[wip fix] ensure parse error on empty block scalars with missing chomp or 0 indentation

changelog/current.md

@@ -12,19 +12,33 @@
     :barbar: b    # was ok
     :barbarbar: c # was ol
   ```
-- Maps: ensure parse error on keyless trailing scalars:
-  ```yaml
-  --- # test suite case 236B
-  foo:
-    bar
-  invalid
-  --- # test suite case 7MNF
-  top1:
-    key1: val1
-  top2
-  --- # test suite case 9CWY
-  key:
-   - item1
-   - item2
-  invalid
-  ```
+- Scalars:
+  - Do not accept block scalars with 0 indentation:
+    ```yaml
+    # test suite case 2G84
+    --- |0   # error
+    --- |10  # error
+    --- |1-  # ok
+    --- |1+  # ok
+    ```
+- Maps
+  - ensure parse error on keyless terminating scalars:
+    ```yaml
+    ---
+    foo:
+      bar
+    valid:    # ok
+    --- # test suite case 236B
+    foo:
+      bar
+    invalid   # parse error expected
+    --- # test suite case 7MNF
+    top1:
+      key1: val1
+    invalid   # parse error expected
+    --- # test suite case 9CWY
+    key:
+     - item1
+     - item2
+    invalid   # parse error expected
+    ```

src/c4/yml/parse.cpp

@@ -4003,6 +4003,7 @@ csubstr Parser::_scan_block()
     BlockStyle_e newline = s.begins_with('>') ? BLOCK_FOLD : BLOCK_LITERAL;
     BlockChomp_e chomp = CHOMP_CLIP; // default to clip unless + or - are used
     size_t indentation = npos; // have to find out if no spec is given
+    bool explicit_chomp = false;
     csubstr digits;
     if(s.len > 1)
     {
@@ -4014,6 +4015,7 @@ csubstr Parser::_scan_block()
         _c4dbgpf("scanning block: spec chomp char at {}", pos);
         if(pos != npos)
         {
+            explicit_chomp = true;
             if(t[pos] == '-')
                 chomp = CHOMP_STRIP;
             else if(t[pos] == '+')
@@ -4028,7 +4030,9 @@ csubstr Parser::_scan_block()
         if( ! digits.empty())
         {
             if( ! c4::atou(digits, &indentation))
-                _c4err("parse error: could not read decimal");
+                _c4err("could not read block scalar indentation");
+            if( ! indentation)
+                _c4err("invalid block scalar indentation");
             _c4dbgpf("scanning block: indentation specified: {}. add {} from curr state -> {}", indentation, m_state->indref, indentation+m_state->indref);
             indentation += m_state->indref;
         }
@@ -4160,7 +4164,6 @@ csubstr Parser::_scan_block()
         ++num_lines;
     }
     _RYML_CB_ASSERT(m_stack.m_callbacks, m_state->pos.line == (first + num_lines));
-    C4_UNUSED(num_lines);
     C4_UNUSED(first);
 
     if(indentation == npos)
@@ -4171,6 +4174,8 @@ csubstr Parser::_scan_block()
 
     if(num_lines)
         _line_ended_undo();
+    else if(!explicit_chomp)
+        _c4err("block scalars must not be empty");
 
     _c4dbgpf("scanning block: raw=~~~{}~~~", raw_block);
 

test/test_suite/test_suite_parts.cpp

@@ -62,8 +62,6 @@ constexpr const AllowedFailure allowed_failures[] = {
     _("Y79Y_004-error", "should not accept tab after -"),
     _("Y79Y_005-error", "should not accept tab after -"),
     // block scalars
-    _("2G84_00-error" , "should not accept the block literal spec"),
-    _("2G84_01-error" , "should not accept the block literal spec"),
     _("5LLU-error"    , "should not accept folded scalar with wrong indented line after spaces only"),
     _("S4GJ-error"    , "should not accept text after block scalar indicator"),
     _("S98Z-error"    , "should not accept block scalar with more spaces than first content line"),
@@ -112,7 +110,8 @@ constexpr const AllowedFailure allowed_failures[] = {
     // These tests are skipped because they cover parts of YAML that
     // are deliberately not implemented by ryml.
 
-    #ifndef RYML_WITH_TAB_TOKENS // -<tab> or :<tab> are supported only when the above macro is defined
+    // -<tab> or :<tab> are supported only when the above macro is defined
+    #ifndef RYML_WITH_TAB_TOKENS
     _("A2M4-in_yaml-events"    , "tabs tokens"),
     _("6BCT-in_yaml"           , "tabs tokens"),
     _("J3BT-in_yaml"           , "tabs tokens"),