Merge pull request #502 from biojppm/ysfixes

biojppm · Feb 26, 2025 · 0bfb286 · 0bfb286
2 parents 8c37616 + 3d66090
commit 0bfb286
Show file tree

Hide file tree

Showing 5 changed files with 949 additions and 13 deletions.
diff --git a/changelog/current.md b/changelog/current.md
@@ -2,3 +2,27 @@
 ## Fixes
 
 - [PR#501](https://github.com/biojppm/rapidyaml/pull/501): fix missing tag in `- !!seq []`
+- [PR#502](https://github.com/biojppm/rapidyaml/pull/502): fix parse errors or missing tags:
+  - missing tags in empty documents:
+    ```yaml
+    !yamlscript/v0/bare
+    --- !code
+    42
+    ```
+  - trailing empty keys or vals:
+    ```yaml
+    a:
+       :
+    ```
+  - missing tags in trailing empty keys or vals:
+    ```yaml
+    a: !tag
+      !tag : !tag
+    ```
+  - missing tags in complex maps:
+    ```yaml
+    ? a: !tag
+      !tag : !tag
+    :
+      !tag : !tag
+    ```
diff --git a/src/c4/yml/parse_engine.def.hpp b/src/c4/yml/parse_engine.def.hpp
@@ -1611,9 +1611,10 @@ void ParseEngine<EventHandler>::_end2_doc()
 {
     _c4dbgp("doc: end");
     _RYML_CB_ASSERT(m_evt_handler->m_stack.m_callbacks, has_any(RDOC));
-    if(m_doc_empty)
+    if(m_doc_empty || (m_pending_tags.num_entries || m_pending_anchors.num_entries))
     {
         _c4dbgp("doc was empty; add empty val");
+        _handle_annotations_before_blck_val_scalar();
         m_evt_handler->set_val_scalar_plain_empty();
     }
     m_evt_handler->end_doc();
@@ -1623,9 +1624,10 @@ template<class EventHandler>
 void ParseEngine<EventHandler>::_end2_doc_expl()
 {
     _c4dbgp("doc: end");
-    if(m_doc_empty)
+    if(m_doc_empty || (m_pending_tags.num_entries || m_pending_anchors.num_entries))
     {
         _c4dbgp("doc: no children; add empty val");
+        _handle_annotations_before_blck_val_scalar();
         m_evt_handler->set_val_scalar_plain_empty();
     }
     m_evt_handler->end_doc_expl();
@@ -1648,6 +1650,14 @@ void ParseEngine<EventHandler>::_maybe_end_doc()
         _c4dbgp("doc must be finished");
         _end2_doc();
     }
+    else if(m_doc_empty && (m_pending_tags.num_entries || m_pending_anchors.num_entries))
+    {
+        _c4dbgp("no doc to finish, but pending annotations");
+        m_evt_handler->begin_doc();
+        _handle_annotations_before_blck_val_scalar();
+        m_evt_handler->set_val_scalar_plain_empty();
+        m_evt_handler->end_doc();
+    }
 }
 
 template<class EventHandler>
@@ -6800,14 +6810,26 @@ void ParseEngine<EventHandler>::_handle_map_block()
                 m_evt_handler->set_val_scalar_plain_empty();
                 m_evt_handler->add_sibling();
                 m_evt_handler->set_key_scalar_plain_empty();
-                _line_progressed(1);
-                _maybe_skip_whitespace_tokens();
-                goto mapblck_again;
+            }
+            else if(startindent > m_evt_handler->m_curr->indref)
+            {
+                _c4dbgp("mapblck[RVAL]: start val mapblck");
+                addrem_flags(RNXT, RVAL);
+                _handle_annotations_before_start_mapblck(startline);
+                m_evt_handler->begin_map_val_block();
+                _handle_annotations_and_indentation_after_start_mapblck(startindent, startline);
+                m_evt_handler->set_key_scalar_plain_empty();
+                _set_indentation(m_evt_handler->m_curr->line_contents.indentation);
+                // keep the child state on RVAL
+                addrem_flags(RVAL, RNXT);
             }
             else
             {
                 _c4err("parse error");
             }
+            _line_progressed(1);
+            _maybe_skip_whitespace_tokens();
+            goto mapblck_again;
         }
         else if(first == '.')
         {
@@ -7160,6 +7182,7 @@ void ParseEngine<EventHandler>::_handle_map_block()
             {
                 _c4dbgp("mapblck[QMRK]: start child seqblck (!)");
                 addrem_flags(RKCL, RKEY|QMRK);
+                _handle_annotations_before_blck_key_scalar();
                 m_evt_handler->begin_seq_key_block();
                 addrem_flags(RVAL|RSEQ, RMAP|RKCL|QMRK);
                 _set_indentation(startindent);
@@ -7521,13 +7544,16 @@ void ParseEngine<EventHandler>::_handle_unk()
         if(m_doc_empty)
         {
             _c4dbgp("it's a map with an empty key");
+            const size_t startindent = m_evt_handler->m_curr->line_contents.indentation; // save
+            const size_t startline = m_evt_handler->m_curr->pos.line; // save
             m_evt_handler->check_trailing_doc_token();
             _maybe_begin_doc();
-            _handle_annotations_before_blck_val_scalar();
+            _handle_annotations_before_start_mapblck(startline);
             m_evt_handler->begin_map_val_block();
+            _handle_annotations_and_indentation_after_start_mapblck(startindent, startline);
             m_evt_handler->set_key_scalar_plain_empty();
             m_doc_empty = false;
-            _save_indentation();
+            _set_indentation(startindent);
         }
         else
         {