fix: Support globs in MatchSpec build strings (#3735)

Signed-off-by: Julien Jerphanion <[email protected]>
Co-authored-by: Johan Mabille <[email protected]>
Co-authored-by: jaimergp <[email protected]>

libmamba/include/mamba/specs/regex_spec.hpp

@@ -31,7 +31,7 @@ namespace mamba::specs
         [[nodiscard]] static auto parse(std::string pattern) -> expected_parse_t<RegexSpec>;
 
         RegexSpec();
-        RegexSpec(std::regex pattern, std::string raw_pattern);
+        explicit RegexSpec(std::string raw_pattern);
 
         [[nodiscard]] auto contains(std::string_view str) const -> bool;
 
@@ -61,8 +61,8 @@ namespace mamba::specs
 
     private:
 
-        std::regex m_pattern;
         std::string m_raw_pattern;
+        std::regex m_pattern;
     };
 }
 

libmamba/src/specs/regex_spec.cpp

@@ -6,6 +6,7 @@
 
 #include <algorithm>
 #include <cassert>
+#include <sstream>
 
 #include <fmt/format.h>
 
@@ -25,39 +26,78 @@ namespace mamba::specs
 
     auto RegexSpec::parse(std::string pattern) -> expected_parse_t<RegexSpec>
     {
-        // No other mean of getting parse result with ``std::regex``, but parse error need
-        // to be handled by ``tl::expected`` to be managed down the road.
+        // Parse error need to be handled by ``tl::expected`` to be managed down the road.
         try
         {
-            auto regex = std::regex(pattern);
-            return { { std::move(regex), std::move(pattern) } };
+            return RegexSpec{ std::move(pattern) };
         }
         catch (const std::regex_error& e)
         {
             return make_unexpected_parse(e.what());
         }
     }
 
-    RegexSpec::RegexSpec()
-        : RegexSpec(std::regex(free_pattern.data(), free_pattern.size()), std::string(free_pattern))
+    auto regexify(std::string raw_pattern) -> std::string
     {
-    }
+        // raw_pattern can be a regex or a glob pattern. We need to convert it to a regex.
 
-    RegexSpec::RegexSpec(std::regex pattern, std::string raw_pattern)
-        : m_pattern(std::move(pattern))
-        , m_raw_pattern(std::move(raw_pattern))
-    {
+        // If the string is wrapped in `^` and `$`, `conda.model.MatchSpec` considers it a regex.
+        // See:
+        // https://github.com/conda/conda/blob/52b6393d6331e8aa36b2e23ab65766a980f381d2/conda/models/match_spec.py#L134-L139.
+        // See:
+        // https://github.com/conda/conda/blob/52b6393d6331e8aa36b2e23ab65766a980f381d2/conda/models/match_spec.py#L889-L894
+        if (util::starts_with(raw_pattern, RegexSpec::pattern_start)
+            && util::ends_with(raw_pattern, RegexSpec::pattern_end))
+        {
+            return raw_pattern;
+        }
+
+        // Construct the regex progressively from raw_pattern, in particular make sure to replace
+        // all `*` by `.*` in the pattern if they are not preceded by a `.`.
+        //
         // We force regex to start with `^` and end with `$` to simplify the multiple
         // possible representations, and because this is the safest way we can make sure it is
         // not a glob when serializing it.
-        if (!util::starts_with(m_raw_pattern, pattern_start))
-        {
-            m_raw_pattern.insert(m_raw_pattern.begin(), pattern_start);
-        }
-        if (!util::ends_with(m_raw_pattern, pattern_end))
+        std::ostringstream ss;
+        ss << RegexSpec::pattern_start;
+
+        auto first_character_it = raw_pattern.cbegin();
+        auto last_character_it = raw_pattern.cend() - 1;
+
+        for (auto it = first_character_it; it != raw_pattern.cend(); ++it)
         {
-            m_raw_pattern.push_back(pattern_end);
+            if (it == first_character_it && *it == RegexSpec::pattern_start)
+            {
+                continue;
+            }
+            if (it == last_character_it && *it == RegexSpec::pattern_end)
+            {
+                continue;
+            }
+            if (*it == '*' && (it == first_character_it || *(it - 1) != '.'))
+            {
+                ss << ".*";
+            }
+            else
+            {
+                ss << *it;
+            }
         }
+
+        ss << RegexSpec::pattern_end;
+
+        return ss.str();
+    }
+
+    RegexSpec::RegexSpec()
+        : RegexSpec(std::string(free_pattern))
+    {
+    }
+
+    RegexSpec::RegexSpec(std::string raw_pattern)
+        : m_raw_pattern(regexify(std::move(raw_pattern)))
+        , m_pattern(std::regex(m_raw_pattern))
+    {
     }
 
     auto RegexSpec::contains(std::string_view str) const -> bool

libmamba/tests/src/specs/test_match_spec.cpp

@@ -981,6 +981,25 @@ namespace
                 /* .track_features =*/{ "openssl", "mkl" },
             }));
         }
+
+        SECTION("pytorch=2.3.1=py3.10_cuda11.8*")
+        {
+            // Check that it contains `pytorch=2.3.1=py3.10_cuda11.8_cudnn8.7.0_0`
+
+            const auto ms = "pytorch=2.3.1=py3.10_cuda11.8*"_ms;
+
+            REQUIRE(ms.contains_except_channel(Pkg{
+                /* .name= */ "pytorch",
+                /* .version= */ "2.3.1"_v,
+                /* .build_string= */ "py3.10_cuda11.8_cudnn8.7.0_0",
+                /* .build_number= */ 0,
+                /* .md5= */ "lemd5",
+                /* .sha256= */ "somesha256",
+                /* .license= */ "GPL",
+                /* .platform= */ "linux-64",
+                /* .track_features =*/{},
+            }));
+        }
     }
 
     TEST_CASE("MatchSpec comparability and hashability")

libmamba/tests/src/specs/test_regex_spec.cpp

@@ -80,4 +80,20 @@ namespace
         REQUIRE(hash_fn(spec1) == hash_fn(spec2));
         REQUIRE(hash_fn(spec1) != hash_fn(spec3));
     }
+
+    TEST_CASE("RegexSpec py3.10_cuda11.8*")
+    {
+        auto spec = RegexSpec::parse("py3.10_cuda11.8*").value();
+        REQUIRE(spec.contains("py3.10_cuda11.8_cudnn8.7.0_0"));
+    }
+
+    TEST_CASE("RegexSpec * semantic")
+    {
+        auto spec = RegexSpec::parse("py3.*").value();
+
+        REQUIRE(spec.contains("py3."));
+        REQUIRE(spec.contains("py3.10"));
+        REQUIRE(spec.contains("py3.10_cuda11.8_cudnn8.7.0_0"));
+    }
+
 }

micromamba/tests/test_create.py

@@ -1639,6 +1639,42 @@ def test_ca_certificates(tmp_path):
     assert root_prefix_ca_certificates_used or fall_back_certificates_used
 
 
+def test_glob_in_build_string(monkeypatch, tmp_path):
+    # Non-regression test for https://github.com/mamba-org/mamba/issues/3699
+    env_prefix = tmp_path / "test_glob_in_build_string"
+
+    pytorch_match_spec = "pytorch=2.3.1=py3.10_cuda11.8*"
+
+    # Export CONDA_OVERRIDE_GLIBC=2.17 to force the solver to use the glibc 2.17 package
+    monkeypatch.setenv("CONDA_OVERRIDE_GLIBC", "2.17")
+
+    # Should run without error
+    out = helpers.create(
+        "-p",
+        env_prefix,
+        pytorch_match_spec,
+        "-c",
+        "pytorch",
+        "-c",
+        "nvidia/label/cuda-11.8.0",
+        "-c",
+        "nvidia",
+        "-c",
+        "conda-forge",
+        "--platform",
+        "linux-64",
+        "--dry-run",
+        "--json",
+    )
+
+    # Check that a build of pytorch 2.3.1 with `py3.10_cuda11.8_cudnn8.7.0_0` as a build string is found
+    assert any(
+        package["name"] == "pytorch"
+        and package["version"] == "2.3.1"
+        and package["build_string"] == "py3.10_cuda11.8_cudnn8.7.0_0"
+        for package in out["actions"]["FETCH"]
+    )
+
 def test_non_url_encoding(tmp_path):
     # Non-regression test for https://github.com/mamba-org/mamba/issues/3737
     env_prefix = tmp_path / "env-non_url_encoding"
@@ -1656,3 +1692,4 @@ def test_non_url_encoding(tmp_path):
     non_encoded_url_start = "https://conda.anaconda.org/conda-forge/linux-64/x264-1!"
     out = helpers.run_env("export", "-p", env_prefix, "--explicit")
     assert non_encoded_url_start in out
+