Add parsing utilities (#3090)

* Add util/parsers skeletton * Add find_matching_parentheses * conditional refactor * Add find_not_in_parentheses * Add some doc * Add find_not_in_parentheses substr overload * Add glob_match * Remove warning
mamba-org · Jan 2, 2024 · 1b56e38 · 1b56e38
1 parent 30d62af
commit 1b56e38
Show file tree

Hide file tree

Showing 7 changed files with 571 additions and 19 deletions.
diff --git a/libmamba/CMakeLists.txt b/libmamba/CMakeLists.txt
@@ -155,6 +155,7 @@ set(
     ${LIBMAMBA_SOURCE_DIR}/util/encoding.cpp
     ${LIBMAMBA_SOURCE_DIR}/util/environment.cpp
     ${LIBMAMBA_SOURCE_DIR}/util/os_win.cpp
+    ${LIBMAMBA_SOURCE_DIR}/util/parsers.cpp
     ${LIBMAMBA_SOURCE_DIR}/util/path_manip.cpp
     ${LIBMAMBA_SOURCE_DIR}/util/string.cpp
     ${LIBMAMBA_SOURCE_DIR}/util/url.cpp
@@ -259,6 +260,7 @@ set(
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/build.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/cast.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/compare.hpp
+    ${LIBMAMBA_INCLUDE_DIR}/mamba/util/conditional.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/cryptography.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/deprecation.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/encoding.hpp
@@ -271,6 +273,7 @@ set(
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/iterator.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/json.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/os_win.hpp
+    ${LIBMAMBA_INCLUDE_DIR}/mamba/util/parsers.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/path_manip.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/string.hpp
     ${LIBMAMBA_INCLUDE_DIR}/mamba/util/tuple_hash.hpp

diff --git a/libmamba/include/mamba/util/conditional.hpp b/libmamba/include/mamba/util/conditional.hpp
@@ -0,0 +1,40 @@
+// Copyright (c) 2023, QuantStack and Mamba Contributors
+//
+// Distributed under the terms of the BSD 3-Clause License.
+//
+// The full license is in the file LICENSE, distributed with this software.
+
+#ifndef MAMBA_UTIL_CONDITIONAL_HPP
+#define MAMBA_UTIL_CONDITIONAL_HPP
+
+#include <type_traits>
+
+namespace mamba::util
+{
+
+    template <typename Int>
+    [[nodiscard]] auto if_else(bool condition, Int true_val, Int false_val) noexcept -> Int;
+
+    /********************
+     *  Implementation  *
+     ********************/
+
+    template <typename Int>
+    auto if_else(bool condition, Int true_val, Int false_val) noexcept -> Int
+    {
+        if constexpr (std::is_enum_v<Int>)
+        {
+            using int_t = std::underlying_type_t<Int>;
+            return static_cast<Int>(
+                if_else(condition, static_cast<int_t>(true_val), static_cast<int_t>(false_val))
+            );
+        }
+        else
+        {
+            using int_t = Int;
+            return static_cast<int_t>(condition) * true_val
+                   + (int_t(1) - static_cast<int_t>(condition)) * false_val;
+        }
+    }
+}
+#endif
diff --git a/libmamba/include/mamba/util/parsers.hpp b/libmamba/include/mamba/util/parsers.hpp
@@ -0,0 +1,108 @@
+// Copyright (c) 2023, QuantStack and Mamba Contributors
+//
+// Distributed under the terms of the BSD 3-Clause License.
+//
+// The full license is in the file LICENSE, distributed with this software.
+
+#ifndef MAMBA_UTIL_PARSERS_HPP
+#define MAMBA_UTIL_PARSERS_HPP
+
+#include <string_view>
+#include <utility>
+
+#include <tl/expected.hpp>
+
+namespace mamba::util
+{
+
+    enum struct ParseError
+    {
+        Ok,
+        InvalidInput,
+        NotFound,
+    };
+
+    /**
+     * Find the next matching parenthesese pair.
+     *
+     * Correctly matches parenteses together so that inner parentheses pairs are skipped.
+     * Open and closing pairs need not be differents.
+     * If an error is encountered, @p err is modified to contain the error, otherwise it is left
+     * as it is.
+     */
+    auto find_matching_parentheses_idx(  //
+        std::string_view text,
+        ParseError& err,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> std::pair<std::size_t, std::size_t>;
+
+    /**
+     * Find the next matching parenthesese pair.
+     *
+     * Correctly matches parentheses together so that inner parentheses pairs are skipped.
+     * Open and closing pairs need not be differents.
+     */
+    [[nodiscard]] auto find_matching_parentheses_idx(  //
+        std::string_view text,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> tl::expected<std::pair<std::size_t, std::size_t>, ParseError>;
+
+    auto find_matching_parentheses_str(  //
+        std::string_view text,
+        ParseError& err,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> std::string_view;
+
+    [[nodiscard]] auto find_matching_parentheses_str(  //
+        std::string_view text,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> tl::expected<std::string_view, ParseError>;
+
+    /**
+     * Find a character, except in mathcing parentheses pairs.
+     *
+     * Find the first occurence of the given character, except if such character is inside a valid
+     * pair of parentheses.
+     * Open and closing pairs need not be differents.
+     */
+    auto find_not_in_parentheses(  //
+        std::string_view text,
+        char c,
+        ParseError& err,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> std::size_t;
+
+    auto find_not_in_parentheses(  //
+        std::string_view text,
+        std::string_view val,
+        ParseError& err,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> std::size_t;
+
+    [[nodiscard]] auto find_not_in_parentheses(  //
+        std::string_view text,
+        char c,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> tl::expected<std::size_t, ParseError>;
+
+    [[nodiscard]] auto find_not_in_parentheses(  //
+        std::string_view text,
+        std::string_view val,
+        char open = '(',
+        char close = ')'
+    ) noexcept -> tl::expected<std::size_t, ParseError>;
+
+    /**
+     * Test wether the glob pattern @p pattern matches the string @p str.
+     */
+    [[nodiscard]] auto glob_match(std::string_view pattern, std::string_view str, char glob = '*')
+        -> bool;
+}
+#endif
diff --git a/libmamba/src/util/encoding.cpp b/libmamba/src/util/encoding.cpp
@@ -7,11 +7,11 @@
 #include <array>
 #include <cstdint>
 #include <cstring>
-#include <type_traits>
 
 #include <openssl/evp.h>
 
 #include "mamba/util/compare.hpp"
+#include "mamba/util/conditional.hpp"
 #include "mamba/util/encoding.hpp"
 #include "mamba/util/string.hpp"
 
@@ -37,24 +37,6 @@ namespace mamba::util
             high |= low & nibble_low_mask;
             return high;
         }
-
-        template <typename Int>
-        [[nodiscard]] auto if_else(bool condition, Int true_val, Int false_val) noexcept -> Int
-        {
-            if constexpr (std::is_enum_v<Int>)
-            {
-                using int_t = std::underlying_type_t<Int>;
-                return static_cast<Int>(
-                    if_else(condition, static_cast<int_t>(true_val), static_cast<int_t>(false_val))
-                );
-            }
-            else
-            {
-                using int_t = Int;
-                return static_cast<int_t>(condition) * true_val
-                       + (int_t(1) - static_cast<int_t>(condition)) * false_val;
-            }
-        }
     }
 
     auto nibble_to_hex(std::byte b) noexcept -> char