branch master updated: * tp/Texinfo/ParserNonXS.pm (_merge_text), tp/Texinfo/XS/parsetexi/parser.c (merge_text): handle especially the case of no last element, as there cannot be paragraph or special text elements handling in that case.

[Patrice Dumas]

This is an automated email from the git hooks/post-receive script.

pertusus pushed a commit to branch master
in repository texinfo.

The following commit(s) were added to refs/heads/master by this push:
     new 650de228a0 * tp/Texinfo/ParserNonXS.pm (_merge_text), 
tp/Texinfo/XS/parsetexi/parser.c (merge_text): handle especially the case of no 
last element, as there cannot be paragraph or special text elements handling in 
that case.
650de228a0 is described below

commit 650de228a0d2335f58becab413218526c6dfc9a1
Author: Patrice Dumas <pertusus@free.fr>
AuthorDate: Sun Jun 23 17:58:14 2024 +0200

    * tp/Texinfo/ParserNonXS.pm (_merge_text),
    tp/Texinfo/XS/parsetexi/parser.c (merge_text): handle especially the
    case of no last element, as there cannot be paragraph or special text
    elements handling in that case.
    
    * tp/Texinfo/ParserNonXS.pm (_close_current)
    (_move_last_space_to_element, _abort_empty_line, _end_line),
    tp/Texinfo/XS/parsetexi/close.c (close_current),
    tp/Texinfo/XS/parsetexi/end_line.c (end_line),
    tp/Texinfo/XS/parsetexi/parser.c (move_last_space_to_element)
    (do_abort_empty_line): separate code moving a last space children as
    info spaces_before_argument in element out of abort_empty_line to
    move_last_space_to_element, and call directly
    move_last_space_to_element when relevant.
---
 ChangeLog                          | 17 +++++++++
 tp/Texinfo/ParserNonXS.pm          | 53 ++++++++++++++++++---------
 tp/Texinfo/XS/parsetexi/close.c    |  2 +-
 tp/Texinfo/XS/parsetexi/end_line.c |  2 +-
 tp/Texinfo/XS/parsetexi/parser.c   | 73 ++++++++++++++++++++++----------------
 tp/Texinfo/XS/parsetexi/parser.h   |  1 +
 6 files changed, 98 insertions(+), 50 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 6b56fa0ba7..7a67934fe4 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,20 @@
+2024-06-23  Patrice Dumas  <pertusus@free.fr>
+
+       * tp/Texinfo/ParserNonXS.pm (_merge_text),
+       tp/Texinfo/XS/parsetexi/parser.c (merge_text): handle especially the
+       case of no last element, as there cannot be paragraph or special text
+       elements handling in that case.
+
+       * tp/Texinfo/ParserNonXS.pm (_close_current)
+       (_move_last_space_to_element, _abort_empty_line, _end_line),
+       tp/Texinfo/XS/parsetexi/close.c (close_current),
+       tp/Texinfo/XS/parsetexi/end_line.c (end_line),
+       tp/Texinfo/XS/parsetexi/parser.c (move_last_space_to_element)
+       (do_abort_empty_line): separate code moving a last space children as
+       info spaces_before_argument in element out of abort_empty_line to
+       move_last_space_to_element, and call directly
+       move_last_space_to_element when relevant.
+
 2024-06-23  Patrice Dumas  <pertusus@free.fr>
 
        * tp/Texinfo/ParserNonXS.pm (_abort_empty_line): define more variables
diff --git a/tp/Texinfo/ParserNonXS.pm b/tp/Texinfo/ParserNonXS.pm
index 8616d05b26..f758a42e81 100644
--- a/tp/Texinfo/ParserNonXS.pm
+++ b/tp/Texinfo/ParserNonXS.pm
@@ -2126,7 +2126,7 @@ sub _close_current($$$;$$)
           and $current->{'contents'}->[0]->{'type'}
                         eq 'internal_spaces_before_argument') {
         # remove spaces element from tree and update extra values
-        _abort_empty_line($self, $current);
+        _move_last_space_to_element($self, $current);
       }
       $current = $current->{'parent'};
     } elsif ($current->{'type'} eq 'balanced_braces') {
@@ -2226,12 +2226,25 @@ sub _close_commands($$$;$$)
 sub _merge_text {
   my ($self, $current, $text, $transfer_marks_element) = @_;
 
-  my $last_element;
-
-  if ($current->{'contents'}) {
-    $last_element = $current->{'contents'}->[-1];
+  # paragraphs are only started in empty lines or in context brace
+  # commands, if there is nothing in the current element, cannot
+  # be in a case where a paragraph is started.
+  # Also, elements without anything in them are only brace_container
+  # or menu_entry_name, otherwise there is always some kind of element
+  # leading added for leading spaces when the element is created
+  if (!$current->{'contents'}) {
+    my $new_element = { 'text' => $text, 'parent' => $current };
+    _transfer_source_marks($transfer_marks_element, $new_element)
+      if ($transfer_marks_element);
+    $current->{'contents'} = [];
+    push @{$current->{'contents'}}, $new_element;
+    print STDERR "NEW TEXT (merge): $text|||\n"
+                         if ($self->{'conf'}->{'DEBUG'});
+    return $current;
   }
 
+  my $last_element = $current->{'contents'}->[-1];
+
   my $paragraph;
 
   my $no_merge_with_following_text = 0;
@@ -2957,10 +2970,25 @@ sub _pop_element_from_contents($$)
   return $popped_element;
 }
 
+sub _move_last_space_to_element($$) {
+  my ($self, $current) = @_;
+
+  # Remove element from main tree. It will still be referenced in
+  # the 'info' hash as 'spaces_before_argument'.
+  my $spaces_before_argument = _pop_element_from_contents($self, $current);
+  delete $spaces_before_argument->{'type'};
+  delete $spaces_before_argument->{'parent'};
+  my $owning_element = $self->{'internal_space_holder'};
+  $owning_element->{'info'} = {} if (!exists($owning_element->{'info'}));
+  $owning_element->{'info'}->{'spaces_before_argument'}
+    = $spaces_before_argument;
+  $self->{'internal_space_holder'} = undef;
+}
+
 # each time a new line appeared, a container is opened to hold the text
 # consisting only of spaces.  This container is removed here, typically
 # this is called when non-space happens on a line.
-sub _abort_empty_line {
+sub _abort_empty_line($$) {
   my ($self, $current) = @_;
 
   if ($current->{'contents'}) {
@@ -3011,16 +3039,7 @@ sub _abort_empty_line {
         } elsif ($type eq 'internal_spaces_after_command'
                  or $type eq 'internal_spaces_before_argument'
                  or $type eq 'internal_spaces_before_context_argument') {
-          # Remove element from main tree. It will still be referenced in
-          # the 'info' hash as 'spaces_before_argument'.
-          my $spaces_before_argument = _pop_element_from_contents($self, 
$current);
-          delete $spaces_before_argument->{'type'};
-          delete $spaces_before_argument->{'parent'};
-          my $owning_element = $self->{'internal_space_holder'};
-          $owning_element->{'info'} = {} if 
(!exists($owning_element->{'info'}));
-          $owning_element->{'info'}->{'spaces_before_argument'}
-            = $spaces_before_argument;
-          $self->{'internal_space_holder'} = undef;
+          _move_last_space_to_element($self, $current);
         }
         return 1;
       }
@@ -4637,7 +4656,7 @@ sub _end_line($$$)
                           eq 'internal_spaces_before_context_argument')) {
     # Empty spaces after brace or comma till the end of line.
     # Remove this element and update 'extra' values.
-    _abort_empty_line($self, $current);
+    _move_last_space_to_element($self, $current);
   }
 
   # this happens if there is a nesting of line @-commands on a line.
diff --git a/tp/Texinfo/XS/parsetexi/close.c b/tp/Texinfo/XS/parsetexi/close.c
index d666b521c8..ba33f99bd9 100644
--- a/tp/Texinfo/XS/parsetexi/close.c
+++ b/tp/Texinfo/XS/parsetexi/close.c
@@ -443,7 +443,7 @@ close_current (ELEMENT *current,
                  == ET_internal_spaces_before_argument)
             {
               /* remove spaces element from tree and update extra values */
-              abort_empty_line (current);
+              move_last_space_to_element (current);
             }
           current = current->parent;
           break;
diff --git a/tp/Texinfo/XS/parsetexi/end_line.c 
b/tp/Texinfo/XS/parsetexi/end_line.c
index ff12fb3051..162dfe02dd 100644
--- a/tp/Texinfo/XS/parsetexi/end_line.c
+++ b/tp/Texinfo/XS/parsetexi/end_line.c
@@ -1936,7 +1936,7 @@ end_line (ELEMENT *current)
     {
       /* Empty spaces after brace or comma till the end of line.
          Remove this element and update 'extra' values. */
-      abort_empty_line (current);
+      move_last_space_to_element (current);
     }
 
   /* 'line' or 'def' at top of "context stack" - this happens when
diff --git a/tp/Texinfo/XS/parsetexi/parser.c b/tp/Texinfo/XS/parsetexi/parser.c
index 6924e402cb..4256e271e6 100644
--- a/tp/Texinfo/XS/parsetexi/parser.c
+++ b/tp/Texinfo/XS/parsetexi/parser.c
@@ -703,6 +703,23 @@ end_preformatted (ELEMENT *current,
  */
 ELEMENT *internal_space_holder;
 
+void
+move_last_space_to_element (ELEMENT *current)
+{
+  /* Remove element from main tree. It will still be referenced in
+     the 'info' hash as 'spaces_before_argument'. */
+  ELEMENT *owning_element;
+  ELEMENT *e = pop_element_from_contents (current);
+  owning_element = internal_space_holder;
+  e->type = ET_other_text;
+  e->parent = 0;
+  if (owning_element->type != ET_context_brace_command)
+    owning_element->elt_info[eit_spaces_before_argument] = e;
+  else
+    owning_element->elt_info[eit_brace_content_spaces_before_argument] = e;
+  internal_space_holder = 0;
+}
+
 static void
 do_abort_empty_line (ELEMENT *current, ELEMENT *last_elt)
 {
@@ -740,19 +757,7 @@ do_abort_empty_line (ELEMENT *current, ELEMENT *last_elt)
            || last_elt->type == ET_internal_spaces_before_argument
            || last_elt->type == ET_internal_spaces_before_context_argument)
     {
-      /* Remove element from main tree. It will still be referenced in
-         the 'info' hash as 'spaces_before_argument'. */
-      ELEMENT *owning_element;
-      ELEMENT *e = pop_element_from_contents (current);
-      owning_element = internal_space_holder;
-      e->type = ET_other_text;
-      e->parent = 0;
-      if (owning_element->type != ET_context_brace_command)
-        owning_element->elt_info[eit_spaces_before_argument] = e;
-      else
-        owning_element->elt_info[eit_brace_content_spaces_before_argument]
-           = e;
-      internal_space_holder = 0;
+      move_last_space_to_element (current);
     }
 }
 
@@ -770,6 +775,17 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
   ELEMENT *e;
   ELEMENT *last_element = last_contents_child (current);
 
+  /* paragraphs are only started in empty lines or in context brace
+     commands, if there is nothing in the current element, cannot
+     be in a case where a paragraph is started.
+     Also, elements without anything in them are only brace_container
+     or menu_entry_name, otherwise there is always some kind of element
+     leading added for leading spaces when the element is created */
+  if (!last_element)
+    goto new_text;
+
+  enum element_type last_elt_type = last_element->type;
+
   /* determine the number of leading characters in whitespace_chars */
   for (; leading_spaces < len_text
          && strchr (whitespace_chars, text[leading_spaces]);
@@ -779,17 +795,13 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
   if (leading_spaces < len_text)
     {
       ELEMENT *paragraph;
-      if (last_element
-          && (last_element->type == ET_empty_line
-              || last_element->type == ET_ignorable_spaces_after_command
-              || last_element->type == ET_internal_spaces_after_command
-              || last_element->type == ET_internal_spaces_before_argument
-              || last_element->type
-                           == ET_internal_spaces_before_context_argument
-              || last_element->type == ET_spaces_after_close_brace))
-        {
-          int no_merge_with_following_text
-               = (last_element->type != ET_empty_line);
+      if ((last_elt_type == ET_empty_line
+           || last_elt_type == ET_ignorable_spaces_after_command
+           || last_elt_type == ET_internal_spaces_after_command
+           || last_elt_type == ET_internal_spaces_before_argument
+           || last_elt_type == ET_internal_spaces_before_context_argument
+           || last_elt_type == ET_spaces_after_close_brace))
+        {
           if (leading_spaces)
             {
               if (global_parser_conf.debug)
@@ -797,7 +809,7 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
                   char *additional_text_dbg = strndup (text, leading_spaces);
                   debug ("MERGE_TEXT ADD leading empty |%s| to %s",
                          additional_text_dbg,
-                         type_data[last_element->type].name);
+                         type_data[last_elt_type].name);
                   free (additional_text_dbg);
                 }
               text_append_n (last_element->e.text, text, leading_spaces);
@@ -812,6 +824,7 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
                */
               e = pop_element_from_contents (current);
               e->type = ET_normal_text;
+
               paragraph = begin_paragraph (current);
               if (paragraph)
                 {
@@ -823,10 +836,10 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
           /* since last_element cannot be empty as this case is
              handled just above, the last_element is
              always kept in current in do_abort_empty_line
-             for an empty_line; its type may have changed */
+             for an empty_line; its type may change */
           do_abort_empty_line (current, last_element);
 
-          if (no_merge_with_following_text)
+          if (last_elt_type != ET_empty_line)
          /* we do not merge these special types, unset last_element */
             last_element = 0;
         }
@@ -838,9 +851,7 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
              space as handled just above, or after a no_paragraph
              command outside of a paragraph or after a non expanded @value
              outside of a paragraph (here), because
-              - there is always an element where a paragraph can begin, since
-                elements without anything in them are only brace_container
-                or menu/preformatted contexts
+              - there is always an element where a paragraph can begin
               - if the element is not a special space nor a no_paragraph
                 command, we are already in a paragraph if a paragraph can
                 be opened.
@@ -854,7 +865,7 @@ merge_text (ELEMENT *current, const char *text, size_t 
len_text,
 
   if (last_element
       /* can actually be normal_text, and some space elements */
-      && type_data[last_element->type].flags & TF_text
+      && type_data[last_elt_type].flags & TF_text
       && !strchr (last_element->e.text->text, '\n'))
     {
       /* Transfer source marks */
diff --git a/tp/Texinfo/XS/parsetexi/parser.h b/tp/Texinfo/XS/parsetexi/parser.h
index 7eb48e7400..c9145841f1 100644
--- a/tp/Texinfo/XS/parsetexi/parser.h
+++ b/tp/Texinfo/XS/parsetexi/parser.h
@@ -72,6 +72,7 @@ CONDITIONAL_STACK_ITEM *pop_conditional_stack (void);
 CONDITIONAL_STACK_ITEM *top_conditional_stack (void);
 extern size_t conditional_number;
 
+void move_last_space_to_element (ELEMENT *current);
 int abort_empty_line (ELEMENT *current);
 ELEMENT *end_paragraph (ELEMENT *current,
                         enum command_id closed_block_command,