Account for rich text strings in BIFF5 files

Fixes #611 See libxls/libxls@b6d9d87
tidyverse · Mar 19, 2022 · a019b23 · a019b23
1 parent d9ea739
commit a019b23
Show file tree

Hide file tree

Showing 5 changed files with 35 additions and 3 deletions.
diff --git a/debug/debug.R b/debug/debug.R
@@ -1,7 +1,20 @@
 devtools::clean_dll()
 devtools::load_all()
 
-files <- fs::dir_ls("investigations/Data", recurse = TRUE, glob = "*.XLS")
-#files
-read_xls(files[[1]])
+# success <- read_excel(
+#   "investigations/sample_data/success.xls",
+#   col_names = FALSE,
+#   range = "A1"
+# )
+failure <- read_excel("investigations/sample_data/failure.xls", col_names=F)
 
+# cell_is_readable
+# cell->id is 214 for the failure
+# 214 in hexadecimal is 0xD6
+# in xlsstruct.h:
+# #define XLS_RECORD_RSTRING      0x00D6
+
+# cell->id is 516 for the success
+# 516 in hexadecimal is 0x204
+# in xlsstruct.h:
+# #define XLS_RECORD_LABEL        0x0204
diff --git a/src/XlsCell.h b/src/XlsCell.h
@@ -130,6 +130,7 @@ class XlsCell {
     switch(cell_->id) {
     case XLS_RECORD_LABELSST:
     case XLS_RECORD_LABEL:
+    case XLS_RECORD_RSTRING:
     {
       std::string s = cell_->str == NULL ? "" : cell_->str;
       ct = na.contains(s, trimWs) ? CELL_BLANK : CELL_TEXT;

diff --git a/src/XlsCellSet.h b/src/XlsCellSet.h
@@ -152,6 +152,7 @@ class XlsCellSet {
   bool cell_is_readable(const xls::xlsCell* cell) {
     return cell && (
         cell->id == XLS_RECORD_MULRK ||
+        cell->id == XLS_RECORD_RSTRING ||
         cell->id == XLS_RECORD_NUMBER ||
         cell->id == XLS_RECORD_RK ||
         cell->id == XLS_RECORD_LABELSST ||

diff --git a/tests/testthat/sheets/biff5-rich-text-string.xls b/tests/testthat/sheets/biff5-rich-text-string.xls
diff --git a/tests/testthat/test-compatibility.R b/tests/testthat/test-compatibility.R
@@ -63,3 +63,20 @@ test_that("we can read LAPD arrest sheets", {
   expect_match(lapd$ARR_LOC[9], "HOLLYWOOD")
   expect_identical(lapd$CHG_DESC[27], "EX CON W/ A GUN")
 })
+
+# https://github.com/tidyverse/readxl/issues/611
+# xls file produced by ABBYY FineReader (OCR of PDFs)
+# inspired libxls to add support for rich-text strings in BIFF5
+# https://github.com/libxls/libxls/commit/b6d9d872756f69780b743dbaec9cd2ec30c37740
+test_that("we can read xls from ABBYY FineReader", {
+  expect_error_free(
+    abbyy <- read_excel(
+      test_sheet("biff5-rich-text-string.xls"),
+      col_names = FALSE,
+      n_max = 1
+    )
+  )
+  expect_equal(nrow(abbyy), 1)
+  expect_equal(ncol(abbyy), 1)
+  expect_match(abbyy[[1,1]], "^ELECTORAL")
+})