Refactor: create new Paragraphs.h/cpp

11 files changed, 199 insertions, 193 deletions

diff --git a/toolsrc/src/Paragraphs.cpp b/toolsrc/src/Paragraphs.cpp
new file mode 100644
index 000000000..5efa9a7b3
--- /dev/null
+++ b/toolsrc/src/Paragraphs.cpp
@@ -0,0 +1,163 @@
+#include "Paragraphs.h"
+#include "vcpkg_Files.h"
+
+namespace vcpkg { namespace Paragraphs
+{
+    struct Parser
+    {
+        Parser(const char* c, const char* e) : cur(c), end(e)
+        {
+        }
+
+    private:
+        const char* cur;
+        const char* const end;
+
+        void peek(char& ch) const
+        {
+            if (cur == end)
+                ch = 0;
+            else
+                ch = *cur;
+        }
+
+        void next(char& ch)
+        {
+            if (cur == end)
+                ch = 0;
+            else
+            {
+                ++cur;
+                peek(ch);
+            }
+        }
+
+        void skip_spaces(char& ch)
+        {
+            while (ch == ' ' || ch == '\t')
+                next(ch);
+        }
+
+        static bool is_alphanum(char ch)
+        {
+            return (ch >= 'A' && ch <= 'Z')
+                || (ch >= 'a' && ch <= 'z')
+                || (ch >= '0' && ch <= '9');
+        }
+
+        static bool is_lineend(char ch)
+        {
+            return ch == '\r' || ch == '\n' || ch == 0;
+        }
+
+        void get_fieldvalue(char& ch, std::string& fieldvalue)
+        {
+            fieldvalue.clear();
+
+            auto beginning_of_line = cur;
+            do
+            {
+                // scan to end of current line (it is part of the field value)
+                while (!is_lineend(ch))
+                    next(ch);
+
+                fieldvalue.append(beginning_of_line, cur);
+
+                if (ch == '\r')
+                    next(ch);
+                if (ch == '\n')
+                    next(ch);
+
+                if (is_alphanum(ch))
+                {
+                    // Line begins a new field.
+                    return;
+                }
+
+                beginning_of_line = cur;
+
+                // Line may continue the current field with data or terminate the paragraph,
+                // depending on first nonspace character.
+                skip_spaces(ch);
+
+                if (is_lineend(ch))
+                {
+                    // Line was whitespace or empty.
+                    // This terminates the field and the paragraph.
+                    // We leave the blank line's whitespace consumed, because it doesn't matter.
+                    return;
+                }
+
+                // First nonspace is not a newline. This continues the current field value.
+                // We forcibly convert all newlines into single '\n' for ease of text handling later on.
+                fieldvalue.push_back('\n');
+            }
+            while (true);
+        }
+
+        void get_fieldname(char& ch, std::string& fieldname)
+        {
+            auto begin_fieldname = cur;
+            while (is_alphanum(ch) || ch == '-')
+                next(ch);
+            Checks::check_throw(ch == ':', "Expected ':'");
+            fieldname = std::string(begin_fieldname, cur);
+
+            // skip ': '
+            next(ch);
+            skip_spaces(ch);
+        }
+
+        void get_paragraph(char& ch, std::unordered_map<std::string, std::string>& fields)
+        {
+            fields.clear();
+            std::string fieldname;
+            std::string fieldvalue;
+            do
+            {
+                get_fieldname(ch, fieldname);
+
+                auto it = fields.find(fieldname);
+                Checks::check_throw(it == fields.end(), "Duplicate field");
+
+                get_fieldvalue(ch, fieldvalue);
+
+                fields.emplace(fieldname, fieldvalue);
+            }
+            while (!is_lineend(ch));
+        }
+
+    public:
+        std::vector<std::unordered_map<std::string, std::string>> get_paragraphs()
+        {
+            std::vector<std::unordered_map<std::string, std::string>> paragraphs;
+
+            char ch;
+            peek(ch);
+
+            while (ch != 0)
+            {
+                if (ch == '\n' || ch == '\r' || ch == ' ' || ch == '\t')
+                {
+                    next(ch);
+                    continue;
+                }
+
+                paragraphs.emplace_back();
+                get_paragraph(ch, paragraphs.back());
+            }
+
+            return paragraphs;
+        }
+    };
+
+    std::vector<std::unordered_map<std::string, std::string>> get_paragraphs(const fs::path& control_path)
+    {
+        return parse_paragraphs(Files::get_contents(control_path).get_or_throw());
+    }
+
+    std::vector<std::unordered_map<std::string, std::string>> parse_paragraphs(const std::string& str)
+    {
+        return Parser(str.c_str(), str.c_str() + str.size()).get_paragraphs();
+    }
+}}
diff --git a/toolsrc/src/commands_cache.cpp b/toolsrc/src/commands_cache.cpp
index 5e20a39af..634a72a64 100644
--- a/toolsrc/src/commands_cache.cpp
+++ b/toolsrc/src/commands_cache.cpp
@@ -2,6 +2,7 @@
 #include "vcpkg_System.h"
 #include "vcpkg_Files.h"
 #include "vcpkg.h"
+#include "Paragraphs.h"
 
 namespace vcpkg
 {
@@ -17,7 +18,7 @@ namespace vcpkg
                 auto file_contents = Files::get_contents(path / "CONTROL");
                 if (auto text = file_contents.get())
                 {
-                    auto pghs = parse_paragraphs(*text);
+                    auto pghs = Paragraphs::parse_paragraphs(*text);
                     if (pghs.size() != 1)
                         continue;
 
diff --git a/toolsrc/src/commands_import.cpp b/toolsrc/src/commands_import.cpp
index 9cfc53d6c..7a6139085 100644
--- a/toolsrc/src/commands_import.cpp
+++ b/toolsrc/src/commands_import.cpp
@@ -1,5 +1,6 @@
 #include "vcpkg_Commands.h"
 #include "vcpkg.h"
+#include "Paragraphs.h"
 
 namespace vcpkg
 {
@@ -12,7 +13,7 @@ namespace vcpkg
         const fs::path include_directory(args.command_arguments[1]);
         const fs::path project_directory(args.command_arguments[2]);
 
-        auto pghs = get_paragraphs(control_file_path);
+        auto pghs = Paragraphs::get_paragraphs(control_file_path);
         Checks::check_throw(pghs.size() == 1, "Invalid control file for package");
 
         StatusParagraph spgh;
diff --git a/toolsrc/src/commands_installation.cpp b/toolsrc/src/commands_installation.cpp
index f6aeafa02..b5f4e1a5d 100644
--- a/toolsrc/src/commands_installation.cpp
+++ b/toolsrc/src/commands_installation.cpp
@@ -9,6 +9,7 @@
 #include "vcpkg_Dependencies.h"
 #include "vcpkg_Input.h"
 #include "vcpkg_Maps.h"
+#include "Paragraphs.h"
 
 namespace vcpkg
 {
@@ -21,7 +22,7 @@ namespace vcpkg
 
     static void build_internal(const package_spec& spec, const vcpkg_paths& paths, const fs::path& port_dir)
     {
-        auto pghs = get_paragraphs(port_dir / "CONTROL");
+        auto pghs = Paragraphs::get_paragraphs(port_dir / "CONTROL");
         Checks::check_exit(pghs.size() == 1, "Error: invalid control file");
         SourceParagraph source_paragraph(pghs[0]);
 
@@ -111,7 +112,7 @@ namespace vcpkg
                     }
                 }
 
-                auto pghs = parse_paragraphs(file_contents.get_or_throw());
+                auto pghs = Paragraphs::parse_paragraphs(file_contents.get_or_throw());
                 Checks::check_throw(pghs.size() == 1, "multiple paragraphs in control file");
                 install_package(paths, BinaryParagraph(pghs[0]), status_db);
                 System::println(System::color::success, "Package %s is installed", spec);
diff --git a/toolsrc/src/commands_portsdiff.cpp b/toolsrc/src/commands_portsdiff.cpp
index e15db3567..fdbaec0ac 100644
--- a/toolsrc/src/commands_portsdiff.cpp
+++ b/toolsrc/src/commands_portsdiff.cpp
@@ -7,6 +7,7 @@
 #include <iostream>
 #include <iomanip>
 #include <set>
+#include "Paragraphs.h"
 
 namespace vcpkg
 {
@@ -52,7 +53,7 @@ namespace vcpkg
 
             try
             {
-                auto pghs = get_paragraphs(path / "CONTROL");
+                auto pghs = Paragraphs::get_paragraphs(path / "CONTROL");
                 if (pghs.empty())
                     continue;
                 auto srcpgh = SourceParagraph(pghs[0]);
diff --git a/toolsrc/src/commands_search.cpp b/toolsrc/src/commands_search.cpp
index d59a338ac..399709cf3 100644
--- a/toolsrc/src/commands_search.cpp
+++ b/toolsrc/src/commands_search.cpp
@@ -1,6 +1,7 @@
 #include "vcpkg_Commands.h"
 #include "vcpkg_System.h"
 #include "vcpkg.h"
+#include "Paragraphs.h"
 
 namespace fs = std::tr2::sys;
 
@@ -15,7 +16,7 @@ namespace vcpkg
 
             try
             {
-                auto pghs = get_paragraphs(path / "CONTROL");
+                auto pghs = Paragraphs::get_paragraphs(path / "CONTROL");
                 if (pghs.empty())
                 {
                     continue;
diff --git a/toolsrc/src/commands_update.cpp b/toolsrc/src/commands_update.cpp
index 5d531ef39..b448091f7 100644
--- a/toolsrc/src/commands_update.cpp
+++ b/toolsrc/src/commands_update.cpp
@@ -2,6 +2,7 @@
 #include "vcpkg.h"
 #include "vcpkg_System.h"
 #include "vcpkg_Files.h"
+#include "Paragraphs.h"
 
 namespace vcpkg
 {
@@ -21,7 +22,7 @@ namespace vcpkg
             const auto& path = begin_it->path();
             try
             {
-                auto pghs = get_paragraphs(path / "CONTROL");
+                auto pghs = Paragraphs::get_paragraphs(path / "CONTROL");
                 if (pghs.empty())
                     continue;
                 auto srcpgh = SourceParagraph(pghs[0]);
diff --git a/toolsrc/src/lib.cpp b/toolsrc/src/lib.cpp
index 3c844ac3f..5de9f9639 100644
--- a/toolsrc/src/lib.cpp
+++ b/toolsrc/src/lib.cpp
@@ -11,6 +11,7 @@
 #include <cassert>
 #include "vcpkg_Files.h"
 #include "vcpkg_System.h"
+#include "Paragraphs.h"
 
 using namespace vcpkg;
 
@@ -51,7 +52,7 @@ static StatusParagraphs load_current_database(const fs::path& vcpkg_dir_status_f
     }
 
     auto text = Files::get_contents(vcpkg_dir_status_file).get_or_throw();
-    auto pghs = parse_paragraphs(text);
+    auto pghs = Paragraphs::parse_paragraphs(text);
 
     std::vector<std::unique_ptr<StatusParagraph>> status_pghs;
     for (auto&& p : pghs)
@@ -94,7 +95,7 @@ StatusParagraphs vcpkg::database_load_check(const vcpkg_paths& paths)
             continue;
 
         auto text = Files::get_contents(b->path()).get_or_throw();
-        auto pghs = parse_paragraphs(text);
+        auto pghs = Paragraphs::parse_paragraphs(text);
         for (auto&& p : pghs)
         {
             current_status_db.insert(std::make_unique<StatusParagraph>(p));
@@ -217,7 +218,7 @@ std::vector<std::string> vcpkg::get_unmet_package_dependencies(const vcpkg_paths
         std::vector<std::unordered_map<std::string, std::string>> pghs;
         try
         {
-            pghs = parse_paragraphs(*control_contents);
+            pghs = Paragraphs::parse_paragraphs(*control_contents);
         }
         catch (std::runtime_error)
         {
@@ -238,7 +239,7 @@ std::vector<std::string> vcpkg::get_unmet_package_build_dependencies(const vcpkg
         std::vector<std::unordered_map<std::string, std::string>> pghs;
         try
         {
-            pghs = parse_paragraphs(*control_contents);
+            pghs = Paragraphs::parse_paragraphs(*control_contents);
         }
         catch (std::runtime_error)
         {
diff --git a/toolsrc/src/main.cpp b/toolsrc/src/main.cpp
index 2200cd105..22c45e5ab 100644
--- a/toolsrc/src/main.cpp
+++ b/toolsrc/src/main.cpp
@@ -12,6 +12,7 @@
 #include "vcpkg_Files.h"
 #include "vcpkg_System.h"
 #include "vcpkg_Input.h"
+#include "Paragraphs.h"
 
 using namespace vcpkg;
 
@@ -115,7 +116,7 @@ static void loadConfig()
         std::string config_contents = Files::get_contents(localappdata / "vcpkg" / "config").get_or_throw();
 
         std::unordered_map<std::string, std::string> keys;
-        auto pghs = parse_paragraphs(config_contents);
+        auto pghs = Paragraphs::parse_paragraphs(config_contents);
         if (pghs.size() > 0)
             keys = pghs[0];
 
diff --git a/toolsrc/src/tests_paragraph.cpp b/toolsrc/src/tests_paragraph.cpp
index b06359b90..28309fd31 100644
--- a/toolsrc/src/tests_paragraph.cpp
+++ b/toolsrc/src/tests_paragraph.cpp
@@ -1,5 +1,6 @@
 #include "CppUnitTest.h"
 #include "vcpkg.h"
+#include "Paragraphs.h"
 
 #pragma comment(lib,"version")
 #pragma comment(lib,"winhttp")
@@ -81,9 +82,9 @@ namespace UnitTest1
         TEST_METHOD(SourceParagraph_Construct_Qualified_Depends)
         {
             vcpkg::SourceParagraph pgh({
-                { "Source", "zlib" },
-                { "Version", "1.2.8" },
-                { "Build-Depends", "libA [windows], libB [uwp]" }
+                {"Source", "zlib"},
+                {"Version", "1.2.8"},
+                {"Build-Depends", "libA [windows], libB [uwp]"}
             });
 
             Assert::AreEqual("zlib", pgh.name.c_str());
@@ -97,7 +98,6 @@ namespace UnitTest1
             Assert::AreEqual("uwp", pgh.depends[1].qualifier.c_str());
         }
 
-
         TEST_METHOD(BinaryParagraph_Construct_Minimum)
         {
             vcpkg::BinaryParagraph pgh({
@@ -153,14 +153,14 @@ namespace UnitTest1
         TEST_METHOD(parse_paragraphs_empty)
         {
             const char* str = "";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::IsTrue(pghs.empty());
         }
 
         TEST_METHOD(parse_paragraphs_one_field)
         {
             const char* str = "f1: v1";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(1), pghs[0].size());
             Assert::AreEqual("v1", pghs[0]["f1"].c_str());
@@ -171,7 +171,7 @@ namespace UnitTest1
             const char* str =
                 "f1: v1\n"
                 "f2: v2";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(2), pghs[0].size());
             Assert::AreEqual("v1", pghs[0]["f1"].c_str());
@@ -186,7 +186,7 @@ namespace UnitTest1
                 "\n"
                 "f3: v3\n"
                 "f4: v4";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(2), pghs.size());
             Assert::AreEqual(size_t(2), pghs[0].size());
             Assert::AreEqual("v1", pghs[0]["f1"].c_str());
@@ -204,7 +204,7 @@ namespace UnitTest1
                 "F:\n"
                 "0:\n"
                 "F-2:\n";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(5), pghs[0].size());
         }
@@ -218,7 +218,7 @@ namespace UnitTest1
                 "\n"
                 "f3: v3\n"
                 "f4: v4";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(2), pghs.size());
         }
 
@@ -227,7 +227,7 @@ namespace UnitTest1
             const char* str =
                 "f1:\n"
                 "f2: ";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(2), pghs[0].size());
             Assert::AreEqual("", pghs[0]["f1"].c_str());
@@ -243,7 +243,7 @@ namespace UnitTest1
                 "f2:\r\n"
                 " f2\r\n"
                 " continue\r\n";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual("simple\n f1", pghs[0]["f1"].c_str());
             Assert::AreEqual("\n f2\n continue", pghs[0]["f2"].c_str());
@@ -257,7 +257,7 @@ namespace UnitTest1
                 "\r\n"
                 "f3: v3\r\n"
                 "f4: v4";
-            auto pghs = vcpkg::parse_paragraphs(str);
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(str);
             Assert::AreEqual(size_t(2), pghs.size());
             Assert::AreEqual(size_t(2), pghs[0].size());
             Assert::AreEqual("v1", pghs[0]["f1"].c_str());
@@ -277,7 +277,7 @@ namespace UnitTest1
                 {"Multi-Arch", "same"},
             });
             ss << pgh;
-            auto pghs = vcpkg::parse_paragraphs(ss.str());
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(ss.str());
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(4), pghs[0].size());
             Assert::AreEqual("zlib", pghs[0]["Package"].c_str());
@@ -299,7 +299,7 @@ namespace UnitTest1
                 {"Multi-Arch", "same"},
             });
             ss << pgh;
-            auto pghs = vcpkg::parse_paragraphs(ss.str());
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(ss.str());
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual(size_t(7), pghs[0].size());
             Assert::AreEqual("zlib", pghs[0]["Package"].c_str());
@@ -321,7 +321,7 @@ namespace UnitTest1
                 {"Depends", "a, b, c"},
             });
             ss << pgh;
-            auto pghs = vcpkg::parse_paragraphs(ss.str());
+            auto pghs = vcpkg::Paragraphs::parse_paragraphs(ss.str());
             Assert::AreEqual(size_t(1), pghs.size());
             Assert::AreEqual("a, b, c", pghs[0]["Depends"].c_str());
         }
diff --git a/toolsrc/src/vcpkg.cpp b/toolsrc/src/vcpkg.cpp
index f705858cc..a9ef963e7 100644
--- a/toolsrc/src/vcpkg.cpp
+++ b/toolsrc/src/vcpkg.cpp
@@ -1,160 +1,5 @@
 #include "vcpkg.h"
 #include <regex>
-#include "vcpkg_Files.h"
-#include "vcpkglib_helpers.h"
-
-namespace
-{
-    using namespace vcpkg;
-
-    struct Parser
-    {
-        Parser(const char* c, const char* e) : cur(c), end(e)
-        {
-        }
-
-    private:
-        const char* cur;
-        const char* const end;
-
-        void peek(char& ch) const
-        {
-            if (cur == end)
-                ch = 0;
-            else
-                ch = *cur;
-        }
-
-        void next(char& ch)
-        {
-            if (cur == end)
-                ch = 0;
-            else
-            {
-                ++cur;
-                peek(ch);
-            }
-        }
-
-        void skip_spaces(char& ch)
-        {
-            while (ch == ' ' || ch == '\t')
-                next(ch);
-        }
-
-        static bool is_alphanum(char ch)
-        {
-            return (ch >= 'A' && ch <= 'Z')
-                || (ch >= 'a' && ch <= 'z')
-                || (ch >= '0' && ch <= '9');
-        }
-
-        static bool is_lineend(char ch)
-        {
-            return ch == '\r' || ch == '\n' || ch == 0;
-        }
-
-        void get_fieldvalue(char& ch, std::string& fieldvalue)
-        {
-            fieldvalue.clear();
-
-            auto beginning_of_line = cur;
-            do
-            {
-                // scan to end of current line (it is part of the field value)
-                while (!is_lineend(ch))
-                    next(ch);
-
-                fieldvalue.append(beginning_of_line, cur);
-
-                if (ch == '\r')
-                    next(ch);
-                if (ch == '\n')
-                    next(ch);
-
-                if (is_alphanum(ch))
-                {
-                    // Line begins a new field.
-                    return;
-                }
-
-                beginning_of_line = cur;
-
-                // Line may continue the current field with data or terminate the paragraph,
-                // depending on first nonspace character.
-                skip_spaces(ch);
-
-                if (is_lineend(ch))
-                {
-                    // Line was whitespace or empty.
-                    // This terminates the field and the paragraph.
-                    // We leave the blank line's whitespace consumed, because it doesn't matter.
-                    return;
-                }
-
-                // First nonspace is not a newline. This continues the current field value.
-                // We forcibly convert all newlines into single '\n' for ease of text handling later on.
-                fieldvalue.push_back('\n');
-            }
-            while (true);
-        }
-
-        void get_fieldname(char& ch, std::string& fieldname)
-        {
-            auto begin_fieldname = cur;
-            while (is_alphanum(ch) || ch == '-')
-                next(ch);
-            Checks::check_throw(ch == ':', "Expected ':'");
-            fieldname = std::string(begin_fieldname, cur);
-
-            // skip ': '
-            next(ch);
-            skip_spaces(ch);
-        }
-
-        void get_paragraph(char& ch, std::unordered_map<std::string, std::string>& fields)
-        {
-            fields.clear();
-            std::string fieldname;
-            std::string fieldvalue;
-            do
-            {
-                get_fieldname(ch, fieldname);
-
-                auto it = fields.find(fieldname);
-                Checks::check_throw(it == fields.end(), "Duplicate field");
-
-                get_fieldvalue(ch, fieldvalue);
-
-                fields.emplace(fieldname, fieldvalue);
-            }
-            while (!is_lineend(ch));
-        }
-
-    public:
-        std::vector<std::unordered_map<std::string, std::string>> get_paragraphs()
-        {
-            std::vector<std::unordered_map<std::string, std::string>> paragraphs;
-
-            char ch;
-            peek(ch);
-
-            while (ch != 0)
-            {
-                if (ch == '\n' || ch == '\r' || ch == ' ' || ch == '\t')
-                {
-                    next(ch);
-                    continue;
-                }
-
-                paragraphs.emplace_back();
-                get_paragraph(ch, paragraphs.back());
-            }
-
-            return paragraphs;
-        }
-    };
-}
 
 namespace vcpkg
 {
@@ -165,14 +10,4 @@ namespace vcpkg
             simple_desc.append("...");
         return simple_desc;
     }
-
-    std::vector<std::unordered_map<std::string, std::string>> get_paragraphs(const fs::path& control_path)
-    {
-        return parse_paragraphs(Files::get_contents(control_path).get_or_throw());
-    }
-
-    std::vector<std::unordered_map<std::string, std::string>> parse_paragraphs(const std::string& str)
-    {
-        return Parser(str.c_str(), str.c_str() + str.size()).get_paragraphs();
-    }
 }