merge Merge branch 'main' of https://github.com/Alex-Monahan/duckdb-gsheets

Author: Alex-Monahan

docs/pages/index.md

@@ -143,7 +143,6 @@ This will also require an additional API request approximately every 30 minutes.
 
 - DuckDB WASM is not (yet) supported.
 - Google Sheets has a limit of 10,000,000 cells per spreadsheet.
-- Writing data to a sheet starting from a cell other than A1 is not yet supported.
 - Sheets must already exist to COPY TO them.
 
 ## Support 

src/gsheets_auth.cpp

@@ -93,6 +93,9 @@ namespace duckdb
         auto filepath = (input.options.find(filepath_key)->second).ToString();
 
         std::ifstream ifs(filepath);
+        if (!ifs.is_open()) {
+            throw IOException("Could not open JSON key file at: " + filepath);
+        }
         json credentials_file = json::parse(ifs);
         std::string email = credentials_file["client_email"].get<std::string>();
         std::string secret = credentials_file["private_key"].get<std::string>();

src/gsheets_read.cpp

@@ -218,16 +218,17 @@ unique_ptr<FunctionData> ReadSheetBind(ClientContext &context, TableFunctionBind
     json cleanJson = parseJson(bind_data->response);
     SheetData sheet_data = getSheetData(cleanJson);
 
-    // Prefering early return style to reduce nesting
+    // Throw error ourselves to give user a better error message
     if (sheet_data.values.empty()) {
-        return bind_data;
+        throw duckdb::InvalidInputException("Sheet %s is empty", sheet_name);
     }
+
     idx_t start_index = header ? 1 : 0;
-    if (start_index >= sheet_data.values.size()) {
-        return bind_data;
-    }
 
-    const auto& first_data_row = sheet_data.values[start_index];
+    // Use empty row for first row if results are header-only
+    const std::vector<string> empty_row = {};
+    const auto& first_data_row = start_index >= sheet_data.values.size() ? empty_row : sheet_data.values[start_index];
+
     // If we have a header, we want the width of the result to be the max of:
     //      the width of the header row
     //      or the width of the first row of data

src/gsheets_utils.cpp

@@ -111,18 +111,18 @@ json parseJson(const std::string& json_str) {
 
 SheetData getSheetData(const json& j) {
     SheetData result;
-    if (j.contains("range") && j.contains("majorDimension") && j.contains("values")) {
+    if (j.contains("range") && j.contains("majorDimension")) {
         result.range = j["range"].get<std::string>();
         result.majorDimension = j["majorDimension"].get<std::string>();
-        result.values = j["values"].get<std::vector<std::vector<std::string>>>();
+        // NOTE: no need to hard fail on empty sheet values. We can handle this more naturally further down the call chain.
+        //       Just default to empty 2D vec.
+        result.values = j.contains("values") ? j["values"].get<std::vector<std::vector<std::string>>>() : std::vector<std::vector<std::string>>{};
     } else if (j.contains("error")) {
         string message = j["error"]["message"].get<std::string>();
-            int code = j["error"]["code"].get<int>();
-            throw std::runtime_error("Google Sheets API error: " + std::to_string(code) + " - " + message);
-        } else {
-        std::cerr << "JSON does not contain expected fields" << std::endl;
-        std::cerr << "Raw JSON string: " << j.dump() << std::endl;
-        throw;
+        int code = j["error"]["code"].get<int>();
+        throw std::runtime_error("Google Sheets API error: " + std::to_string(code) + " - " + message);
+    } else {
+        throw duckdb::InvalidInputException("JSON does not contain expected fields.\nRaw JSON string: %s", j.dump());
     }
     return result;
 }

test/sql/read_gsheet.test

@@ -109,7 +109,11 @@ NULL	NULL
 Archie	99.0
 
 # Test single value from range
-# NOTE: *must* use `header=false` to avoid uncaught bind error
+query I
+FROM read_gsheet('11QdEasMWbETbFVxry-SsD8jVcdYIT1zBQszcF84MdE8', sheet='Sheet1', range='A2');
+----
+
+# Test single value from range (no headers)
 query I
 FROM read_gsheet('11QdEasMWbETbFVxry-SsD8jVcdYIT1zBQszcF84MdE8', sheet='Sheet1', range='A2', header=false);
 ----
@@ -168,6 +172,17 @@ FROM read_gsheet('https://docs.google.com/spreadsheets/d/11QdEasMWbETbFVxry-SsD8
 woot	blah	NULL	should get this!
 more wooting	more blah	NULL	should get this!
 
+# Header only results
+query II
+FROM read_gsheet('11QdEasMWbETbFVxry-SsD8jVcdYIT1zBQszcF84MdE8', sheet='62-header_only');
+----
+
+# Empty sheet
+statement error
+FROM read_gsheet('11QdEasMWbETbFVxry-SsD8jVcdYIT1zBQszcF84MdE8', sheet='62-empty');
+----
+Invalid Input Error: Sheet 62-empty is empty
+
 # Drop the secret
 statement ok
 drop secret test_secret;