(internal): speed up extraction of global props, headlines, and title (#1061)

Avoids usage of org-element-parse-buffer, preferring simple regex searches.

CHANGELOG.md

@@ -10,6 +10,7 @@ In this release we support fuzzy links of the form `[[Title]]`, `[[*Headline]]`
 
 ### Features
 
+- [#1061](https://github.com/org-roam/org-roam/pull/1061) Speed up the extraction of file properties, headlines, and titles
 - [#1046](https://github.com/org-roam/org-roam/pull/1046) New user option to exclude files from Org-roam
 - [#1032](https://github.com/org-roam/org-roam/pull/1032) File changes are now propagated to the database on idle timer. This prevents large wait times during file saves.
 - [#974](https://github.com/org-roam/org-roam/pull/974) Protect region targeted by `org-roam-insert`

org-roam.el

@@ -495,6 +495,10 @@ Use external shell commands if defined in `org-roam-list-files-commands'."
 (defun org-roam--extract-global-props (props)
   "Extract PROPS from the current org buffer.
 The search terminates when the first property is encountered."
+  (if (functionp 'org-collect-keywords)
+      (->> (org-collect-keywords props)
+           ;; convert (("TITLE" "my title")) to (("TITLE" . "my title"))
+           (mapcar (pcase-lambda (`(,k ,v)) (cons k v))))
     (let ((buf (org-element-parse-buffer))
           res)
       (dolist (prop props)
@@ -504,7 +508,7 @@ The search terminates when the first property is encountered."
                        (org-element-property :value kw)))
                    :first-match t)))
           (push (cons prop p) res)))
-    res))
+      res)))
 
 (defun org-roam--expand-links (content path)
   "Crawl CONTENT for relative links and expand them.
@@ -631,15 +635,15 @@ it as FILE-PATH."
 If FILE-PATH is nil, use the current file."
   (let ((file-path (or file-path
                        (file-truename (buffer-file-name)))))
-    (org-element-map (org-element-parse-buffer) 'node-property
+    ;; Use `org-map-region' instead of `org-map-entries' as the latter
-      (lambda (node-property)
+    ;; would require another step to remove all nil values.
-        (let ((key (org-element-property :key node-property))
+    (let ((result nil))
-              (value (org-element-property :value node-property)))
+      (org-map-region
-          (when (string= key "ID")
+       (lambda ()
-            (let* ((id value)
+         (when-let ((id (org-entry-get nil "ID")))
-                   (data (vector id
+           (push (vector id file-path) result)))
-                                 file-path)))
+       (point-min) (point-max))
-              data)))))))
+      result)))
 
 (defun org-roam--extract-titles-title ()
   "Return title from \"#+title\" of the current buffer."
@@ -665,12 +669,15 @@ Reads from the \"roam_alias\" property."
 
 (defun org-roam--extract-titles-headline ()
   "Return the first headline of the current buffer."
-  (let ((headline (org-element-map
+  (let ((headline (save-excursion
-                      (org-element-parse-buffer)
+                    (goto-char (point-min))
-                      'headline
+                    ;; "What happens if a heading star was quoted
-                    (lambda (h)
+                    ;; before the first heading?"
-                      (org-no-properties (org-element-property :raw-value h)))
+                    ;; - `org-map-region' also does this
-                    :first-match t)))
+                    ;; - Org already breaks badly when you do that;
+                    ;; precede the heading star with a ",".
+                    (re-search-forward org-outline-regexp-bol nil t)
+                    (org-entry-get nil "ITEM"))))
     (when headline
       (list headline))))
 

tests/test-org-roam.el

@@ -255,7 +255,7 @@
     (it "extracts headlines"
       (expect (test #'org-roam--extract-headlines
                     "headlines/headline.org")
-              :to-equal
+              :to-have-same-items-as
               `(["e84d0630-efad-4017-9059-5ef917908823" ,(test-org-roam--abs-path "headlines/headline.org")]
                 ["801b58eb-97e2-435f-a33e-ff59a2f0c213" ,(test-org-roam--abs-path "headlines/headline.org")])))))