From eeac21fb347a8f49edadf90d8f9e0c7ed3011a0b Mon Sep 17 00:00:00 2001
From: Stephane Del Pino <stephane.delpino44@gmail.com>
Date: Sat, 15 May 2021 19:30:07 +0200
Subject: [PATCH] git subrepo clone --branch=master
git@github.com:taocpp/PEGTL.git packages/PEGTL
subrepo:
subdir: "packages/PEGTL"
merged: "c131c2e2a"
upstream:
origin: "git@github.com:taocpp/PEGTL.git"
branch: "master"
commit: "c131c2e2a"
git-subrepo:
version: "0.4.3"
origin: "git@github.com:ingydotnet/git-subrepo.git"
commit: "2f68596"
---
.../PEGTL/.github/workflows/clang-analyze.yml | 2 +-
.../PEGTL/.github/workflows/clang-tidy.yml | 2 +-
packages/PEGTL/.github/workflows/linux.yml | 56 +-
.../PEGTL/.github/workflows/no-exceptions.yml | 4 +-
.../PEGTL/.github/workflows/sanitizer.yml | 2 +-
packages/PEGTL/.gitrepo | 6 +-
packages/PEGTL/README.md | 1 +
packages/PEGTL/doc/Actions-and-States.md | 67 +-
packages/PEGTL/doc/Changelog.md | 6 +
packages/PEGTL/doc/Errors-and-Exceptions.md | 2 +-
packages/PEGTL/doc/Grammar-Analysis.md | 2 +-
packages/PEGTL/doc/Parse-Tree.md | 5 +
packages/PEGTL/doc/README.md | 2 +
packages/PEGTL/doc/Rule-Reference.md | 4 +-
.../PEGTL/include/tao/pegtl/buffer_input.hpp | 7 +-
.../include/tao/pegtl/contrib/analyze.hpp | 4 +-
.../include/tao/pegtl/contrib/check_bytes.hpp | 55 ++
.../include/tao/pegtl/contrib/instantiate.hpp | 2 +-
.../contrib/internal/set_stack_guard.hpp | 6 +-
.../contrib/internal/vector_stack_guard.hpp | 6 +-
.../include/tao/pegtl/contrib/limit_bytes.hpp | 88 +++
.../include/tao/pegtl/contrib/limit_depth.hpp | 83 +++
.../PEGTL/include/tao/pegtl/contrib/skip.hpp | 59 --
.../tao/pegtl/internal/file_mapper_posix.hpp | 8 +-
.../tao/pegtl/internal/file_mapper_win32.hpp | 12 +-
.../tao/pegtl/internal/file_reader.hpp | 4 +-
.../include/tao/pegtl/internal/marker.hpp | 12 +-
.../include/tao/pegtl/internal/ranges.hpp | 51 +-
.../PEGTL/include/tao/pegtl/memory_input.hpp | 29 +-
.../PEGTL/include/tao/pegtl/mmap_input.hpp | 8 +-
packages/PEGTL/include/tao/pegtl/parse.hpp | 52 +-
.../PEGTL/include/tao/pegtl/parse_error.hpp | 5 +
.../PEGTL/include/tao/pegtl/read_input.hpp | 4 +-
.../PEGTL/include/tao/pegtl/string_input.hpp | 8 +-
packages/PEGTL/include/tao/pegtl/version.hpp | 4 +-
.../PEGTL/src/example/pegtl/CMakeLists.txt | 2 +-
.../PEGTL/src/example/pegtl/expression.cpp | 613 ++++++++++++++++++
.../PEGTL/src/example/pegtl/json_classes.hpp | 4 +-
.../PEGTL/src/example/pegtl/json_parse.cpp | 14 +-
packages/PEGTL/src/example/pegtl/skipper.cpp | 88 ---
.../PEGTL/src/example/pegtl/token_input.cpp | 8 +-
packages/PEGTL/src/test/pegtl/CMakeLists.txt | 3 +
packages/PEGTL/src/test/pegtl/check_bytes.cpp | 50 ++
.../src/test/pegtl/contrib_instantiate.cpp | 4 +-
packages/PEGTL/src/test/pegtl/limit_bytes.cpp | 50 ++
packages/PEGTL/src/test/pegtl/limit_depth.cpp | 50 ++
46 files changed, 1244 insertions(+), 310 deletions(-)
create mode 100644 packages/PEGTL/include/tao/pegtl/contrib/check_bytes.hpp
create mode 100644 packages/PEGTL/include/tao/pegtl/contrib/limit_bytes.hpp
create mode 100644 packages/PEGTL/include/tao/pegtl/contrib/limit_depth.hpp
delete mode 100644 packages/PEGTL/include/tao/pegtl/contrib/skip.hpp
create mode 100644 packages/PEGTL/src/example/pegtl/expression.cpp
delete mode 100644 packages/PEGTL/src/example/pegtl/skipper.cpp
create mode 100644 packages/PEGTL/src/test/pegtl/check_bytes.cpp
create mode 100644 packages/PEGTL/src/test/pegtl/limit_bytes.cpp
create mode 100644 packages/PEGTL/src/test/pegtl/limit_depth.cpp
diff --git a/packages/PEGTL/.github/workflows/clang-analyze.yml b/packages/PEGTL/.github/workflows/clang-analyze.yml
index 2226ea7b0..523b10556 100644
--- a/packages/PEGTL/.github/workflows/clang-analyze.yml
+++ b/packages/PEGTL/.github/workflows/clang-analyze.yml
@@ -4,7 +4,7 @@ on: [push, pull_request]
jobs:
clang-analyze:
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
diff --git a/packages/PEGTL/.github/workflows/clang-tidy.yml b/packages/PEGTL/.github/workflows/clang-tidy.yml
index 41e1003c4..4da8af6ff 100644
--- a/packages/PEGTL/.github/workflows/clang-tidy.yml
+++ b/packages/PEGTL/.github/workflows/clang-tidy.yml
@@ -4,7 +4,7 @@ on: [push, pull_request]
jobs:
clang-tidy:
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
diff --git a/packages/PEGTL/.github/workflows/linux.yml b/packages/PEGTL/.github/workflows/linux.yml
index ba30d3d98..139982af1 100644
--- a/packages/PEGTL/.github/workflows/linux.yml
+++ b/packages/PEGTL/.github/workflows/linux.yml
@@ -8,16 +8,14 @@ jobs:
fail-fast: false
matrix:
compiler:
- - g++-7
- - g++-8
- g++-9
- g++-10
- - clang++-8
- clang++-9
- clang++-10
+ - clang++-11
build_type: [Debug, Release]
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
env:
CXX: ${{ matrix.compiler }}
@@ -36,22 +34,34 @@ jobs:
- working-directory: build/
run: ctest --output-on-failure
- linux-gcc-extra:
+ linux-old:
strategy:
fail-fast: false
matrix:
- flags: ["-fno-rtti"]
+ compiler:
+ - g++-7
+ - g++-8
+ - clang++-6.0
+ - clang++-7
+ - clang++-8
build_type: [Debug, Release]
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
+
+ env:
+ CXX: ${{ matrix.compiler }}
steps:
- uses: actions/checkout@v2
+ - run: sudo apt-get update
+
+ - run: sudo apt-get install -y ${{ matrix.compiler }}
+
- run: cmake -E make_directory build
- working-directory: build/
- run: cmake $GITHUB_WORKSPACE -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} -DCMAKE_CXX_FLAGS="${{ matrix.flags }}"
+ run: cmake $GITHUB_WORKSPACE -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
- working-directory: build/
run: cmake --build .
@@ -59,17 +69,14 @@ jobs:
- working-directory: build/
run: ctest --output-on-failure
- linux-clang-extra:
+ linux-gcc-extra:
strategy:
fail-fast: false
matrix:
- flags: ["-fno-rtti", "-fms-extensions"]
+ flags: ["-fno-rtti"]
build_type: [Debug, Release]
- runs-on: ubuntu-20.04
-
- env:
- CXX: clang++
+ runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
@@ -85,32 +92,25 @@ jobs:
- working-directory: build/
run: ctest --output-on-failure
- clang-conanio:
+ linux-clang-extra:
strategy:
fail-fast: false
matrix:
- image:
- # List: https://github.com/conan-io/conan-docker-tools
- - clang50
- - clang60
- - clang7
- - clang9-x86
- - clang11
+ flags: ["-fno-rtti", "-fms-extensions"]
build_type: [Debug, Release]
- container:
- image: conanio/${{ matrix.image }}
- options: --user root
-
runs-on: ubuntu-latest
+ env:
+ CXX: clang++
+
steps:
- - uses: actions/checkout@v1
+ - uses: actions/checkout@v2
- run: cmake -E make_directory build
- working-directory: build/
- run: cmake $GITHUB_WORKSPACE -DCMAKE_BUILD_TYPE=${{ matrix.build_type }}
+ run: cmake $GITHUB_WORKSPACE -DCMAKE_BUILD_TYPE=${{ matrix.build_type }} -DCMAKE_CXX_FLAGS="${{ matrix.flags }}"
- working-directory: build/
run: cmake --build .
diff --git a/packages/PEGTL/.github/workflows/no-exceptions.yml b/packages/PEGTL/.github/workflows/no-exceptions.yml
index 3dd3d74b8..eeb964f25 100644
--- a/packages/PEGTL/.github/workflows/no-exceptions.yml
+++ b/packages/PEGTL/.github/workflows/no-exceptions.yml
@@ -7,10 +7,10 @@ jobs:
strategy:
fail-fast: false
matrix:
- compiler: [g++-10, clang++-10]
+ compiler: [g++, clang++]
build_type: [Debug, Release]
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
env:
CXX: ${{ matrix.compiler }}
diff --git a/packages/PEGTL/.github/workflows/sanitizer.yml b/packages/PEGTL/.github/workflows/sanitizer.yml
index 1f478cb06..61d61aebd 100644
--- a/packages/PEGTL/.github/workflows/sanitizer.yml
+++ b/packages/PEGTL/.github/workflows/sanitizer.yml
@@ -10,7 +10,7 @@ jobs:
cxx: [g++, clang++]
sanitizer: [address, undefined]
- runs-on: ubuntu-20.04
+ runs-on: ubuntu-latest
env:
CXX: ${{ matrix.cxx }}
diff --git a/packages/PEGTL/.gitrepo b/packages/PEGTL/.gitrepo
index 393af709f..d60e16e83 100644
--- a/packages/PEGTL/.gitrepo
+++ b/packages/PEGTL/.gitrepo
@@ -6,7 +6,7 @@
[subrepo]
remote = git@github.com:taocpp/PEGTL.git
branch = master
- commit = 57f8ebe0045d7e35cbb251536146a57bc0cf9db5
- parent = 0a259f7e3e4fe2364b8d45b641c7f48ff3bc7341
- cmdver = 0.4.3
+ commit = c131c2e2aad67037285ef39d11ec4f1d28d4fc73
+ parent = 2f2fa0e22bd114f44f78c5bee89bc13bd0959d1d
method = merge
+ cmdver = 0.4.3
diff --git a/packages/PEGTL/README.md b/packages/PEGTL/README.md
index 64d090c99..e613c97f9 100644
--- a/packages/PEGTL/README.md
+++ b/packages/PEGTL/README.md
@@ -112,6 +112,7 @@ In appreciation of all contributions here are the people that have [directly con
[<img alt="pauloscustodio" src="https://avatars.githubusercontent.com/u/70773" width="120">](https://github.com/pauloscustodio)
[<img alt="pleroux0" src="https://avatars.githubusercontent.com/u/39619854" width="120">](https://github.com/pleroux0)
[<img alt="quadfault" src="https://avatars.githubusercontent.com/u/30195320" width="120">](https://github.com/quadfault)
+[<img alt="quarticcat" src="https://avatars.githubusercontent.com/u/70888415" width="120">](https://github.com/quarticcat)
[<img alt="ras0219" src="https://avatars.githubusercontent.com/u/533828" width="120">](https://github.com/ras0219)
[<img alt="redmercury" src="https://avatars.githubusercontent.com/u/4424222" width="120">](https://github.com/redmercury)
[<img alt="robertcampion" src="https://avatars.githubusercontent.com/u/4220569" width="120">](https://github.com/robertcampion)
diff --git a/packages/PEGTL/doc/Actions-and-States.md b/packages/PEGTL/doc/Actions-and-States.md
index 951720b7d..4c8101682 100644
--- a/packages/PEGTL/doc/Actions-and-States.md
+++ b/packages/PEGTL/doc/Actions-and-States.md
@@ -24,6 +24,7 @@ When an action is *applied*, the corresponding function receives the *states*, a
* [Changing Actions and States](#changing-actions-and-states)
* [Match](#match)
* [Nothing](#nothing)
+* [Backtracking](#backtracking)
* [Troubleshooting](#troubleshooting)
* [Boolean Return](#boolean-return)
* [State Mismatch](#state-mismatch)
@@ -285,7 +286,7 @@ std::string unescape( const std::string& escaped )
At the end of the parsing run, the complete unescaped string can be found in the aptly named variable `unescaped`.
-A more complete example of how to unescape strings can be found in `src/examples/pegtl/unescape.cpp`.
+A more complete example of how to unescape strings can be found in `src/example/pegtl/unescape.cpp`.
## Specialising
@@ -506,8 +507,72 @@ For example when a class `b` is derived from `change_state`, it also gains that
At this point `b` is allowed to either have or not have an `apply()` or `apply0()`.
By letting `b` also derive from one of the three mentioned classes, the `maybe_nothing` will be ignored and `b` will be checked to have or not have the functions as dictated by the respective additional base class.
+## Backtracking
+
+Sometimes there can be *backtracking* during a parsing run which can lead to Actions being called in places where their effects are undesired.
+While it might be intuitively clear what backtracking is, for the purpose of the following discussion we give a slightly more formal definition.
+
+We speak of *backtracking* across a rule `S` when there is a rule `R` of which `S` is a (direct or indirect) sub-rule and during a parsing run
+1. `R` returns local failure after
+2. `S` succeeded and its success is a requirement for the success of `R` and
+3. it is "still possible" for the top-level grammar rule of the parsing run to succeed.
+
+In this case the input will have been rewound to the point at which `R` was attempted to match and all effects of `S` on the Input will have been undone, however, and this is the subject of this section, any action attached to `S` will have been already performed without there being an automatic "undo".
+
+#### The AAC-Problem
+
+In some cases it is easy to rewrite the grammar in a way that prevents backtracking.
+This simultaneously removes the issue of having to undo actions and improves parsing performance.
+
+The prototypical case for which such a rewrite can be done is `R = sor< seq< A, B >, seq< A, C > >` where `A`, `B` and `C` are arbitrary rules.
+
+If during a parsing run there are actions attached to `A` and `C`, and the input matches `seq< A, C >` but not `seq< A, B >`, then the action for `A` will be called *twice* before the action for `C`, which gives this problem its "AAC" name, given that what happens is:
+
+* Begin `sor< seq< A, B >, seq< A, C > >`
+* Begin `seq< A, B >`
+* Begin `A`
+* Success `A` with action called
+* Begin `B`
+* Failure `B`
+* Failure `seq< A, B >`
+* Begin `seq< A, C >`
+* Begin `A` at the same position as the begin `A` above
+* Success `A` with action called again on the same input
+* Begin `C`
+* Success `C`
+* Success `seq< A, C >`
+* Success `sor< seq< A, B >, seq< A, C > >`
+
+#### Rewriting
+
+In practice the structure of the rule might be more complicated than the pure AAC-problem which will make it harder to recognise the pattern.
+One solution is to rewrite `R` as `R' = seq< A, sor< B, C > >` where of course any action for `A` will be called at most once for every successful match of `R'`.
+
+#### Manual Undo
+
+Another solution is to undo the effects of the Action attached to `A` in case the encompassing `seq< A, B >` (or `seq< A, C >`) fail.
+
+The advantage of this approach is that the implementation of the Action for `A` can pretend that is only called when really needed.
+The disadvantage is that there is no function on the Action that is called in the case of failure which requires the user to either write a custom `match()` function in the Action for `seq< A, B >` or to implement the `failure()` function in a custom [Control class](Control-and-Debug.md).
+
+#### Manual Commit
+
+A further solution is to let the Action for `A` perform its job "to the side", and only "commit" the effects to the target data structure in the Action for `seq< A, B >`.
+
+For example if the Action attached to `A` takes the matched portion of the Input as `std::string` and appends it to `std::vector< std::string >` one could change said Action for `A` to only fill some temporary string in one of the States, and create an Action for `seq< A, B >` that, after it is called on success of that rule, appends the aforementioned temporary string to the target vector.
+
+#### Looking Ahead
+
+When everything else fails and a quick-and-dirty solution to Actions being called too often in the presence of backtracking is required and/or performance is not of prime importance it is relatively easy to solve the problem by employing the infinite look-ahead capability of PEGs.
+
+When backtracking across `S` is a problem because an Action attached to `S` can be called when `S` succeeds even though there is a higher-up rule `R` that can still fail then simply replace `R` with `seq< at< R >, R >` in the grammar.
+
+Remembering that `at` disables all Actions explains how this solves the problem; we first verify without Actions that `R` will indeed match at this point and only then match `R` again with Actions enabled.
+
## Troubleshooting
+The following lists a couple of frequently encountered Action-related errors and how to fix them.
+
### Boolean Return
Actions returning `bool` are an advanced use case that should be used with caution.
diff --git a/packages/PEGTL/doc/Changelog.md b/packages/PEGTL/doc/Changelog.md
index d1b57f131..cc194318f 100644
--- a/packages/PEGTL/doc/Changelog.md
+++ b/packages/PEGTL/doc/Changelog.md
@@ -1,5 +1,11 @@
# Changelog
+## 3.2.1
+
+**Not yet released**
+
+* Added an optional limiter to guard against infinite recursion.
+
## 3.2.0
Released 2021-01-15
diff --git a/packages/PEGTL/doc/Errors-and-Exceptions.md b/packages/PEGTL/doc/Errors-and-Exceptions.md
index f080c0917..7b8f018d8 100644
--- a/packages/PEGTL/doc/Errors-and-Exceptions.md
+++ b/packages/PEGTL/doc/Errors-and-Exceptions.md
@@ -162,7 +162,7 @@ This is often insufficient and one would like to provide more meaningful error m
A practical technique to provide customised error messages for all `must<>` error points uses the `must_if<>` helper.
-For an example of this method see `src/examples/pegtl/json_errors.hpp`, where all errors that might occur in the supplied JSON grammar are customised like this:
+For an example of this method see `src/example/pegtl/json_errors.hpp`, where all errors that might occur in the supplied JSON grammar are customised like this:
```c++
template< typename > inline constexpr const char* error_message = nullptr;
diff --git a/packages/PEGTL/doc/Grammar-Analysis.md b/packages/PEGTL/doc/Grammar-Analysis.md
index 964adb9c1..629982d8e 100644
--- a/packages/PEGTL/doc/Grammar-Analysis.md
+++ b/packages/PEGTL/doc/Grammar-Analysis.md
@@ -56,7 +56,7 @@ This support automatically extends to all custom rules built "the usual way" via
For true custom rules, i.e. rules that implement their own `match()` function, the following steps need to be taken for them to work with the grammar analysis.
-1. The rule needs a `rule_t` that, usually for true custom rules, is a type alias for the grammar rule itself.
+1. The rule needs a [`rule_t`](Meta-Data-and-Visit.md#rule-type) that, usually for true custom rules, is a type alias for the grammar rule itself.
2. There needs to be a specialisation of the `analyze_traits<>` for the custom rule, with an additional first template parameter:
Assuming a custom rule like the following
diff --git a/packages/PEGTL/doc/Parse-Tree.md b/packages/PEGTL/doc/Parse-Tree.md
index 039208294..baa3b24a7 100644
--- a/packages/PEGTL/doc/Parse-Tree.md
+++ b/packages/PEGTL/doc/Parse-Tree.md
@@ -18,6 +18,7 @@ It provides the basic infrastructure to build a parse tree that
* [Transformers](#transformers)
* [`tao::pegtl::parse_tree::node`](#taopegtlparse_treenode)
* [Custom Node Class](#custom-node-class)
+* [Requirements](#requirements)
## Full Parse Tree
@@ -235,4 +236,8 @@ struct my_node
};
```
+## Requirements
+
+The parse tree uses a rule's meta data supplied by [`subs_t`](Meta-Data-and-Visit.md#sub-rules) for internal optimizations.
+
Copyright (c) 2018-2021 Dr. Colin Hirsch and Daniel Frey
diff --git a/packages/PEGTL/doc/README.md b/packages/PEGTL/doc/README.md
index 09b6fbcf1..7c028ccae 100644
--- a/packages/PEGTL/doc/README.md
+++ b/packages/PEGTL/doc/README.md
@@ -44,6 +44,7 @@
* [Changing Actions and States](Actions-and-States.md#changing-actions-and-states)
* [Match](Actions-and-States.md#match)
* [Nothing](Actions-and-States.md#nothing)
+ * [Backtracking](Actions-and-States.md#backtracking)
* [Troubleshooting](Actions-and-States.md#troubleshooting)
* [Boolean Return](Actions-and-States.md#boolean-return)
* [State Mismatch](Actions-and-States.md#state-mismatch)
@@ -103,6 +104,7 @@
* [Transformer](Parse-Tree.md#transformer)
* [`tao::pegtl::parse_tree::node`](Parse-Tree.md#taopegtlparse_treenode)
* [Custom Node Class](Parse-Tree.md#custom-node-class)
+ * [Requirements](Parse-Tree.md#requirements)
* [Meta Data and Visit](Meta-Data-and-Visit.md)
* [Internals](Meta-Data-and-Visit.md#internals)
* [Rule Type](Meta-Data-and-Visit.md#rule-type)
diff --git a/packages/PEGTL/doc/Rule-Reference.md b/packages/PEGTL/doc/Rule-Reference.md
index a39700edf..79dd46ccf 100644
--- a/packages/PEGTL/doc/Rule-Reference.md
+++ b/packages/PEGTL/doc/Rule-Reference.md
@@ -100,7 +100,7 @@ These rules are in namespace `tao::pegtl`.
* Enables all actions (if any).
* [Meta data] and [implementation] mapping:
- `enable<>::rule_t` is `internal::success`
- - `enable< R >::rule_t` is `internal::enable<, R >`
+ - `enable< R >::rule_t` is `internal::enable< R >`
- `enable< R >::subs_t` is `type_list< R >`
- `enable< R... >::rule_t` is `internal::enable< internal::seq< R... > >`
- `enable< R... >::subs_t` is `type_list< internal::seq< R... > >`
@@ -368,6 +368,8 @@ Note that the `true` template parameter to `internal::if_must` corresponds to th
- `rematch< R, S... >::rule_t` is `internal::rematch< R, S... >`
- `rematch< R, S... >::subs_t` is `type_list< R, S... >`
+Note that the `S` do *not* need to match *all* of the input matched by `R` (which is why `minus` uses `eof` in its implementation).
+
###### `rep< Num, R... >`
* Matches `seq< R... >` for `Num` times without checking for further matches.
diff --git a/packages/PEGTL/include/tao/pegtl/buffer_input.hpp b/packages/PEGTL/include/tao/pegtl/buffer_input.hpp
index 0003442b0..60d39d349 100644
--- a/packages/PEGTL/include/tao/pegtl/buffer_input.hpp
+++ b/packages/PEGTL/include/tao/pegtl/buffer_input.hpp
@@ -66,8 +66,8 @@ namespace TAO_PEGTL_NAMESPACE
~buffer_input() = default;
- void operator=( const buffer_input& ) = delete;
- void operator=( buffer_input&& ) = delete;
+ buffer_input& operator=( const buffer_input& ) = delete;
+ buffer_input& operator=( buffer_input&& ) = delete;
[[nodiscard]] bool empty()
{
@@ -216,6 +216,9 @@ namespace TAO_PEGTL_NAMESPACE
iterator_t m_current;
char* m_end;
const Source m_source;
+
+ public:
+ std::size_t private_depth = 0;
};
} // namespace TAO_PEGTL_NAMESPACE
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/analyze.hpp b/packages/PEGTL/include/tao/pegtl/contrib/analyze.hpp
index 586710735..0bfbd18c2 100644
--- a/packages/PEGTL/include/tao/pegtl/contrib/analyze.hpp
+++ b/packages/PEGTL/include/tao/pegtl/contrib/analyze.hpp
@@ -44,8 +44,8 @@ namespace TAO_PEGTL_NAMESPACE
~analyze_cycles_impl() = default;
- void operator=( analyze_cycles_impl&& ) = delete;
- void operator=( const analyze_cycles_impl& ) = delete;
+ analyze_cycles_impl& operator=( analyze_cycles_impl&& ) = delete;
+ analyze_cycles_impl& operator=( const analyze_cycles_impl& ) = delete;
[[nodiscard]] std::size_t problems()
{
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/check_bytes.hpp b/packages/PEGTL/include/tao/pegtl/contrib/check_bytes.hpp
new file mode 100644
index 000000000..84333c7d7
--- /dev/null
+++ b/packages/PEGTL/include/tao/pegtl/contrib/check_bytes.hpp
@@ -0,0 +1,55 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#ifndef TAO_PEGTL_CONTRIB_CHECK_BYTES_HPP
+#define TAO_PEGTL_CONTRIB_CHECK_BYTES_HPP
+
+#include "../apply_mode.hpp"
+#include "../config.hpp"
+#include "../match.hpp"
+#include "../nothing.hpp"
+#include "../rewind_mode.hpp"
+
+#if defined( __cpp_exceptions )
+#include "../parse_error.hpp"
+#else
+#include <cstdio>
+#include <exception>
+#endif
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ template< std::size_t Maximum >
+ struct check_bytes
+ : maybe_nothing
+ {
+ template< typename Rule,
+ pegtl::apply_mode A,
+ pegtl::rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename... States >
+ static bool match( ParseInput& in, States&&... st )
+ {
+ const auto* start = in.current();
+ if( TAO_PEGTL_NAMESPACE::match< Rule, A, M, Action, Control >( in, st... ) ) {
+ if( std::size_t( in.current() - start ) > Maximum ) {
+#if defined( __cpp_exceptions )
+ throw TAO_PEGTL_NAMESPACE::parse_error( "maximum allowed rule consumption exceeded", in );
+#else
+ std::fputs( "maximum allowed rule consumption exceeded\n", stderr );
+ std::terminate();
+#endif
+ }
+ return true;
+ }
+ return false;
+ }
+ };
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#endif
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/instantiate.hpp b/packages/PEGTL/include/tao/pegtl/contrib/instantiate.hpp
index c04db3f54..ebf6e7f1c 100644
--- a/packages/PEGTL/include/tao/pegtl/contrib/instantiate.hpp
+++ b/packages/PEGTL/include/tao/pegtl/contrib/instantiate.hpp
@@ -26,7 +26,7 @@ namespace TAO_PEGTL_NAMESPACE
class Control,
typename ParseInput,
typename... States >
- [[nodiscard]] static bool match( ParseInput& in, States&... st )
+ [[nodiscard]] static bool match( ParseInput& in, States&&... st )
{
const T t( static_cast< const ParseInput& >( in ), st... );
return TAO_PEGTL_NAMESPACE::match< Rule, A, M, Action, Control >( in, st... );
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/internal/set_stack_guard.hpp b/packages/PEGTL/include/tao/pegtl/contrib/internal/set_stack_guard.hpp
index 94df74cab..c0071e959 100644
--- a/packages/PEGTL/include/tao/pegtl/contrib/internal/set_stack_guard.hpp
+++ b/packages/PEGTL/include/tao/pegtl/contrib/internal/set_stack_guard.hpp
@@ -12,7 +12,7 @@
namespace TAO_PEGTL_NAMESPACE::internal
{
template< typename... Cs >
- class set_stack_guard
+ class [[nodiscard]] set_stack_guard
{
public:
template< typename... Ts >
@@ -24,8 +24,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
set_stack_guard( set_stack_guard&& ) = delete;
set_stack_guard( const set_stack_guard& ) = delete;
- void operator=( set_stack_guard&& ) = delete;
- void operator=( const set_stack_guard& ) = delete;
+ set_stack_guard& operator=( set_stack_guard&& ) = delete;
+ set_stack_guard& operator=( const set_stack_guard& ) = delete;
~set_stack_guard()
{
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/internal/vector_stack_guard.hpp b/packages/PEGTL/include/tao/pegtl/contrib/internal/vector_stack_guard.hpp
index 146da0f1a..4b0cfbe8b 100644
--- a/packages/PEGTL/include/tao/pegtl/contrib/internal/vector_stack_guard.hpp
+++ b/packages/PEGTL/include/tao/pegtl/contrib/internal/vector_stack_guard.hpp
@@ -12,7 +12,7 @@
namespace TAO_PEGTL_NAMESPACE::internal
{
template< typename... Cs >
- class vector_stack_guard
+ class [[nodiscard]] vector_stack_guard
{
public:
template< typename... Ts >
@@ -25,8 +25,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
vector_stack_guard( vector_stack_guard&& ) = delete;
vector_stack_guard( const vector_stack_guard& ) = delete;
- void operator=( vector_stack_guard&& ) = delete;
- void operator=( const vector_stack_guard& ) = delete;
+ vector_stack_guard& operator=( vector_stack_guard&& ) = delete;
+ vector_stack_guard& operator=( const vector_stack_guard& ) = delete;
~vector_stack_guard()
{
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/limit_bytes.hpp b/packages/PEGTL/include/tao/pegtl/contrib/limit_bytes.hpp
new file mode 100644
index 000000000..cea99fc6f
--- /dev/null
+++ b/packages/PEGTL/include/tao/pegtl/contrib/limit_bytes.hpp
@@ -0,0 +1,88 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#ifndef TAO_PEGTL_CONTRIB_LIMIT_BYTES_HPP
+#define TAO_PEGTL_CONTRIB_LIMIT_BYTES_HPP
+
+#include <algorithm>
+
+#include "../apply_mode.hpp"
+#include "../config.hpp"
+#include "../match.hpp"
+#include "../nothing.hpp"
+#include "../rewind_mode.hpp"
+
+#if defined( __cpp_exceptions )
+#include "../parse_error.hpp"
+#else
+#include <cstdio>
+#include <exception>
+#endif
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ namespace internal
+ {
+ template< std::size_t Maximum, typename MemoryInput >
+ struct [[nodiscard]] bytes_guard
+ {
+ MemoryInput& m_in;
+ const char* m_end;
+
+ explicit bytes_guard( MemoryInput& in_in ) noexcept
+ : m_in( in_in ),
+ m_end( in_in.end() )
+ {
+ m_in.private_set_end( m_in.begin() + std::min( m_in.size(), Maximum ) );
+ }
+
+ bytes_guard( bytes_guard&& ) = delete;
+ bytes_guard( const bytes_guard& ) = delete;
+
+ ~bytes_guard()
+ {
+ m_in.private_set_end( m_end );
+ }
+
+ bytes_guard& operator=( bytes_guard&& ) = delete;
+ bytes_guard& operator=( const bytes_guard& ) = delete;
+ };
+
+ // C++17 does not allow for partial deduction guides.
+
+ } // namespace internal
+
+ template< std::size_t Maximum >
+ struct limit_bytes
+ : maybe_nothing
+ {
+ template< typename Rule,
+ apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename... States >
+ [[nodiscard]] static bool match( ParseInput& in, States&&... st )
+ {
+ internal::bytes_guard< Maximum, ParseInput > bg( in );
+ if( TAO_PEGTL_NAMESPACE::match< Rule, A, M, Action, Control >( in, st... ) ) {
+ if( in.empty() && ( bg.m_end != in.current() ) ) {
+#if defined( __cpp_exceptions )
+ throw TAO_PEGTL_NAMESPACE::parse_error( "maximum allowed rule consumption reached", in );
+#else
+ std::fputs( "maximum allowed rule consumption reached\n", stderr );
+ std::terminate();
+#endif
+ }
+ return true;
+ }
+ return false;
+ }
+ };
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#endif
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/limit_depth.hpp b/packages/PEGTL/include/tao/pegtl/contrib/limit_depth.hpp
new file mode 100644
index 000000000..447bebff7
--- /dev/null
+++ b/packages/PEGTL/include/tao/pegtl/contrib/limit_depth.hpp
@@ -0,0 +1,83 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#ifndef TAO_PEGTL_CONTRIB_LIMIT_DEPTH_HPP
+#define TAO_PEGTL_CONTRIB_LIMIT_DEPTH_HPP
+
+#include "../apply_mode.hpp"
+#include "../config.hpp"
+#include "../match.hpp"
+#include "../nothing.hpp"
+#include "../rewind_mode.hpp"
+
+#if defined( __cpp_exceptions )
+#include "../parse_error.hpp"
+#else
+#include <cstdio>
+#include <exception>
+#endif
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ namespace internal
+ {
+ struct [[nodiscard]] depth_guard
+ {
+ std::size_t& m_depth;
+
+ explicit depth_guard( std::size_t& depth ) noexcept
+ : m_depth( depth )
+ {
+ ++m_depth;
+ }
+
+ depth_guard( depth_guard&& ) = delete;
+ depth_guard( const depth_guard& ) = delete;
+
+ ~depth_guard()
+ {
+ --m_depth;
+ }
+
+ depth_guard& operator=( depth_guard&& ) = delete;
+ depth_guard& operator=( const depth_guard& ) = delete;
+ };
+
+ } // namespace internal
+
+ template< std::size_t Maximum >
+ struct limit_depth
+ : maybe_nothing
+ {
+ template< typename Rule,
+ apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename... States >
+ [[nodiscard]] static bool match( ParseInput& in, States&&... st )
+ {
+ if constexpr( Control< Rule >::enable ) {
+ const internal::depth_guard dg( in.private_depth );
+ if( in.private_depth > Maximum ) {
+#if defined( __cpp_exceptions )
+ throw TAO_PEGTL_NAMESPACE::parse_error( "maximum parser rule nesting depth exceeded", in );
+#else
+ std::fputs( "maximum parser rule nesting depth exceeded\n", stderr );
+ std::terminate();
+#endif
+ }
+ return TAO_PEGTL_NAMESPACE::match< Rule, A, M, Action, Control >( in, st... );
+ }
+ else {
+ return TAO_PEGTL_NAMESPACE::match< Rule, A, M, Action, Control >( in, st... );
+ }
+ }
+ };
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#endif
diff --git a/packages/PEGTL/include/tao/pegtl/contrib/skip.hpp b/packages/PEGTL/include/tao/pegtl/contrib/skip.hpp
deleted file mode 100644
index 8aea252e8..000000000
--- a/packages/PEGTL/include/tao/pegtl/contrib/skip.hpp
+++ /dev/null
@@ -1,59 +0,0 @@
-// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
-// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
-
-#ifndef TAO_PEGTL_CONTRIB_SKIP_HPP
-#define TAO_PEGTL_CONTRIB_SKIP_HPP
-
-#include "../apply_mode.hpp"
-#include "../config.hpp"
-#include "../match.hpp"
-#include "../normal.hpp"
-#include "../nothing.hpp"
-#include "../rewind_mode.hpp"
-
-namespace TAO_PEGTL_NAMESPACE
-{
- // this is currently experimental and may change at any time
- template< typename How, typename Where, template< typename... > class Base = normal >
- struct skip
- {
- template< typename Rule >
- struct control
- : Base< Rule >
- {
- template< apply_mode A,
- rewind_mode M,
- template< typename... >
- class Action,
- template< typename... >
- class Control,
- typename ParseInput,
- typename... States >
- [[nodiscard]] static bool match( ParseInput& in, States&&... st )
- {
- // TODO: if we only skip after but not before the actual rule,
- // we would not need this marker.
- auto m = in.template mark< M >();
-
- // TODO: different conditions for before/after skipping?
- if( Where::template value< Rule > ) {
- // TODO: assert on result to be successful?
- (void)TAO_PEGTL_NAMESPACE::match< How, apply_mode::nothing, M, nothing, normal >( in );
- }
-
- const bool result = Base< Rule >::template match< A, M, Action, Control >( in, st... );
-
- // TODO: different conditions for before/after skipping?
- if( result && Where::template value< Rule > ) {
- // TODO: assert on result to be successful?
- (void)TAO_PEGTL_NAMESPACE::match< How, apply_mode::nothing, M, nothing, normal >( in );
- }
-
- return m( result );
- }
- };
- };
-
-} // namespace TAO_PEGTL_NAMESPACE
-
-#endif
diff --git a/packages/PEGTL/include/tao/pegtl/internal/file_mapper_posix.hpp b/packages/PEGTL/include/tao/pegtl/internal/file_mapper_posix.hpp
index ad8147a7d..0c60e3a4a 100644
--- a/packages/PEGTL/include/tao/pegtl/internal/file_mapper_posix.hpp
+++ b/packages/PEGTL/include/tao/pegtl/internal/file_mapper_posix.hpp
@@ -38,8 +38,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
::close( m_fd );
}
- void operator=( const file_opener& ) = delete;
- void operator=( file_opener&& ) = delete;
+ file_opener& operator=( const file_opener& ) = delete;
+ file_opener& operator=( file_opener&& ) = delete;
[[nodiscard]] std::size_t size() const
{
@@ -118,8 +118,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
::munmap( const_cast< char* >( m_data ), m_size );
}
- void operator=( const file_mapper& ) = delete;
- void operator=( file_mapper&& ) = delete;
+ file_mapper& operator=( const file_mapper& ) = delete;
+ file_mapper& operator=( file_mapper&& ) = delete;
[[nodiscard]] bool empty() const noexcept
{
diff --git a/packages/PEGTL/include/tao/pegtl/internal/file_mapper_win32.hpp b/packages/PEGTL/include/tao/pegtl/internal/file_mapper_win32.hpp
index a8a471fd9..73c27dec9 100644
--- a/packages/PEGTL/include/tao/pegtl/internal/file_mapper_win32.hpp
+++ b/packages/PEGTL/include/tao/pegtl/internal/file_mapper_win32.hpp
@@ -52,8 +52,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
::CloseHandle( m_handle );
}
- void operator=( const file_opener& ) = delete;
- void operator=( file_opener&& ) = delete;
+ file_opener& operator=( const file_opener& ) = delete;
+ file_opener& operator=( file_opener&& ) = delete;
[[nodiscard]] std::size_t size() const
{
@@ -134,8 +134,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
::CloseHandle( m_handle );
}
- void operator=( const win32_file_mapper& ) = delete;
- void operator=( win32_file_mapper&& ) = delete;
+ win32_file_mapper& operator=( const win32_file_mapper& ) = delete;
+ win32_file_mapper& operator=( win32_file_mapper&& ) = delete;
const size_t m_size;
const HANDLE m_handle;
@@ -202,8 +202,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
::UnmapViewOfFile( LPCVOID( m_data ) );
}
- void operator=( const file_mapper& ) = delete;
- void operator=( file_mapper&& ) = delete;
+ file_mapper& operator=( const file_mapper& ) = delete;
+ file_mapper& operator=( file_mapper&& ) = delete;
[[nodiscard]] bool empty() const noexcept
{
diff --git a/packages/PEGTL/include/tao/pegtl/internal/file_reader.hpp b/packages/PEGTL/include/tao/pegtl/internal/file_reader.hpp
index 7b846e1bf..7c2b05494 100644
--- a/packages/PEGTL/include/tao/pegtl/internal/file_reader.hpp
+++ b/packages/PEGTL/include/tao/pegtl/internal/file_reader.hpp
@@ -78,8 +78,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
~file_reader() = default;
- void operator=( const file_reader& ) = delete;
- void operator=( file_reader&& ) = delete;
+ file_reader& operator=( const file_reader& ) = delete;
+ file_reader& operator=( file_reader&& ) = delete;
[[nodiscard]] std::size_t size() const
{
diff --git a/packages/PEGTL/include/tao/pegtl/internal/marker.hpp b/packages/PEGTL/include/tao/pegtl/internal/marker.hpp
index 4beb9293a..a9ffa824f 100644
--- a/packages/PEGTL/include/tao/pegtl/internal/marker.hpp
+++ b/packages/PEGTL/include/tao/pegtl/internal/marker.hpp
@@ -10,7 +10,7 @@
namespace TAO_PEGTL_NAMESPACE::internal
{
template< typename Iterator, rewind_mode M >
- class marker
+ class [[nodiscard]] marker
{
public:
static constexpr rewind_mode next_rewind_mode = M;
@@ -23,8 +23,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
~marker() = default;
- void operator=( const marker& ) = delete;
- void operator=( marker&& ) = delete;
+ marker& operator=( const marker& ) = delete;
+ marker& operator=( marker&& ) = delete;
[[nodiscard]] bool operator()( const bool result ) const noexcept
{
@@ -33,7 +33,7 @@ namespace TAO_PEGTL_NAMESPACE::internal
};
template< typename Iterator >
- class marker< Iterator, rewind_mode::required >
+ class [[nodiscard]] marker< Iterator, rewind_mode::required >
{
public:
static constexpr rewind_mode next_rewind_mode = rewind_mode::active;
@@ -53,8 +53,8 @@ namespace TAO_PEGTL_NAMESPACE::internal
}
}
- void operator=( const marker& ) = delete;
- void operator=( marker&& ) = delete;
+ marker& operator=( const marker& ) = delete;
+ marker& operator=( marker&& ) = delete;
[[nodiscard]] bool operator()( const bool result ) noexcept
{
diff --git a/packages/PEGTL/include/tao/pegtl/internal/ranges.hpp b/packages/PEGTL/include/tao/pegtl/internal/ranges.hpp
index 1af27d755..bf57db611 100644
--- a/packages/PEGTL/include/tao/pegtl/internal/ranges.hpp
+++ b/packages/PEGTL/include/tao/pegtl/internal/ranges.hpp
@@ -6,6 +6,8 @@
#include "../config.hpp"
+#include <utility>
+
#include "bump_help.hpp"
#include "enable_control.hpp"
#include "failure.hpp"
@@ -16,37 +18,12 @@
namespace TAO_PEGTL_NAMESPACE::internal
{
- template< typename Char, Char... Cs >
- struct ranges_impl;
-
- template< typename Char >
- struct ranges_impl< Char >
- {
- [[nodiscard]] static constexpr bool test( const Char /*unused*/ ) noexcept
- {
- return false;
- }
- };
-
- template< typename Char, Char Eq >
- struct ranges_impl< Char, Eq >
- {
- [[nodiscard]] static constexpr bool test( const Char c ) noexcept
- {
- return c == Eq;
- }
- };
-
- template< typename Char, Char Lo, Char Hi, Char... Cs >
- struct ranges_impl< Char, Lo, Hi, Cs... >
+ template< typename Char, Char Lo, Char Hi >
+ constexpr bool validate_range( Char c ) noexcept
{
- static_assert( Lo <= Hi, "invalid range detected" );
-
- [[nodiscard]] static constexpr bool test( const Char c ) noexcept
- {
- return ( ( Lo <= c ) && ( c <= Hi ) ) || ranges_impl< Char, Cs... >::test( c );
- }
- };
+ static_assert( Lo <= Hi, "invalid range" );
+ return ( Lo <= c ) && ( c <= Hi );
+ }
template< typename Peek, typename Peek::data_t... Cs >
struct ranges
@@ -57,9 +34,21 @@ namespace TAO_PEGTL_NAMESPACE::internal
using rule_t = ranges;
using subs_t = empty_list;
+ template< std::size_t... Is >
+ [[nodiscard]] static constexpr bool test( std::index_sequence< Is... > /*unused*/, const data_t c ) noexcept
+ {
+ constexpr const data_t cs[] = { Cs... };
+ if constexpr( sizeof...( Cs ) % 2 == 0 ) {
+ return ( validate_range< data_t, cs[ 2 * Is ], cs[ 2 * Is + 1 ] >( c ) || ... );
+ }
+ else {
+ return ( validate_range< data_t, cs[ 2 * Is ], cs[ 2 * Is + 1 ] >( c ) || ... ) || ( c == cs[ sizeof...( Cs ) - 1 ] );
+ }
+ }
+
[[nodiscard]] static constexpr bool test( const data_t c ) noexcept
{
- return ranges_impl< data_t, Cs... >::test( c );
+ return test( std::make_index_sequence< sizeof...( Cs ) / 2 >(), c );
}
template< int Eol >
diff --git a/packages/PEGTL/include/tao/pegtl/memory_input.hpp b/packages/PEGTL/include/tao/pegtl/memory_input.hpp
index 675e8b32f..7cdc9f9f8 100644
--- a/packages/PEGTL/include/tao/pegtl/memory_input.hpp
+++ b/packages/PEGTL/include/tao/pegtl/memory_input.hpp
@@ -62,8 +62,8 @@ namespace TAO_PEGTL_NAMESPACE
~memory_input_base() = default;
- memory_input_base operator=( const memory_input_base& ) = delete;
- memory_input_base operator=( memory_input_base&& ) = delete;
+ memory_input_base& operator=( const memory_input_base& ) = delete;
+ memory_input_base& operator=( memory_input_base&& ) = delete;
[[nodiscard]] const char* current() const noexcept
{
@@ -124,13 +124,17 @@ namespace TAO_PEGTL_NAMESPACE
m_current.byte = in_byte;
m_current.line = in_line;
m_current.column = in_column;
+ private_depth = 0;
}
protected:
const char* const m_begin;
iterator_t m_current;
- const char* const m_end;
+ const char* m_end;
const Source m_source;
+
+ public:
+ std::size_t private_depth = 0;
};
template< typename Eol, typename Source >
@@ -160,8 +164,8 @@ namespace TAO_PEGTL_NAMESPACE
~memory_input_base() = default;
- memory_input_base operator=( const memory_input_base& ) = delete;
- memory_input_base operator=( memory_input_base&& ) = delete;
+ memory_input_base& operator=( const memory_input_base& ) = delete;
+ memory_input_base& operator=( memory_input_base&& ) = delete;
[[nodiscard]] const char* current() const noexcept
{
@@ -208,13 +212,17 @@ namespace TAO_PEGTL_NAMESPACE
void restart()
{
m_current = m_begin.data;
+ private_depth = 0;
}
protected:
const internal::iterator m_begin;
iterator_t m_current;
- const char* const m_end;
+ const char* m_end;
const Source m_source;
+
+ public:
+ std::size_t private_depth = 0;
};
} // namespace internal
@@ -268,8 +276,8 @@ namespace TAO_PEGTL_NAMESPACE
~memory_input() = default;
- memory_input operator=( const memory_input& ) = delete;
- memory_input operator=( memory_input&& ) = delete;
+ memory_input& operator=( const memory_input& ) = delete;
+ memory_input& operator=( memory_input&& ) = delete;
[[nodiscard]] const Source& source() const noexcept
{
@@ -355,6 +363,11 @@ namespace TAO_PEGTL_NAMESPACE
const char* b = begin_of_line( p );
return std::string_view( b, static_cast< std::size_t >( end_of_line( p ) - b ) );
}
+
+ void private_set_end( const char* new_end ) noexcept
+ {
+ this->m_end = new_end;
+ }
};
template< typename... Ts >
diff --git a/packages/PEGTL/include/tao/pegtl/mmap_input.hpp b/packages/PEGTL/include/tao/pegtl/mmap_input.hpp
index 1bd4dce19..2362b6c76 100644
--- a/packages/PEGTL/include/tao/pegtl/mmap_input.hpp
+++ b/packages/PEGTL/include/tao/pegtl/mmap_input.hpp
@@ -42,8 +42,8 @@ namespace TAO_PEGTL_NAMESPACE
~mmap_holder() = default;
- void operator=( const mmap_holder& ) = delete;
- void operator=( mmap_holder&& ) = delete;
+ mmap_holder& operator=( const mmap_holder& ) = delete;
+ mmap_holder& operator=( mmap_holder&& ) = delete;
};
} // namespace internal
@@ -67,8 +67,8 @@ namespace TAO_PEGTL_NAMESPACE
~mmap_input() = default;
- void operator=( const mmap_input& ) = delete;
- void operator=( mmap_input&& ) = delete;
+ mmap_input& operator=( const mmap_input& ) = delete;
+ mmap_input& operator=( mmap_input&& ) = delete;
};
template< typename... Ts >
diff --git a/packages/PEGTL/include/tao/pegtl/parse.hpp b/packages/PEGTL/include/tao/pegtl/parse.hpp
index 9634655c5..3888fb3fc 100644
--- a/packages/PEGTL/include/tao/pegtl/parse.hpp
+++ b/packages/PEGTL/include/tao/pegtl/parse.hpp
@@ -4,7 +4,7 @@
#ifndef TAO_PEGTL_PARSE_HPP
#define TAO_PEGTL_PARSE_HPP
-#include <utility>
+#include <type_traits>
#include "apply_mode.hpp"
#include "config.hpp"
@@ -16,17 +16,20 @@
namespace TAO_PEGTL_NAMESPACE
{
- template< typename Rule,
- template< typename... > class Action = nothing,
- template< typename... > class Control = normal,
- apply_mode A = apply_mode::action,
- rewind_mode M = rewind_mode::required,
- typename ParseInput,
- typename... States >
- auto parse( ParseInput&& in, States&&... st )
+ namespace internal
{
- return Control< Rule >::template match< A, M, Action, Control >( in, st... );
- }
+ [[nodiscard]] inline auto get_position( const position& p ) noexcept( std::is_nothrow_copy_constructible_v< position > )
+ {
+ return p;
+ }
+
+ template< typename ParseInput >
+ [[nodiscard]] position get_position( const ParseInput& in ) noexcept( noexcept( position( in.position() ) ) )
+ {
+ return in.position();
+ }
+
+ } // namespace internal
template< typename Rule,
template< typename... > class Action = nothing,
@@ -35,46 +38,31 @@ namespace TAO_PEGTL_NAMESPACE
rewind_mode M = rewind_mode::required,
typename ParseInput,
typename... States >
- auto parse_nested( position op, ParseInput&& in, States&&... st )
+ auto parse( ParseInput&& in, States&&... st )
{
-#if defined( __cpp_exceptions )
- try {
- return parse< Rule, Action, Control, A, M >( in, st... );
- }
- catch( parse_error& e ) {
- e.add_position( std::move( op ) );
- throw;
- }
-#else
- (void)op;
- return parse< Rule, Action, Control, A, M >( in, st... );
-#endif
+ return Control< Rule >::template match< A, M, Action, Control >( in, st... );
}
- // NOTE: The oi.position() in the version below can be expensive for lazy
- // inputs, which is why the version below does not simply call the version
- // above with said oi.position() as first parameter.
-
template< typename Rule,
template< typename... > class Action = nothing,
template< typename... > class Control = normal,
apply_mode A = apply_mode::action,
rewind_mode M = rewind_mode::required,
- typename OuterInput,
+ typename Outer,
typename ParseInput,
typename... States >
- auto parse_nested( const OuterInput& oi, ParseInput&& in, States&&... st )
+ auto parse_nested( const Outer& o, ParseInput&& in, States&&... st )
{
#if defined( __cpp_exceptions )
try {
return parse< Rule, Action, Control, A, M >( in, st... );
}
catch( parse_error& e ) {
- e.add_position( oi.position() );
+ e.add_position( internal::get_position( o ) );
throw;
}
#else
- (void)oi;
+ (void)o;
return parse< Rule, Action, Control, A, M >( in, st... );
#endif
}
diff --git a/packages/PEGTL/include/tao/pegtl/parse_error.hpp b/packages/PEGTL/include/tao/pegtl/parse_error.hpp
index f471fb420..df62ce681 100644
--- a/packages/PEGTL/include/tao/pegtl/parse_error.hpp
+++ b/packages/PEGTL/include/tao/pegtl/parse_error.hpp
@@ -107,6 +107,11 @@ namespace TAO_PEGTL_NAMESPACE
}
m_impl->add_position( std::move( p ) );
}
+
+ void add_position( const position& p )
+ {
+ add_position( position( p ) );
+ }
};
} // namespace TAO_PEGTL_NAMESPACE
diff --git a/packages/PEGTL/include/tao/pegtl/read_input.hpp b/packages/PEGTL/include/tao/pegtl/read_input.hpp
index 7b748d093..019c9cd20 100644
--- a/packages/PEGTL/include/tao/pegtl/read_input.hpp
+++ b/packages/PEGTL/include/tao/pegtl/read_input.hpp
@@ -42,8 +42,8 @@ namespace TAO_PEGTL_NAMESPACE
~read_input() = default;
- void operator=( const read_input& ) = delete;
- void operator=( read_input&& ) = delete;
+ read_input& operator=( const read_input& ) = delete;
+ read_input& operator=( read_input&& ) = delete;
};
template< typename... Ts >
diff --git a/packages/PEGTL/include/tao/pegtl/string_input.hpp b/packages/PEGTL/include/tao/pegtl/string_input.hpp
index 2d83a6334..43349eb6f 100644
--- a/packages/PEGTL/include/tao/pegtl/string_input.hpp
+++ b/packages/PEGTL/include/tao/pegtl/string_input.hpp
@@ -30,8 +30,8 @@ namespace TAO_PEGTL_NAMESPACE
~string_holder() = default;
- void operator=( const string_holder& ) = delete;
- void operator=( string_holder&& ) = delete;
+ string_holder& operator=( const string_holder& ) = delete;
+ string_holder& operator=( string_holder&& ) = delete;
};
} // namespace internal
@@ -52,8 +52,8 @@ namespace TAO_PEGTL_NAMESPACE
~string_input() = default;
- void operator=( const string_input& ) = delete;
- void operator=( string_input&& ) = delete;
+ string_input& operator=( const string_input& ) = delete;
+ string_input& operator=( string_input&& ) = delete;
};
template< typename... Ts >
diff --git a/packages/PEGTL/include/tao/pegtl/version.hpp b/packages/PEGTL/include/tao/pegtl/version.hpp
index 4d460c536..2ffc6b953 100644
--- a/packages/PEGTL/include/tao/pegtl/version.hpp
+++ b/packages/PEGTL/include/tao/pegtl/version.hpp
@@ -4,10 +4,10 @@
#ifndef TAO_PEGTL_VERSION_HPP
#define TAO_PEGTL_VERSION_HPP
-#define TAO_PEGTL_VERSION "3.2.0"
+#define TAO_PEGTL_VERSION "3.2.1"
#define TAO_PEGTL_VERSION_MAJOR 3
#define TAO_PEGTL_VERSION_MINOR 2
-#define TAO_PEGTL_VERSION_PATCH 0
+#define TAO_PEGTL_VERSION_PATCH 1
#endif
diff --git a/packages/PEGTL/src/example/pegtl/CMakeLists.txt b/packages/PEGTL/src/example/pegtl/CMakeLists.txt
index 46ffeb068..64bb7380d 100644
--- a/packages/PEGTL/src/example/pegtl/CMakeLists.txt
+++ b/packages/PEGTL/src/example/pegtl/CMakeLists.txt
@@ -8,6 +8,7 @@ set(example_sources
csv1.cpp
csv2.cpp
dynamic_match.cpp
+ expression.cpp
hello_world.cpp
indent_aware.cpp
json_analyze.cpp
@@ -29,7 +30,6 @@ set(example_sources
peg2pegtl.cpp
proto3.cpp
recover.cpp
- skipper.cpp
s_expression.cpp
sum.cpp
symbol_table.cpp
diff --git a/packages/PEGTL/src/example/pegtl/expression.cpp b/packages/PEGTL/src/example/pegtl/expression.cpp
new file mode 100644
index 000000000..5d490f93f
--- /dev/null
+++ b/packages/PEGTL/src/example/pegtl/expression.cpp
@@ -0,0 +1,613 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#if !defined( __cpp_exceptions )
+#include <iostream>
+int main()
+{
+ std::cerr << "Exception support required, example unavailable." << std::endl;
+ return 1;
+}
+#else
+
+#include <algorithm>
+#include <cassert>
+#include <cstdint>
+#include <cstring>
+#include <iomanip>
+#include <iostream>
+#include <stdexcept>
+#include <tuple>
+#include <variant>
+#include <vector>
+
+#include <tao/pegtl.hpp>
+
+namespace TAO_PEGTL_NAMESPACE::expression
+{
+ // Expression parsing with prefix, postfix and infix operators, ternary
+ // operator and a couple of other special cases supported.
+
+ // The handling of operator precedences with left and right binding power is
+ // based on https://github.com/matklad/minipratt/blob/master/src/bin/pratt.rs
+
+ // It correctly recognises all operators with their precedence and associativity,
+ // however is still very much work-in-progress regarding a lot of details...
+
+ // TODO: Fix missing whitespace-skip before infix/postfix operators.
+ // TODO: Decide whether to use must everywhere or nowhere?
+ // TODO: Decide whether to suppress actions for sub-rules.
+ // TODO: Finalise the event-style interface or change to fake actions or actions with ops?
+ // TODO: Decide on where to use config vs. where to use grammar template parameters.
+ // TODO: Choose customisation points vs. copy-n-paste customisation.
+ // TODO: Constexpr-ify where possible with C++20.
+
+ namespace internal
+ {
+ struct prefix_info
+ {
+ prefix_info( const std::string_view n, const std::uint8_t pbp ) noexcept
+ : name( n ),
+ prefix_binding_power( pbp )
+ {
+ assert( pbp );
+ }
+
+ std::string name;
+
+ std::uint8_t prefix_binding_power;
+ };
+
+ struct infix_postfix_info
+ {
+ infix_postfix_info( const std::string_view n, const std::uint8_t lbp, const std::uint8_t rbp = 0 ) noexcept
+ : infix_postfix_info( n, std::string_view(), lbp, rbp )
+ {}
+
+ infix_postfix_info( const std::string_view n, const std::string_view o, const std::uint8_t lbp, const std::uint8_t rbp = 0 ) noexcept
+ : name( n ),
+ other( o ),
+ left_binding_power( lbp ),
+ right_binding_power( rbp )
+ {
+ if( right_binding_power > 0 ) {
+ assert( std::min( left_binding_power, right_binding_power ) & 1 );
+ assert( 2 * std::min( left_binding_power, right_binding_power ) + 1 == left_binding_power + right_binding_power );
+ }
+ assert( left_binding_power > 0 );
+ }
+
+ [[nodiscard]] bool is_infix() const noexcept
+ {
+ return right_binding_power != 0;
+ }
+
+ [[nodiscard]] bool is_postfix() const noexcept
+ {
+ return right_binding_power == 0;
+ }
+
+ std::string name;
+ std::string other; // Used for the ':' of the ternary operator etc.
+
+ std::uint8_t left_binding_power;
+ std::uint8_t right_binding_power;
+ };
+
+ template< typename ParseInput >
+ [[nodiscard]] bool match_string_view( ParseInput& in, const std::string_view sv )
+ {
+ if( in.size( sv.size() ) >= sv.size() ) {
+ if( std::memcmp( in.current(), sv.data(), sv.size() ) == 0 ) {
+ in.bump( sv.size() );
+ return true;
+ }
+ }
+ return false;
+ }
+
+ template< typename ParseInput, typename OperatorInfo >
+ [[nodiscard]] const OperatorInfo* match_prefix( ParseInput& in, const std::size_t max_length, const std::vector< OperatorInfo >& ops )
+ {
+ const std::size_t max = std::min( max_length, in.size( max_length ) );
+ for( std::string op( in.current(), max ); !op.empty(); op.pop_back() ) {
+ if( const auto i = std::find_if( ops.begin(), ops.end(), [ = ]( const OperatorInfo& info ) { return info.name == op; } ); i != ops.end() ) {
+ in.bump( op.size() );
+ return &*i;
+ }
+ }
+ return nullptr;
+ }
+
+ template< typename ParseInput, typename OperatorInfo >
+ [[nodiscard]] const OperatorInfo* match_infix_postfix( ParseInput& in, const std::size_t max_length, const std::vector< OperatorInfo >& ops, const std::uint8_t min_precedence )
+ {
+ const std::size_t max = std::min( max_length, in.size( max_length ) );
+ for( std::string op( in.current(), max ); !op.empty(); op.pop_back() ) {
+ if( const auto i = std::find_if( ops.begin(), ops.end(), [ = ]( const OperatorInfo& info ) { return info.name == op; } ); ( i != ops.end() ) && ( i->left_binding_power >= min_precedence ) ) {
+ in.bump( op.size() );
+ return &*i;
+ }
+ }
+ return nullptr;
+ }
+
+ template< typename T >
+ [[nodiscard]] std::vector< T > sorted_operator_vector( const std::initializer_list< T >& t )
+ {
+ std::vector< T > v{ t };
+ const auto less = []( const auto& l, const auto& r ) { return l.name < r.name; };
+ std::sort( v.begin(), v.end(), less );
+ return v;
+ }
+
+ struct operator_maps
+ {
+ // clang-format off
+ operator_maps()
+ : prefix( sorted_operator_vector( {
+ prefix_info( "!", 80 ),
+ prefix_info( "+", 80 ),
+ prefix_info( "-", 80 ),
+ prefix_info( "~", 80 ),
+ prefix_info( "*", 80 ),
+ prefix_info( "&", 80 ),
+ prefix_info( "++", 80 ),
+ prefix_info( "--", 80 )
+ } ) ),
+ infix_postfix( sorted_operator_vector( {
+ infix_postfix_info( "::", 99, 100 ), // Special: Followed by identifier (or template-space-identifer, which we don't support yet).
+ infix_postfix_info( ".*", 37, 38 ),
+ infix_postfix_info( "->*", 37, 38 ),
+ infix_postfix_info( "*", 35, 36 ),
+ infix_postfix_info( "/", 35, 36 ),
+ infix_postfix_info( "%", 35, 36 ),
+ infix_postfix_info( "+", 33, 34 ),
+ infix_postfix_info( "-", 33, 34 ),
+ infix_postfix_info( "<<", 31, 32 ),
+ infix_postfix_info( ">>", 31, 32 ),
+ infix_postfix_info( "<=>", 29, 30 ),
+ infix_postfix_info( "<", 27, 28 ),
+ infix_postfix_info( "<=", 27, 28 ),
+ infix_postfix_info( ">", 27, 28 ),
+ infix_postfix_info( ">=", 27, 28 ),
+ infix_postfix_info( "==", 25, 26 ),
+ infix_postfix_info( "!=", 25, 26 ),
+ infix_postfix_info( "&", 23, 24 ),
+ infix_postfix_info( "^", 21, 22 ),
+ infix_postfix_info( "|", 19, 20 ),
+ infix_postfix_info( "&&", 17, 18 ),
+ infix_postfix_info( "||", 15, 16 ),
+ infix_postfix_info( "?", ":", 14, 13 ), // Special: Ternary operator.
+ infix_postfix_info( "=", 12, 11 ),
+ infix_postfix_info( "+=", 12, 11 ),
+ infix_postfix_info( "-=", 12, 11 ),
+ infix_postfix_info( "*=", 12, 11 ),
+ infix_postfix_info( "/=", 12, 11 ),
+ infix_postfix_info( "%=", 12, 11 ),
+ infix_postfix_info( "<<=", 12, 11 ),
+ infix_postfix_info( ">>=", 12, 11 ),
+ infix_postfix_info( "&=", 12, 11 ),
+ infix_postfix_info( "^=", 12, 11 ),
+ infix_postfix_info( "|=", 12, 11 ),
+ // infix_postfix_info( ",", 9, 10 ), // TODO: Enable, but forbid in function argument list.
+ infix_postfix_info( "[", "]", 90 ), // Special: Argument list.
+ infix_postfix_info( "(", ")", 90 ), // Special: Argument list.
+ infix_postfix_info( ".", 90 ), // Special: Followed by identifier.
+ infix_postfix_info( "->", 90 ), // Special: Followed by identifier.
+ infix_postfix_info( "++", 90 ),
+ infix_postfix_info( "--", 90 )
+ } ) ),
+ max_prefix_length( std::max_element( prefix.begin(), prefix.end(), []( const auto& l, const auto& r ) { return l.name.size() < r.name.size(); } )->name.size() ),
+ max_infix_postfix_length( std::max_element( infix_postfix.begin(), infix_postfix.end(), []( const auto& l, const auto& r ) { return l.name.size() < r.name.size(); } )->name.size() )
+ {
+ // These are C++20 operators with the correct associativity and relative precedence, however some are still missing:
+ // TODO: Compound literal (C99), _Alignof (C11), Functional cast, sizeof, co_await, co_yield, throw, new, new[], delete, delete[], C-style casts.
+ }
+ // clang-format on
+
+ const std::vector< prefix_info > prefix;
+ const std::vector< infix_postfix_info > infix_postfix;
+
+ const std::size_t max_prefix_length;
+ const std::size_t max_infix_postfix_length;
+ };
+
+ struct string_view_rule
+ {
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput >
+ [[nodiscard]] static bool match( ParseInput& in, const std::string_view sv ) noexcept( noexcept( match_string_view( in, sv ) ) )
+ {
+ return match_string_view( in, sv );
+ }
+ };
+
+ struct comment
+ : seq< one< '#' >, until< eolf > >
+ {};
+
+ struct ignored
+ : sor< space, comment >
+ {};
+
+ template< typename Literal, typename Identifier >
+ struct expression;
+
+ template< typename Literal, typename Identifier >
+ struct bracket_expression
+ {
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename Result,
+ typename Config >
+ [[nodiscard]] static bool match( ParseInput& in, Result& res, const Config& cfg, const std::uint8_t /*unused*/ )
+ {
+ return Control< if_must< one< '(' >, star< ignored >, expression< Literal, Identifier >, star< ignored >, one< ')' > > >::template match< A, M, Action, Control >( in, res, cfg, 0 );
+ }
+ };
+
+ template< typename Literal, typename Identifier >
+ struct prefix_expression
+ {
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename Result,
+ typename Config >
+ [[nodiscard]] static bool match( ParseInput& in, Result& res, const Config& cfg, const std::uint8_t /*unused*/ )
+ {
+ if( const auto* info = match_prefix( in, cfg.max_prefix_length, cfg.prefix ) ) {
+ (void)Control< must< star< ignored >, expression< Literal, Identifier > > >::template match< A, M, Action, Control >( in, res, cfg, info->prefix_binding_power );
+ if constexpr( A == apply_mode::action ) {
+ res.prefix( info->name );
+ }
+ return true;
+ }
+ return false;
+ }
+ };
+
+ template< typename Literal, typename Identifier >
+ struct infix_postfix_expression
+ {
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename Result,
+ typename Config >
+ [[nodiscard]] static bool match( ParseInput& in, Result& res, const Config& cfg, const std::uint8_t min )
+ {
+ if( const auto* info = match_infix_postfix( in, cfg.max_infix_postfix_length, cfg.infix_postfix, min ) ) {
+ if( info->name == "?" ) {
+ (void)Control< must< star< ignored >, expression< Literal, Identifier > > >::template match< A, M, Action, Control >( in, res, cfg, 0 );
+ (void)Control< must< star< ignored >, string_view_rule > >::template match< A, M, Action, Control >( in, info->other );
+ (void)Control< must< star< ignored >, expression< Literal, Identifier > > >::template match< A, M, Action, Control >( in, res, cfg, info->right_binding_power );
+ if constexpr( A == apply_mode::action ) {
+ res.ternary( info->name, info->other );
+ }
+ return true;
+ }
+ if( ( info->name == "." ) || ( info->name == "::" ) || ( info->name == "->" ) ) {
+ (void)Control< must< star< ignored >, Identifier > >::template match< A, M, Action, Control >( in, res, cfg, 0 );
+ if constexpr( A == apply_mode::action ) {
+ res.infix( info->name );
+ }
+ return true;
+ }
+ if( ( info->name == "(" ) || ( info->name == "[" ) ) {
+ const std::size_t size = res.term_stack.size(); // TODO: Determine number of arguments without relying on res!?
+ (void)Control< must< star< ignored >, opt< list_must< expression< Literal, Identifier >, one< ',' >, ignored > > > >::template match< A, M, Action, Control >( in, res, cfg, 0 );
+ (void)Control< must< star< ignored >, string_view_rule > >::template match< A, M, Action, Control >( in, info->other );
+ if constexpr( A == apply_mode::action ) {
+ res.call( info->name, info->other, res.term_stack.size() - size );
+ }
+ return true;
+ }
+ if( info->is_infix() ) {
+ (void)Control< must< star< ignored >, expression< Literal, Identifier > > >::template match< A, M, Action, Control >( in, res, cfg, info->right_binding_power );
+ if constexpr( A == apply_mode::action ) {
+ res.infix( info->name );
+ }
+ return true;
+ }
+ if( info->is_postfix() ) {
+ if constexpr( A == apply_mode::action ) {
+ res.postfix( info->name );
+ }
+ return true;
+ }
+ }
+ return false;
+ }
+ };
+
+ template< typename Literal, typename Identifier >
+ struct first_expression
+ : sor< Literal, Identifier, bracket_expression< Literal, Identifier >, prefix_expression< Literal, Identifier > >
+ {};
+
+ template< typename Literal, typename Identifier >
+ struct expression
+ : seq< first_expression< Literal, Identifier >, star< infix_postfix_expression< Literal, Identifier > > >
+ {};
+
+ } // namespace internal
+
+ template< typename Literal, typename Identifier >
+ struct grammar
+ {
+ using rule_t = grammar;
+ using subs_t = type_list< internal::expression< Literal, Identifier > >;
+
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename Result >
+ [[nodiscard]] static bool match( ParseInput& in, Result& res )
+ {
+ const internal::operator_maps cfg;
+ return match< A, M, Action, Control >( in, res, cfg );
+ }
+
+ template< apply_mode A,
+ rewind_mode M,
+ template< typename... >
+ class Action,
+ template< typename... >
+ class Control,
+ typename ParseInput,
+ typename Result,
+ typename Config >
+ [[nodiscard]] static bool match( ParseInput& in, Result& res, const Config& cfg )
+ {
+ return Control< internal::expression< Literal, Identifier > >::template match< A, M, Action, Control >( in, res, cfg, 0 );
+ }
+ };
+
+} // namespace TAO_PEGTL_NAMESPACE::expression
+
+namespace application
+{
+ namespace pegtl = TAO_PEGTL_NAMESPACE;
+
+ struct term_t;
+
+ using tuple_t = std::tuple< std::string, std::vector< term_t > >;
+ using variant_t = std::variant< std::int64_t, std::string, tuple_t >;
+
+ struct term_t
+ {
+ explicit term_t( const std::int64_t l ) noexcept
+ : variant( l )
+ {}
+
+ explicit term_t( std::string&& s ) noexcept
+ : variant( std::move( s ) )
+ {}
+
+ explicit term_t( variant_t&& v ) noexcept
+ : variant( std::move( v ) )
+ {}
+
+ variant_t variant;
+ };
+
+ [[nodiscard]] inline std::string operator+( const char* l, const std::string_view r )
+ {
+ return std::string( l ) + " '" + std::string( r ) + "'";
+ }
+
+ struct result
+ {
+ void infix( const std::string_view op )
+ {
+ assert( term_stack.size() >= 2 );
+ {
+ variant_t tmp = tuple_t( "infix" + op, { std::move( term_stack.at( term_stack.size() - 2 ) ), std::move( term_stack.at( term_stack.size() - 1 ) ) } );
+ term_stack.pop_back();
+ term_stack.back().variant = std::move( tmp );
+ }
+ assert( string_stack.size() >= 2 );
+ {
+ std::string tmp = "( " + string_stack.at( string_stack.size() - 2 ) + " " + std::string( op ) + " " + string_stack.at( string_stack.size() - 1 ) + " )";
+ string_stack.pop_back();
+ string_stack.back() = std::move( tmp );
+ }
+ }
+
+ void prefix( const std::string_view op )
+ {
+ assert( term_stack.size() >= 1 ); // NOLINT(readability-container-size-empty)
+ {
+ variant_t tmp = tuple_t( "prefix" + op, { std::move( term_stack.at( term_stack.size() - 1 ) ) } );
+ term_stack.back().variant = std::move( tmp );
+ }
+ assert( string_stack.size() >= 1 ); // NOLINT(readability-container-size-empty)
+ {
+ std::string tmp = std::string( op ) + "( " + string_stack.at( string_stack.size() - 1 ) + " )";
+ string_stack.back() = std::move( tmp );
+ }
+ }
+
+ void postfix( const std::string_view op )
+ {
+ assert( term_stack.size() >= 1 ); // NOLINT(readability-container-size-empty)
+ {
+ variant_t tmp = tuple_t( "postfix" + op, { std::move( term_stack.at( term_stack.size() - 1 ) ) } );
+ term_stack.back().variant = std::move( tmp );
+ }
+ assert( string_stack.size() >= 1 ); // NOLINT(readability-container-size-empty)
+ {
+ std::string tmp = "( " + string_stack.at( string_stack.size() - 1 ) + " )" + std::string( op );
+ string_stack.back() = std::move( tmp );
+ }
+ }
+
+ void ternary( const std::string_view op, const std::string_view o2 )
+ {
+ assert( term_stack.size() >= 2 );
+ {
+ variant_t tmp = tuple_t( "ternary", { std::move( term_stack.at( term_stack.size() - 3 ) ), std::move( term_stack.at( term_stack.size() - 2 ) ), std::move( term_stack.at( term_stack.size() - 1 ) ) } );
+ term_stack.pop_back();
+ term_stack.pop_back();
+ term_stack.back().variant = std::move( tmp );
+ }
+ assert( string_stack.size() >= 2 );
+ {
+ std::string tmp = "( " + string_stack.at( string_stack.size() - 3 ) + " " + std::string( op ) + " " + string_stack.at( string_stack.size() - 2 ) + " " + std::string( o2 ) + " " + string_stack.at( string_stack.size() - 1 ) + " )";
+ string_stack.pop_back();
+ string_stack.pop_back();
+ string_stack.back() = std::move( tmp );
+ }
+ }
+
+ void call( const std::string_view op, const std::string_view o2, const std::size_t args )
+ {
+ assert( term_stack.size() > args );
+ {
+ variant_t tmp = tuple_t( "call '" + std::string( op ) + std::string( o2 ) + "'", std::vector< term_t >( term_stack.end() - args - 1, term_stack.end() ) );
+ for( std::size_t i = 0; i < args; ++i ) {
+ term_stack.pop_back();
+ }
+ term_stack.back().variant = ( std::move( tmp ) );
+ }
+ assert( string_stack.size() > args );
+ {
+ std::string tmp = *( string_stack.end() - args - 1 ) + std::string( op ) + " ";
+ for( std::size_t i = 0; i < args; ++i ) {
+ if( i > 0 ) {
+ tmp += ", ";
+ }
+ tmp += *( string_stack.end() - args + i );
+ }
+ tmp += " " + std::string( o2 );
+ string_stack.resize( string_stack.size() - args );
+ string_stack.back() = std::move( tmp );
+ }
+ }
+
+ void number( const std::int64_t l )
+ {
+ term_stack.emplace_back( l );
+ string_stack.emplace_back( std::to_string( l ) );
+ }
+
+ void identifier( const std::string& id )
+ {
+ term_stack.emplace_back( id );
+ string_stack.emplace_back( id );
+ }
+
+ std::vector< term_t > term_stack;
+ std::vector< std::string > string_stack;
+ };
+
+ inline std::ostream& operator<<( std::ostream& o, const term_t& t );
+
+ inline std::ostream& operator<<( std::ostream& o, const tuple_t& t )
+ {
+ o << "{ " << std::get< 0 >( t );
+ for( const auto& res : std::get< 1 >( t ) ) {
+ o << " " << res;
+ }
+ o << " }";
+ return o;
+ }
+
+ inline std::ostream& operator<<( std::ostream& o, const variant_t& v )
+ {
+ std::visit( [ & ]( const auto& t ) { o << t; }, v );
+ return o;
+ }
+
+ inline std::ostream& operator<<( std::ostream& o, const term_t& t )
+ {
+ o << t.variant;
+ return o;
+ }
+
+ struct literal
+ : pegtl::plus< pegtl::digit >
+ {};
+
+ struct grammar
+ : pegtl::must< pegtl::expression::grammar< literal, pegtl::identifier >, pegtl::eof >
+ {};
+
+ template< typename Rule >
+ struct action
+ : pegtl::nothing< Rule >
+ {};
+
+ template<>
+ struct action< literal >
+ {
+ template< typename Input, typename... States >
+ static void apply( const Input& in, result& res, States&&... /*unused*/ )
+ {
+ res.number( std::stoll( in.string() ) );
+ }
+ };
+
+ template<>
+ struct action< pegtl::identifier >
+ {
+ template< typename Input, typename... States >
+ static void apply( const Input& in, result& res, States&&... /*unused*/ )
+ {
+ res.identifier( in.string() );
+ }
+ };
+
+} // namespace application
+
+int main( int argc, char** argv )
+{
+ // if( TAO_PEGTL_NAMESPACE::analyze< application::grammar >() != 0 ) {
+ // return 1;
+ // }
+ for( int i = 1; i < argc; ++i ) {
+ TAO_PEGTL_NAMESPACE::argv_input in( argv, i );
+ try {
+ application::result res;
+ TAO_PEGTL_NAMESPACE::parse< application::grammar, application::action >( in, res );
+ std::cout << "Input: " << argv[ i ] << std::endl;
+ assert( res.term_stack.size() == 1 );
+ assert( res.string_stack.size() == 1 );
+ std::cout << "Result: " << res.string_stack.at( 0 ) << std::endl;
+ std::cout << "Result: " << res.term_stack.at( 0 ) << std::endl;
+ }
+ catch( const TAO_PEGTL_NAMESPACE::parse_error& e ) {
+ const auto p = e.positions().front();
+ std::cerr << e.what() << '\n'
+ << in.line_at( p ) << '\n'
+ << std::setw( p.column ) << '^' << '\n';
+ }
+ }
+ return 0;
+}
+
+#endif
diff --git a/packages/PEGTL/src/example/pegtl/json_classes.hpp b/packages/PEGTL/src/example/pegtl/json_classes.hpp
index c9f26de16..3e69515b0 100644
--- a/packages/PEGTL/src/example/pegtl/json_classes.hpp
+++ b/packages/PEGTL/src/example/pegtl/json_classes.hpp
@@ -35,8 +35,8 @@ namespace example
json_base( const json_base& ) = delete;
json_base( json_base&& ) = delete;
- void operator=( const json_base& ) = delete;
- void operator=( json_base&& ) = delete;
+ json_base& operator=( const json_base& ) = delete;
+ json_base& operator=( json_base&& ) = delete;
virtual void stream( std::ostream& ) const = 0;
diff --git a/packages/PEGTL/src/example/pegtl/json_parse.cpp b/packages/PEGTL/src/example/pegtl/json_parse.cpp
index f3c572d12..240a8898c 100644
--- a/packages/PEGTL/src/example/pegtl/json_parse.cpp
+++ b/packages/PEGTL/src/example/pegtl/json_parse.cpp
@@ -6,6 +6,7 @@
#include <tao/pegtl.hpp>
#include <tao/pegtl/contrib/json.hpp>
+#include <tao/pegtl/contrib/limit_depth.hpp>
#include <tao/pegtl/contrib/trace.hpp>
#include "json_errors.hpp"
@@ -16,6 +17,15 @@ namespace example
{
using grammar = pegtl::seq< pegtl::json::text, pegtl::eof >;
+ template< typename >
+ struct action
+ {};
+
+ template<>
+ struct action< pegtl::json::value >
+ : pegtl::limit_depth< 42 >
+ {};
+
} // namespace example
int main( int argc, char** argv ) // NOLINT(bugprone-exception-escape)
@@ -30,7 +40,7 @@ int main( int argc, char** argv ) // NOLINT(bugprone-exception-escape)
pegtl::argv_input in( argv, 1 );
#if defined( __cpp_exceptions )
try {
- pegtl::parse< example::grammar, pegtl::nothing, example::control >( in );
+ pegtl::parse< example::grammar, example::action, example::control >( in );
}
catch( const pegtl::parse_error& e ) {
const auto p = e.positions().front();
@@ -40,7 +50,7 @@ int main( int argc, char** argv ) // NOLINT(bugprone-exception-escape)
return 1;
}
#else
- if( !pegtl::parse< example::grammar, pegtl::nothing, example::control >( in ) ) {
+ if( !pegtl::parse< example::grammar, example::action, example::control >( in ) ) {
std::cerr << "error occurred" << std::endl;
return 1;
}
diff --git a/packages/PEGTL/src/example/pegtl/skipper.cpp b/packages/PEGTL/src/example/pegtl/skipper.cpp
deleted file mode 100644
index fd11200bb..000000000
--- a/packages/PEGTL/src/example/pegtl/skipper.cpp
+++ /dev/null
@@ -1,88 +0,0 @@
-// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
-// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
-
-#include <iostream>
-#include <string>
-
-#include <tao/pegtl.hpp>
-#include <tao/pegtl/contrib/skip.hpp>
-
-namespace pegtl = TAO_PEGTL_NAMESPACE;
-
-namespace demo
-{
- // define your grammar without concerning yourself with skipping whitespace
-
- // clang-format off
- struct key : pegtl::identifier {};
- struct value : pegtl::identifier {};
- struct assign : pegtl::one< '=' > {};
- struct grammar : pegtl::seq< key, assign, value, pegtl::eof > {};
- // clang-format on
-
- // define your actions as usual
-
- template< typename Rule >
- struct action
- {};
-
- template<>
- struct action< key >
- {
- template< typename ActionInput >
- static void apply( const ActionInput& in )
- {
- std::cout << "key: '" << in.string() << "'\n";
- }
- };
-
- template<>
- struct action< value >
- {
- template< typename ActionInput >
- static void apply( const ActionInput& in )
- {
- std::cout << "value: '" << in.string() << "'\n";
- }
- };
-
- // now specify how and where to skip whitespace
-
- using skip_how = pegtl::star< pegtl::sor< pegtl::space, pegtl::eol > >;
-
- // clang-format off
- template< typename > inline constexpr bool where = false;
- template<> inline constexpr bool where< key > = true;
- template<> inline constexpr bool where< value > = true;
- // clang-format on
-
- // as 'skip<>' can not take 'where' as a template parameter directly, we need to wrap it.
- struct skip_where
- {
- // when to skip
- template< typename Rule >
- static constexpr auto value = where< Rule >;
- };
-
- template< typename Rule >
- using control = pegtl::skip< skip_how, skip_where >::control< Rule >;
-
- // The above is a first step, the helper (skip<>) may change in the future
- // to allow separate skipping before and after and also support different skip
- // rules for each rule/set-of-rules. Also, a more convenient way to specify
- // the whole thing.
-
-} // namespace demo
-
-int main( int argc, char** argv ) // NOLINT(bugprone-exception-escape)
-{
- if( argc > 1 ) {
- pegtl::argv_input in( argv, 1 );
- if( pegtl::parse< demo::grammar, demo::action, demo::control >( in ) ) {
- std::cout << "success!" << std::endl;
- }
- else {
- std::cerr << "failure." << std::endl;
- }
- }
-}
diff --git a/packages/PEGTL/src/example/pegtl/token_input.cpp b/packages/PEGTL/src/example/pegtl/token_input.cpp
index 6f03dad7e..7d2795a3f 100644
--- a/packages/PEGTL/src/example/pegtl/token_input.cpp
+++ b/packages/PEGTL/src/example/pegtl/token_input.cpp
@@ -35,8 +35,8 @@ namespace TAO_PEGTL_NAMESPACE
~token_action_input() = default;
- token_action_input operator=( const token_action_input& ) = delete;
- token_action_input operator=( token_action_input&& ) = delete;
+ token_action_input& operator=( const token_action_input& ) = delete;
+ token_action_input& operator=( token_action_input&& ) = delete;
[[nodiscard]] const iterator_t& iterator() const noexcept
{
@@ -101,8 +101,8 @@ namespace TAO_PEGTL_NAMESPACE
~token_parse_input() = default;
- token_parse_input operator=( const token_parse_input& ) = delete;
- token_parse_input operator=( token_parse_input&& ) = delete;
+ token_parse_input& operator=( const token_parse_input& ) = delete;
+ token_parse_input& operator=( token_parse_input&& ) = delete;
void discard() const noexcept {}
diff --git a/packages/PEGTL/src/test/pegtl/CMakeLists.txt b/packages/PEGTL/src/test/pegtl/CMakeLists.txt
index 889b0520a..6bfacb4a7 100644
--- a/packages/PEGTL/src/test/pegtl/CMakeLists.txt
+++ b/packages/PEGTL/src/test/pegtl/CMakeLists.txt
@@ -23,6 +23,7 @@ set(test_sources
change_action_and_states.cpp
change_state.cpp
change_states.cpp
+ check_bytes.cpp
contains.cpp
contrib_alphabet.cpp
contrib_analyze.cpp
@@ -65,6 +66,8 @@ set(test_sources
internal_endian.cpp
internal_file_mapper.cpp
internal_file_opener.cpp
+ limit_bytes.cpp
+ limit_depth.cpp
parse_error.cpp
pegtl_string_t.cpp
position.cpp
diff --git a/packages/PEGTL/src/test/pegtl/check_bytes.cpp b/packages/PEGTL/src/test/pegtl/check_bytes.cpp
new file mode 100644
index 000000000..b773abed9
--- /dev/null
+++ b/packages/PEGTL/src/test/pegtl/check_bytes.cpp
@@ -0,0 +1,50 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#include <tao/pegtl/contrib/check_bytes.hpp>
+
+#include "test.hpp"
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ struct test_rule
+ : star< alpha >
+ {};
+
+ struct test_grammar
+ : seq< test_rule, eof >
+ {};
+
+ template< typename Rule >
+ struct test_action
+ : nothing< Rule >
+ {};
+
+ template<>
+ struct test_action< test_rule >
+ : check_bytes< 5 >
+ {};
+
+ void unit_test()
+ {
+ memory_input<> i1( "aaa", __FUNCTION__ );
+ const auto r1 = pegtl::parse< test_grammar >( i1 );
+ TAO_PEGTL_TEST_ASSERT( r1 );
+
+ memory_input<> i2( "aaaaaaaaaaa", __FUNCTION__ );
+ const auto r2 = pegtl::parse< test_grammar >( i2 );
+ TAO_PEGTL_TEST_ASSERT( r2 );
+
+ memory_input<> i3( "aaa", __FUNCTION__ );
+ const auto r3 = pegtl::parse< test_grammar, test_action >( i3 );
+ TAO_PEGTL_TEST_ASSERT( r3 );
+
+#if defined( __cpp_exceptions )
+ memory_input<> i4( "aaaaaaaaaaa", __FUNCTION__ );
+ TAO_PEGTL_TEST_THROWS( pegtl::parse< test_grammar, test_action >( i4 ) );
+#endif
+ }
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#include "main.hpp"
diff --git a/packages/PEGTL/src/test/pegtl/contrib_instantiate.cpp b/packages/PEGTL/src/test/pegtl/contrib_instantiate.cpp
index 5c6c268cb..fc85dec75 100644
--- a/packages/PEGTL/src/test/pegtl/contrib_instantiate.cpp
+++ b/packages/PEGTL/src/test/pegtl/contrib_instantiate.cpp
@@ -32,8 +32,8 @@ namespace TAO_PEGTL_NAMESPACE
dtor = true;
}
- void operator=( test_class&& ) = delete;
- void operator=( const test_class& ) = delete;
+ test_class& operator=( test_class&& ) = delete;
+ test_class& operator=( const test_class& ) = delete;
};
using test_grammar = sor< alpha, digit >;
diff --git a/packages/PEGTL/src/test/pegtl/limit_bytes.cpp b/packages/PEGTL/src/test/pegtl/limit_bytes.cpp
new file mode 100644
index 000000000..94e66c3f6
--- /dev/null
+++ b/packages/PEGTL/src/test/pegtl/limit_bytes.cpp
@@ -0,0 +1,50 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#include <tao/pegtl/contrib/limit_bytes.hpp>
+
+#include "test.hpp"
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ struct test_rule
+ : star< alpha >
+ {};
+
+ struct test_grammar
+ : seq< test_rule, eof >
+ {};
+
+ template< typename Rule >
+ struct test_action
+ : nothing< Rule >
+ {};
+
+ template<>
+ struct test_action< test_rule >
+ : limit_bytes< 5 >
+ {};
+
+ void unit_test()
+ {
+ memory_input<> i1( "aaa", __FUNCTION__ );
+ const auto r1 = pegtl::parse< test_grammar >( i1 );
+ TAO_PEGTL_TEST_ASSERT( r1 );
+
+ memory_input<> i2( "aaaaaaaaaaa", __FUNCTION__ );
+ const auto r2 = pegtl::parse< test_grammar >( i2 );
+ TAO_PEGTL_TEST_ASSERT( r2 );
+
+ memory_input<> i3( "aaa", __FUNCTION__ );
+ const auto r3 = pegtl::parse< test_grammar, test_action >( i3 );
+ TAO_PEGTL_TEST_ASSERT( r3 );
+
+#if defined( __cpp_exceptions )
+ memory_input<> i4( "aaaaaaaaaaa", __FUNCTION__ );
+ TAO_PEGTL_TEST_THROWS( pegtl::parse< test_grammar, test_action >( i4 ) );
+#endif
+ }
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#include "main.hpp"
diff --git a/packages/PEGTL/src/test/pegtl/limit_depth.cpp b/packages/PEGTL/src/test/pegtl/limit_depth.cpp
new file mode 100644
index 000000000..8fb1b2713
--- /dev/null
+++ b/packages/PEGTL/src/test/pegtl/limit_depth.cpp
@@ -0,0 +1,50 @@
+// Copyright (c) 2021 Dr. Colin Hirsch and Daniel Frey
+// Please see LICENSE for license or visit https://github.com/taocpp/PEGTL/
+
+#include <tao/pegtl/contrib/limit_depth.hpp>
+
+#include "test.hpp"
+
+namespace TAO_PEGTL_NAMESPACE
+{
+ struct test_recursive
+ : seq< alpha, opt< test_recursive > >
+ {};
+
+ struct test_grammar
+ : seq< test_recursive, eof >
+ {};
+
+ template< typename Rule >
+ struct test_action
+ : nothing< Rule >
+ {};
+
+ template<>
+ struct test_action< test_recursive >
+ : limit_depth< 5 >
+ {};
+
+ void unit_test()
+ {
+ memory_input<> i1( "aaa", __FUNCTION__ );
+ const auto r1 = pegtl::parse< test_grammar >( i1 );
+ TAO_PEGTL_TEST_ASSERT( r1 );
+
+ memory_input<> i2( "aaaaaaaaaaa", __FUNCTION__ );
+ const auto r2 = pegtl::parse< test_grammar >( i2 );
+ TAO_PEGTL_TEST_ASSERT( r2 );
+
+ memory_input<> i3( "aaa", __FUNCTION__ );
+ const auto r3 = pegtl::parse< test_grammar, test_action >( i3 );
+ TAO_PEGTL_TEST_ASSERT( r3 );
+
+#if defined( __cpp_exceptions )
+ memory_input<> i4( "aaaaaaaaaaa", __FUNCTION__ );
+ TAO_PEGTL_TEST_THROWS( pegtl::parse< test_grammar, test_action >( i4 ) );
+#endif
+ }
+
+} // namespace TAO_PEGTL_NAMESPACE
+
+#include "main.hpp"
--
GitLab