My own corner of monopam
linkedin: match LinkedIn's current public-Pulse DOM
Article body now lives under [data-test-id="publishing-text-block"]
inside <article class="article-main">, and engagement counts are
exposed as data-num-reactions / data-num-comments attributes on
<div data-test-id="social-actions__*"> elements. The old class-based
selectors no longer match, so the body extractor 404'd on real
articles.
Reorder body selectors to try the current LinkedIn shape first, fall
back to historical class names. Swap the engagement extractor to read
the attribute values directly (taking the max across the page since
each article has several social-actions widgets). Update the HTML
fixture to use the modern shape so tests exercise the primary chain.
Also reuse Vlog's --json flag instead of defining a second one;
cmdliner was rejecting the duplicate registration at startup with
"option name --json defined twice".
+51
-30
+2
-6
ocaml-linkedin/bin/main.ml
+2
-6
ocaml-linkedin/bin/main.ml
···
305
305
print_string s.url;
306
306
print_char '\n'
307
307
308
-
let json_flag =
309
-
let doc = "Output as JSON." in
310
-
Arg.(value & flag & info [ "json" ] ~doc)
311
-
312
308
let print_item_json i =
313
309
print_string (Json.to_string Linkedin.Item.json i);
314
310
print_char '\n'
···
347
343
let info = Cmd.info "get" ~doc ~man in
348
344
Cmd.v info
349
345
Term.(
350
-
const run' $ setup $ li_at_t $ jsessionid_t $ url_t $ json_flag
346
+
const run' $ setup $ li_at_t $ jsessionid_t $ url_t $ Vlog.json
351
347
$ const ())
352
348
353
349
let feed_cmd =
···
389
385
Cmd.v info
390
386
Term.(
391
387
const run' $ setup $ li_at_t $ jsessionid_t $ id_or_url_t $ count_t
392
-
$ json_flag $ const ())
388
+
$ Vlog.json $ const ())
393
389
394
390
let cookies_cmd =
395
391
let run' () () =
+44
-19
ocaml-linkedin/lib/pulse.ml
+44
-19
ocaml-linkedin/lib/pulse.ml
···
263
263
let parts = Soup.trimmed_texts el in
264
264
String.concat " " parts
265
265
266
+
let max_count_from_attr soup attr =
267
+
let nodes = Soup.($$) soup (Fmt.str "[%s]" attr) |> Soup.to_list in
268
+
List.fold_left
269
+
(fun acc el ->
270
+
match Soup.attribute attr el with
271
+
| Some s -> max acc (first_int s)
272
+
| None -> acc)
273
+
0 nodes
274
+
266
275
let extract_count_from_selectors soup selectors =
267
276
List.find_map
268
277
(fun sel ->
···
280
289
selectors
281
290
|> Option.value ~default:0
282
291
292
+
(** The article body has several [data-num-reactions] attributes (one per
293
+
social-actions widget, at article top and bottom, plus nested ones inside
294
+
comments). The article's own count is the max across the page. *)
283
295
let extract_num_likes soup =
284
-
extract_count_from_selectors soup
285
-
[
286
-
"[data-test-id=\"social-counts-reactions\"]";
287
-
".social-details-social-counts__reactions-count";
288
-
".social-counts-reactions";
289
-
"[aria-label*=\"reactions\"]";
290
-
"[aria-label*=\"Reactions\"]";
291
-
]
296
+
let from_attr = max_count_from_attr soup "data-num-reactions" in
297
+
if from_attr > 0 then from_attr
298
+
else
299
+
extract_count_from_selectors soup
300
+
[
301
+
"[data-test-id=\"social-actions__reaction-count\"]";
302
+
"[data-test-id=\"social-actions__reactions\"]";
303
+
".social-details-social-counts__reactions-count";
304
+
"[aria-label*=\"reactions\"]";
305
+
"[aria-label*=\"Reactions\"]";
306
+
]
292
307
293
308
let extract_num_comments soup =
294
-
extract_count_from_selectors soup
295
-
[
296
-
"[data-test-id=\"social-counts-comments\"]";
297
-
".social-details-social-counts__comments";
298
-
".social-counts-comments";
299
-
"[aria-label*=\"comments\"]";
300
-
"[aria-label*=\"Comments\"]";
301
-
]
309
+
let from_attr = max_count_from_attr soup "data-num-comments" in
310
+
if from_attr > 0 then from_attr
311
+
else
312
+
extract_count_from_selectors soup
313
+
[
314
+
"[data-test-id=\"social-actions__comments\"]";
315
+
".social-details-social-counts__comments";
316
+
"[aria-label*=\"comments\"]";
317
+
"[aria-label*=\"Comments\"]";
318
+
]
302
319
303
-
(** Find the article body element. Tries a ranked list of selectors; returns the
304
-
first match. *)
320
+
(** Find the article body element. Tries a ranked list of selectors, starting
321
+
with LinkedIn's current public-Pulse markup and falling back to historical
322
+
shapes. *)
305
323
let extract_body soup =
306
324
let selectors =
307
325
[
326
+
(* Current public Pulse (2026): <article class="article-main"><div
327
+
class="article-main__content" data-test-id="publishing-text-block">
328
+
...body... </div></article> *)
329
+
"[data-test-id=\"publishing-text-block\"]";
330
+
"article.article-main .article-main__content";
331
+
".article-main__content";
332
+
"article.article-main";
333
+
(* Historical / in-app shapes *)
308
334
"article .reader-article-content";
309
335
"div.reader-article-content";
310
336
"article .article__body";
311
337
"div.article-body";
312
338
"main article";
313
339
"article";
314
-
"div[data-testid=\"article-body\"]";
315
340
]
316
341
in
317
342
List.find_map (fun sel -> Soup.( $? ) soup sel) selectors
+5
-5
ocaml-linkedin/test/fixtures/pulse_article.html
+5
-5
ocaml-linkedin/test/fixtures/pulse_article.html
···
13
13
<header>
14
14
<a href="https://www.linkedin.com/in/parsimoni" class="author-link">Parsimoni</a>
15
15
</header>
16
-
<article>
17
-
<div class="article__body">
16
+
<article class="article-main relative flex-grow pulse">
17
+
<div class="article-main__content" data-test-id="publishing-text-block">
18
18
<p>Satellite operators are leaving significant <strong>economic value</strong>
19
19
in orbit. Three patterns recur:</p>
20
20
<ul>
···
42
42
<p>See <code>orbit_value</code> above.</p>
43
43
</div>
44
44
</article>
45
-
<div class="social-details-social-counts">
46
-
<span class="social-details-social-counts__reactions-count" aria-label="142 reactions">142</span>
47
-
<span class="social-details-social-counts__comments">37 comments</span>
45
+
<div class="social-actions">
46
+
<div data-test-id="social-actions__reactions" data-num-reactions="142" aria-label="142 Reactions">Reactions</div>
47
+
<div data-test-id="social-actions__comments" data-num-comments="37" aria-label="37 Comments">Comments</div>
48
48
</div>
49
49
</body>
50
50
</html>