branch: elpa/logview
commit fb7762b0a9cc52f786f15230c171e1b88e1e1a03
Author: Paul Pogonyshev <pogonys...@gmail.com>
Commit: Paul Pogonyshev <pogonys...@gmail.com>
Add support for "really special" log entry parts, where format string
specifies full regexp for a subpart (only name, thread or "ignored"); issue #48.
---
logview.el | 102 +++++++++++++++++++++++++++++-------------------------
test/custom/2.log | 3 ++
test/logview.el | 10 ++++++
3 files changed, 67 insertions(+), 48 deletions(-)
diff --git a/logview.el b/logview.el
index 3bdd33323c..b036d95d42 100644
--- a/logview.el
+++ b/logview.el
@@ -640,13 +640,11 @@ settings) with this face.")
(defconst logview--final-levels '(error warning information debug trace)
"List of final (submode-independent) levels, most to least severe.")
-(defconst logview--entry-part-regexp (rx bow (or (group "TIMESTAMP")
- (group "LEVEL")
- (group "NAME")
- (group "THREAD")
- (group "IGNORED")
- (group "MESSAGE"))
- eow))
+(defconst logview--entry-part-regexp (rx (or (seq bow (or (group "TIMESTAMP")
(group "LEVEL") (group "NAME")
+ (group "THREAD")
(group "IGNORED") (group "MESSAGE")) ;; 1--6, see above
+ eow)
+ (seq "<<RX:" (or (group "NAME")
(group "THREAD") (group "IGNORED")) ;; 7--9
+ ":" (group (+? anychar))
">>")))) ;; 10
(defconst logview--timestamp-entry-part-regexp (rx bow "TIMESTAMP" eow))
(defvar logview--datetime-matching-options '(:second-fractional-extension t
@@ -2931,48 +2929,56 @@ returns non-nil."
(setq have-explicit-message t))
(t
(dolist (k (list logview--name-group logview--thread-group
logview--ignored-group))
- (when (match-beginning k)
- (push (format "\\(?%s:%s\\)"
- (if (/= k logview--ignored-group)
- (number-to-string k)
- "")
- (cond ((and starter terminator
- (or (and (= starter ?\() (=
terminator ?\)))
- (and (= starter ?\[) (=
terminator ?\]))))
- ;; See
https://github.com/doublep/logview/issues/2
- ;; We allow _one_ level of nested parens
inside
- ;; parenthesized THREAD or NAME.
Allowing more would
- ;; complicate regexp even further.
Unlimited nesting
- ;; level is not possible with regexps at
all.
- ;;
- ;; 'rx-to-string' is used to avoid
escaping things
- ;; ourselves.
- (rx-to-string `(seq (* (not (any
,starter ,terminator ?\n)))
- (* ,starter (* (not
(any ?\n))) ,terminator
- (* (not (any
,starter ,terminator ?\n)))))
- t))
- ((and terminator (/= terminator ? ))
- (format "[^%c\n]*" terminator))
- (terminator
- "[^ \t\n]+")
- (t
- ".+")))
- parts)
- (push (if (= k logview--name-group) 'name 'thread)
features)))))
+ ;; See definition of `logview--entry-part-regexp' for the
meaning of 4 and 10.
+ (let ((special-regexp (match-beginning (+ k 4))))
+ (when (or (match-beginning k) special-regexp)
+ (push (format "\\(?%s:%s\\)"
+ (if (/= k logview--ignored-group)
+ (number-to-string k)
+ "")
+ (cond (special-regexp
+ (let ((forced-regexp (match-string 10
format)))
+ (unless (logview--valid-regexp-p
forced-regexp)
+ ;; Ideally would also ensure that
there are no catching groups,
+ ;; but for this we'd need `xr' as
dependency. Not now.
+ (warn "In format specifier `%s':
`%s' is not a valid regexp" format forced-regexp)
+ (setf cannot-match t))
+ forced-regexp))
+ ((and starter terminator
+ (or (and (= starter ?\() (=
terminator ?\)))
+ (and (= starter ?\[) (=
terminator ?\]))))
+ ;; See
https://github.com/doublep/logview/issues/2 We allow _one_
+ ;; level of nested parens inside
parenthesized THREAD or NAME.
+ ;; Allowing more would complicate
regexp even further. Unlimited
+ ;; nesting level is not possible with
regexps at all.
+ ;;
+ ;; 'rx-to-string' is used to avoid
escaping things ourselves.
+ (rx-to-string `(seq (* (not (any
,starter ,terminator ?\n)))
+ (* ,starter (*
(not (any ?\n))) ,terminator
+ (* (not (any
,starter ,terminator ?\n)))))
+ t))
+ ((and terminator (/= terminator ? ))
+ (format "[^%c\n]*" terminator))
+ (terminator
+ "[^ \t\n]+")
+ (t
+ ".+")))
+ parts)
+ (push (if (= k logview--name-group) 'name 'thread)
features))))))
(setq search-from end))
- (when (< search-from (length format))
- (funcall add-text-part search-from nil))
- ;; Unless `MESSAGE' field is used explicitly, behave as if format string
ends with whitespace.
- (unless (or have-explicit-message (string-match-p "[ \t]$" format))
- (push "\\(?:[ \t]+\\|$\\)" parts))
- (setq parts (nreverse parts))
- (when timestamp-at
- ;; Speed optimization: if the submode includes a timestamp, but
- ;; the test line doesn't have even two digits at the expected
- ;; place, don't even loop through all the timestamp options.
- (setcar timestamp-at ".*[0-9][0-9].*")
- (when (and test-line (not (string-match-p (apply #'concat parts)
test-line)))
- (setq cannot-match t)))
+ (unless cannot-match
+ (when (< search-from (length format))
+ (funcall add-text-part search-from nil))
+ ;; Unless `MESSAGE' field is used explicitly, behave as if format string
ends with whitespace.
+ (unless (or have-explicit-message (string-match-p "[ \t]$" format))
+ (push "\\(?:[ \t]+\\|$\\)" parts))
+ (setq parts (nreverse parts))
+ (when timestamp-at
+ ;; Speed optimization: if the submode includes a timestamp, but the
test line doesn't have even two
+ ;; digits at the expected place, don't even loop through all the
timestamp options.
+ (setcar timestamp-at ".*[0-9][0-9].*")
+ (when (and test-line (not (string-match-p (apply #'concat parts)
test-line)))
+ (setq cannot-match t))))
(unless cannot-match
(dolist (timestamp-option (if timestamp-at timestamp-options '(nil)))
(let* ((timestamp-pattern (assq 'java-pattern timestamp-option))
diff --git a/test/custom/2.log b/test/custom/2.log
new file mode 100644
index 0000000000..3f9c091450
--- /dev/null
+++ b/test/custom/2.log
@@ -0,0 +1,3 @@
+2023-07-19 14:10:02.736 GMT+08:00 INFO T: Fake Thread WhateverName - see
https://github.com/doublep/logview/issues/48
+2023-07-19 14:10:02.736 GMT+08:00 DEBUG T: Subscription Manager Consumer
Thread LottieLockView - loading from filesystem
+2023-07-19 14:10:02.739 GMT+08:00 DEBUG T: pool-40-thread-1 LottieLockView -
starting to play t4.json
diff --git a/test/logview.el b/test/logview.el
index 9e4549a4fc..7bea5df334 100644
--- a/test/logview.el
+++ b/test/logview.el
@@ -148,6 +148,16 @@
(logview-difference-to-current-entry)
(logview-go-to-difference-base-entry)))
+;; See https://github.com/doublep/logview/issues/48 for rationale to have this
at all.
+(ert-deftest logview-test-custom-submode-with-special-regexp ()
+ (logview--test-with-file "custom/2.log"
+ :extra-customizations '((logview-additional-submodes
+ '(("custom" (format . "TIMESTAMP IGNORED LEVEL T:
<<RX:THREAD:[^-]+>> NAME - MESSAGE") (levels . "SLF4J")))))
+ (should (equal logview--submode-name "custom"))
+ (logview--locate-current-entry entry start
+ (should (and entry (equal start 1)))
+ (should (equal (logview--entry-group entry start logview--name-group)
"WhateverName")))))
+
;; Bug: Logview would ignore entry lines if they didn't contain a space at the
end. This
;; would e.g. happen if you had code like 'log.info ("\n...");' in your
program.
(ert-deftest logview-test-multiline-entries ()
