Add more tests and benchmarks for the Markdown module

benchmark/CMakeLists.txt

@@ -36,6 +36,10 @@ if(SOURCEMETA_CORE_HTML)
   list(APPEND BENCHMARK_SOURCES html.cc)
 endif()
 
+if(SOURCEMETA_CORE_MARKDOWN)
+  list(APPEND BENCHMARK_SOURCES markdown.cc)
+endif()
+
 if(SOURCEMETA_CORE_GZIP)
   list(APPEND BENCHMARK_SOURCES gzip.cc)
 endif()
@@ -102,6 +106,11 @@ if(BENCHMARK_SOURCES)
       PRIVATE sourcemeta::core::html)
   endif()
 
+  if(SOURCEMETA_CORE_MARKDOWN)
+    target_link_libraries(sourcemeta_core_benchmark
+      PRIVATE sourcemeta::core::markdown)
+  endif()
+
   if(SOURCEMETA_CORE_GZIP)
     target_link_libraries(sourcemeta_core_benchmark
       PRIVATE sourcemeta::core::gzip sourcemeta::core::io)

benchmark/markdown.cc

@@ -0,0 +1,103 @@
+#include <benchmark/benchmark.h>
+
+#include <sourcemeta/core/markdown.h>
+
+#include <cstddef> // std::size_t
+#include <string>  // std::string, std::to_string
+
+static auto repeat(const std::string &pattern, const std::size_t count)
+    -> std::string {
+  std::string result;
+  result.reserve(pattern.size() * count);
+  for (std::size_t index = 0; index < count; ++index) {
+    result.append(pattern);
+  }
+
+  return result;
+}
+
+static auto increasing_backtick_runs(const std::size_t count) -> std::string {
+  std::string result;
+  for (std::size_t length = 1; length <= count; ++length) {
+    result.push_back('x');
+    result.append(length, '`');
+  }
+
+  return result;
+}
+
+static auto staircase_list(const std::size_t depth) -> std::string {
+  std::string result;
+  for (std::size_t level = 0; level < depth; ++level) {
+    result.append(level * 2, ' ');
+    result.append("- level\n");
+  }
+
+  return result;
+}
+
+// NOLINTNEXTLINE(readability-identifier-naming)
+static void Markdown_To_HTML_Realistic_Document(benchmark::State &state) {
+  std::string input;
+  for (std::size_t index = 0; index < 500; ++index) {
+    const auto number{std::to_string(index)};
+    input.append("# Section ").append(number).append("\n\n");
+    input.append("This paragraph has *emphasis*, **strong emphasis**, ")
+        .append("`inline code`, a [link](https://sourcemeta.com/")
+        .append(number)
+        .append(" \"Title\"), an autolink to www.sourcemeta.com, and ")
+        .append("~~deleted text~~ with an entity &copy; and a footnote[^")
+        .append(number)
+        .append("].\n\n");
+    input.append("> A block quote with a [reference link][ref")
+        .append(number)
+        .append("]\n> spanning two lines.\n\n");
+    input.append("- First item\n  - Nested *item*\n  - Another `item`\n"
+                 "- [x] Completed task\n- [ ] Pending task\n\n");
+    input.append("1. One\n2. Two\n3. Three\n\n");
+    input.append("| Name | Value | Notes |\n| :--- | ---: | :---: |\n"
+                 "| alpha | 1 | *first* |\n| beta | 2 | `second` |\n"
+                 "| gamma | 3 | [third](/third) |\n\n");
+    input.append("```cpp\nauto main() -> int {\n  return 0;\n}\n```\n\n");
+    input.append("    indented code line\n\n---\n\n");
+    input.append("[ref")
+        .append(number)
+        .append("]: https://sourcemeta.com/reference/")
+        .append(number)
+        .append("\n[^")
+        .append(number)
+        .append("]: Footnote ")
+        .append(number)
+        .append(".\n\n");
+  }
+
+  for (auto iteration : state) {
+    auto result{sourcemeta::core::markdown_to_html(input)};
+    benchmark::DoNotOptimize(result);
+  }
+}
+
+// Inputs that make naive delimiter, bracket, and container handling
+// quadratic or worse, which a linear parser renders almost instantly
+// NOLINTNEXTLINE(readability-identifier-naming)
+static void Markdown_To_HTML_Pathological(benchmark::State &state) {
+  const std::string input{
+      repeat("_x **y ", 5000) + "z" + repeat(" y** x_", 5000) + "\n\n" +
+      repeat("x_. ", 10000) + "\n\n" + repeat("__y ", 10000) + "\n\n" +
+      repeat("]b", 10000) + "\n\n" + repeat("![b", 10000) + "\n\n" +
+      repeat("_b* ", 10000) + "\n\n" + "x***y" + repeat("z* ", 10000) + "\n\n" +
+      repeat("[ b*", 10000) + "\n\n" + repeat("[ x](", 10000) + "\n\n" +
+      repeat("![", 5000) + "x" + repeat("]", 5000) + "\n\n" +
+      repeat("![x](<y", 5000) + "\n\nx" + repeat("<![CDATA[", 5000) + "\n\nx" +
+      repeat("<?", 5000) + "\n\n" + increasing_backtick_runs(2000) + "\n\n" +
+      repeat("> - ", 50) + "deep\n\n" + staircase_list(200) + "\n" +
+      "| a | b | c |\n|---|---|---|\n" + repeat("| x | y | z |\n", 5000)};
+
+  for (auto iteration : state) {
+    auto result{sourcemeta::core::markdown_to_html(input)};
+    benchmark::DoNotOptimize(result);
+  }
+}
+
+BENCHMARK(Markdown_To_HTML_Realistic_Document);
+BENCHMARK(Markdown_To_HTML_Pathological);

test/markdown/CMakeLists.txt

@@ -1,5 +1,28 @@
 sourcemeta_test(NAMESPACE sourcemeta PROJECT core NAME markdown
-  SOURCES markdown_test.cc)
+  SOURCES
+    markdown_autolinks_test.cc
+    markdown_blockquotes_test.cc
+    markdown_characters_test.cc
+    markdown_code_blocks_test.cc
+    markdown_code_spans_test.cc
+    markdown_concurrency_test.cc
+    markdown_emphasis_test.cc
+    markdown_entities_test.cc
+    markdown_escapes_test.cc
+    markdown_footnotes_test.cc
+    markdown_headings_test.cc
+    markdown_images_test.cc
+    markdown_line_breaks_test.cc
+    markdown_links_test.cc
+    markdown_lists_test.cc
+    markdown_paragraphs_test.cc
+    markdown_pathological_test.cc
+    markdown_safety_test.cc
+    markdown_strikethrough_test.cc
+    markdown_tables_test.cc
+    markdown_task_lists_test.cc
+    markdown_thematic_breaks_test.cc
+    markdown_unsafe_test.cc)
 
 target_link_libraries(sourcemeta_core_markdown_unit
   PRIVATE sourcemeta::core::markdown)