From 6254448640530240dc9199bed60cd5568cbaf601 Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Tue, 1 Aug 2017 16:42:22 +0300
Subject: Add manifest_parser::split_comment() and
 manifest_serializer::merge_comment()

---
 libbutl/manifest-parser.cxx          | 42 ++++++++++++++++++++++++++++++++++++
 libbutl/manifest-parser.hxx          |  8 +++++++
 libbutl/manifest-serializer.cxx      | 25 +++++++++++++++++++++
 libbutl/manifest-serializer.hxx      |  7 ++++++
 tests/manifest-parser/driver.cxx     | 17 +++++++++++++++
 tests/manifest-serializer/driver.cxx |  8 +++++++
 6 files changed, 107 insertions(+)

diff --git a/libbutl/manifest-parser.cxx b/libbutl/manifest-parser.cxx
index ae92d79..304815a 100644
--- a/libbutl/manifest-parser.cxx
+++ b/libbutl/manifest-parser.cxx
@@ -125,6 +125,48 @@ namespace butl
     return r;
   }
 
+  pair<string, string> manifest_parser::
+  split_comment (const string& v)
+  {
+    using iterator = string::const_iterator;
+
+    auto space = [] (char c) -> bool {return c == ' ' || c == '\t';};
+
+    iterator i (v.begin ());
+    iterator e (v.end ());
+
+    string r;
+    size_t n (0);
+    for (char c; i != e && (c = *i) != ';'; ++i)
+    {
+      // Unescape ';' character.
+      //
+      if (c == '\\' && i + 1 != e && *(i + 1) == ';')
+        c = *++i;
+
+      r += c;
+
+      if (!space (c))
+        n = r.size ();
+    }
+
+    // Strip the value trailing spaces.
+    //
+    if (r.size () != n)
+      r.resize (n);
+
+    // Find beginning of a comment (i).
+    //
+    if (i != e)
+    {
+      // Skip spaces.
+      //
+      for (++i; i != e && space (*i); ++i);
+    }
+
+    return make_pair (move (r), string (i, e));
+  }
+
   void manifest_parser::
   parse_name (name_value& r)
   {
diff --git a/libbutl/manifest-parser.hxx b/libbutl/manifest-parser.hxx
index fd0cc4e..f84badd 100644
--- a/libbutl/manifest-parser.hxx
+++ b/libbutl/manifest-parser.hxx
@@ -8,6 +8,7 @@
 #include <string>
 #include <iosfwd>
 #include <cstdint>   // uint64_t
+#include <utility>   // pair
 #include <stdexcept> // runtime_error
 
 #include <libbutl/export.hxx>
@@ -71,6 +72,13 @@ namespace butl
     manifest_name_value
     next ();
 
+    // Split the manifest value, optionally followed by ';' character and a
+    // comment into the value/comment pair. Note that ';' characters in the
+    // value must be escaped by the backslash.
+    //
+    static std::pair<std::string, std::string>
+    split_comment (const std::string&);
+
   private:
     void
     parse_name (manifest_name_value&);
diff --git a/libbutl/manifest-serializer.cxx b/libbutl/manifest-serializer.cxx
index adb7f13..d8f640b 100644
--- a/libbutl/manifest-serializer.cxx
+++ b/libbutl/manifest-serializer.cxx
@@ -133,6 +133,31 @@ namespace butl
     os_ << endl;
   }
 
+  string manifest_serializer::
+  merge_comment (const string& value, const string& comment)
+  {
+    string r;
+    for (char c: value)
+    {
+      // Escape ';' character.
+      //
+      if (c == ';')
+        r += '\\';
+
+      r += c;
+    }
+
+    // Add the comment.
+    //
+    if (!comment.empty ())
+    {
+      r += "; ";
+      r += comment;
+    }
+
+    return r;
+  }
+
   void manifest_serializer::
   check_name (const string& n)
   {
diff --git a/libbutl/manifest-serializer.hxx b/libbutl/manifest-serializer.hxx
index 89c162c..d95ddf1 100644
--- a/libbutl/manifest-serializer.hxx
+++ b/libbutl/manifest-serializer.hxx
@@ -51,6 +51,13 @@ namespace butl
     void
     comment (const std::string&);
 
+    // Merge the manifest value and a comment into the single string, having
+    // the '<value>; <comment>' form. Escape ';' characters in the value with
+    // the backslash.
+    //
+    static std::string
+    merge_comment (const std::string& value, const std::string& comment);
+
   private:
     void
     check_name (const std::string&);
diff --git a/tests/manifest-parser/driver.cxx b/tests/manifest-parser/driver.cxx
index e480508..037df96 100644
--- a/tests/manifest-parser/driver.cxx
+++ b/tests/manifest-parser/driver.cxx
@@ -130,6 +130,23 @@ main ()
                 {{"","1"},{"a","x\\"},{"b",""},{"",""},{"",""}}));
   assert (test (":1\na:\\\nx\\\\\\\n\\\nb:",
                 {{"","1"},{"a","x\\\\"},{"b",""},{"",""},{"",""}}));
+
+  // Manifest value splitting (into the value/comment pair).
+  //
+  {
+    auto p (manifest_parser::split_comment ("value\\; text ; comment text"));
+    assert (p.first == "value; text" && p.second == "comment text");
+  }
+
+  {
+    auto p (manifest_parser::split_comment ("value"));
+    assert (p.first == "value" && p.second == "");
+  }
+
+  {
+    auto p (manifest_parser::split_comment ("; comment"));
+    assert (p.first == "" && p.second == "comment");
+  }
 }
 
 static ostream&
diff --git a/tests/manifest-serializer/driver.cxx b/tests/manifest-serializer/driver.cxx
index 69c1c71..dad96bd 100644
--- a/tests/manifest-serializer/driver.cxx
+++ b/tests/manifest-serializer/driver.cxx
@@ -213,6 +213,14 @@ main ()
                 ": 1\na: c:\\\\\n"));
   assert (test ({{"","1"},{"a","c:\\\nd:\\"},{"",""},{"",""}},
                 ": 1\na: \\\nc:\\\\\nd:\\\\\n\\\n"));
+
+  // Manifest value/comment merging.
+  //
+  assert (manifest_serializer::merge_comment ("value; text", "comment") ==
+          "value\\; text; comment");
+
+  assert (manifest_serializer::merge_comment ("value text", "") ==
+          "value text");
 }
 
 static string
-- 
cgit v1.1