enh: harvest URLs from Hypothes.is comments & higleted text

src/promnesia/sources/hypothesis.py

@@ -1,12 +1,23 @@
-'''
+"""
 Uses HPI [[https://github.com/karlicoss/HPI/blob/master/doc/MODULES.org#myhypothesis][hypothesis]] module
-'''
-from ..common import Visit, Results, logger, Loc
+"""
+from ..common import Loc, Results, Visit, extract_urls
+
+
+def _harvest_text(text, part_name, visit_base) -> Results:
+    if text and text.strip():
+        urls = extract_urls(text)
+        for url in urls:
+            yield visit_base._replace(
+                url=url,
+                locator=visit_base.locator._replace(title=f"hypothesis-{part_name}"),
+            )
 
 
 def index() -> Results:
     from . import hpi
     import my.hypothesis as hyp
+
     for h in hyp.get_highlights():
         if isinstance(h, Exception):
             yield h
@@ -18,15 +29,23 @@ def index() -> Results:
         if hl is not None:
             cparts.append(hl)
         if ann is not None:
-            cparts.extend(['comment: ' + ann])
+            cparts.append(f"comment: {ann}")
         if tags:
             cparts.append(" ".join(f"#{t}" for t in tags))
-        yield Visit(
+        visit = Visit(
             url=h.url,
             dt=h.created,
-            context='\n\n'.join(cparts),
+            context="\n\n".join(cparts),
             locator=Loc.make(
-                title='hypothesis',
+                title="hypothesis",
                 href=h.hyp_link,
-            )
+            ),
         )
+
+        yield visit
+
+        for text, part_name in (
+            (hl, "highlighted"),
+            (ann, "comment"),
+        ):
+            yield from _harvest_text(text, part_name, visit)