Merge pull request #135 from hanickadot/multiline

multiline support + new asserts
hanickadot · Nov 12, 2020 · d0f3778 · d0f3778
2 parents a947f4f + e3d95c7
commit d0f3778
Show file tree

Hide file tree

Showing 28 changed files with 2,322 additions and 1,941 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,5 +1,6 @@
 *.o
 *.d
+**/*.tmp
 test
 result
 tests/benchmark-exec/*

diff --git a/README.md b/README.md
@@ -19,11 +19,12 @@ ctre::match<"REGEX">(subject); // C++20
 * Searching (`search` or `starts_with`)
 * Capturing content (named captures are supported too)
 * Back-Reference (\g{N} syntax, and \1...\9 syntax too)
+* Multiline support (with `multi_`) functions
+* Unicode properties and UTF-8 support
 
 The library is implementing most of the PCRE syntax with a few exceptions:
 
 * atomic groups
-* boundaries other than `^$`
 * callouts
 * comments
 * conditional patterns

diff --git a/include/ctll/fixed_string.hpp.orig b/include/ctll/fixed_string.hpp.orig
diff --git a/include/ctre/actions/asserts.inc.hpp b/include/ctre/actions/asserts.inc.hpp
@@ -3,12 +3,27 @@
 
 // push_assert_begin
 template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_assert_begin, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
-	return pcre_context{ctll::push_front(assert_begin(), subject.stack), subject.parameters};
+	return pcre_context{ctll::push_front(assert_line_begin(), subject.stack), subject.parameters};
 }
 
 // push_assert_end
 template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_assert_end, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
-	return pcre_context{ctll::push_front(assert_end(), subject.stack), subject.parameters};
+	return pcre_context{ctll::push_front(assert_line_end(), subject.stack), subject.parameters};
+}
+
+// push_assert_begin
+template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_assert_subject_begin, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
+	return pcre_context{ctll::push_front(assert_subject_begin(), subject.stack), subject.parameters};
+}
+
+// push_assert_subject_end
+template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_assert_subject_end, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
+	return pcre_context{ctll::push_front(assert_subject_end(), subject.stack), subject.parameters};
+}
+
+// push_assert_subject_end_with_lineend
+template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_assert_subject_end_with_lineend, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
+	return pcre_context{ctll::push_front(assert_subject_end_line(), subject.stack), subject.parameters};
 }
 
 #endif
diff --git a/include/ctre/actions/boundaries.inc.hpp b/include/ctre/actions/boundaries.inc.hpp
@@ -0,0 +1,14 @@
+#ifndef CTRE__ACTIONS__BOUNDARIES__HPP
+#define CTRE__ACTIONS__BOUNDARIES__HPP
+
+// push_word_boundary
+template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_word_boundary, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
+	return pcre_context{ctll::push_front(boundary<word_chars>(), subject.stack), subject.parameters};
+}
+
+// push_not_word_boundary
+template <auto V, typename... Ts, typename Parameters> static constexpr auto apply(pcre::push_not_word_boundary, ctll::term<V>, pcre_context<ctll::list<Ts...>, Parameters> subject) {
+	return pcre_context{ctll::push_front(boundary<negative_set<word_chars>>(), subject.stack), subject.parameters};
+}
+
+#endif
diff --git a/include/ctre/atoms.hpp b/include/ctre/atoms.hpp
@@ -1,6 +1,7 @@
 #ifndef CTRE__ATOMS__HPP
 #define CTRE__ATOMS__HPP
 
+#include "atoms_characters.hpp"
 #include <cstdint>
 
 namespace ctre {
@@ -14,6 +15,8 @@ struct end_cycle_mark { };
 struct end_lookahead_mark { };
 template <size_t Id> struct numeric_mark { };
 
+struct any { };
+
 // actual AST of regexp
 template <auto... Str> struct string { };
 template <typename... Opts> struct select { };
@@ -51,9 +54,17 @@ struct atomic_start { };
 
 template <typename... Content> struct atomic_group { };
 
-struct assert_begin { };
-struct assert_end { };
+template <typename... Content> struct boundary { };
+template <typename... Content> struct not_boundary { };
+
+using word_boundary = boundary<word_chars>;
+using not_word_boundary = not_boundary<word_chars>;
 
+struct assert_subject_begin { };
+struct assert_subject_end { };
+struct assert_subject_end_line{ };
+struct assert_line_begin { };
+struct assert_line_end { };
 
 }
 

diff --git a/include/ctre/atoms_characters.hpp b/include/ctre/atoms_characters.hpp
@@ -23,26 +23,22 @@ template <auto V> struct character {
 	}
 };
 
-struct any {
-	template <typename CharT> CTRE_FORCE_INLINE static constexpr bool match_char(CharT) noexcept { return true; }
-};
-
 template <typename... Content> struct negative_set {
-	template <typename CharT> inline static constexpr bool match_char(CharT value) noexcept {
+	template <typename CharT> CTRE_FORCE_INLINE static constexpr bool match_char(CharT value) noexcept {
 		return !(Content::match_char(value) || ... || false);
 	}
 };
 
 template <typename... Content> struct set {
-	template <typename CharT> inline static constexpr bool match_char(CharT value) noexcept {
+	template <typename CharT> CTRE_FORCE_INLINE static constexpr bool match_char(CharT value) noexcept {
 		return (Content::match_char(value) || ... || false);
 	}
 };
 
 template <auto... Cs> struct enumeration : set<character<Cs>...> { };
 
 template <typename... Content> struct negate {
-	template <typename CharT> inline static constexpr bool match_char(CharT value) noexcept {
+	template <typename CharT> CTRE_FORCE_INLINE static constexpr bool match_char(CharT value) noexcept {
 		return !(Content::match_char(value) || ... || false);
 	}
 };