Spreadsheet: Make the CSV reader more lenient

This adds an option "Lenient" that makes the reader conform to what
appears to be the norm in spreadsheet-land:
- Treat missing values as empty ones
- Update previously read rows if another row with more columns are seen
  afterwards

3 files changed, 28 insertions, 5 deletions

diff --git a/Userland/Applications/Spreadsheet/ImportDialog.cpp b/Userland/Applications/Spreadsheet/ImportDialog.cpp
index d607a1e4d8..3d920c3357 100644
--- a/Userland/Applications/Spreadsheet/ImportDialog.cpp
+++ b/Userland/Applications/Spreadsheet/ImportDialog.cpp
@@ -158,7 +158,7 @@ auto CSVImportDialogPage::make_reader() -> Optional<Reader::XSV>
         quote_escape,
     };
 
-    auto behaviours = Reader::default_behaviours();
+    auto behaviours = Reader::default_behaviours() | Reader::ParserBehaviour::Lenient;
 
     if (should_read_headers)
         behaviours = behaviours | Reader::ParserBehaviour::ReadHeaders;
diff --git a/Userland/Applications/Spreadsheet/Readers/XSV.cpp b/Userland/Applications/Spreadsheet/Readers/XSV.cpp
index 8ed6ce9dac..ca9c3eb4a1 100644
--- a/Userland/Applications/Spreadsheet/Readers/XSV.cpp
+++ b/Userland/Applications/Spreadsheet/Readers/XSV.cpp
@@ -103,10 +103,29 @@ Vector<XSV::Field> XSV::read_row(bool header_row)
         }
     }
 
-    if (!header_row && (m_behaviours & ParserBehaviour::ReadHeaders) != ParserBehaviour::None && row.size() != m_names.size())
-        set_error(ReadError::NonConformingColumnCount);
-    else if (!header_row && !has_explicit_headers() && !m_rows.is_empty() && m_rows.first().size() != row.size())
-        set_error(ReadError::NonConformingColumnCount);
+    auto is_lenient = (m_behaviours & ParserBehaviour::Lenient) != ParserBehaviour::None;
+    if (is_lenient) {
+        if (m_rows.is_empty())
+            return row;
+
+        auto& last_row = m_rows.last();
+        if (row.size() < last_row.size()) {
+            if (!m_names.is_empty())
+                row.resize(m_names.size());
+            else
+                row.resize(last_row.size());
+        } else if (row.size() > last_row.size()) {
+            auto new_size = row.size();
+            for (auto& row : m_rows)
+                row.resize(new_size);
+        }
+    } else {
+        auto should_read_headers = (m_behaviours & ParserBehaviour::ReadHeaders) != ParserBehaviour::None;
+        if (!header_row && should_read_headers && row.size() != m_names.size())
+            set_error(ReadError::NonConformingColumnCount);
+        else if (!header_row && !has_explicit_headers() && !m_rows.is_empty() && m_rows.first().size() != row.size())
+            set_error(ReadError::NonConformingColumnCount);
+    }
 
     return row;
 }
diff --git a/Userland/Applications/Spreadsheet/Readers/XSV.h b/Userland/Applications/Spreadsheet/Readers/XSV.h
index 309e0ec384..444c81992d 100644
--- a/Userland/Applications/Spreadsheet/Readers/XSV.h
+++ b/Userland/Applications/Spreadsheet/Readers/XSV.h
@@ -41,6 +41,10 @@ enum class ParserBehaviour : u32 {
     TrimLeadingFieldSpaces = ReadHeaders << 2,
     TrimTrailingFieldSpaces = ReadHeaders << 3,
     QuoteOnlyInFieldStart = ReadHeaders << 4,
+    Lenient = ReadHeaders << 5, // This is the typical "spreadsheet import" behavior
+                                // Currently, it:
+                                // - fills in missing fields with empty values
+                                // - updates previous rows with extra columns
 };
 
 ParserBehaviour operator&(ParserBehaviour left, ParserBehaviour right);