a particular format is `NA`, the
+#' # overall cell format applies.
+#' tidy_xlsx(examples)$data$Sheet1$character_formatted[77]
 #' }
 tidy_xlsx <- function(path, sheets = NA) {
   .Deprecated(msg = paste("'tidy_xlsx()' is deprecated.",

R/xlsx_cells.R

@@ -99,6 +99,12 @@
 #'   it as an index into the format structure.  E.g. to look up the font size,
 #'   `my_formats$local$font$size[local_format_id]`.  To see all available
 #'   formats, type `str(my_formats$local)`.
+#'
+#'   Strings can be formatted within a cell, so that a single cell can contain
+#'   substrings with different formatting.  This in-cell formatting is available
+#'   in the column `character_formatted`, which is a list-column of data frames.
+#'   Each row of each data frame describes a substring and its formatting.
+#'   When a particular format is `NA`, the overall cell format applies.
 #' }
 #'
 #' @export
@@ -124,6 +130,11 @@
 #' # the relevant indices, and then filter the cells by those indices.
 #' bold_indices <- which(formats$local$font$bold)
 #' Sheet1[Sheet1$local_format_id %in% bold_indices, ]
+#'
+#' # In-cell formatting is available in the `character_formatted` column as a
+#' # data frame, one row per substring.  When a particular format is `NA`, the
+#' # overall cell format applies.
+#' xlsx_cells(examples)$character_formatted[77]
 xlsx_cells <- function(path, sheets = NA) {
   path <- check_file(path)
   all_sheets <- utils_xlsx_sheet_files(path)

src/utils.h

@@ -3,6 +3,7 @@
 
 #include <Rcpp.h>
 #include "rapidxml.h"
+#include "xlsxstyles.h"
 #include <R_ext/GraphicsDevice.h> // Rf_ucstoutf8 is exported in R_ext/GraphicsDevice.h
 
 // Follow tidyverse/readxl
@@ -145,4 +146,122 @@ inline bool parseString(const rapidxml::xml_node<>* string, std::string& out) {
   return found;
 }
 
+// Return a dataframe, one row per substring, columns for formatting
+inline Rcpp::List parseFormattedString(
+    const rapidxml::xml_node<>* string,
+    xlsxstyles& styles) {
+  int n(0); // number of substrings
+  for (rapidxml::xml_node<>* node = string->first_node();
+       node != NULL;
+       node = node->next_sibling()) {
+    ++n;
+  }
+
+  std::vector<std::string> character;
+  std::vector<int> bold;
+  std::vector<int> italic;
+  std::vector<int> underline;
+  std::vector<double> size;
+  Rcpp::CharacterVector color_rgb(n, NA_STRING);
+  std::vector<int> color_theme;
+  std::vector<int> color_indexed;
+  Rcpp::CharacterVector font(n, NA_STRING);
+  std::vector<int> family;
+  Rcpp::CharacterVector scheme(n, NA_STRING);
+
+  int i(0); // ith substring
+  for (rapidxml::xml_node<>* node = string->first_node();
+       node != NULL;
+       node = node->next_sibling()) {
+    std::string node_name = node->name();
+    if (node_name == "t") {
+        character.push_back(node->value());
+        bold.push_back(NA_LOGICAL);
+        italic.push_back(NA_LOGICAL);
+        underline.push_back(NA_LOGICAL);
+        size.push_back(NA_REAL);
+        color_theme.push_back(NA_INTEGER);
+        color_indexed.push_back(NA_INTEGER);
+        family.push_back(NA_INTEGER);
+    } else {
+      character.push_back(node->first_node("t")->value());
+      rapidxml::xml_node<>* rPr = node->first_node("rPr");
+      if (rPr != NULL) {
+        bold.push_back(rPr->first_node("b") != NULL);
+        italic.push_back(rPr->first_node("i") != NULL);
+        underline.push_back(rPr->first_node("u") != NULL);
+        rapidxml::xml_node<>* sz = rPr->first_node("sz");
+        if (sz != NULL) {
+          size.push_back(strtod(sz->value(), NULL));
+        } else {
+          size.push_back(NA_REAL);
+        }
+        rapidxml::xml_node<>* color = rPr->first_node("color");
+        if (color != NULL) {
+          rapidxml::xml_attribute<>* color_attr = color->first_attribute();
+          std::string attr_name = color_attr->name();
+          if (attr_name == "rgb") {
+            color_rgb[i] = color_attr->value();
+            color_theme.push_back(NA_INTEGER);
+            color_indexed.push_back(NA_INTEGER);
+          } else if (attr_name == "theme") {
+            int theme_int = strtol(color_attr->value(), NULL, 10) + 1;
+            color_rgb[i] = styles.theme_[theme_int - 1];
+            color_theme.push_back(theme_int);
+            color_indexed.push_back(NA_INTEGER);
+          } else if (attr_name == "indexed") {
+            // I haven't been able to force this in real life
+            int indexed_int = strtol(color_attr->value(), NULL, 10) + 1;
+            color_rgb[i] = styles.indexed_[indexed_int - 1];
+            color_theme.push_back(NA_INTEGER);
+            color_indexed.push_back(indexed_int);
+          }
+        } else {
+          color_theme.push_back(NA_INTEGER);
+          color_indexed.push_back(NA_INTEGER);
+        }
+        rapidxml::xml_node<>* rFont = rPr->first_node("rFont");
+        if (rFont != NULL) {
+          font[i] = rFont->first_attribute("val")->value();
+        }
+        rapidxml::xml_node<>* family_node = rPr->first_node("family");
+        if (family_node != NULL) {
+          family.push_back(strtol(family_node->first_attribute("val")->value(), NULL, 10));
+        } else {
+          family.push_back(NA_INTEGER);
+        }
+        rapidxml::xml_node<>* scheme_node = rPr->first_node("scheme");
+        if (scheme_node != NULL) {
+          scheme[i] = scheme_node->first_attribute("val")->value();
+        }
+      } else {
+        bold.push_back(NA_LOGICAL);
+        italic.push_back(NA_LOGICAL);
+        underline.push_back(NA_LOGICAL);
+        size.push_back(NA_REAL);
+        color_theme.push_back(NA_INTEGER);
+        color_indexed.push_back(NA_INTEGER);
+        family.push_back(NA_INTEGER);
+      }
+    }
+    ++i;
+  }
+  Rcpp::List out = Rcpp::List::create(
+      Rcpp::_["character"] = character,
+      Rcpp::_["bold"] = bold,
+      Rcpp::_["italic"] = italic,
+      Rcpp::_["underline"] = underline,
+      Rcpp::_["size"] = size,
+      Rcpp::_["color_rgb"] = color_rgb,
+      Rcpp::_["color_theme"] = color_theme,
+      Rcpp::_["color_indexed"] = color_indexed,
+      Rcpp::_["font"] = font,
+      Rcpp::_["family"] = family,
+      Rcpp::_["scheme"] = scheme);
+  // Turn list of vectors into a data frame without checking anything
+  out.attr("class") = Rcpp::CharacterVector::create("tbl_df", "tbl", "data.frame");
+  out.attr("row.names") = Rcpp::IntegerVector::create(NA_INTEGER, -n); // Dunno how this works (the -n part)
+  return out;
+}
+
 #endif

src/xlsxbook.cpp

@@ -68,6 +68,9 @@ void xlsxbook::cacheStrings() {
     std::string out;
     parseString(string, out);    // missing strings are treated as empty ""
     strings_.push_back(out);
+
+    Rcpp::List out_df = parseFormattedString(string, styles_);
+    strings_formatted_.push_back(out_df);
   }
 }
 
@@ -147,6 +150,7 @@ void xlsxbook::initializeColumns() {
   formula_ref_     = CharacterVector(cellcount_, NA_STRING);
   formula_group_   = IntegerVector(cellcount_,   NA_INTEGER);
   comment_         = CharacterVector(cellcount_, NA_STRING);
+  character_formatted_ = List(cellcount_);
   height_          = NumericVector(cellcount_,   NA_REAL);
   width_           = NumericVector(cellcount_,   NA_REAL);
   style_format_    = CharacterVector(cellcount_, NA_STRING);
@@ -200,7 +204,7 @@ void xlsxbook::cacheInformation() {
   /*     _["style_format"] = style_format_, */
   /*     _["local_format_id"] = local_format_id_); */
 
-  information_ = List(20);
+  information_ = List(21);
   information_[0] = sheet_;
   information_[1] = address_;
   information_[2] = row_;
@@ -212,17 +216,18 @@ void xlsxbook::cacheInformation() {
   information_[8] = numeric_;
   information_[9] = date_;
   information_[10] = character_;
-  information_[11] = formula_;
-  information_[12] = is_array_;
-  information_[13] = formula_ref_;
-  information_[14] = formula_group_;
-  information_[15] = comment_;
-  information_[16] = height_;
-  information_[17] = width_;
-  information_[18] = style_format_;
-  information_[19] = local_format_id_;
-
-  std::vector<std::string> names(20);
+  information_[11] = character_formatted_;
+  information_[12] = formula_;
+  information_[13] = is_array_;
+  information_[14] = formula_ref_;
+  information_[15] = formula_group_;
+  information_[16] = comment_;
+  information_[17] = height_;
+  information_[18] = width_;
+  information_[19] = style_format_;
+  information_[20] = local_format_id_;
+
+  std::vector<std::string> names(21);
   names[0]  = "sheet";
   names[1]  = "address";
   names[2]  = "row";
@@ -234,15 +239,16 @@ void xlsxbook::cacheInformation() {
   names[8]  = "numeric";
   names[9]  = "date";
   names[10] = "character";
-  names[11] = "formula";
-  names[12] = "is_array";
-  names[13] = "formula_ref";
-  names[14] = "formula_group";
-  names[15] = "comment";
-  names[16] = "height";
-  names[17] = "width";
-  names[18] = "style_format";
-  names[19] = "local_format_id";
+  names[11] = "character_formatted";
+  names[12] = "formula";
+  names[13] = "is_array";
+  names[14] = "formula_ref";
+  names[15] = "formula_group";
+  names[16] = "comment";
+  names[17] = "height";
+  names[18] = "width";
+  names[19] = "style_format";
+  names[20] = "local_format_id";
 
   information_.attr("names") = names;
 

src/xlsxbook.h

@@ -15,7 +15,8 @@ class xlsxbook {
     Rcpp::CharacterVector sheet_names_;    // worksheet names
     Rcpp::CharacterVector comments_paths_; // comments files
     std::vector<std::string> strings_;     // strings table
-
+    std::vector<Rcpp::List> strings_formatted_; // strings with inline formatting
+                                                // list of data frames
     xlsxstyles styles_;
 
     int dateSystem_; // 1900 or 1904
@@ -44,6 +45,7 @@ class xlsxbook {
     Rcpp::CharacterVector formula_ref_;   // If present
     Rcpp::IntegerVector   formula_group_; // If present
     Rcpp::CharacterVector comment_;
+    Rcpp::List            character_formatted_; // data frame
     Rcpp::NumericVector   height_;          // Provided to cell constructor
     Rcpp::NumericVector   width_;           // Provided to cell constructor
     Rcpp::CharacterVector style_format_;    // cellXfs xfId links to cellStyleXfs entry

src/xlsxcell.cpp

@@ -138,6 +138,7 @@ void xlsxcell::cacheValue(
     // into the string table.
     book.data_type_[i] = "character";
     SET_STRING_ELT(book.character_, i, Rf_mkCharCE(book.strings_[strtol(vvalue.c_str(), NULL, 10)].c_str(), CE_UTF8));
+    book.character_formatted_[i] = book.strings_formatted_[strtol(vvalue.c_str(), NULL, 10)];
     return;
   } else if (tvalue == "str") {
     // Formula, which could have evaluated to anything, so only a string is safe