22281039592

Committed 22 Feb 2026 04:37PM UTC coverage: 93.5% (+0.04%) from 93.464%

Build # 22281039592

Build Type

Pull #641

github

Committed by

web-flow

Commit Message

Merge 74cb9d0b3 into 672fb743f

Pull Request Pull Request #641: Feat/various improvements

Run Details

180 of 187 new or added lines in 10 files covered. (96.26%)

8 existing lines in 2 files now uncovered.

9250 of 9893 relevant lines covered (93.5%)

267283.45 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

77.38

/src/arkreactor/State.cpp

#include <Ark/State.hpp>

#include <Ark/Constants.hpp>
#include <Ark/Utils/Files.hpp>
#include <Ark/Compiler/Welder.hpp>

#ifdef _MSC_VER
#    pragma warning(push)
#    pragma warning(disable : 4996)
#endif

#include <Proxy/Picosha2.hpp>
#include <Ark/Compiler/BytecodeReader.hpp>
#include <fmt/core.h>
#include <fmt/color.h>

namespace Ark
{
    State::State(const std::vector<std::filesystem::path>& libenv) noexcept :
        m_debug_level(0),
        m_features(0),
        m_libenv(libenv),
        m_filename(ARK_NO_NAME_FILE),
        m_max_page_size(0)
    {
        // default value for builtin__sys:args is empty list
        const Value val(ValueType::List);
        m_bound[std::string(internal::Language::SysArgs)] = val;

        m_bound[std::string(internal::Language::SysProgramName)] = Value("");
    }

    bool State::feed(const std::string& bytecode_filename, const bool fail_with_exception)
    {
        if (!Utils::fileExists(bytecode_filename))
            return false;

        return feed(Utils::readFileAsBytes(bytecode_filename), fail_with_exception);
    }

    bool State::feed(const bytecode_t& bytecode, const bool fail_with_exception)
    {
        BytecodeReader bcr;
        bcr.feed(bytecode);
        if (!bcr.checkMagic())
            return false;

        m_bytecode = bytecode;

        try
        {
            configure(bcr);
            return true;
        }
        catch (const std::exception& e)
        {
            if (fail_with_exception)
                throw;

            fmt::println("{}", e.what());
            return false;
        }
    }

    bool State::compile(const std::string& file, const std::string& output)
    {
        Welder welder(m_debug_level, m_libenv, m_features);
        for (const auto& key : m_bound | std::views::keys)
            welder.registerSymbol(key);

        if (!welder.computeASTFromFile(file))
            return false;
        if (!welder.generateBytecode())
            return false;

        const std::string destination = output.empty() ? (file.substr(0, file.find_last_of('.')) + ".arkc") : output;
        if ((m_features & DisableCache) == 0 && !welder.saveBytecodeToFile(destination))
            return false;
        if (!feed(welder.bytecode()))
            return false;

        return true;
    }

    bool State::doFile(const std::string& file_path, const uint16_t features)
    {
        m_features = features;

        if (!Utils::fileExists(file_path))
        {
            fmt::print(fmt::fg(fmt::color::red), "Can not find file '{}'\n", file_path);
            return false;
        }
        m_filename = file_path;
        m_bound[std::string(internal::Language::SysProgramName)] = Value(std::filesystem::path(m_filename).filename().string());

        const bytecode_t bytecode = Utils::readFileAsBytes(file_path);
        BytecodeReader bcr;
        bcr.feed(bytecode);
        if (!bcr.checkMagic())  // couldn't read magic number, it's a source file
        {
            // check if it's in the arkscript cache
            const std::string filename = std::filesystem::path(file_path).filename().replace_extension(".arkc").string();
            const std::filesystem::path cache_directory = std::filesystem::path(file_path).parent_path() / ARK_CACHE_DIRNAME;
            const std::string bytecode_path = (cache_directory / filename).string();

            if (!exists(cache_directory) && (m_features & DisableCache) == 0)
            {
                try
                {
                    create_directory(cache_directory);
                }
                catch (const std::filesystem::filesystem_error&)
                {
                    m_features |= DisableCache;
                }
            }

            if (compile(file_path, bytecode_path))
                return true;
        }
        else if (feed(bytecode))  // it's a bytecode file
            return true;
        return false;
    }

    bool State::doString(const std::string& code, const uint16_t features)
    {
        m_features = features;

        Welder welder(m_debug_level, m_libenv, m_features);
        for (const auto& p : m_bound)
            welder.registerSymbol(p.first);

        if (!welder.computeASTFromString(code))
            return false;
        if (!welder.generateBytecode())
            return false;
        return feed(welder.bytecode());
    }

    void State::loadFunction(const std::string& name, Procedure::CallbackType&& function) noexcept
    {
        m_bound[name] = Value(std::move(function));
    }

    void State::setArgs(const std::vector<std::string>& args) noexcept
    {
        Value val(ValueType::List);
        std::ranges::transform(args, std::back_inserter(val.list()), [](const std::string& arg) {
            return Value(arg);
        });

        m_bound[std::string(internal::Language::SysArgs)] = val;
    }

    void State::setDebug(const unsigned level) noexcept
    {
        m_debug_level = level;
    }

    void State::setLibDirs(const std::vector<std::filesystem::path>& libenv) noexcept
    {
        m_libenv = libenv;
    }

    void State::configure(const BytecodeReader& bcr)
    {
        using namespace internal;

        const auto [major, minor, patch] = bcr.version();
        if (major != ARK_VERSION_MAJOR)
        {
            const std::string str_version = fmt::format("{}.{}.{}", major, minor, patch);
            throwStateError(fmt::format("Compiler and VM versions don't match: got {} while running {}", str_version, ARK_VERSION));
        }

        const auto bytecode_hash = bcr.sha256();

        std::vector<unsigned char> hash(picosha2::k_digest_size);
        picosha2::hash256(m_bytecode.begin() + bytecode::HeaderSize + picosha2::k_digest_size, m_bytecode.end(), hash);
        // checking integrity
        for (std::size_t j = 0; j < picosha2::k_digest_size; ++j)
        {
#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
            if (hash[j] != bytecode_hash[j])
                throwStateError("Integrity check failed");
#endif
        }

        const auto syms = bcr.symbols();
        const auto vals = bcr.values(syms);
        const auto files = bcr.filenames(vals);
        const auto inst_locs = bcr.instLocations(files);
        const auto [pages, _] = bcr.code(inst_locs);

        m_symbols = syms.symbols;
        m_constants = vals.values;
        m_filenames = files.filenames;
        m_inst_locations = inst_locs.locations;
        m_pages = pages;
        m_max_page_size = maxPageSize(m_pages);

        // Make m_code as a big contiguous chunk of instructions,
        // aligned on the biggest page size.
        // This might have a downside when we have a single big page and
        // a bunch of smaller ones, though I couldn't measure it while testing.
        m_code.resize(m_max_page_size * pages.size(), Instruction::NOP);
        addPagesToContiguousBytecode(pages, /* start= */ 0);
    }

    void State::reset() noexcept
    {
        m_symbols.clear();
        m_constants.clear();
        m_filenames.clear();
        m_inst_locations.clear();
        m_max_page_size = 0;
        m_code.clear();
        m_bound.clear();

        // default value for builtin__sys:args is empty list
        const Value val(ValueType::List);
        m_bound[std::string(internal::Language::SysArgs)] = val;

        m_bound[std::string(internal::Language::SysProgramName)] = Value("");
    }

    void State::addPagesToContiguousBytecode(const std::vector<bytecode_t>& pages, const std::size_t start)
    {
        for (std::size_t i = 0, end = pages.size(); i < end; ++i)
        {
            for (std::size_t j = 0, end_j = pages[i].size(); j < end_j; ++j)
                m_code[(start + i) * m_max_page_size + j] = pages[i][j];
        }
    }

    std::size_t State::maxPageSize(const std::vector<bytecode_t>& pages)
    {
        return std::ranges::max(pages, {}, &bytecode_t::size).size();
    }

    void State::extendBytecode(const std::vector<bytecode_t>& pages, const std::vector<std::string>& symbols, const std::vector<Value>& constants)
    {
        m_symbols = symbols;
        m_constants = constants;

        // do not modify m_pages so that we can start over
        m_max_page_size = std::max(m_max_page_size, maxPageSize(pages));

        m_code.resize(m_max_page_size * (m_pages.size() + pages.size()), internal::Instruction::NOP);
        addPagesToContiguousBytecode(m_pages, /* start= */ 0);
        addPagesToContiguousBytecode(pages, /* start= */ m_pages.size());
    }
}

#ifdef _MSC_VER
#    pragma warning(pop)
#endif

1	#include <Ark/State.hpp>
2
3	#include <Ark/Constants.hpp>
4	#include <Ark/Utils/Files.hpp>
5	#include <Ark/Compiler/Welder.hpp>
6
7	#ifdef _MSC_VER
8	# pragma warning(push)
9	# pragma warning(disable : 4996)
10	#endif
11
12	#include <Proxy/Picosha2.hpp>
13	#include <Ark/Compiler/BytecodeReader.hpp>
14	#include <fmt/core.h>
15	#include <fmt/color.h>
16
17	namespace Ark
18	{
19	State::State(const std::vector<std::filesystem::path>& libenv) noexcept :	1,188✔
20	m_debug_level(0),	396✔
21	m_features(0),	396✔
22	m_libenv(libenv),	396✔
23	m_filename(ARK_NO_NAME_FILE),	396✔
24	m_max_page_size(0)	396✔
25	{	396✔
26	// default value for builtin__sys:args is empty list
27	const Value val(ValueType::List);	396✔
28	m_bound[std::string(internal::Language::SysArgs)] = val;	396✔
29
30	m_bound[std::string(internal::Language::SysProgramName)] = Value("");	396✔
31	}	396✔
32
UNCOV 33	bool State::feed(const std::string& bytecode_filename, const bool fail_with_exception)	×
UNCOV 34	{	×
UNCOV 35	if (!Utils::fileExists(bytecode_filename))	×
36	return false;	×
37
UNCOV 38	return feed(Utils::readFileAsBytes(bytecode_filename), fail_with_exception);	×
UNCOV 39	}	×
40
41	bool State::feed(const bytecode_t& bytecode, const bool fail_with_exception)	266✔
42	{	266✔
43	BytecodeReader bcr;	266✔
44	bcr.feed(bytecode);	266✔
45	if (!bcr.checkMagic())	266✔
46	return false;	×
47
48	m_bytecode = bytecode;	266✔
49
50	try
51	{
52	configure(bcr);	266✔
53	return true;	264✔
54	}	2✔
55	catch (const std::exception& e)
56	{
57	if (fail_with_exception)	2✔
58	throw;	2✔
59
60	fmt::println("{}", e.what());	×
61	return false;	×
62	}	2✔
63	}	270✔
64
65	bool State::compile(const std::string& file, const std::string& output)	372✔
66	{	372✔
67	Welder welder(m_debug_level, m_libenv, m_features);	372✔
68	for (const auto& key : m_bound \| std::views::keys)	1,121✔
69	welder.registerSymbol(key);	749✔
70
71	if (!welder.computeASTFromFile(file))	372✔
72	return false;	×
73	if (!welder.generateBytecode())	293✔
74	return false;	×
75
76	const std::string destination = output.empty() ? (file.substr(0, file.find_last_of('.')) + ".arkc") : output;	250✔
77	if ((m_features & DisableCache) == 0 && !welder.saveBytecodeToFile(destination))	250✔
NEW 78	return false;	×
79	if (!feed(welder.bytecode()))	250✔
UNCOV 80	return false;	×
81
82	return true;	250✔
83	}	372✔
84
85	bool State::doFile(const std::string& file_path, const uint16_t features)	372✔
86	{	372✔
87	m_features = features;	372✔
88
89	if (!Utils::fileExists(file_path))	372✔
90	{
91	fmt::print(fmt::fg(fmt::color::red), "Can not find file '{}'\n", file_path);	×
92	return false;	×
93	}
94	m_filename = file_path;	372✔
95	m_bound[std::string(internal::Language::SysProgramName)] = Value(std::filesystem::path(m_filename).filename().string());	494✔
96
97	const bytecode_t bytecode = Utils::readFileAsBytes(file_path);	372✔
98	BytecodeReader bcr;	372✔
99	bcr.feed(bytecode);	372✔
100	if (!bcr.checkMagic()) // couldn't read magic number, it's a source file	372✔
101	{
102	// check if it's in the arkscript cache
103	const std::string filename = std::filesystem::path(file_path).filename().replace_extension(".arkc").string();	372✔
104	const std::filesystem::path cache_directory = std::filesystem::path(file_path).parent_path() / ARK_CACHE_DIRNAME;	372✔
105	const std::string bytecode_path = (cache_directory / filename).string();	372✔
106
107	if (!exists(cache_directory) && (m_features & DisableCache) == 0)	372✔
108	{
109	try
110	{
111	create_directory(cache_directory);	18✔
112	}	18✔
113	catch (const std::filesystem::filesystem_error&)
114	{
NEW 115	m_features \|= DisableCache;	×
116	}	18✔
117	}	18✔
118
119	if (compile(file_path, bytecode_path))	372✔
120	return true;	250✔
121	}	372✔
122	else if (feed(bytecode)) // it's a bytecode file	×
123	return true;	×
124	return false;	×
125	}	494✔
126
127	bool State::doString(const std::string& code, const uint16_t features)	13✔
128	{	13✔
129	m_features = features;	13✔
130
131	Welder welder(m_debug_level, m_libenv, m_features);	13✔
132	for (const auto& p : m_bound)	47✔
133	welder.registerSymbol(p.first);	34✔
134
135	if (!welder.computeASTFromString(code))	13✔
136	return false;	×
137	if (!welder.generateBytecode())	12✔
138	return false;	×
139	return feed(welder.bytecode());	12✔
140	}	13✔
141
142	void State::loadFunction(const std::string& name, Procedure::CallbackType&& function) noexcept	13✔
143	{	13✔
144	m_bound[name] = Value(std::move(function));	13✔
145	}	13✔
146
147	void State::setArgs(const std::vector<std::string>& args) noexcept	10✔
148	{	10✔
149	Value val(ValueType::List);	10✔
150	std::ranges::transform(args, std::back_inserter(val.list()), [](const std::string& arg) {	13✔
151	return Value(arg);	3✔
152	});
153
154	m_bound[std::string(internal::Language::SysArgs)] = val;	10✔
155	}	10✔
156
157	void State::setDebug(const unsigned level) noexcept	1✔
158	{	1✔
159	m_debug_level = level;	1✔
160	}	1✔
161
162	void State::setLibDirs(const std::vector<std::filesystem::path>& libenv) noexcept	×
163	{	×
164	m_libenv = libenv;	×
165	}	×
166
167	void State::configure(const BytecodeReader& bcr)	266✔
168	{	266✔
169	using namespace internal;
170
171	const auto [major, minor, patch] = bcr.version();	266✔
172	if (major != ARK_VERSION_MAJOR)	266✔
173	{
174	const std::string str_version = fmt::format("{}.{}.{}", major, minor, patch);	2✔
175	throwStateError(fmt::format("Compiler and VM versions don't match: got {} while running {}", str_version, ARK_VERSION));	1✔
176	}	1✔
177
178	const auto bytecode_hash = bcr.sha256();	265✔
179
180	std::vector<unsigned char> hash(picosha2::k_digest_size);	265✔
181	picosha2::hash256(m_bytecode.begin() + bytecode::HeaderSize + picosha2::k_digest_size, m_bytecode.end(), hash);	265✔
182	// checking integrity
183	for (std::size_t j = 0; j < picosha2::k_digest_size; ++j)	8,716✔
184	{
185	#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
186	if (hash[j] != bytecode_hash[j])	8,451✔
187	throwStateError("Integrity check failed");	1✔
188	#endif
189	}	8,450✔
190
191	const auto syms = bcr.symbols();	264✔
192	const auto vals = bcr.values(syms);	264✔
193	const auto files = bcr.filenames(vals);	264✔
194	const auto inst_locs = bcr.instLocations(files);	264✔
195	const auto [pages, _] = bcr.code(inst_locs);	792✔
196
197	m_symbols = syms.symbols;	264✔
198	m_constants = vals.values;	264✔
199	m_filenames = files.filenames;	264✔
200	m_inst_locations = inst_locs.locations;	264✔
201	m_pages = pages;	264✔
202	m_max_page_size = maxPageSize(m_pages);	264✔
203
204	// Make m_code as a big contiguous chunk of instructions,
205	// aligned on the biggest page size.
206	// This might have a downside when we have a single big page and
207	// a bunch of smaller ones, though I couldn't measure it while testing.
208	m_code.resize(m_max_page_size * pages.size(), Instruction::NOP);	264✔
209	addPagesToContiguousBytecode(pages, /* start= */ 0);	264✔
210	}	268✔
211
212	void State::reset() noexcept	×
213	{	×
214	m_symbols.clear();	×
215	m_constants.clear();	×
216	m_filenames.clear();	×
217	m_inst_locations.clear();	×
218	m_max_page_size = 0;	×
219	m_code.clear();	×
220	m_bound.clear();	×
221
222	// default value for builtin__sys:args is empty list
223	const Value val(ValueType::List);	×
224	m_bound[std::string(internal::Language::SysArgs)] = val;	×
225
226	m_bound[std::string(internal::Language::SysProgramName)] = Value("");	×
227	}	×
228
229	void State::addPagesToContiguousBytecode(const std::vector<bytecode_t>& pages, const std::size_t start)	292✔
230	{	292✔
231	for (std::size_t i = 0, end = pages.size(); i < end; ++i)	4,343✔
232	{
233	for (std::size_t j = 0, end_j = pages[i].size(); j < end_j; ++j)	400,107✔
234	m_code[(start + i) * m_max_page_size + j] = pages[i][j];	396,056✔
235	}	4,051✔
236	}	292✔
237
238	std::size_t State::maxPageSize(const std::vector<bytecode_t>& pages)	278✔
239	{	278✔
240	return std::ranges::max(pages, {}, &bytecode_t::size).size();	278✔
241	}
242
243	void State::extendBytecode(const std::vector<bytecode_t>& pages, const std::vector<std::string>& symbols, const std::vector<Value>& constants)	14✔
244	{	14✔
245	m_symbols = symbols;	14✔
246	m_constants = constants;	14✔
247
248	// do not modify m_pages so that we can start over
249	m_max_page_size = std::max(m_max_page_size, maxPageSize(pages));	14✔
250
251	m_code.resize(m_max_page_size * (m_pages.size() + pages.size()), internal::Instruction::NOP);	14✔
252	addPagesToContiguousBytecode(m_pages, /* start= */ 0);	14✔
253	addPagesToContiguousBytecode(pages, /* start= */ m_pages.size());	14✔
254	}	14✔
255	}
256
257	#ifdef _MSC_VER
258	# pragma warning(pop)
259	#endif

ArkScript-lang / Ark / 22281039592

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous