Add support for filtering during manifest parsing and serialization

8 files changed, 133 insertions, 32 deletions

diff --git a/libbutl/manifest-parser.cxx b/libbutl/manifest-parser.cxx
index 20932e3..9aa35b7 100644
--- a/libbutl/manifest-parser.cxx
+++ b/libbutl/manifest-parser.cxx
@@ -41,13 +41,17 @@ namespace butl
   using parsing = manifest_parsing;
   using name_value = manifest_name_value;
 
-  name_value manifest_parser::
-  next ()
+  void manifest_parser::
+  parse_next (name_value& r)
   {
     if (s_ == end)
-      return name_value {
+    {
+      r = name_value {
         "", "", line, column, line, column, position, position, position};
 
+      return;
+    }
+
     auto clp (skip_spaces ());
     xchar c (clp.first);
     uint64_t start_pos (clp.second);
@@ -67,15 +71,18 @@ namespace butl
     {
       s_ = start;
 
-      return name_value {"", "",
-                         c.line, c.column, c.line, c.column,
-                         start_pos, c.position, c.position};
+      r = name_value {"", "",
+                      c.line, c.column, c.line, c.column,
+                      start_pos, c.position, c.position};
+      return;
     }
 
+    r.name.clear ();
+    r.value.clear ();
+
     // Regardless of the state, what should come next is a name,
     // potentially the special empty one.
     //
-    name_value r;
     r.start_pos = start_pos;
 
     parse_name (r);
@@ -98,7 +105,7 @@ namespace butl
       r.value_column = r.name_column;
       r.colon_pos = r.start_pos;
       r.end_pos = r.start_pos;
-      return r;
+      return;
     }
 
     if (c != ':')
@@ -162,8 +169,6 @@ namespace butl
       //
       assert (!r.name.empty ());
     }
-
-    return r;
   }
 
   pair<string, string> manifest_parser::
diff --git a/libbutl/manifest-parser.ixx b/libbutl/manifest-parser.ixx
new file mode 100644
index 0000000..b308264
--- /dev/null
+++ b/libbutl/manifest-parser.ixx
@@ -0,0 +1,14 @@
+// file      : libbutl/manifest-parser.ixx -*- C++ -*-
+// copyright : Copyright (c) 2014-2018 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+namespace butl
+{
+  inline manifest_name_value manifest_parser::
+  next ()
+  {
+    manifest_name_value r;
+    do { parse_next (r); } while (filter_ && !filter_ (r));
+    return r;
+  }
+}
diff --git a/libbutl/manifest-parser.mxx b/libbutl/manifest-parser.mxx
index 7fc4ee3..0289d6f 100644
--- a/libbutl/manifest-parser.mxx
+++ b/libbutl/manifest-parser.mxx
@@ -11,9 +11,10 @@
 #ifndef __cpp_lib_modules
 #include <string>
 #include <iosfwd>
-#include <cstdint>   // uint64_t
-#include <utility>   // pair
-#include <stdexcept> // runtime_error
+#include <cstdint>    // uint64_t
+#include <utility>    // pair, move()
+#include <stdexcept>  // runtime_error
+#include <functional>
 #endif
 
 // Other includes.
@@ -52,8 +53,19 @@ LIBBUTL_MODEXPORT namespace butl
   class LIBBUTL_SYMEXPORT manifest_parser: protected butl::char_scanner
   {
   public:
-    manifest_parser (std::istream& is, const std::string& name)
-        : char_scanner (is), name_ (name) {}
+    // The filter, if specified, is called by next() prior to returning the
+    // pair to the caller. If the filter returns false, then the pair is
+    // discarded.
+    //
+    // Note that the filter should handle the end-of-manifest pairs (see
+    // below) carefully, so next() doesn't end up with an infinite cycle.
+    //
+    using filter_function = bool (manifest_name_value&);
+
+    manifest_parser (std::istream& is,
+                     const std::string& name,
+                     std::function<filter_function> filter = {})
+      : char_scanner (is), name_ (name), filter_ (std::move (filter)) {}
 
     const std::string&
     name () const {return name_;}
@@ -83,6 +95,9 @@ LIBBUTL_MODEXPORT namespace butl
 
   private:
     void
+    parse_next (manifest_name_value&);
+
+    void
     parse_name (manifest_name_value&);
 
     void
@@ -98,8 +113,11 @@ LIBBUTL_MODEXPORT namespace butl
 
   private:
     const std::string name_;
+    const std::function<filter_function> filter_;
 
     enum {start, body, end} s_ = start;
     std::string version_; // Current format version.
   };
 }
+
+#include <libbutl/manifest-parser.ixx>
diff --git a/libbutl/manifest-serializer.cxx b/libbutl/manifest-serializer.cxx
index 2d2e722..dfbd9e1 100644
--- a/libbutl/manifest-serializer.cxx
+++ b/libbutl/manifest-serializer.cxx
@@ -39,7 +39,7 @@ namespace butl
   using serialization = manifest_serialization;
 
   void manifest_serializer::
-  next (const string& n, const string& v)
+  write_next (const string& n, const string& v)
   {
     switch (s_)
     {
diff --git a/libbutl/manifest-serializer.ixx b/libbutl/manifest-serializer.ixx
new file mode 100644
index 0000000..1108ba1
--- /dev/null
+++ b/libbutl/manifest-serializer.ixx
@@ -0,0 +1,13 @@
+// file      : libbutl/manifest-serializer.ixx -*- C++ -*-
+// copyright : Copyright (c) 2014-2018 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+namespace butl
+{
+  inline void manifest_serializer::
+  next (const std::string& n, const std::string& v)
+  {
+    if (!filter_ || filter_ (n, v))
+      write_next (n, v);
+  }
+}
diff --git a/libbutl/manifest-serializer.mxx b/libbutl/manifest-serializer.mxx
index 66ca398..6a87656 100644
--- a/libbutl/manifest-serializer.mxx
+++ b/libbutl/manifest-serializer.mxx
@@ -11,8 +11,9 @@
 #ifndef __cpp_lib_modules
 #include <string>
 #include <iosfwd>
-#include <cstddef>   // size_t
-#include <stdexcept> // runtime_error
+#include <cstddef>    // size_t
+#include <stdexcept>  // runtime_error
+#include <functional>
 #endif
 
 // Other includes.
@@ -42,8 +43,23 @@ LIBBUTL_MODEXPORT namespace butl
   class LIBBUTL_SYMEXPORT manifest_serializer
   {
   public:
-    manifest_serializer (std::ostream& os, const std::string& name)
-        : os_ (os), name_ (name) {}
+    // The filter, if specified, is called by next() prior to serializing the
+    // pair into the stream. If the filter returns false, then the pair is
+    // discarded.
+    //
+    // Note that currently there is no way for the filter to modify the name
+    // or value. If we ever need this functionality, then we can add an
+    // "extended" filter alternative with two "receiving" arguments:
+    //
+    // bool (..., optional<string>& n, optional<string>& v);
+    //
+    using filter_function = bool (const std::string& name,
+                                  const std::string& value);
+
+    manifest_serializer (std::ostream& os,
+                         const std::string& name,
+                         std::function<filter_function> filter = {})
+        : os_ (os), name_ (name), filter_ (std::move (filter)) {}
 
     const std::string&
     name () const {return name_;}
@@ -77,6 +93,9 @@ LIBBUTL_MODEXPORT namespace butl
   private:
     friend class manifest_rewriter;
 
+    void
+    write_next (const std::string& name, const std::string& value);
+
     // Validate and write a name.
     //
     void
@@ -105,5 +124,8 @@ LIBBUTL_MODEXPORT namespace butl
   private:
     std::ostream& os_;
     const std::string name_;
+    const std::function<filter_function> filter_;
   };
 }
+
+#include <libbutl/manifest-serializer.ixx>
diff --git a/tests/manifest-parser/driver.cxx b/tests/manifest-parser/driver.cxx
index 30b754f..d2d54e9 100644
--- a/tests/manifest-parser/driver.cxx
+++ b/tests/manifest-parser/driver.cxx
@@ -7,7 +7,7 @@
 #ifndef __cpp_lib_modules
 #include <vector>
 #include <string>
-#include <utility> // pair
+#include <utility> // pair, move()
 #include <sstream>
 #include <iostream>
 #endif
@@ -30,7 +30,9 @@ using namespace butl;
 using pairs = vector<pair<string, string>>;
 
 static bool
-test (const char* manifest, const pairs& expected);
+test (const char* manifest,
+      const pairs& expected,
+      manifest_parser::filter_function = {});
 
 static bool
 fail (const char* manifest);
@@ -160,6 +162,25 @@ main ()
     auto p (manifest_parser::split_comment ("; comment"));
     assert (p.first == "" && p.second == "comment");
   }
+
+  // Filtering.
+  //
+  assert (test (":1\na: abc\nb: bca\nc: cab",
+                {{"","1"},{"a","abc"},{"c","cab"},{"",""},{"",""}},
+                [] (manifest_name_value& nv) {return nv.name != "b";}));
+
+  assert (test (":1\na: abc\nb: bca",
+                {{"","1"},{"ax","abc."},{"bx","bca."},{"",""},{"",""}},
+                [] (manifest_name_value& nv)
+                {
+                  if (!nv.name.empty ())
+                  {
+                    nv.name += 'x';
+                    nv.value += '.';
+                  }
+
+                  return true;
+                }));
 }
 
 static ostream&
@@ -177,11 +198,11 @@ operator<< (ostream& os, const pairs& ps)
 }
 
 static pairs
-parse (const char* m)
+parse (const char* m, manifest_parser::filter_function f = {})
 {
   istringstream is (m);
   is.exceptions (istream::failbit | istream::badbit);
-  manifest_parser p (is, "");
+  manifest_parser p (is, "", move (f));
 
   pairs r;
 
@@ -197,16 +218,16 @@ parse (const char* m)
     else
       eom = false;
 
-    r.emplace_back (nv.name, nv.value); // move
+    r.emplace_back (move (nv.name), move (nv.value));
   }
 
   return r;
 }
 
 static bool
-test (const char* m, const pairs& e)
+test (const char* m, const pairs& e, manifest_parser::filter_function f)
 {
-  pairs r (parse (m));
+  pairs r (parse (m, move (f)));
 
   if (r != e)
   {
diff --git a/tests/manifest-serializer/driver.cxx b/tests/manifest-serializer/driver.cxx
index d9a7255..be00106 100644
--- a/tests/manifest-serializer/driver.cxx
+++ b/tests/manifest-serializer/driver.cxx
@@ -30,7 +30,9 @@ using namespace butl;
 using pairs = vector<pair<string, string>>;
 
 static bool
-test (const pairs& manifest, const string& expected);
+test (const pairs& manifest,
+      const string& expected,
+      manifest_serializer::filter_function f = {});
 
 static bool
 fail (const pairs& manifest);
@@ -234,14 +236,20 @@ main ()
 
   assert (manifest_serializer::merge_comment ("value text", "") ==
           "value text");
+
+  // Filtering.
+  //
+  assert (test ({{"","1"},{"a","abc"},{"b","bca"},{"c","cab"},{"",""},{"",""}},
+                ": 1\na: abc\nc: cab\n",
+                [] (const string& n, const string&) {return n != "b";}));
 }
 
 static string
-serialize (const pairs& m)
+serialize (const pairs& m, manifest_serializer::filter_function f = {})
 {
   ostringstream os;
   os.exceptions (istream::failbit | istream::badbit);
-  manifest_serializer s (os, "");
+  manifest_serializer s (os, "", f);
 
   for (const auto& p: m)
   {
@@ -255,9 +263,9 @@ serialize (const pairs& m)
 }
 
 static bool
-test (const pairs& m, const string& e)
+test (const pairs& m, const string& e, manifest_serializer::filter_function f)
 {
-  string r (serialize (m));
+  string r (serialize (m, f));
 
   if (r != e)
   {