Implement low-level manifest parser

1 files changed, 141 insertions, 0 deletions

diff --git a/bpkg/manifest-parser b/bpkg/manifest-parser
new file mode 100644
index 0000000..2b1e4a5
--- /dev/null
+++ b/bpkg/manifest-parser
@@ -0,0 +1,141 @@
+// file      : bpkg/manifest-parser -*- C++ -*-
+// copyright : Copyright (c) 2014-2015 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#ifndef BPKG_MANIFEST_PARSER
+#define BPKG_MANIFEST_PARSER
+
+#include <string>
+#include <iosfwd>
+#include <cstdint>   // uint64_t
+#include <stdexcept> // runtime_error
+
+namespace bpkg
+{
+  class manifest_parsing: public std::runtime_error
+  {
+  public:
+    manifest_parsing (const std::string& name,
+                      std::uint64_t line,
+                      std::uint64_t column,
+                      const std::string& description);
+
+    std::string name;
+    std::uint64_t line;
+    std::uint64_t column;
+    std::string description;
+  };
+
+  class manifest_parser
+  {
+  public:
+    manifest_parser (std::istream& is, const std::string& name)
+        : is_ (is), name_ (name) {}
+
+    const std::string&
+    name () const {return name_;}
+
+    struct name_value_type
+    {
+      std::string name;
+      std::string value;
+
+      std::uint64_t name_line;
+      std::uint64_t name_column;
+
+      std::uint64_t value_line;
+      std::uint64_t value_column;
+    };
+
+    // The first returned pair is special "start-of-manifest" with
+    // empty name and value being the format version: {"", "<ver>"}.
+    // After that we have a sequence of ordinary pairs which are
+    // the manifest. At the end of the manifest we have the special
+    // "end-of-manifest" pair with empty name and value: {"", ""}.
+    // After that we can either get another start-of-manifest pair
+    // (in which case the whole sequence repeats from the beginning)
+    // or we get another end-of-manifest pair which signals the end
+    // of stream (aka EOF). To put it another way, the parse sequence
+    // always has the following form:
+    //
+    // ({"", "<ver>"} {"<name>", "<value>"}* {"", ""})* {"", ""}
+    //
+    name_value_type
+    next ();
+
+  private:
+    class xchar
+    {
+    public:
+      typedef std::char_traits<char> traits_type;
+      typedef traits_type::int_type int_type;
+      typedef traits_type::char_type char_type;
+
+      xchar (int_type v, std::uint64_t l, std::uint64_t c)
+          : v_ (v), l_ (l), c_ (c) {}
+
+      operator char_type () const {return static_cast<char_type> (v_);}
+
+      int_type
+      value () const {return v_;}
+
+      std::uint64_t line () const {return l_;}
+      std::uint64_t column () const {return c_;}
+
+    private:
+      int_type v_;
+      std::uint64_t l_;
+      std::uint64_t c_;
+    };
+
+  private:
+    void
+    parse_name (name_value_type&);
+
+    void
+    parse_value (name_value_type&);
+
+    // Skip spaces and return the first peeked non-space character.
+    //
+    xchar
+    skip_spaces ();
+
+    // Character interface.
+    //
+  private:
+    xchar
+    peek ();
+
+    xchar
+    get ();
+
+    void
+    unget (const xchar&);
+
+    // Tests.
+    //
+    bool
+    is_eos (const xchar& c) const
+    {
+      return c.value () == xchar::traits_type::eof ();
+    }
+
+  private:
+    enum {start, body, eos} s_ = start;
+    std::string version_; // Current format version.
+
+  private:
+    std::istream& is_;
+    const std::string name_;
+
+    std::uint64_t l_ {1};
+    std::uint64_t c_ {1};
+
+    bool unget_ {false};
+    xchar buf_ {0, 0, 0};
+
+    bool eos_ {false};
+  };
+}
+
+#endif // BPKG_MANIFEST_PARSER