From 405dfa3e28ab71d4f6b5210faba0e3600070a0f3 Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Tue, 23 Nov 2021 19:24:00 +0300
Subject: Add support for new-fashioned multi-line manifest value introducer

---
 libbutl/manifest-parser.cxx     | 35 ++++++++++++++++++++++++++++++++++-
 libbutl/manifest-parser.hxx     |  4 ++--
 libbutl/manifest-serializer.cxx |  3 +++
 libbutl/manifest-serializer.hxx | 10 ++++++++--
 4 files changed, 47 insertions(+), 5 deletions(-)

(limited to 'libbutl')

diff --git a/libbutl/manifest-parser.cxx b/libbutl/manifest-parser.cxx
index 1cb0ec7..ae0d43d 100644
--- a/libbutl/manifest-parser.cxx
+++ b/libbutl/manifest-parser.cxx
@@ -222,7 +222,8 @@ namespace butl
     string& v (r.value);
     string::size_type n (0); // Size of last non-space character (simple mode).
 
-    // Detect the multi-line mode introducer.
+    // Detect the old-fashioned multi-line mode introducer (like in
+    // 'foo:\<newline>').
     //
     bool ml (false);
     if (c == '\\')
@@ -245,6 +246,38 @@ namespace butl
         unget (c);
     }
 
+    // Detect the new-fashioned multi-line mode introducer (like in
+    // 'foo:<newline>\<newline>').
+    //
+    if (!ml && c == '\n')
+    {
+      get ();
+      xchar p1 (peek ());
+
+      if (p1 == '\\')
+      {
+        get ();
+        xchar p2 (peek ());
+
+        if (p2 == '\n')
+        {
+          get (); // Newline is not part of the value so skip it.
+          c = peek ();
+          ml = true;
+        }
+        else if (eos (p2))
+        {
+          c = p2;    // Set to EOF.
+          ml = true;
+        }
+        else
+          unget (p1);  // Unget '\\'. Note: '\n' will be ungot below.
+      }
+
+      if (!ml)
+        unget (c); // Unget '\n'.
+    }
+
     // Multi-line value starts from the line that follows the name.
     //
     if (ml)
diff --git a/libbutl/manifest-parser.hxx b/libbutl/manifest-parser.hxx
index d53eb42..601fb2d 100644
--- a/libbutl/manifest-parser.hxx
+++ b/libbutl/manifest-parser.hxx
@@ -37,7 +37,7 @@ namespace butl
   };
 
   class LIBBUTL_SYMEXPORT manifest_parser:
-    protected char_scanner<utf8_validator>
+    protected char_scanner<utf8_validator, 2>
   {
   public:
     // The filter, if specified, is called by next() prior to returning the
@@ -83,7 +83,7 @@ namespace butl
     split_comment (const std::string&);
 
   private:
-    using base = char_scanner<utf8_validator>;
+    using base = char_scanner<utf8_validator, 2>;
 
     void
     parse_next (manifest_name_value&);
diff --git a/libbutl/manifest-serializer.cxx b/libbutl/manifest-serializer.cxx
index 5875052..b0d0324 100644
--- a/libbutl/manifest-serializer.cxx
+++ b/libbutl/manifest-serializer.cxx
@@ -297,6 +297,9 @@ namespace butl
         v.back () == ' '      ||
         v.back () == '\t')
     {
+      if (multiline_v2_)
+        os_ << endl;
+
       os_ << "\\" << endl; // Multi-line mode introducer.
 
       // Chunk the value into fragments separated by newlines.
diff --git a/libbutl/manifest-serializer.hxx b/libbutl/manifest-serializer.hxx
index 43924e7..2159901 100644
--- a/libbutl/manifest-serializer.hxx
+++ b/libbutl/manifest-serializer.hxx
@@ -45,14 +45,19 @@ namespace butl
     // Unless long_lines is true, break lines in values (including multi-line)
     // so that their length does not exceed 78 codepoints (including '\n').
     //
+    // Note that the multiline_v2 flag is temporary and should not be used
+    // except by the implementation for testing.
+    //
     manifest_serializer (std::ostream& os,
                          const std::string& name,
                          bool long_lines = false,
-                         std::function<filter_function> filter = {})
+                         std::function<filter_function> filter = {},
+                         bool multiline_v2 = false)
       : os_ (os),
         name_ (name),
         long_lines_ (long_lines),
-        filter_ (std::move (filter))
+        filter_ (std::move (filter)),
+        multiline_v2_ (multiline_v2)
     {
     }
 
@@ -123,6 +128,7 @@ namespace butl
     const std::string name_;
     bool long_lines_;
     const std::function<filter_function> filter_;
+    bool multiline_v2_;
   };
 
   // Serialize a manifest to a stream adding the leading format version pair
-- 
cgit v1.1