[fix] re #205: escaping of special chars in double quoted scalars

src/c4/yml/parse.cpp

@@ -4302,30 +4302,25 @@ csubstr Parser::_filter_dquot_scalar(substr s)
             }
             else if(next == 'r')
             {
-                //m_filter_arena.str[pos++] = '\r';
+                m_filter_arena.str[pos++] = '\r';
                 ++i;
             }
             else if(next == 't')
             {
                 m_filter_arena.str[pos++] = '\t';
                 ++i;
             }
-            else if(next == 'b')
-            {
-                m_filter_arena.str[pos++] = '\b';
-                ++i;
-            }
             else if(next == 'x')
             {
                 m_filter_arena.str[pos++] = '\\';
                 m_filter_arena.str[pos++] = 'x';
-                ++i; // loop will increment next
+                ++i;
             }
             else if(next == 'u')
             {
                 m_filter_arena.str[pos++] = '\\';
                 m_filter_arena.str[pos++] = 'u';
-                ++i; // loop will increment next
+                ++i;
             }
             else if(next == '\\')
             {
@@ -4337,6 +4332,21 @@ csubstr Parser::_filter_dquot_scalar(substr s)
                 m_filter_arena.str[pos++] = next;
                 ++i;
             }
+            else if(next == 'b')
+            {
+                m_filter_arena.str[pos++] = '\b';
+                ++i;
+            }
+            else if(next == 'f')
+            {
+                m_filter_arena.str[pos++] = '\f';
+                ++i;
+            }
+            else if(next == '0')
+            {
+                m_filter_arena.str[pos++] = '\0';
+                ++i;
+            }
             _c4dbgfdq("[%zu]: backslash...sofar=[%zu]~~~%.*s~~~", i, pos, _c4prsp(m_filter_arena.first(pos)));
         }
         else
@@ -4576,7 +4586,7 @@ csubstr Parser::_filter_block_scalar(substr s, BlockStyle_e style, BlockChomp_e
     case CHOMP_STRIP: // strip all newlines from the end
     {
         _c4dbgp("filt_block: chomp=STRIP (-)");
-        r = r.trimr("\r\n");
+        r = r.trimr("\n\r");
         break;
     }
     case CHOMP_CLIP: // clip to a single newline

test/CMakeLists.txt

@@ -69,6 +69,7 @@ ryml_add_test(json)
 ryml_add_test(preprocess)
 ryml_add_test(merge)
 ryml_add_test(location)
+ryml_add_test(yaml_events)
 ryml_add_test_case_group(empty_file)
 ryml_add_test_case_group(empty_map)
 ryml_add_test_case_group(empty_seq)

test/test_suite/test_suite_events.hpp

@@ -30,6 +30,14 @@ void emit_events(CharContainer *container, Tree const& C4_RESTRICT tree)
     container->resize(ret);
 }
 
+template<class CharContainer>
+CharContainer emit_events(Tree const& C4_RESTRICT tree)
+{
+    CharContainer result;
+    emit_events(&result, tree);
+    return result;
+}
+
 } // namespace yml
 } // namespace c4
 

test/test_suite/test_suite_events_emitter.cpp

@@ -35,38 +35,39 @@ struct EventsEmitter
             buf[pos] = c;
         ++pos;
     }
+    C4_ALWAYS_INLINE size_t emit_to_esc(csubstr val, size_t i, size_t prev, char c)
+    {
+        pr(val.range(prev, i));
+        pr('\\');
+        pr(c);
+        return i+1;
+    }
 };
 
 void EventsEmitter::emit_scalar(csubstr val, bool quoted)
 {
-    static constexpr const char openscalar[] = {':', '\''};
-    pr(openscalar[quoted]);
+    constexpr const char openchar[] = {':', '\''};
+    pr(openchar[quoted]);
     size_t prev = 0;
     for(size_t i = 0; i < val.len; ++i)
     {
-        if(val[i] == '\n')
-        {
-            pr(val.range(prev, i));
-            pr('\\');
-            pr('n');
-            prev = i+1;
-        }
-        else if(val[i] == '\t')
+        switch(val[i])
         {
-            pr(val.range(prev, i));
-            pr('\\');
-            pr('t');
-            prev = i+1;
+        case '\n':
+            prev = emit_to_esc(val, i, prev, 'n'); break;
+        case '\t':
+            prev = emit_to_esc(val, i, prev, 't'); break;
+        case '\\':
+            prev = emit_to_esc(val, i, prev, '\\'); break;
+        case '\r':
+            prev = emit_to_esc(val, i, prev, 'r'); break;
+        case '\b':
+            prev = emit_to_esc(val, i, prev, 'b'); break;
+        case '\f':
+            prev = emit_to_esc(val, i, prev, 'f'); break;
+        case '\0':
+            prev = emit_to_esc(val, i, prev, '0'); break;
         }
-        else if(val[i] == '\\')
-        {
-            pr(val.range(prev, i));
-            pr('\\');
-            pr('\\');
-            prev = i+1;
-        }
-        else if(val[i] == '\r')
-            continue;  // not really sure about this
     }
     pr(val.sub(prev)); // print remaining portion
 }
@@ -171,11 +172,7 @@ void EventsEmitter::emit_events(size_t node)
 
 void EventsEmitter::emit_doc(size_t node)
 {
-    bool use_triple = !m_tree->is_root(node) && m_tree->is_doc(node);
-    if(use_triple)
-        pr("+DOC ---");
-    else
-        pr("+DOC");
+    pr("+DOC");
     if(m_tree->is_val(node))
     {
         pr("\n=VAL");
@@ -190,10 +187,7 @@ void EventsEmitter::emit_doc(size_t node)
         pr('\n');
         emit_events(node);
     }
-    if(use_triple)
-        pr("-DOC ...\n");
-    else
-        pr("-DOC\n");
+    pr("-DOC\n");
 }
 
 void EventsEmitter::emit_events()

test/test_yaml_events.cpp

@@ -0,0 +1,125 @@
+#ifndef RYML_SINGLE_HEADER
+#include <c4/yml/std/std.hpp>
+#include <c4/yml/yml.hpp>
+#endif
+#include <gtest/gtest.h>
+
+#include "./test_case.hpp"
+#include "./test_suite/test_suite_events.hpp"
+#include "./test_suite/test_suite_events_emitter.cpp" // HACK
+
+namespace c4 {
+namespace yml {
+
+void test_evts(csubstr src, std::string expected)
+{
+    Tree tree = parse_in_arena(src);
+    auto actual = emit_events<std::string>(tree);
+    EXPECT_EQ(actual, expected);
+}
+
+TEST(events, docval)
+{
+    test_evts(
+        R"('quoted val'
+)",
+        R"(+STR
++DOC
+=VAL 'quoted val
+-DOC
+-STR
+)"
+        );
+}
+
+TEST(events, docsep)
+{
+    test_evts(
+        R"(--- 'quoted val'
+--- another
+...
+--- and yet another
+...
+---
+...
+)",
+        R"(+STR
++DOC
+=VAL 'quoted val
+-DOC
++DOC
+=VAL :another
+-DOC
++DOC
+=VAL :and yet another
+-DOC
++DOC
+=VAL :
+-DOC
+-STR
+)"
+        );
+}
+
+TEST(events, basic_map)
+{
+    test_evts(
+        "{foo: bar}",
+        R"(+STR
++DOC
++MAP
+=VAL :foo
+=VAL :bar
+-MAP
+-DOC
+-STR
+)"
+        );
+}
+
+TEST(events, basic_seq)
+{
+    test_evts(
+        "[foo, bar]",
+        R"(+STR
++DOC
++SEQ
+=VAL :foo
+=VAL :bar
+-SEQ
+-DOC
+-STR
+)"
+        );
+}
+
+TEST(events, escapes)
+{
+    test_evts(
+        R"("\b\r\n\0\f\/")",
+        "+STR\n"
+        "+DOC\n"
+        "=VAL '\\b\\r\\n\\0\\f/\n"
+        "-DOC\n"
+        "-STR\n"
+        );
+}
+
+
+//-----------------------------------------------------------------------------
+//-----------------------------------------------------------------------------
+//-----------------------------------------------------------------------------
+
+// The other test executables are written to contain the declarative-style
+// YmlTestCases. This executable does not have any but the build setup
+// assumes it does, and links with the test lib, which requires an existing
+// get_case() function. So this is here to act as placeholder until (if?)
+// proper test cases are added here. This was detected in #47 (thanks
+// @cburgard).
+Case const* get_case(csubstr)
+{
+    return nullptr;
+}
+
+} // namespace yml
+} // namespace c4