More unit tests

Author: vincentlaucsb

.github/workflows/codeql.yml

@@ -47,6 +47,9 @@ jobs:
         cmake .. \
           -DCMAKE_BUILD_TYPE=Release \
           -DCSV_CXX_STANDARD=20 \
+          -DCSV_BUILD_PROGRAMS=OFF \
+          -DCSV_BUILD_TESTS=OFF \
+          -DBUILD_PYTHON=OFF \
           -DCMAKE_CXX_COMPILER=g++ \
           -DCMAKE_C_COMPILER=gcc
     

CMakeLists.txt

@@ -104,6 +104,9 @@ if (CSV_BUILD_PROGRAMS)
     add_subdirectory("programs")
 endif()
 
+## Tests
+option(CSV_BUILD_TESTS "Allow to disable building of tests" ON)
+
 ## Developer settings
 if (CSV_DEVELOPER)
     # Allow for performance profiling
@@ -180,6 +183,8 @@ if (CSV_DEVELOPER)
     endif()
 
     ## Tests
-    enable_testing()
-    add_subdirectory("tests")
+    if (CSV_BUILD_TESTS)
+      enable_testing()
+      add_subdirectory("tests")
+    endif()
 endif()

include/internal/basic_csv_parser.cpp

@@ -11,6 +11,7 @@
 
 namespace csv {
     namespace internals {
+#if defined(__EMSCRIPTEN__)
         // Opens the file and delegates to the template overload to avoid duplicating the read/resize logic.
         CSV_INLINE std::string get_csv_head_stream(csv::string_view filename) {
             std::ifstream infile(std::string(filename), std::ios::binary);
@@ -19,6 +20,7 @@ namespace csv {
             }
             return get_csv_head_stream(infile);
         }
+#endif
 
 #if !defined(__EMSCRIPTEN__)
         CSV_INLINE std::pair<std::string, size_t> get_csv_head_mmap(csv::string_view filename) {

include/internal/basic_csv_parser.hpp

@@ -123,6 +123,11 @@ namespace csv {
             return buf;
         }
 
+    #if defined(__EMSCRIPTEN__)
+        /** Open a file-backed source and read the first 500KB through the stream path. */
+        CSV_INLINE std::string get_csv_head_stream(csv::string_view filename);
+    #endif
+
     #if !defined(__EMSCRIPTEN__)
         /** Read the first 500KB from a filename using mmap.
          *  Also returns the total file size so callers avoid a second mmap open.

tests/test_error_handling.cpp

@@ -117,6 +117,53 @@ TEST_CASE("Worker thread exceptions propagate to main thread", "[error_handling]
     }
 }
 
+/* Ensure reading empty CSVs does not cause errors.
+ *
+ * Reported in:
+ *  - https://github.com/vincentlaucsb/csv-parser/issues/116
+ *  - https://github.com/vincentlaucsb/csv-parser/issues/121
+ *  - empty mmap EOF guard in issue #267
+ */
+TEST_CASE("Empty CSV does not crash parser entry points", "[error_handling][empty_csv]") {
+    SECTION("Stream path handles empty inputs and header-only inputs") {
+        auto csv_string = GENERATE(as<std::string>{},
+            "A,B,C,D\r\n", // Header row only
+            ""             // No content
+        );
+
+        std::stringstream source(csv_string);
+        CSVReader reader(source);
+        REQUIRE(reader.empty());
+
+        for (auto& row : reader) {
+            (void)row;
+        }
+
+        REQUIRE(reader.n_rows() == 0);
+    }
+
+#ifndef __EMSCRIPTEN__
+    SECTION("Filename path reports a zero-byte file as an open failure") {
+        bool caught = false;
+        std::string error_message;
+
+        try {
+            CSVReader reader("./tests/data/fake_data/empty.csv");
+            for (auto& row : reader) {
+                (void)row;
+            }
+        }
+        catch (const std::exception& e) {
+            caught = true;
+            error_message = e.what();
+        }
+
+        REQUIRE(caught);
+        REQUIRE(error_message == "Cannot open file ./tests/data/fake_data/empty.csv");
+    }
+#endif
+}
+
 #ifndef __EMSCRIPTEN__
 TEST_CASE("Fields at chunk boundaries are not corrupted", "[chunking][data_integrity]") {
     SECTION("Large file with known values around chunk boundary") {

tests/test_read_csv.cpp

@@ -681,32 +681,6 @@ TEST_CASE("Single Column CSV", "[read_single_col_direct]") {
     }
 }
 
-/* Ensure reading empty CSVs does not cause errors
- * 
- * Reported in:
- *  - https://github.com/vincentlaucsb/csv-parser/issues/116
- *  - https://github.com/vincentlaucsb/csv-parser/issues/121
- */
-TEST_CASE("Empty CSV", "[read_empty_csv]") {
-    auto csv_string = GENERATE(as<std::string>{},
-        "A,B,C,D\r\n", // Header row only
-        ""             // No content
-    );
-
-    SECTION("Read Empty CSV") {
-        std::stringstream source(csv_string);
-        CSVReader reader(source);
-        REQUIRE(reader.empty());
-
-        for (auto& row : reader) {
-            (void)row;
-        }
-
-        // We want to make sure that no exceptions are thrown
-        REQUIRE(reader.n_rows() == 0);
-    }
-}
-
 // Regression test for issue #149: trailing newline at EOF must not produce a spurious empty row
 TEST_CASE("Trailing newline at EOF (stringstream)", "[trailing_newline_stringstream]") {
     auto check = [](const std::string& csv_text) {