[no subject]

[Patrice Dumas]

branch: master
commit 62df9dda7d34ad5ca286f92c02fedca73b113a51
Author: Patrice Dumas <pertusus@free.fr>
AuthorDate: Fri Jun 7 11:29:50 2024 +0200

    * tp/Texinfo/XS/main/element_types.awk: block_line_arg and line_arg
    hold spaces_after_cmd_before_arg and comment_at_end, not the commands.
    
    * tp/Texinfo/XS/main/build_perl_info.c (store_info_element)
    (element_to_perl_hash), tp/Texinfo/XS/main/convert_to_texinfo.c
    (convert_to_texinfo_internal), tp/Texinfo/XS/main/manipulate_tree.c
    (copy_element, copy_associated_info, copy_tree_internal)
    (copy_extra_info), tp/Texinfo/XS/main/tree.c (new_element)
    (destroy_element), tp/Texinfo/XS/main/tree_types.h
    (enum elt_info_type), tp/Texinfo/XS/parsetexi/parser.c
    (isolate_last_space),
    tp/Texinfo/XS/structuring_transfo/transformations.c (new_node):
    initialize, free, copy elt_info list of elements meant to replace
    the elements in info.  Put comment_at_end in that list instead of
    in the info structure and pass it to Perl.
    
    * tp/Texinfo/XS/main/tree_types.h (enum string_info_type, ELEMENT):
    prepare string information in element to replace information in info.
---
 ChangeLog                                          | 21 +++++++
 tp/TODO                                            |  1 +
 tp/Texinfo/XS/main/build_perl_info.c               | 43 ++++++++++++-
 tp/Texinfo/XS/main/convert_to_texinfo.c            |  9 ++-
 tp/Texinfo/XS/main/element_types.awk               | 10 ++-
 tp/Texinfo/XS/main/element_types.c                 |  8 +--
 tp/Texinfo/XS/main/manipulate_tree.c               | 71 ++++++++++++++++++----
 tp/Texinfo/XS/main/tree.c                          | 14 +++++
 tp/Texinfo/XS/main/tree_types.h                    | 19 ++++++
 tp/Texinfo/XS/parsetexi/parser.c                   |  4 +-
 .../XS/structuring_transfo/transformations.c       |  3 +-
 11 files changed, 177 insertions(+), 26 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index a313be4dc4..8b62764249 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,24 @@
+2024-06-07  Patrice Dumas  <pertusus@free.fr>
+
+       * tp/Texinfo/XS/main/element_types.awk: block_line_arg and line_arg
+       hold spaces_after_cmd_before_arg and comment_at_end, not the commands.
+
+       * tp/Texinfo/XS/main/build_perl_info.c (store_info_element)
+       (element_to_perl_hash), tp/Texinfo/XS/main/convert_to_texinfo.c
+       (convert_to_texinfo_internal), tp/Texinfo/XS/main/manipulate_tree.c
+       (copy_element, copy_associated_info, copy_tree_internal)
+       (copy_extra_info), tp/Texinfo/XS/main/tree.c (new_element)
+       (destroy_element), tp/Texinfo/XS/main/tree_types.h
+       (enum elt_info_type), tp/Texinfo/XS/parsetexi/parser.c
+       (isolate_last_space),
+       tp/Texinfo/XS/structuring_transfo/transformations.c (new_node):
+       initialize, free, copy elt_info list of elements meant to replace
+       the elements in info.  Put comment_at_end in that list instead of
+       in the info structure and pass it to Perl.
+
+       * tp/Texinfo/XS/main/tree_types.h (enum string_info_type, ELEMENT):
+       prepare string information in element to replace information in info.
+
 2024-06-05  Patrice Dumas  <pertusus@free.fr>
 
        Do not use unnamed union for KEY_PAIR in rebased code
diff --git a/tp/TODO b/tp/TODO
index 239804414f..60e78bfeba 100644
--- a/tp/TODO
+++ b/tp/TODO
@@ -76,6 +76,7 @@ Gavin and Patrice
 char *alias_of
 char *command_info_string (for arg_line or command_name or @verb delimiter)
 ELEMENT *elements[3]
+in block_command line_command
  spaces_after_cmd_before_arg (brace commands) or comment_at_end (line/block 
commands)
 for type brace_arg, line and block commands, BRACE_context commands
  spaces_before_argument, spaces_after_argument
diff --git a/tp/Texinfo/XS/main/build_perl_info.c 
b/tp/Texinfo/XS/main/build_perl_info.c
index 37ef6beea1..14d84b4800 100644
--- a/tp/Texinfo/XS/main/build_perl_info.c
+++ b/tp/Texinfo/XS/main/build_perl_info.c
@@ -541,6 +541,37 @@ store_source_mark_list (const ELEMENT *e)
     }
 }
 
+/* FIXME it is inefficient to store and get every time an element
+   is stored, better use additional_info_hv or a pointer on it as an argument 
*/
+void
+store_info_element (ELEMENT *e, ELEMENT *info_element, const char *type_key,
+                    const char *key, int avoid_recursion, int *nr_info)
+{
+  HV *additional_info_hv;
+
+  dTHX;
+
+  if (*nr_info == 0)
+    {
+      additional_info_hv = (HV *) newHV ();
+      hv_store (e->hv, type_key, strlen (type_key),
+                newRV_inc ((SV *)additional_info_hv), 0);
+    }
+  else
+    {
+      SV **additional_info_sv = hv_fetch (e->hv, type_key,
+                                          strlen (type_key), 0);
+      additional_info_hv = (HV *)SvRV (*additional_info_sv);
+    }
+
+  (*nr_info)++;
+  if (!info_element->hv || !avoid_recursion)
+    element_to_perl_hash (info_element, avoid_recursion);
+
+  hv_store (additional_info_hv, key, strlen (key),
+            newRV_inc ((SV *)info_element->hv), 0);
+}
+
 static int hashes_ready = 0;
 static U32 HSH_parent = 0;
 static U32 HSH_type = 0;
@@ -640,7 +671,7 @@ element_to_perl_hash (ELEMENT *e, int avoid_recursion)
       hv_store (info_hv, "inserted", strlen ("inserted"),
                 newSViv (1), 0);
       hv_store (e->hv, key, strlen (key),
-              newRV_inc ((SV *)info_hv), 0);
+                newRV_inc ((SV *)info_hv), 0);
       nr_info++;
     }
 
@@ -651,6 +682,16 @@ element_to_perl_hash (ELEMENT *e, int avoid_recursion)
       return;
     }
 
+  if (e->type == ET_block_line_arg || e->type == ET_line_arg)
+    {
+      ELEMENT *f = e->elt_info[eit_comment_at_end];
+      if (f)
+        {
+          store_info_element (e, f, "info", "comment_at_end",
+                              avoid_recursion, &nr_info);
+        }
+    }
+
   /* non-text elements */
 
   store_additional_info (e, &e->c->info_info, "info", &nr_info,
diff --git a/tp/Texinfo/XS/main/convert_to_texinfo.c 
b/tp/Texinfo/XS/main/convert_to_texinfo.c
index 1903502e1a..6d120de0c7 100644
--- a/tp/Texinfo/XS/main/convert_to_texinfo.c
+++ b/tp/Texinfo/XS/main/convert_to_texinfo.c
@@ -172,9 +172,12 @@ convert_to_texinfo_internal (const ELEMENT *e, TEXT 
*result)
           ADD((char *)elt->text->text);
         }
 
-      elt = lookup_info_element (e, "comment_at_end");
-      if (elt)
-        convert_to_texinfo_internal (elt, result);
+      if (e->type == ET_block_line_arg || e->type == ET_line_arg)
+        {
+          elt = e->elt_info[eit_comment_at_end];
+          if (elt)
+            convert_to_texinfo_internal (elt, result);
+        }
 
       if (e->type == ET_bracketed_arg || e->type == ET_bracketed_linemacro_arg)
         ADD("}");
diff --git a/tp/Texinfo/XS/main/element_types.awk 
b/tp/Texinfo/XS/main/element_types.awk
index d66455259b..acdb231982 100644
--- a/tp/Texinfo/XS/main/element_types.awk
+++ b/tp/Texinfo/XS/main/element_types.awk
@@ -59,11 +59,15 @@ END {
         t = types[line_idx]
         flags_str = ""
         elt_info_number = 0;
+        # line_command: spaces_before_argument
+        # block_line_arg: spaces_before_argument
         if (t == "macro_call" || t == "rmacro_call" \
-                   || t == "brace_noarg_command" || t == "brace_command") {
+                   || t == "brace_noarg_command" || t == "brace_command" \
+                   || t == "block_command" || t == "line_command" ) {
           elt_info_number = 1;
-        } else if (t == "line_command" || t == "block_command" \
-                   || t == "brace_arg" || t == "bracketed_arg") {
+        # block_line_arg and line_arg: comment_at_end, spaces_after_argument
+        } else if (t == "brace_arg" || t == "bracketed_arg" \
+                   || t == "block_line_arg" || t == "line_arg") {
           elt_info_number = 2;
         } else if (t == "context_brace_command") {
           elt_info_number = 3;
diff --git a/tp/Texinfo/XS/main/element_types.c 
b/tp/Texinfo/XS/main/element_types.c
index 49721843fa..e0a9b358b1 100644
--- a/tp/Texinfo/XS/main/element_types.c
+++ b/tp/Texinfo/XS/main/element_types.c
@@ -10,8 +10,8 @@ TYPE_DATA type_data[] = {
 "brace_noarg_command", 0, 1,
 "container_command", 0, 0,
 "lineraw_command", 0, 0,
-"line_command", 0, 2,
-"block_command", 0, 2,
+"line_command", 0, 1,
+"block_command", 0, 1,
 "brace_command", 0, 1,
 "brace_args_command", 0, 0,
 "context_brace_command", 0, 3,
@@ -41,8 +41,8 @@ TYPE_DATA type_data[] = {
 "brace_container", 0, 0,
 "brace_command_context", 0, 0,
 "brace_arg", 0, 2,
-"block_line_arg", 0, 0,
-"line_arg", 0, 0,
+"block_line_arg", 0, 2,
+"line_arg", 0, 2,
 "following_arg", 0, 0,
 "rawline_arg", TF_text, 0,
 "menu_entry", 0, 0,
diff --git a/tp/Texinfo/XS/main/manipulate_tree.c 
b/tp/Texinfo/XS/main/manipulate_tree.c
index 1820212465..4ba6619ed2 100644
--- a/tp/Texinfo/XS/main/manipulate_tree.c
+++ b/tp/Texinfo/XS/main/manipulate_tree.c
@@ -53,6 +53,24 @@ increase_ref_counter (ELEMENT *element)
   (*counter_ptr) ++;
 }
 
+ELEMENT *
+copy_element (ELEMENT *f)
+{
+  KEY_PAIR *k_copy;
+  ELEMENT *e = 0;
+  k_copy = lookup_extra_by_index (f, "_copy", -1);
+  if (k_copy)
+    {
+      e = k_copy->k.element;
+    }
+  else
+    {
+      increase_ref_counter (f);
+    }
+  copy_tree_internal (f, 0);
+  return e;
+}
+
 void
 copy_associated_info (ASSOCIATED_INFO *info, ASSOCIATED_INFO* new_info)
 {
@@ -76,18 +94,15 @@ copy_associated_info (ASSOCIATED_INFO *info, 
ASSOCIATED_INFO* new_info)
         case extra_element_oot:
           if (!strcmp (key, "_copy"))
             break;
-          k_copy = lookup_extra_by_index (f, "_copy", -1);
-          if (k_copy)
-            {
-              KEY_PAIR *k
-                = get_associated_info_key (new_info, key, k_ref->type);
-              k->k.element = k_copy->k.element;
-            }
-          else
-            {
-              increase_ref_counter (f);
-            }
-          copy_tree_internal (f, 0);
+          {
+            ELEMENT *copy = copy_element (f);
+            if (copy)
+              {
+                KEY_PAIR *k
+                  = get_associated_info_key (new_info, key, k_ref->type);
+                k->k.element = copy;
+              }
+          }
           break;
         case extra_contents:
         case extra_directions:
@@ -192,6 +207,17 @@ copy_tree_internal (ELEMENT* current, ELEMENT *parent)
   for (i = 0; i < current->c->contents.number; i++)
     add_to_element_contents (new,
                 copy_tree_internal (current->c->contents.list[i], new));
+
+  if (type_data[current->type].elt_info_number > 0)
+    {
+      for (i = 0; i < type_data[current->type].elt_info_number; i++)
+        if (current->elt_info[i])
+          {
+            ELEMENT *copy = copy_element (current->elt_info[i]);
+            if (copy)
+              new->elt_info[i] = copy;
+          }
+    }
   copy_associated_info (&current->c->info_info, &new->c->info_info);
   copy_associated_info (&current->extra_info, &new->extra_info);
   return new;
@@ -378,6 +404,27 @@ copy_extra_info (ELEMENT *current, ELEMENT *new)
       for (i = 0; i < current->c->contents.number; i++)
         copy_extra_info (current->c->contents.list[i], 
new->c->contents.list[i]);
 
+      if (type_data[current->type].elt_info_number > 0)
+        {
+          int j;
+          for (j = 0; j < type_data[current->type].elt_info_number; j++)
+            {
+              if (current->elt_info[j])
+                {
+                  ELEMENT *f = current->elt_info[j];
+                  const KEY_PAIR *k_copy;
+
+                  if (!new->elt_info[j])
+                    {
+                      ELEMENT *e = get_copy_ref (f);
+                      new->elt_info[j] = e;
+                    }
+                  k_copy = lookup_extra_by_index (f, "_copy", -1);
+                  if (k_copy)
+                    copy_extra_info (f, k_copy->k.element);
+                }
+            }
+        }
       associate_info_references (&current->c->info_info, &new->c->info_info);
 
       /* text element have _copy and _counter only in extra, not to be copied 
*/
diff --git a/tp/Texinfo/XS/main/tree.c b/tp/Texinfo/XS/main/tree.c
index 7c69617405..6790ba3bbf 100644
--- a/tp/Texinfo/XS/main/tree.c
+++ b/tp/Texinfo/XS/main/tree.c
@@ -89,6 +89,14 @@ new_element (enum element_type type)
   e->c = (CONTAINER *) malloc (sizeof (CONTAINER));
   memset (e->c, 0, sizeof (CONTAINER));
 
+  if (type_data[type].elt_info_number > 0)
+    {
+      e->elt_info = (ELEMENT **)
+        malloc (sizeof (ELEMENT *) * type_data[type].elt_info_number);
+      memset (e->elt_info, 0,
+         sizeof (ELEMENT *) * type_data[type].elt_info_number);
+    }
+
   return e;
 }
 
@@ -209,11 +217,17 @@ destroy_element (ELEMENT *e)
     }
   else
     {
+      int i;
       destroy_associated_info (&e->c->info_info);
   /* Note the pointers in these lists are not themselves freed. */
       free (e->c->contents.list);
       free (e->c->args.list);
 
+      for (i = 0; i < type_data[e->type].elt_info_number; i++)
+        if (e->elt_info[i])
+          destroy_element_and_children (e->elt_info[i]);
+      free (e->elt_info);
+
       free (e->c);
     }
 
diff --git a/tp/Texinfo/XS/main/tree_types.h b/tp/Texinfo/XS/main/tree_types.h
index c36cb70714..0db1ea9253 100644
--- a/tp/Texinfo/XS/main/tree_types.h
+++ b/tp/Texinfo/XS/main/tree_types.h
@@ -217,6 +217,23 @@ typedef struct CONTAINER {
     OUTPUT_UNIT *associated_unit;
 } CONTAINER;
 
+/* indices in ELEMENT elt_info */
+enum elt_info_type {
+  eit_spaces_after_cmd_before_arg,
+  eit_comment_at_end = 0,
+  eit_types_spaces_before_argument = 0,
+  eit_spaces_before_argument,
+  eit_types_spaces_after_argument = 1,
+  eit_spaces_after_argument,
+};
+
+/* indices in ELEMENT string_info */
+enum string_info_type {
+   sit_alias_of,
+   sit_arg_line,
+   sit_delimiter = 1,
+};
+
 typedef struct ELEMENT {
     /* Used when building Perl tree only. This should be HV *hv,
        but we don't want to include the Perl headers everywhere; */
@@ -227,6 +244,8 @@ typedef struct ELEMENT {
     struct ELEMENT *parent;
     /* depends on the element, can be space elements, comments */
     struct ELEMENT **elt_info;
+    /* depends on the element */
+    char **string_info;
     SOURCE_MARK_LIST source_mark_list;
 
     ASSOCIATED_INFO extra_info;
diff --git a/tp/Texinfo/XS/parsetexi/parser.c b/tp/Texinfo/XS/parsetexi/parser.c
index 3cd5d4d483..88b8b7a75f 100644
--- a/tp/Texinfo/XS/parsetexi/parser.c
+++ b/tp/Texinfo/XS/parsetexi/parser.c
@@ -972,8 +972,8 @@ isolate_last_space (ELEMENT *current)
       && (last_contents_child (current)->cmd == CM_c
           || last_contents_child (current)->cmd == CM_comment))
     {
-      add_info_element_oot (current, "comment_at_end",
-                            pop_element_from_contents (current));
+      current->elt_info[eit_comment_at_end]
+         = pop_element_from_contents (current);
     }
 
   if (current->c->contents.number == 0)
diff --git a/tp/Texinfo/XS/structuring_transfo/transformations.c 
b/tp/Texinfo/XS/structuring_transfo/transformations.c
index dd2f53899d..febec05ec6 100644
--- a/tp/Texinfo/XS/structuring_transfo/transformations.c
+++ b/tp/Texinfo/XS/structuring_transfo/transformations.c
@@ -639,7 +639,8 @@ new_node (ERROR_MESSAGE_LIST *error_messages, ELEMENT 
*node_tree,
                             spaces_after);
 
       if (comment_at_end)
-        add_info_element_oot (node_line_arg, "comment_at_end", comment_at_end);
+        node_line_arg->elt_info[eit_comment_at_end] = comment_at_end;
+
       insert_slice_into_contents (node_line_arg, 0, node_tree, 0,
                                   node_tree->c->contents.number);