build!: raise the standard to C++23, and spend it

.github/config/conan/profiles/android.jinja

@@ -24,7 +24,7 @@ arch={{arch}}
 build_type=Release
 compiler=clang
 compiler.version=17
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=c++_shared
 
 [conf]

.github/config/conan/profiles/apple.jinja

@@ -19,7 +19,7 @@ os.version={{os_version}}
 build_type=Release
 compiler=apple-clang
 compiler.version=17
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=libc++
 
 [conf]

.github/config/conan/profiles/emscripten-wasm

@@ -17,7 +17,7 @@ build_type=Release
 compiler=emcc
 compiler.version={{emsdk_version}}
 compiler.libcxx=libc++
-compiler.cppstd=20
+compiler.cppstd=23
 
 [options]
 # No sockets and no threads in a browser; the CLI has no meaning here either.

.github/config/conan/profiles/macos-15-armv8-clang-14

@@ -3,7 +3,7 @@ arch=armv8
 build_type=Release
 compiler=apple-clang
 compiler.version=14
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=libc++
 os=Macos
 

.github/config/conan/profiles/macos-26-armv8-clang-14

@@ -3,7 +3,7 @@ arch=armv8
 build_type=Release
 compiler=apple-clang
 compiler.version=14
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=libc++
 os=Macos
 

.github/config/conan/profiles/ubuntu-24.04-clang-18

@@ -3,7 +3,7 @@ arch=x86_64
 build_type=Release
 compiler=clang
 compiler.version=18
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=libstdc++11
 os=Linux
 

.github/config/conan/profiles/ubuntu-24.04-gcc-14

@@ -3,7 +3,7 @@ arch=x86_64
 build_type=Release
 compiler=gcc
 compiler.version=14
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.libcxx=libstdc++11
 os=Linux
 

.github/config/conan/profiles/windows-2022-msvc-1940

@@ -3,7 +3,7 @@ arch=x86_64
 build_type=Release
 compiler=msvc
 compiler.version=194
-compiler.cppstd=20
+compiler.cppstd=23
 compiler.runtime=dynamic
 os=Windows
 

AGENTS.md

@@ -5,10 +5,10 @@ user-facing docs see [`README.md`](README.md) and [`docs/`](docs/README.md).
 
 ## What this is
 
-`odr` (a.k.a. `odrcore`) is a **C++20 library that decodes documents and renders
+`odr` (a.k.a. `odrcore`) is a **C++23 library that decodes documents and renders
 them to HTML**. It reads many formats (ODF, OOXML, legacy MS binary, PDF, CSV, …)
 behind one abstract document model and a generic HTML renderer. It is the backend
-for OpenDocument.droid / .ios. Build: **CMake + Conan**; standard: **C++20**.
+for OpenDocument.droid / .ios. Build: **CMake + Conan**; standard: **C++23**.
 
 ## Big picture: how a file becomes HTML
 
@@ -202,6 +202,28 @@ Dispatch `release.yml` against main, publish the draft that appears —
 - **Formatting**: clang-format (LLVM-based, `.clang-format`); run `scripts/format`
   or use the `scripts/setup` git hook. `clang-tidy` per `.clang-tidy`. CI enforces
   both.
+- **C++23, but three separate ceilings sit under it.** Check a facility against
+  all three before reaching for it, and check it by building an **object file**
+  — `-fsyntax-only` misses the codegen bugs.
+  - **The standard library is capped by `emsdk 3.1.73`'s libc++ 18.1**, the
+    oldest in the profile matrix and the newest emsdk conan-center packages.
+    `std::ranges::to`, `std::expected`, `std::string::resize_and_overwrite`,
+    `views::zip`, `ranges::fold_left` and the monadic `std::optional` are there;
+    **`views::enumerate`, `std::generator`, `std::move_only_function` and
+    `std::flat_map` are not**. `std::mdspan` fails from the other side —
+    libstdc++ has no `<mdspan>`, so the gcc-14 job would not build it.
+  - **`std::format` is unusable, on every slice** — the apple deployment target
+    gates it, and `fmt` is what we format with instead. See *Format numbers*
+    below.
+  - **Deducing `this` is fine on accessors, not on a capturing recursive
+    lambda.** NDK 28.1's clang 19 segfaults on `[&](this auto self, …)` that
+    recurses (`cannot compile this l-value expression yet`). The Y-combinator
+    form — pass the lambda to itself — is what `ooxml_text_list` still uses.
+- **The public headers stay C++20**: nothing propagates the standard to a
+  consumer — no `target_compile_features(odr PUBLIC …)`, no `cppstd` in
+  `package_info` — so `src/odr/*.hpp` must keep compiling under C++20. C++23 is
+  for `internal/`, `cli/` and the bindings. `conanfile.py`'s `check_min_cppstd`
+  sits in `validate_build`, which constrains building `odr` and not using it.
 - **Fail fast**: where the spec dictates what to expect, **throw** on unexpected
   input (`std::runtime_error` or the typed exceptions in `src/odr/exceptions.hpp`)
   rather than silently degrading. Only pass through (return empty / skip) values

CHANGELOG.md

@@ -16,6 +16,10 @@ The release run heads these entries with the version and opens a fresh
 
 ## Unreleased
 
+- **Breaking**: building `odr` from source now needs a **C++23** compiler. Using
+  it does not — the public headers still compile as C++20, and the package
+  forces no standard on a consumer.
+
 - **Breaking**: the drawing elements `Rect`, `Line`, `Circle` and `CustomShape`
   are gone, with their `ElementType` values and `Element::as_rect`/`as_line`/
   `as_circle`/`as_custom_shape`. Every shape is a `Frame` now, naming itself

CMakeLists.txt

@@ -6,7 +6,7 @@ if (CMAKE_VERSION VERSION_GREATER_EQUAL "3.24.0")
 endif ()
 
 project(odr LANGUAGES C CXX)
-set(CMAKE_CXX_STANDARD 20)
+set(CMAKE_CXX_STANDARD 23)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_CXX_EXTENSIONS OFF)
 

conanfile.py

@@ -70,7 +70,7 @@ def build_requirements(self):
 
     def validate_build(self):
         if self.settings.get_safe("compiler.cppstd"):
-            check_min_cppstd(self, 20)
+            check_min_cppstd(self, 23)
 
     def configure(self):
         if self.options.shared:

src/odr/internal/common/element_registry.hpp

@@ -33,37 +33,21 @@ template <typename T> class SideTable final {
     return m_entries.insert_or_assign(id, std::move(value)).first->second;
   }
 
-  [[nodiscard]] T *find(const ElementIdentifier id) {
-    return find_(m_entries, id);
-  }
-  [[nodiscard]] const T *find(const ElementIdentifier id) const {
-    return find_(m_entries, id);
-  }
-
-  [[nodiscard]] T &at(const ElementIdentifier id) { return at_(*this, id); }
-  [[nodiscard]] const T &at(const ElementIdentifier id) const {
-    return at_(*this, id);
-  }
-
-private:
-  std::unordered_map<ElementIdentifier, T> m_entries;
-
-  /// One body for both constnesses: the argument carries the const and the
-  /// deduced return takes it on.
-  template <typename Entries>
-  static auto *find_(Entries &entries, const ElementIdentifier id) {
-    const auto it = entries.find(id);
-    return it != std::end(entries) ? &it->second : nullptr;
+  [[nodiscard]] auto *find(this auto &self, const ElementIdentifier id) {
+    const auto it = self.m_entries.find(id);
+    return it != std::end(self.m_entries) ? &it->second : nullptr;
   }
 
-  template <typename Self>
-  static auto &at_(Self &self, const ElementIdentifier id) {
+  [[nodiscard]] auto &at(this auto &self, const ElementIdentifier id) {
     auto *entry = self.find(id);
     if (entry == nullptr) {
       throw std::out_of_range("SideTable::at: identifier not found");
     }
     return *entry;
   }
+
+private:
+  std::unordered_map<ElementIdentifier, T> m_entries;
 };
 
 /// A per-type payload appended as its elements are created, so the ids only
@@ -80,44 +64,32 @@ class SortedSideTable final {
     return m_entries.emplace_back(static_cast<Id>(id), std::move(value)).second;
   }
 
-  [[nodiscard]] T *find(const ElementIdentifier id) {
-    return find_(m_entries, id);
-  }
-  [[nodiscard]] const T *find(const ElementIdentifier id) const {
-    return find_(m_entries, id);
+  [[nodiscard]] auto *find(this auto &self, const ElementIdentifier id) {
+    const auto it =
+        std::ranges::lower_bound(self.m_entries, id, {}, &Entry::first);
+    return it != std::end(self.m_entries) && it->first == id ? &it->second
+                                                             : nullptr;
   }
 
-  [[nodiscard]] T &at(const ElementIdentifier id) { return at_(*this, id); }
-  [[nodiscard]] const T &at(const ElementIdentifier id) const {
-    return at_(*this, id);
+  [[nodiscard]] auto &at(this auto &self, const ElementIdentifier id) {
+    auto *entry = self.find(id);
+    if (entry == nullptr) {
+      throw std::out_of_range("SortedSideTable::at: identifier not found");
+    }
+    return *entry;
   }
 
-  [[nodiscard]] auto begin() const noexcept { return m_entries.begin(); }
-  [[nodiscard]] auto end() const noexcept { return m_entries.end(); }
-  [[nodiscard]] auto begin() noexcept { return m_entries.begin(); }
-  [[nodiscard]] auto end() noexcept { return m_entries.end(); }
+  [[nodiscard]] auto begin(this auto &self) noexcept {
+    return self.m_entries.begin();
+  }
+  [[nodiscard]] auto end(this auto &self) noexcept {
+    return self.m_entries.end();
+  }
 
 private:
   using Entry = std::pair<Id, T>;
 
   std::deque<Entry> m_entries;
-
-  /// One body for both constnesses: the argument carries the const and the
-  /// deduced return takes it on.
-  template <typename Entries>
-  static auto *find_(Entries &entries, const ElementIdentifier id) {
-    const auto it = std::ranges::lower_bound(entries, id, {}, &Entry::first);
-    return it != std::end(entries) && it->first == id ? &it->second : nullptr;
-  }
-
-  template <typename Self>
-  static auto &at_(Self &self, const ElementIdentifier id) {
-    auto *entry = self.find(id);
-    if (entry == nullptr) {
-      throw std::out_of_range("SortedSideTable::at: identifier not found");
-    }
-    return *entry;
-  }
 };
 
 /// The flat store an engine builds its element tree in: an id is the index
@@ -132,14 +104,9 @@ class ElementRegistry {
 
   [[nodiscard]] std::size_t size() const noexcept { return m_elements.size(); }
 
-  [[nodiscard]] Element &element_at(const ElementIdentifier id) {
-    check_element_id(id);
-    return m_elements[id - 1];
-  }
-
-  [[nodiscard]] const Element &element_at(const ElementIdentifier id) const {
-    check_element_id(id);
-    return m_elements[id - 1];
+  [[nodiscard]] auto &element_at(this auto &self, const ElementIdentifier id) {
+    self.check_element_id(id);
+    return self.m_elements[id - 1];
   }
 
   void append_child(const ElementIdentifier parent_id,

src/odr/internal/font/cff_builder.cpp

@@ -3,6 +3,7 @@
 #include <odr/internal/util/byte_string.hpp>
 
 #include <cstdint>
+#include <ranges>
 #include <string>
 #include <vector>
 
@@ -108,11 +109,10 @@ std::string cff::build_cff(const std::string_view name,
   // String INDEX: every glyph name gets a custom SID (391 + position). Glyph 0
   // is the implicit `.notdef` (SID 0), so its name is not stored; the charset
   // lists SIDs for glyphs 1..n-1.
-  std::vector<std::string> strings;
-  for (std::size_t i = 1; i < glyphs.size(); ++i) {
-    strings.push_back(glyphs[i].name);
-  }
-  const std::string string_index = build_index(strings);
+  const std::string string_index =
+      build_index(glyphs | std::views::drop(1) |
+                  std::views::transform(&BuilderGlyph::name) |
+                  std::ranges::to<std::vector<std::string>>());
 
   // Format-0 charset: SID per glyph 1..n-1.
   std::string charset;

src/odr/internal/iwork/iwork_element_registry.hpp

@@ -90,46 +90,29 @@ class ElementRegistry final
   std::tuple<ElementIdentifier, Element &, Cell &>
   create_cell_element(ElementType type);
 
-  [[nodiscard]] Text &text_element_at(const ElementIdentifier id) {
-    return m_texts.at(id);
+  [[nodiscard]] auto &text_element_at(this auto &self,
+                                      const ElementIdentifier id) {
+    return self.m_texts.at(id);
   }
-  [[nodiscard]] Frame &frame_element_at(const ElementIdentifier id) {
-    return m_frames.at(id);
+  [[nodiscard]] auto &frame_element_at(this auto &self,
+                                       const ElementIdentifier id) {
+    return self.m_frames.at(id);
   }
-  [[nodiscard]] Slide &slide_element_at(const ElementIdentifier id) {
-    return m_slides.at(id);
+  [[nodiscard]] auto &slide_element_at(this auto &self,
+                                       const ElementIdentifier id) {
+    return self.m_slides.at(id);
   }
-  [[nodiscard]] Table &table_element_at(const ElementIdentifier id) {
-    return m_tables.at(id);
+  [[nodiscard]] auto &table_element_at(this auto &self,
+                                       const ElementIdentifier id) {
+    return self.m_tables.at(id);
   }
-  [[nodiscard]] Sheet &sheet_element_at(const ElementIdentifier id) {
-    return m_sheets.at(id);
+  [[nodiscard]] auto &sheet_element_at(this auto &self,
+                                       const ElementIdentifier id) {
+    return self.m_sheets.at(id);
   }
-  [[nodiscard]] Cell &cell_element_at(const ElementIdentifier id) {
-    return m_cells.at(id);
-  }
-
-  [[nodiscard]] const Text &text_element_at(const ElementIdentifier id) const {
-    return m_texts.at(id);
-  }
-  [[nodiscard]] const Frame &
-  frame_element_at(const ElementIdentifier id) const {
-    return m_frames.at(id);
-  }
-  [[nodiscard]] const Slide &
-  slide_element_at(const ElementIdentifier id) const {
-    return m_slides.at(id);
-  }
-  [[nodiscard]] const Table &
-  table_element_at(const ElementIdentifier id) const {
-    return m_tables.at(id);
-  }
-  [[nodiscard]] const Sheet &
-  sheet_element_at(const ElementIdentifier id) const {
-    return m_sheets.at(id);
-  }
-  [[nodiscard]] const Cell &cell_element_at(const ElementIdentifier id) const {
-    return m_cells.at(id);
+  [[nodiscard]] auto &cell_element_at(this auto &self,
+                                      const ElementIdentifier id) {
+    return self.m_cells.at(id);
   }
 
   /// Links @p column_id into @p table_id's column chain. Columns are not

src/odr/internal/markdown/markdown_element_registry.hpp

@@ -57,16 +57,15 @@ class ElementRegistry final
   std::tuple<ElementIdentifier, Element &, TableCell &>
   create_table_cell_element();
 
-  [[nodiscard]] Text &text_element_at(const ElementIdentifier id) {
-    return m_texts.at(id);
+  [[nodiscard]] auto &text_element_at(this auto &self,
+                                      const ElementIdentifier id) {
+    return self.m_texts.at(id);
   }
-  [[nodiscard]] Table &table_element_at(const ElementIdentifier id) {
-    return m_tables.at(id);
+  [[nodiscard]] auto &table_element_at(this auto &self,
+                                       const ElementIdentifier id) {
+    return self.m_tables.at(id);
   }
 
-  [[nodiscard]] const Text &text_element_at(const ElementIdentifier id) const {
-    return m_texts.at(id);
-  }
   [[nodiscard]] const Link &link_element_at(const ElementIdentifier id) const {
     return m_links.at(id);
   }
@@ -77,10 +76,6 @@ class ElementRegistry final
   list_item_element_at(const ElementIdentifier id) const {
     return m_list_items.at(id);
   }
-  [[nodiscard]] const Table &
-  table_element_at(const ElementIdentifier id) const {
-    return m_tables.at(id);
-  }
   [[nodiscard]] const TableCell &
   table_cell_element_at(const ElementIdentifier id) const {
     return m_table_cells.at(id);