Add support for parsing cpu-vendor-os target triplets

3 files changed, 219 insertions, 0 deletions

diff --git a/butl/buildfile b/butl/buildfile
index 9c96aab..de51ad5 100644
--- a/butl/buildfile
+++ b/butl/buildfile
@@ -15,6 +15,7 @@ lib{butl}:                        \
 {hxx ixx     cxx}{ process      } \
 {hxx     txx    }{ string-table } \
 {hxx         cxx}{ timestamp    } \
+{hxx         cxx}{ triplet      } \
 {hxx            }{ utility      } \
 {hxx            }{ version      }
 
diff --git a/butl/triplet b/butl/triplet
new file mode 100644
index 0000000..c18368c
--- /dev/null
+++ b/butl/triplet
@@ -0,0 +1,97 @@
+// file      : butl/triplet -*- C++ -*-
+// copyright : Copyright (c) 2014-2016 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#ifndef BUTL_TRIPLET
+#define BUTL_TRIPLET
+
+#include <string>
+
+namespace butl
+{
+  // This is the ubiquitous 'target triplet' that loosely has the CPU-VENDOR-OS
+  // form which, these days, quite often takes the CPU-VENDOR-OS-ABI form. Plus
+  // some fields can sometimes be omitted. This looseness makes it hard to base
+  // any kind of decisions on the triplet without canonicalizing it and then
+  // splitting it into components. the way we are going to split it is like
+  // this:
+  //
+  // CPU
+  //
+  // This one is reasonably straightforward. Note that we always expect at
+  // least two components with the first being the CPU. In other words, we
+  // don't try to guess what just 'mingw32' might mean like config.sub does.
+  //
+  // VENDOR
+  //
+  // This can be a machine vendor as in i686-apple-darwin8, a toolchain vendor
+  // as in i686-lfs-linux-gnu, or something else as in arm-softfloat-linux-gnu.
+  // Just as we think vendor is pretty irrelevant and can be ignored, comes
+  // MinGW-W64 and calls itself *-w64-mingw32. While it is tempting to
+  // attribute w64 to OS-ABI, the MinGW-W64 folks insist it is a (presumably
+  // toolchain) vendor.
+  //
+  // To make thing more regular we also convert the information-free vendor
+  // names 'pc', 'unknown' and 'none' to the empty name.
+  //
+  // OS/KERNEL-OS/OS-ABI
+  //
+  // This is where things get really messy and instead of trying to guess, we
+  // call the entire thing SYSTEM. Except, in certain cases, we factor out the
+  // trailing version, again, to make SYSTEM easier to compare to. For example,
+  // *-darwin14.5.0 becomes 'darwin' and '14.5.0'.
+  //
+  // Again, to make things more regular, if the first component in SYSTEM is
+  // none, then it is removed (so *-none-eabi becomes just 'eabi').
+  //
+  // Values for two-component systems (e.g., linux-gnu) that don't specify
+  // VENDOR explicitly are inherently ambiguous: is 'linux' VENDOR or part of
+  // SYSTEM? The only way to handle this is to recognize their specific names
+  // as special cases and this is what we do for some of the more common
+  // ones. The alternative would be to first run such names through config.sub
+  // which adds explicit VENDOR and this could be a reasonable fallback
+  // strategy for (presumably less common) cases were we don't split things
+  // correctly.
+  //
+  // Note also that the version splitting is only done for certain,
+  // commonly-used targets.
+  //
+  // Some examples of canonicalization and splitting:
+  //
+  // x86_64-apple-darwin14.5.0    x86_64 apple     darwin        14.5.0
+  // x86_64-unknown-freebsd10.2   x86_64           freebsd       10.2
+  // i686-elf                     i686             elf
+  // arm-eabi                     arm              eabi
+  // arm-none-eabi                arm              eabi
+  // arm-none-linux-gnueabi       arm              linux-gnueabi
+  // arm-softfloat-linux-gnu      arm    softfloat linux-gnu
+  // i686-pc-mingw32              i686             mingw32
+  // i686-w64-mingw32             i686   w64       mingw32
+  // i686-lfs-linux-gnu           i686   lfs       linux-gnu
+  // x86_64-unknown-linux-gnu     x86_64           linux-gnu
+  // x86_64-linux-gnux32          x86_64           linux-gnux32
+  //
+  // References:
+  //
+  // 1. The libtool repository contains the PLATFORM file that lists many known
+  //    triplets.
+  //
+  // 2. LLVM has the Triple class with similar goals.
+  //
+  struct triplet
+  {
+    std::string cpu;
+    std::string vendor;
+    std::string system;
+    std::string version;
+
+    // Parse the triplet optionally returning the canonicalized string. Throw
+    // std::invalid_argument if the triplet is not recognizable.
+    //
+    explicit
+    triplet (const std::string&, std::string* canon = nullptr);
+    triplet (const std::string& s, std::string& canon): triplet (s, &canon) {}
+  };
+};
+
+#endif // BUTL_TRIPLET
diff --git a/butl/triplet.cxx b/butl/triplet.cxx
new file mode 100644
index 0000000..3208375
--- /dev/null
+++ b/butl/triplet.cxx
@@ -0,0 +1,121 @@
+// file      : butl/triplet.cxx -*- C++ -*-
+// copyright : Copyright (c) 2014-2016 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#include <butl/triplet>
+
+#include <stdexcept> // invalid_argument
+
+using namespace std;
+
+namespace butl
+{
+  triplet::
+  triplet (const string& s, string* c)
+  {
+    auto bad = [](const char* m) {throw invalid_argument (m);};
+
+    // Find the first and the last components. The first is CPU and the last is
+    // (part of) SYSTEM, that we know for sure.
+    //
+    string::size_type f (s.find ('-')), l (s.rfind ('-'));
+
+    if (f == 0 || f == string::npos)
+      bad ("missing cpu");
+
+    cpu.assign (s, 0, f);
+
+    if (c != nullptr)
+      *c = cpu;
+
+    // If we have something in between, then the first component after CPU is
+    // VENDOR. Unless it is a first component of two-component system, as in
+    // i686-linux-gnu.
+    //
+    if (f != l)
+    {
+      // [f, p) is VENDOR.
+      //
+      string::size_type p (s.find ('-', ++f)), n (p - f);
+
+      if (n == 0)
+        bad ("empty vendor");
+
+      // Do we have all four components? If so, then we don't need to do any
+      // special recognition of two-component systems.
+      //
+      if (l != p)
+      {
+        l = s.rfind ('-', --l);
+
+        if (l != p)
+          bad ("too many components");
+
+        // Handle the none-* case here.
+        //
+        if (s.compare (l + 1, 5, "none-") == 0)
+          l += 5;
+      }
+      else
+      {
+        // See if this is one of the well-known non-vendors.
+        //
+        if (s.compare (f, n, "linux") == 0 ||
+            s.compare (f, n, "kfreebsd") == 0)
+        {
+          l = f - 1;
+          n = 0; // No VENDOR.
+        }
+      }
+
+      // Handle special VENDOR values.
+      //
+      if (n != 0)
+      {
+        if (s.compare (f, n, "pc") != 0 &&
+            s.compare (f, n, "none") != 0 &&
+            s.compare (f, n, "unknown") != 0)
+        {
+          vendor.assign (s, f, n);
+
+          if (c != nullptr)
+          {
+            *c += '-';
+            *c += vendor;
+          }
+        }
+      }
+    }
+
+    // (l, npos) is SYSTEM
+    //
+    system.assign (s, ++l, string::npos);
+
+    if (system.empty ())
+      bad ("missing os/kernel/abi");
+
+    if (system.front () == '-' || system.back () == '-')
+      bad ("invalid os/kernel/abi");
+
+    if (c != nullptr)
+    {
+      *c += '-';
+      *c += system;
+    }
+
+    // Finally, extract VERSION for some recognized systems.
+    //
+    string::size_type v (0);
+    if (system.compare (0, (v = 6), "darwin") == 0  ||
+        system.compare (0, (v = 7), "freebsd") == 0 ||
+        system.compare (0, (v = 7), "openbsd") == 0 ||
+        system.compare (0, (v = 7), "netbsd") == 0  ||
+        system.compare (0, (v = 7), "solaris") == 0 ||
+        system.compare (0, (v = 7), "aix") == 0     ||
+        system.compare (0, (v = 7), "hpux") == 0)
+    {
+      version.assign (system, v, string::npos);
+      system.resize (system.size () - version.size ());
+    }
+  }
+}