(feat)db: add customizability of extra element link parsing (#2042)

Adds two customizable variables: org-roam-db-extra-links-elements and
org-roam-db-extra-links-exclude-keys, which govern which elements are to
be considered for link parsing by Org-roam.

Also added sane defaults to org-roam-db-extra-links-exclude-keys.

CHANGELOG.md

@@ -1,6 +1,7 @@
 # Changelog
 ## TBD
 ### Added
+- [#2042](https://github.com/org-roam/org-roam/pull/2042) Add `org-roam-db-extra-links-elements` and `org-roam-db-extra-links-exclude-keys` for fine-grained control over additional link parsing.
 ### Removed
 ### Fixed
 ### Changed

doc/org-roam.org

@@ -490,6 +490,35 @@ all headlines with the ~ATTACH~ tag from the Org-roam database, one can set:
         (not (member "ATTACH" (org-get-tags)))))
 #+end_src
 
+Org-roam relied on the obtained Org AST for the buffer to parse links. However,
+links appearing in some places (e.g. within property drawers) are not considered
+by the Org AST to be links. Therefore, Org-roam takes special care of
+additionally trying to process these links. Use
+~org-roam-db-extra-links-elements~ to specify which additional Org AST element
+types to consider.
+
+- Variable: org-roam-db-extra-links-elements
+
+  The list of Org element types to include for parsing by Org-roam.
+
+  By default, when parsing Org's AST, links within keywords and
+  property drawers are not parsed as links. Sometimes however, it
+  is desirable to parse and cache these links (e.g. hiding links in
+  a property drawer).
+
+Additionally, one may want to ignore certain keys from being excluded within
+property drawers. For example, we would not want ~ROAM_REFS~ links to be
+self-referential. Hence, to exclude specific keys, we use
+~org-roam-db-extra-links-exclude-keys~.
+
+- Variable: org-roam-db-extra-links-exclude-keys
+
+  Keys to ignore when mapping over links.
+
+  The car of the association list is the Org element type (e.g. keyword). The
+  cdr is a list of case-insensitive strings to exclude from being treated as
+  links.
+
 ** When to cache
 
 By default, Org-roam is eager in caching: each time an Org-roam file is modified

doc/org-roam.texi

@@ -791,6 +791,37 @@ all headlines with the @code{ATTACH} tag from the Org-roam database, one can set
         (not (member "ATTACH" (org-get-tags)))))
 @end example
 
+Org-roam relied on the obtained Org AST for the buffer to parse links. However,
+links appearing in some places (e.g. within property drawers) are not considered
+by the Org AST to be links. Therefore, Org-roam takes special care of
+additionally trying to process these links. Use
+@code{org-roam-db-extra-links-elements} to specify which additional Org AST element
+types to consider.
+
+@defvar org-roam-db-extra-links-elements
+
+The list of Org element types to include for parsing by Org-roam.
+
+By default, when parsing Org's AST, links within keywords and
+property drawers are not parsed as links. Sometimes however, it
+is desirable to parse and cache these links (e.g. hiding links in
+a property drawer).
+@end defvar
+
+Additionally, one may want to ignore certain keys from being excluded within
+property drawers. For example, we would not want @code{ROAM_REFS} links to be
+self-referential. Hence, to exclude specific keys, we use
+@code{org-roam-db-extra-links-exclude-keys}.
+
+@defvar org-roam-db-extra-links-exclude-keys
+
+Keys to ignore when mapping over links.
+
+The car of the association list is the Org element type (e.g. keyword). The
+cdr is a list of case-insensitive strings to exclude from being treated as
+links.
+@end defvar
+
 @node When to cache
 @section When to cache
 

org-roam-db.el

@@ -98,6 +98,32 @@ slow."
   :type 'boolean
   :group 'org-roam)
 
+(defcustom org-roam-db-extra-links-elements '(node-property keyword)
+  "The list of Org element types to include for parsing by Org-roam.
+
+By default, when parsing Org's AST, links within keywords and
+property drawers are not parsed as links. Sometimes however, it
+is desirable to parse and cache these links (e.g. hiding links in
+a property drawer)."
+  :package-version '(org-roam . "2.2.0")
+  :group 'org-roam
+  :type '(set (const :tag "keywords" keyword)
+              (const :tag "property drawers" node-property)))
+
+(defcustom org-roam-db-extra-links-exclude-keys '((node-property . ("ROAM_REFS"))
+                                                  (keyword . ("transclude")))
+  "Keys to ignore when mapping over links.
+
+The car of the association list is the Org element type (e.g.
+keyword). The cdr is a list of case-insensitive strings to
+exclude from being treated as links.
+
+For example, we use this to prevent self-referential links in
+ROAM_REFS."
+  :package-version '(org-roam . "2.2.0")
+  :group 'org-roam
+  :type '(alist))
+
 ;;; Variables
 (defconst org-roam-db-version 18)
 
@@ -353,15 +379,12 @@ If UPDATE-P is non-nil, first remove the file in the database."
          ;; Links correctly recognized by Org Mode
          ((eq type 'link)
           (setq link element))
-         ;; Prevent self-referencing links in ROAM_REFS
-         ((and (eq type 'node-property)
-               (org-roam-string-equal (org-element-property :key element) "ROAM_REFS"))
-          nil)
          ;; Links in property drawers and lines starting with #+. Recall that, as for Org Mode v9.4.4, the
          ;; org-element-type of links within properties drawers is "node-property" and for lines starting with
          ;; #+ is "keyword".
-         ((and (or (eq type 'node-property)
+         ((and (member type org-roam-db-extra-links-elements)
-                   (eq type 'keyword))
+               (not (member-ignore-case (org-element-property :key element)
+                                        (cdr (assoc type org-roam-db-extra-links-exclude-keys))))
                (setq bounds (org-in-regexp org-link-any-re))
                (setq link (buffer-substring-no-properties
                            (car bounds)