Implement all of semver

src/semver/build_metadata.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+#include <semver/ident.hpp>
+
+namespace semver {
+
+class build_metadata {
+    std::vector<ident> _ids;
+
+public:
+    build_metadata() = default;
+
+    [[nodiscard]] bool empty() const noexcept { return _ids.empty(); }
+
+    void add_ident(ident id) noexcept { _ids.push_back(id); }
+    void add_ident(std::string_view s) { add_ident(ident(s)); }
+
+    auto& idents() const noexcept { return _ids; }
+
+    static build_metadata parse(std::string_view s) {
+        build_metadata ret;
+        ret._ids = ident::parse_dotted_seq(s);
+        return ret;
+    }
+};
+
+}  // namespace semver

src/semver/ident.cpp

@@ -0,0 +1,114 @@
+#include "./ident.hpp"
+
+#include <cassert>
+#include <charconv>
+
+using namespace semver;
+
+ident::ident(std::string_view str) {
+    const auto str_begin = str.data();
+    auto       ptr       = str_begin;
+    const auto str_end   = str_begin + str.size();
+
+    bool any_alpha = false;
+    if (ptr == str_end) {
+        throw invalid_ident(std::string(str));
+    }
+    for (; ptr != str_end; ++ptr) {
+        any_alpha = any_alpha || (*ptr == '-' || std::isalpha(*ptr));
+        if (!std::isalnum(*ptr) && *ptr != '-') {
+            break;
+        }
+    }
+    if (ptr != str_end) {
+        throw invalid_ident(std::string(str));
+    }
+
+    _str.assign(str_begin, ptr);
+
+    if (any_alpha) {
+        _kind = ident_kind::alphanumeric;
+    } else if (_str.size() > 1 && _str[0] == '0') {
+        _kind = ident_kind::digits;
+    } else {
+        _kind = ident_kind::numeric;
+        // Check that the integer is representable on this system
+        std::uint64_t n;
+        auto          res = std::from_chars(str_begin, str_end, n);
+        if (res.ptr != str_end) {
+            throw invalid_ident(_str);
+        }
+    }
+}
+
+std::vector<ident> ident::parse_dotted_seq(const std::string_view s) {
+    std::vector<ident> acc;
+    auto               remaining = s;
+
+    while (!remaining.empty()) {
+        auto next_dot = remaining.find('.');
+        auto id_sub   = remaining.substr(0, next_dot);
+        if (id_sub.empty()) {
+            throw invalid_ident(std::string(s));
+        }
+        acc.emplace_back(id_sub);
+        if (next_dot == remaining.npos) {
+            break;
+        }
+        remaining = remaining.substr(next_dot + 1);
+        if (remaining.empty()) {
+            throw invalid_ident(std::string(s));
+        }
+    }
+    if (acc.empty()) {
+        throw invalid_ident(std::string(s));
+    }
+    return acc;
+}
+
+order semver::compare(ident_kind lhs, ident_kind rhs) noexcept {
+    assert(lhs != ident_kind::digits && "Ordering with digit identifiers is undefined");
+    assert(rhs != ident_kind::digits && "Ordering with digit identifiers is undefined");
+    if (lhs == rhs) {
+        return order::equivalent;
+    } else if (lhs == ident_kind::numeric) {
+        // numeric is less than alnum
+        return order::less;
+    } else {
+        return order::greater;
+    }
+}
+
+order semver::compare(const ident& lhs, const ident& rhs) noexcept {
+    auto ord = compare(lhs.kind(), rhs.kind());
+    if (ord != order::equivalent) {
+        return ord;
+    }
+    assert(lhs.kind() == rhs.kind() && "[semver library bug]");
+    if (lhs.kind() == ident_kind::numeric) {
+        auto as_int = [](auto& str) {
+            std::uint64_t value;
+            auto          fc_res = std::from_chars(str.data(), str.data() + str.size(), value);
+            assert(fc_res.ptr == str.data() + str.size());
+            return value;
+        };
+        auto lhs_num = as_int(lhs.string());
+        auto rhs_num = as_int(rhs.string());
+        if (lhs_num == rhs_num) {
+            return order::equivalent;
+        } else if (lhs_num < rhs_num) {
+            return order::less;
+        } else {
+            return order::greater;
+        }
+    }
+    assert(lhs.kind() == ident_kind::alphanumeric && "[semver library bug]");
+    auto comp = lhs.string().compare(rhs.string());
+    if (comp == 0) {
+        return order::equivalent;
+    } else if (comp < 0) {
+        return order::less;
+    } else {
+        return order::greater;
+    }
+}

src/semver/ident.hpp

@@ -0,0 +1,62 @@
+#pragma once
+
+#include <semver/order.hpp>
+
+#include <stdexcept>
+#include <string>
+#include <string_view>
+#include <vector>
+
+namespace semver {
+
+class invalid_ident : public std::runtime_error {
+    std::string _str;
+
+public:
+    invalid_ident(std::string s)
+        : std::runtime_error::runtime_error("Invalid metadata identifier: " + s)
+        , _str(s) {}
+
+    auto& string() const noexcept { return _str; }
+};
+
+enum class ident_kind {
+    alphanumeric,
+    numeric,
+    digits,
+};
+
+order compare(ident_kind lhs, ident_kind rhs) noexcept;
+
+class ident;
+order compare(const ident& lhs, const ident& rhs) noexcept;
+
+class ident {
+    std::string _str;
+    ident_kind  _kind;
+
+public:
+    explicit ident(std::string_view str);
+
+    auto        kind() const noexcept { return _kind; }
+    const auto& string() const noexcept { return _str; }
+
+#define DEF_OP(op, expr)                                                                           \
+    inline friend bool operator op(const ident& lhs, const ident& rhs) noexcept {                  \
+        auto o = compare(lhs, rhs);                                                                \
+        return (expr);                                                                             \
+    }                                                                                              \
+    static_assert(true)
+
+    DEF_OP(==, (o == order::equivalent));
+    DEF_OP(!=, (o != order::equivalent));
+    DEF_OP(<, (o == order::less));
+    DEF_OP(>, (o == order::greater));
+    DEF_OP(<=, (o == order::less || o == order::equivalent));
+    DEF_OP(>=, (o == order::greater || o == order::equivalent));
+#undef DEF_OP
+
+    static std::vector<ident> parse_dotted_seq(std::string_view s);
+};
+
+}  // namespace semver

src/semver/ident.test.cpp

@@ -0,0 +1,89 @@
+#include <semver/ident.hpp>
+
+#include <catch2/catch.hpp>
+
+TEST_CASE("Parse identifiers") {
+    struct case_ {
+        std::string        str;
+        semver::ident_kind expect_kind;
+    };
+    case_ cases[] = {
+        {"foo", semver::ident_kind::alphanumeric},
+        {"12", semver::ident_kind::numeric},
+        {"03412", semver::ident_kind::digits},
+        {"0", semver::ident_kind::numeric},
+        {"00", semver::ident_kind::digits},
+    };
+    for (auto [str, expect] : cases) {
+        INFO("Checking parsing of '" << str << "'");
+        auto id = semver::ident(str);
+        CHECK(id.string() == str);
+        CHECK(id.kind() == expect);
+    }
+}
+
+TEST_CASE("Invalid identifiers") {
+#define BAD_IDENT(x) CHECK_THROWS_AS(semver::ident(x), semver::invalid_ident)
+    BAD_IDENT("");
+    BAD_IDENT("=");
+    BAD_IDENT("asdf-_");
+    BAD_IDENT(".");
+    BAD_IDENT("  ");
+    BAD_IDENT("124a[");
+}
+
+TEST_CASE("Ident comparison") {
+    using semver::order;
+    struct comp_test {
+        std::string lhs;
+        std::string rhs;
+        order       expect_ordering;
+    };
+    comp_test comparisons[] = {
+        {"foo", "bar", order::greater},
+        {"foo", "foo", order::equivalent},
+        {"bar", "foo", order::less},
+        {"12", "333", order::less},
+        {"fooood", "3", order::greater},
+        {"0", "0", order::equivalent},
+        {"34", "f", order::less},
+        {"aaaaaaaaaaa", "z", order::less},
+    };
+    for (auto [lhs, rhs, expect] : comparisons) {
+        auto lhs_id = semver::ident(lhs);
+        auto rhs_id = semver::ident(rhs);
+        INFO("Comparing '" << lhs << "' to '" << rhs << "'");
+        auto result = semver::compare(lhs_id, rhs_id);
+        CHECK(result == expect);
+    }
+}
+
+TEST_CASE("Parse dotted sequence") {
+    struct case_ {
+        std::string              str;
+        std::vector<std::string> expected;
+    };
+    case_ cases[] = {
+        {"foo", {"foo"}},
+        {"foo.bar", {"foo", "bar"}},
+    };
+    for (auto& [str, expected] : cases) {
+        INFO("Parsing dotted-ident-sequence '" << str << "'");
+        auto actual = semver::ident::parse_dotted_seq(str);
+        REQUIRE(actual.size() == expected.size());
+    }
+}
+
+TEST_CASE("Invalid dotted sequence") {
+    std::string_view bad_seqs[] = {
+        "",
+        ".",
+        ".foo",
+        "foo.bar.",
+        "foo..bar",
+    };
+    for (auto s : bad_seqs) {
+        INFO("Checking bad ident sequence '" << s << "'");
+        CHECK_THROWS_AS(semver::ident::parse_dotted_seq(s), semver::invalid_ident);
+    }
+}

src/semver/order.hpp

@@ -0,0 +1,11 @@
+#pragma once
+
+namespace semver {
+
+enum class order {
+    less = -1,
+    equivalent = 0,
+    greater = 1,
+};
+
+} // namespace semver

src/semver/prerelease.cpp

@@ -0,0 +1,40 @@
+#include "./prerelease.hpp"
+
+using namespace semver;
+
+prerelease prerelease::parse(std::string_view s) {
+    auto ids = ident::parse_dotted_seq(s);
+    for (auto& id : ids) {
+        if (id.kind() == ident_kind::digits) {
+            throw invalid_ident("Prerelease strings may not have plain-digit identifiers: "
+                                + std::string(s));
+        }
+    }
+    prerelease ret;
+    ret._ids = std::move(ids);
+    return ret;
+}
+
+order semver::compare(const prerelease& lhs, const prerelease& rhs) noexcept {
+    auto       lhs_iter = lhs.idents().cbegin();
+    auto       rhs_iter = rhs.idents().cbegin();
+    const auto lhs_end  = lhs.idents().cend();
+    const auto rhs_end  = rhs.idents().cend();
+
+    for (; lhs_iter != lhs_end && rhs_iter != rhs_end; ++lhs_iter, ++rhs_iter) {
+        auto ord = compare(*lhs_iter, *rhs_iter);
+        if (ord != order::equivalent) {
+            return ord;
+        }
+    }
+    if (lhs_iter != lhs_end) {
+        // Left-hand is longer
+        return order::greater;
+    } else if (rhs_iter != rhs_end) {
+        // Right-hand is longer
+        return order::less;
+    } else {
+        // They are equivalent
+        return order::equivalent;
+    }
+}

src/semver/prerelease.hpp

@@ -0,0 +1,46 @@
+#pragma once
+
+#include <semver/ident.hpp>
+#include <semver/order.hpp>
+
+#include <stdexcept>
+#include <string_view>
+#include <vector>
+
+namespace semver {
+
+class prerelease;
+order compare(const prerelease& lhs, const prerelease& rhs) noexcept;
+
+class prerelease {
+    std::vector<ident> _ids;
+
+public:
+    prerelease() = default;
+
+    [[nodiscard]] bool empty() const noexcept { return _ids.empty(); }
+
+    void add_ident(ident id);
+    void add_ident(std::string_view str) { add_ident(ident(str)); }
+
+    auto& idents() const noexcept { return _ids; }
+
+    static prerelease parse(std::string_view str);
+
+#define DEF_OP(op, expr)                                                                           \
+    inline friend bool operator op(const prerelease& lhs, const prerelease& rhs) noexcept {        \
+        auto o = compare(lhs, rhs);                                                                \
+        return (expr);                                                                             \
+    }                                                                                              \
+    static_assert(true)
+
+    DEF_OP(==, (o == order::equivalent));
+    DEF_OP(!=, (o != order::equivalent));
+    DEF_OP(<, (o == order::less));
+    DEF_OP(>, (o == order::greater));
+    DEF_OP(<=, (o == order::less || o == order::equivalent));
+    DEF_OP(>=, (o == order::greater || o == order::equivalent));
+#undef DEF_OP
+};
+
+}  // namespace semver

src/semver/prerelease.test.cpp

@@ -0,0 +1,48 @@
+#include "./prerelease.hpp"
+
+#include <catch2/catch.hpp>
+
+TEST_CASE("Parse a prerelease") {
+    auto pre = semver::prerelease::parse("foo.bar");
+    CHECK(pre.idents() == semver::ident::parse_dotted_seq("foo.bar"));
+    pre = pre.parse("foo.0");
+    CHECK(pre.idents() == semver::ident::parse_dotted_seq("foo.0"));
+}
+
+TEST_CASE("Invalid prereleases") {
+    std::string_view bad_tags[] = {
+        "foo.",        // Just plain bad
+        "foo.122.03",  // Leading zero
+    };
+    for (auto bad : bad_tags) {
+        CHECK_THROWS_AS(semver::prerelease::parse(bad), semver::invalid_ident);
+    }
+}
+
+TEST_CASE("Compare prereleases") {
+    using semver::order;
+    struct case_ {
+        std::string lhs;
+        std::string rhs;
+        order       expected_ord;
+    };
+
+    case_ cases[] = {
+        {"foo.bar", "foo.bar", order::equivalent},
+        {"foo", "foo.bar", order::less},
+        {"foo.bar", "foo", order::greater},
+        {"foo", "foo", order::equivalent},
+        {"foo.1", "foo.bar", order::less},  // numeric is less
+        {"foo.foo", "foo.bar", order::greater},
+        {"alpha", "beta", order::less},
+        {"alpha.1", "alpha.2", order::less},
+    };
+
+    for (auto& [lhs, rhs, exp] : cases) {
+        INFO("Comparing prerelease '" << lhs << "' to '" << rhs << "'");
+        auto lhs_pre = semver::prerelease::parse(lhs);
+        auto rhs_pre = semver::prerelease::parse(rhs);
+        auto actual  = semver::compare(lhs_pre, rhs_pre);
+        CHECK(actual == exp);
+    }
+}

src/semver/version.cpp

@@ -14,38 +14,57 @@ namespace {
 version parse(const char* ptr, std::size_t size) {
     version ret;
     // const auto str_begin = ptr;
-    const auto str_end = ptr + size;
+    const auto str_begin = ptr;
+    const auto str_end   = ptr + size;
+    auto       get_str   = [=] { return std::string(str_begin, size); };
+    auto       cur_off   = [&] { return ptr - str_begin; };
 
     std::from_chars_result fc_res;
     auto did_error = [&](int elem) { return fc_res.ec == std::errc::invalid_argument || elem < 0; };
 
     // Parse major
     fc_res = std::from_chars(ptr, str_end, ret.major);
+    ptr    = fc_res.ptr;
     if (did_error(ret.major) || fc_res.ptr == str_end || *fc_res.ptr != '.') {
-        throw invalid_version(0);
+        throw invalid_version(get_str(), cur_off());
     }
 
     // Parse minor
     ptr    = fc_res.ptr + 1;
     fc_res = std::from_chars(ptr, str_end, ret.minor);
+    ptr    = fc_res.ptr;
     if (did_error(ret.minor) || fc_res.ptr == str_end || *fc_res.ptr != '.') {
-        throw invalid_version(ptr - str_end);
+        throw invalid_version(get_str(), cur_off());
     }
 
     // Parse patch
     ptr    = fc_res.ptr + 1;
     fc_res = std::from_chars(ptr, str_end, ret.patch);
+    ptr    = fc_res.ptr;
     if (did_error(ret.patch)) {
-        throw invalid_version(ptr - str_end);
+        throw invalid_version(get_str(), cur_off());
     }
 
-    if (fc_res.ptr != str_end) {
-        assert(false && "More complex version numbers are not ready yet!");
-        throw invalid_version(-42);
+    auto remaining = std::string_view(ptr, str_end - ptr);
+    if (!remaining.empty() && remaining[0] == '-') {
+        auto plus_pos       = remaining.find('+');
+        auto prerelease_str = remaining.substr(1, plus_pos - 1);
+        ret.prerelease      = prerelease::parse(prerelease_str);
+        remaining           = remaining.substr(prerelease_str.size() + 1);
+    }
+
+    if (!remaining.empty() && remaining[0] == '+') {
+        auto bmeta_str     = remaining.substr(1);
+        ret.build_metadata = build_metadata::parse(bmeta_str);
+        remaining          = remaining.substr(bmeta_str.size() + 1);
+    }
+
+    if (!remaining.empty()) {
+        throw invalid_version(get_str(), remaining.data() - str_begin);
     }
 
     return ret;
-}
+}  // namespace
 
 }  // namespace
 
@@ -72,5 +91,48 @@ std::string version::to_string() const noexcept {
     *buf_ptr++ = '.';
     conv_one(patch);
 
-    return std::string(buf_begin, (buf_ptr - buf_begin));
+    auto main_ver = std::string(buf_begin, (buf_ptr - buf_begin));
+
+    auto format_ids = [](auto& ids) {
+        std::string acc;
+        auto        it   = ids.cbegin();
+        auto        stop = ids.cend();
+        while (it != stop) {
+            acc += it->string();
+            ++it;
+            if (it != stop) {
+                acc += ".";
+            }
+        }
+        return acc;
+    };
+
+    if (!prerelease.empty()) {
+        main_ver += "-" + format_ids(prerelease.idents());
+    }
+    if (!build_metadata.empty()) {
+        main_ver += "+" + format_ids(build_metadata.idents());
+    }
+    return main_ver;
+}
+
+order semver::compare(const version& lhs, const version& rhs) noexcept {
+    auto lhs_tup = std::tie(lhs.major, lhs.minor, lhs.patch);
+    auto rhs_tup = std::tie(rhs.major, rhs.minor, rhs.patch);
+    if (lhs_tup < rhs_tup) {
+        return order::less;
+    } else if (lhs_tup > rhs_tup) {
+        return order::greater;
+    } else {
+        if (!lhs.is_prerelease() && rhs.is_prerelease()) {
+            // No prerelease is greater than any prerelease
+            return order::greater;
+        } else if (lhs.is_prerelease() && !rhs.is_prerelease()) {
+            // A prerelease version is lesser than any non-prerelease version
+            return order::less;
+        } else {
+            // Compare the prerelease tags
+            return compare(lhs.prerelease, rhs.prerelease);
+        }
+    }
 }

src/semver/version.hpp

@@ -1,5 +1,8 @@
 #pragma once
 
+#include <semver/build_metadata.hpp>
+#include <semver/prerelease.hpp>
+
 #include <stdexcept>
 #include <string>
 #include <string_view>
@@ -9,37 +12,52 @@
 namespace semver {
 
 class invalid_version : public std::runtime_error {
+    std::string    _string;
     std::ptrdiff_t _offset = 0;
 
 public:
-    invalid_version(std::ptrdiff_t n)
-        : runtime_error("Invalid version number")
+    invalid_version(std::string string, std::ptrdiff_t n)
+        : runtime_error("Invalid semantic version: " + string)
+        , _string(string)
         , _offset(n) {}
 
-    auto offset() const noexcept { return _offset; }
+    auto& string() const noexcept { return _string; }
+    auto  offset() const noexcept { return _offset; }
 };
 
+class version;
+order compare(const version& lhs, const version& rhs) noexcept;
+
 struct version {
     int major = 0;
     int minor = 0;
     int patch = 0;
+    // Prerelease tag is optional:
+    class prerelease prerelease = {};
+    // Build metadata is optional:
+    class build_metadata build_metadata = {};
 
     static version parse(std::string_view s);
 
     std::string to_string() const noexcept;
-
-    auto tie() const noexcept { return std::tie(major, minor, patch); }
-    auto tie() noexcept { return std::tie(major, minor, patch); }
+    bool        is_prerelease() const noexcept { return !prerelease.empty(); }
+
+#define DEF_OP(op, expr)                                                                           \
+    inline friend bool operator op(const version& lhs, const version& rhs) noexcept {                  \
+        auto o = compare(lhs, rhs);                                                                \
+        return (expr);                                                                             \
+    }                                                                                              \
+    static_assert(true)
+
+    DEF_OP(==, (o == order::equivalent));
+    DEF_OP(!=, (o != order::equivalent));
+    DEF_OP(<, (o == order::less));
+    DEF_OP(>, (o == order::greater));
+    DEF_OP(<=, (o == order::less || o == order::equivalent));
+    DEF_OP(>=, (o == order::greater || o == order::equivalent));
+#undef DEF_OP
+
+    friend inline std::string to_string(const version& ver) noexcept { return ver.to_string(); }
 };
 
-inline bool operator!=(const version& lhs, const version& rhs) noexcept {
-    return lhs.tie() != rhs.tie();
-}
-
-inline bool operator<(const version& lhs, const version& rhs) noexcept {
-    return lhs.tie() < rhs.tie();
-}
-
-inline std::string to_string(const version& ver) noexcept { return ver.to_string(); }
-
 }  // namespace semver

src/semver/version.test.cpp

@@ -13,4 +13,82 @@ TEST_CASE("Parsing") {
     CHECK(v1.to_string() == "1.2.55");
     v1.major = 999999;
     CHECK(v1.to_string() == "999999.2.55");
+
+    v1 = semver::version::parse("1.2.3-r1");
+    CHECK(v1.prerelease == semver::prerelease::parse("r1"));
+    CHECK(v1.to_string() == "1.2.3-r1");
+
+    v1 = semver::version::parse("1.2.3-3");
+    CHECK(v1.prerelease == semver::prerelease::parse("3"));
+
+    v1 = semver::version::parse("1.2.3-0");
+    CHECK(v1.prerelease == semver::prerelease::parse("0"));
+
+    v1 = semver::version::parse("1.2.3--fasdf");
+    CHECK(v1.prerelease == semver::prerelease::parse("-fasdf"));
+
+    v1 = semver::version::parse("1.2.3-foo.bar");
+    CHECK(v1.prerelease == semver::prerelease::parse("foo.bar"));
+
+    v1 = semver::version::parse("1.2.3-foo.bar+cats");
+    CHECK(v1.prerelease == semver::prerelease::parse("foo.bar"));
+    CHECK(v1.build_metadata.idents() == semver::ident::parse_dotted_seq("cats"));
 }
+
+TEST_CASE("Compare versions") {
+    using semver::order;
+    struct case_ {
+        std::string_view lhs;
+        std::string_view rhs;
+        order            expect_ord;
+    };
+
+    case_ cases[] = {
+        {"1.2.3", "1.2.3", order::equivalent},
+        {"1.2.3-alpha", "1.2.3-alpha", order::equivalent},
+        {"1.2.3-alpha", "1.2.3", order::less},
+        {"1.2.3-alpha.1", "1.2.3-beta", order::less},
+        {"1.2.3-alpha.1", "1.2.3-alpha.2", order::less},
+        {"1.2.3-alpha.4", "1.2.3-alpha.2", order::greater},
+        {"1.2.1", "1.2.2-alpha.2", order::less},
+        {"1.2.1", "1.2.1+foo", order::equivalent}, // Build metadata has no effect
+    };
+
+    for (auto [lhs, rhs, exp] : cases) {
+        INFO("Compare version '" << lhs << "' to '" << rhs << "'");
+        auto lhs_ver = semver::version::parse(lhs);
+        auto rhs_ver = semver::version::parse(rhs);
+        auto actual  = semver::compare(lhs_ver, rhs_ver);
+        CHECK(actual == exp);
+    }
+}
+
+TEST_CASE("Invalid versions") {
+    struct invalid_version {
+        std::string str;
+        int         bad_offset;
+    };
+    invalid_version versions[] = {
+        {"", 0},
+        {"1.", 2},
+        {"1.2", 3},
+        {"1.2.", 4},
+        {"1.e", 2},
+        {"lol", 0},
+        {"1e.3.1", 1},
+        {"1.2.5-", 6},
+        {"1.2.5-02", 6},
+        {"1.2.3-1..3", 8},
+    };
+    for (auto&& [str, bad_offset] : versions) {
+        INFO("Checking for failure while parsing bad version string '" << str << "'");
+        try {
+            auto ver = semver::version::parse(str);
+            FAIL_CHECK("Parsing didn't throw! Produced version: " << ver.to_string());
+        } catch (const semver::invalid_version& e) {
+            CHECK(e.offset() == bad_offset);
+        } catch (const semver::invalid_ident&) {
+            // Nothing to check, but a valid error
+        }
+    }
+}