improve logic expression evaluation (microsoft#7508)

* better logic expression evaluation

Improve the logic expression evaluation currently used when filtering
dependencies.

Biggest improvements:
+  Allow '|' operator
+  Support nested '()'
+  Allow whitespace
+  Useful error message for malformed expressions

Also changed names of types to RawParagraph when that is what the original author was using.

ports/pango/CONTROL

@@ -2,4 +2,4 @@ Source: pango
 Version: 1.40.11-4
 Homepage: https://ftp.gnome.org/pub/GNOME/sources/pango/
 Description: Text and font handling library.
-Build-Depends: glib, gettext, cairo, fontconfig, freetype, harfbuzz[glib] (!windows-static)
+Build-Depends: glib, gettext, cairo, fontconfig, freetype, harfbuzz[glib] (!(windows&static))

toolsrc/include/vcpkg/binaryparagraph.h

@@ -2,6 +2,7 @@
 
 #include <vcpkg/packagespec.h>
 #include <vcpkg/sourceparagraph.h>
+#include <vcpkg/parse.h>
 
 #include <unordered_map>
 
@@ -13,7 +14,7 @@ namespace vcpkg
     struct BinaryParagraph
     {
         BinaryParagraph();
-        explicit BinaryParagraph(std::unordered_map<std::string, std::string> fields);
+        explicit BinaryParagraph(Parse::RawParagraph fields);
         BinaryParagraph(const SourceParagraph& spgh, const Triplet& triplet, const std::string& abi_tag);
         BinaryParagraph(const SourceParagraph& spgh, const FeatureParagraph& fpgh, const Triplet& triplet);
 

toolsrc/include/vcpkg/logicexpression.h

@@ -0,0 +1,10 @@
+#pragma once
+
+#include <string>
+
+namespace vcpkg
+{
+	// Evaluate simple vcpkg logic expressions.  An identifier in the expression is considered 'true'
+	// if it is a substring of the evaluation_context (typically the name of the triplet)
+	bool evaluate_expression(const std::string& expression, const std::string& evaluation_context);
+}

toolsrc/include/vcpkg/paragraphs.h

@@ -12,7 +12,6 @@ namespace vcpkg::Paragraphs
 
     Expected<RawParagraph> get_single_paragraph(const Files::Filesystem& fs, const fs::path& control_path);
     Expected<std::vector<RawParagraph>> get_paragraphs(const Files::Filesystem& fs, const fs::path& control_path);
-    Expected<RawParagraph> parse_single_paragraph(const std::string& str);
     Expected<std::vector<RawParagraph>> parse_paragraphs(const std::string& str);
 
     Parse::ParseExpected<SourceControlFile> try_load_port(const Files::Filesystem& fs, const fs::path& control_path);

toolsrc/include/vcpkg/statusparagraph.h

@@ -30,7 +30,7 @@ namespace vcpkg
     struct StatusParagraph
     {
         StatusParagraph() noexcept;
-        explicit StatusParagraph(std::unordered_map<std::string, std::string>&& fields);
+        explicit StatusParagraph(Parse::RawParagraph&& fields);
 
         bool is_installed() const { return want == Want::INSTALL && state == InstallState::INSTALLED; }
 

toolsrc/src/vcpkg/binaryparagraph.cpp

@@ -27,7 +27,7 @@ namespace vcpkg
 
     BinaryParagraph::BinaryParagraph() = default;
 
-    BinaryParagraph::BinaryParagraph(std::unordered_map<std::string, std::string> fields)
+    BinaryParagraph::BinaryParagraph(Parse::RawParagraph fields)
     {
         using namespace vcpkg::Parse;
 

toolsrc/src/vcpkg/build.cpp

@@ -938,7 +938,7 @@ namespace vcpkg::Build
                                Commands::Version::version());
     }
 
-    static BuildInfo inner_create_buildinfo(std::unordered_map<std::string, std::string> pgh)
+    static BuildInfo inner_create_buildinfo(Parse::RawParagraph pgh)
     {
         Parse::ParagraphParser parser(std::move(pgh));
 
@@ -995,7 +995,7 @@ namespace vcpkg::Build
 
     BuildInfo read_build_info(const Files::Filesystem& fs, const fs::path& filepath)
     {
-        const Expected<std::unordered_map<std::string, std::string>> pghs =
+        const Expected<Parse::RawParagraph> pghs =
             Paragraphs::get_single_paragraph(fs, filepath);
         Checks::check_exit(VCPKG_LINE_INFO, pghs.get() != nullptr, "Invalid BUILD_INFO file for package");
         return inner_create_buildinfo(*pghs.get());

toolsrc/src/vcpkg/commands.cache.cpp

@@ -14,7 +14,7 @@ namespace vcpkg::Commands::Cache
         std::vector<BinaryParagraph> output;
         for (auto&& path : paths.get_filesystem().get_files_non_recursive(paths.packages))
         {
-            const Expected<std::unordered_map<std::string, std::string>> pghs =
+            const Expected<Parse::RawParagraph> pghs =
                 Paragraphs::get_single_paragraph(paths.get_filesystem(), path / "CONTROL");
             if (const auto p = pghs.get())
             {

toolsrc/src/vcpkg/commands.import.cpp

@@ -108,7 +108,7 @@ namespace vcpkg::Commands::Import
         const fs::path include_directory(args.command_arguments[1]);
         const fs::path project_directory(args.command_arguments[2]);
 
-        const Expected<std::unordered_map<std::string, std::string>> pghs =
+        const Expected<Parse::RawParagraph> pghs =
             Paragraphs::get_single_paragraph(paths.get_filesystem(), control_file_path);
         Checks::check_exit(VCPKG_LINE_INFO,
                            pghs.get() != nullptr,

toolsrc/src/vcpkg/logicexpression.cpp

@@ -0,0 +1,285 @@
+
+#include "pch.h"
+
+#include <vcpkg/logicexpression.h>
+#include <vcpkg/base/checks.h>
+#include <vcpkg/base/system.print.h>
+
+#include <string>
+#include <vector>
+
+
+namespace vcpkg
+{
+    struct ParseError
+    {
+        ParseError(int column, std::string line, std::string message)
+            :column(column), line(line), message(message)
+        {}
+
+        const int column;
+        const std::string line;
+        const std::string message;
+
+        void print_error() const
+        {
+            System::print2(System::Color::error,
+                "Error: ", message, "\n"
+                "   on expression: \"", line, "\"\n",
+                "                   ", std::string(column, ' '), "^\n");
+            Checks::exit_fail(VCPKG_LINE_INFO);
+        }
+    };
+
+    // logic expression supports the following :
+    //  primary-expression:
+    //    ( logic-expression )
+    //    identifier
+    //  identifier:
+    //    alpha-numeric string of characters   
+    //  logic-expression: <- this is the entry point
+    //    not-expression
+    //    not-expression | logic-expression
+    //    not-expression & logic-expression
+    //  not-expression:
+    //    ! primary-expression
+    //    primary-expression
+    //
+    // | and & have equal precidence and cannot be used together at the same nesting level
+    //   for example a|b&c is not allowd but (a|b)&c and a|(b&c) are allowed.
+    class ExpressionParser
+    {
+    public:
+        ExpressionParser(const std::string& str, const std::string& evaluation_context)
+            : raw_text(str), evaluation_context(evaluation_context)
+        {
+            go_to_begin();
+
+            final_result = logic_expression();
+
+            if (current_iter != raw_text.end())
+            {
+                add_error("Invalid logic expression");
+            }
+
+            if (err)
+            {
+                err->print_error();
+                final_result = false;
+            }
+        }
+
+        bool get_result() const
+        {
+            return final_result;
+        }
+
+        bool has_error() const
+        {
+            return err == nullptr;
+        }
+
+    private:
+
+        bool final_result;
+
+        std::string::const_iterator current_iter;
+        const std::string& raw_text;
+		char current_char;
+
+		const std::string& evaluation_context;
+
+		std::unique_ptr<ParseError> err;
+
+        void add_error(std::string message, int column = -1)
+        {
+            // avoid castcading errors by only saving the first
+            if (!err)
+            {
+                if (column < 0)
+                {
+                    column = current_column();
+                }
+                err = std::make_unique<ParseError>(column, raw_text, message);
+            }
+
+            // Avoid error loops by skipping to the end
+            skip_to_end();
+        }
+
+        int current_column() const
+        {
+            return static_cast<int>(current_iter - raw_text.begin());
+        }
+
+        void go_to_begin()
+        {
+            current_iter = raw_text.begin();
+            current_char = (current_iter != raw_text.end() ? *current_iter : current_char);
+
+            if (current_char == ' ' || current_char == '\t')
+            {
+                next_skip_whitespace();
+            }
+        }
+        void skip_to_end()
+        {
+            current_iter = raw_text.end();
+            current_char = '\0';
+        }
+        char current() const
+        {
+            return current_char;
+        }
+        char next()
+        {
+            if (current_char != '\0')
+            {
+                current_iter++;
+                current_char = (current_iter != raw_text.end() ? *current_iter : '\0');
+            }
+            return current();
+        }
+        void skip_whitespace()
+        {
+            while (current_char == ' ' || current_char == '\t')
+            {
+                current_char = next();
+            }
+        }
+        char next_skip_whitespace()
+        {
+            next();
+            skip_whitespace();
+            return current_char;
+        }
+
+        static bool is_alphanum(char ch)
+        {
+            return (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9') || (ch == '-');
+        }
+
+        bool evaluate_identifier(const std::string name) const
+        {
+            return evaluation_context.find(name) != std::string::npos;
+        }
+
+        //  identifier:
+        //    alpha-numeric string of characters   
+        bool identifier_expression()
+        {
+            auto curr = current();
+            std::string name;
+
+            for (curr = current(); is_alphanum(curr); curr = next())
+            {
+                name += curr;
+            }
+
+            if (name.empty())
+            {
+                add_error("Invalid logic expression, unexpected character");
+                return false;
+            }
+
+            bool result = evaluate_identifier(name);
+            skip_whitespace();
+            return result;
+        }
+
+        //  not-expression:
+        //    ! primary-expression
+        //    primary-expression
+        bool not_expression()
+        {
+            if (current() == '!')
+            {
+                next_skip_whitespace();
+                return !primary_expression();
+            }
+
+            return primary_expression();
+        }
+
+
+        template <char oper, char other, bool operation(bool, bool)>
+        bool logic_expression_helper(bool seed)
+        {
+            do
+            {
+                // Support chains of the operator to avoid breaking backwards compatability
+                while (next() == oper) {};
+                seed = operation(not_expression(), seed);
+
+            } while (current() == oper);
+
+            if (current() == other)
+            {
+                add_error("Mixing & and | is not allowed, Use () to specify order of operations.");
+            }
+
+			skip_whitespace();
+            return seed;
+        }
+		static bool and_helper(bool left, bool right)
+		{
+			return left && right;
+		}
+		static bool or_helper(bool left, bool right)
+		{
+			return left || right;
+		}
+
+        //  logic-expression: <- entry point
+        //    not-expression
+        //    not-expression | logic-expression
+        //    not-expression & logic-expression
+        bool logic_expression()
+        {
+            auto result = not_expression();
+
+            switch (current())
+            {
+            case '|':
+            {
+                return logic_expression_helper< '|', '&', or_helper > (result);
+            }
+            case '&':
+            {
+                return logic_expression_helper< '&', '|', and_helper > (result);
+            }
+            default:
+                return result;
+            }
+        }
+
+        //  primary-expression:
+        //    ( logic-expression )
+        //    identifier
+        bool primary_expression()
+        {
+            if (current() == '(')
+            {
+                next_skip_whitespace();
+                bool result = logic_expression();
+                if (current() != ')')
+                {
+                    add_error("Error: missing closing )");
+                    return result;
+                }
+                next_skip_whitespace();
+                return result;
+            }
+
+            return identifier_expression();
+        }
+    };
+
+
+    bool evaluate_expression(const std::string& expression, const std::string& evaluation_context)
+    {
+        ExpressionParser parser(expression, evaluation_context);
+
+        return parser.get_result();
+    }
+}