Re: bug#34023: Support double colons in Info index entries

[Juri Linkov]

Hi Gavin,

> In the Info format colons are special, and for this reason, there is 
> limited support for colons in index entries.  The Emacs Info mode 
> supports single colons in index entries as long as they are not followed 
> by a space.

Thanks for the detailed description.

> It doesn't state it, but when I tested it double colons don't work even 
> if they are not followed by a space.
>
> There is a fairly simple solution to this problem that I haven't seen 
> suggested in all the messages posted on this topic in the mailing list 
> archives. In index nodes only (which have a special marker included, 
> address@hidden@^H]), use a colon to terminate the text of the index entry, 
> but instead of looking for the first colon in the line, look for the 
> last.  So this entry:
>
> * a::b:  a colon b.  (line 129)
>
> would refer to line 129 of the node "a colon b".  This is possible 
> because node names cannot contain colons.  This restriction is not too 
> important, whereas the inability to index items containing colons is 
> quite important.  This is what is implemented in the standalone info 
> browser (since change on 2017-04-08).

The following patch handles the cases that you presented,
but it's hard to predict what other cases it might break.

Do you have a sample test file that covers different cases?
We could add such file to Emacs regression tests.

> This change shouldn't be made for all nodes, because the comment after 
> the closing '.' could contain a colon:
>
> * label: node.  comment: with a colon.
>
> This shouldn't be interpreted as refering to a node "with a colon".
>
> However, the "(line ...)" comment can't contain a colon.

The following change is made only for index nodes.

I have to say that the current regexp-based parsing is
an inherently fragile approach.  Do you think it would be possible
to add more markup to Info files instead of relying on regexps?

Like index nodes having a special marker address@hidden@^H]
maybe adding some markers to identify index entries,
node references, line numbers?

Better yet would be to read Info manual in HTML format in Info reader.
That would allow extracting all information unambiguously.

diff --git a/lisp/info.el b/lisp/info.el
index 6038273c37..2f7e293297 100644
--- a/lisp/info.el
+++ b/lisp/info.el
@@ -2664,9 +2664,15 @@ Info-menu-entry-name-re
 Because of ambiguities, this should be concatenated with something like
 `:' and `Info-following-node-name-re'.")
 
+(defconst Info-index-entry-name-re "\\(?:[^:]\\|:[^,.;() \t\n]\\)*"
+  "Regexp that matches an index entry name possibly including a colon.")
+
 (defun Info-extract-menu-node-name (&optional multi-line index-node)
   (skip-chars-forward " \t\n")
-  (when (looking-at (concat Info-menu-entry-name-re ":\\(:\\|"
+  (when (looking-at (concat (if index-node
+                                Info-index-entry-name-re
+                                Info-menu-entry-name-re
+                              ) ":\\(:\\|"
                            (Info-following-node-name-re
                              (cond
                               (index-node "^,\t\n")
@@ -2741,7 +2747,9 @@ Info-complete-menu-item
          (t
           (let ((pattern (concat "\n\\* +\\("
                                  (regexp-quote string)
-                                 Info-menu-entry-name-re "\\):"
+                                 (if (Info-index-node)
+                                     Info-index-entry-name-re
+                                   Info-menu-entry-name-re) "\\):"
                                  Info-node-spec-re))
                 completions
                 (complete-nodes Info-complete-nodes))
@@ -3966,7 +3974,8 @@ Info-try-follow-nearest-node
              (setq node t))
          (setq node nil))))
      ;; menu item: node name
-     ((setq node (Info-get-token (point) "\\* +" "\\* +\\([^:]*\\)::"))
+     ((setq node (unless (Info-index-node)
+                   (Info-get-token (point) "\\* +" "\\* +\\([^:]*\\)::")))
       (Info-goto-node node fork))
      ;; menu item: node name or index entry
      ((Info-get-token (point) "\\* +" "\\* +\\(.*\\): ")
@@ -4929,7 +4938,9 @@ Info-fontify-node
         (let ((n 0)
               cont)
           (while (re-search-forward
-                  (concat "^\\* Menu:\\|\\(?:^\\* +\\(" 
Info-menu-entry-name-re "\\)\\(:"
+                  (concat "^\\* Menu:\\|\\(?:^\\* +\\(" (if (Info-index-node)
+                                                            
Info-index-entry-name-re
+                                                          
Info-menu-entry-name-re) "\\)\\(:"
                           Info-node-spec-re "\\([ \t]*\\)\\)\\)")
                   nil t)
            (when (match-beginning 1)