fix: show source_ref annotations on the cards

scripts/sitesources.py

@@ -16,9 +16,10 @@ from dataclasses import dataclass
 from pathlib import Path
 
 import upstream
-from siterender import _by_mode
+from siterender import _by_mode, _escape_tags
 from sourceref import (
-    _HEX_ADDRESS_RE, _prose_runs, is_binary_citation, ref_role, source_ref_values,
+    _HEX_ADDRESS_RE, _prose_runs, is_binary_citation, ref_annotation, ref_role,
+    source_ref_values,
     split_source_ref,
 )
 
@@ -512,7 +513,7 @@ def source_ref_items(profile: dict, value: str | dict | None) -> list[tuple[str,
         for part in split_source_ref(refs):
             if is_binary_citation(part.path):
                 text, binary = _binary_citation(part.path, binary)
-                rendered.append((text, ""))
+                rendered.append((text + _annotation_markdown(ref_annotation(part.raw)), ""))
                 continue
             binary = ""
             role, bare = ref_role(part.path)
@@ -544,19 +545,31 @@ def source_ref_items(profile: dict, value: str | dict | None) -> list[tuple[str,
 
             role = (role_label(profile, current_role, part.path)
                     if current_role and _two_repositories(profile) else "")
+            note = _annotation_markdown(ref_annotation(part.raw))
             if selected is None:
-                rendered.append((f"`{display}`", role))
+                rendered.append((f"`{display}`{note}", role))
             else:
                 repo, pin = selected
                 url = _source_permalink(
                     repo, pin, real_path, part.start, part.end
                 )
-                rendered.append((f"[`{display}`]({url})", role))
+                rendered.append((f"[`{display}`]({url}){note}", role))
 
         groups.append((label, rendered or [(f"`{refs}`", "")]))
     return groups
 
 
+def _annotation_markdown(text: str) -> str:
+    """A part's annotation after its citation, as written: `(lsf default)`.
+
+    Profile prose, so placeholders keep their brackets and nothing in it
+    opens emphasis or a table cell.
+    """
+    if not text:
+        return ""
+    return " " + _escape_tags(re.sub(r"([*|\\])", r"\\\1", text))
+
+
 def _binary_citation(path: str, previous: str) -> tuple[str, str]:
     """A part citing code inside a shipped binary, and the binary it names.
 

scripts/sourceref.py

@@ -123,6 +123,25 @@ def split_source_ref(ref: str) -> list[RefPart]:
     return parts
 
 
+_LOCATION_RE = re.compile(r":\d+(?:-\d+)?|^\s*\d+(?:-\d+)?(?=\s|$)")
+
+
+def ref_annotation(raw: str) -> str:
+    """The words a part writes beside its location, for a reader.
+
+    split_source_ref drops them because a comparison has no use for them,
+    but `src/loader2.cpp:284 (lsf default), src/netcfg.cpp:204 (lsf
+    override)` says which line wins only through them, and 1,103 of the
+    profiles' annotations were on no page.
+    """
+    paren = raw.find("(")
+    head = raw if paren < 0 else raw[:paren]
+    located = _LOCATION_RE.search(head)
+    if located:
+        return raw[located.end():].strip()
+    return raw[paren:].strip() if paren >= 0 else ""
+
+
 def ref_prose_text(ref: str) -> str:
     """The text of a source_ref its parts do not read, positions kept.
 

tests/test_site_render.py

@@ -255,6 +255,47 @@ class AnAliasTreeIsListedOnItsParentsPage(unittest.TestCase):
         self.assertEqual([t["ref"] for t in trees], ["freeintv-overlays"])
 
 
+class CitationsKeepTheirAnnotations(unittest.TestCase):
+    """split_source_ref drops annotations for profile_sync, and the cards
+    reused it: abuse's two lsf lines lost "(lsf default)" and "(lsf
+    override)", the only words saying which one wins."""
+
+    @mock.patch("sitesources.tree_holds", lambda *_args: True)
+    def test_each_part_shows_what_the_profile_wrote_beside_it(self):
+        import sitesources
+
+        profile = {"source": "https://github.com/Xenoveritas/abuse",
+                   "source_commit": "a" * 40}
+        ref = ("src/loader2.cpp:284 (lsf default), src/netcfg.cpp:204 (lsf override), "
+               "libretro.cpp:1520-1521 candidates_a1200, 30-31 (see <system_dir>), src/x.c:1")
+        (label, items), = sitesources.source_ref_items(profile, ref)
+        texts = [text for text, _role in items]
+        self.assertTrue(texts[0].endswith(") (lsf default)"), texts[0])
+        self.assertTrue(texts[1].endswith(") (lsf override)"), texts[1])
+        self.assertTrue(texts[2].endswith(") candidates_a1200"), texts[2])
+        self.assertTrue(texts[3].endswith(") (see &lt;system_dir>)"), texts[3])
+        self.assertTrue(texts[4].endswith(")"), texts[4])
+
+    def test_the_annotation_is_what_follows_the_location(self):
+        from sourceref import ref_annotation
+
+        self.assertEqual(ref_annotation("munt ROMInfo.cpp:206-213"), "")
+        self.assertEqual(ref_annotation("source atari800/src/sysrom.c:91"), "")
+        self.assertEqual(ref_annotation("doc/notes.txt (read at boot)"), "(read at boot)")
+        self.assertEqual(ref_annotation("a.c (see b.c:12)"), "(see b.c:12)")
+        self.assertEqual(ref_annotation("0x53c338-0x53c36f (dreamBios row)"), "(dreamBios row)")
+
+    def test_a_binary_citation_keeps_its_annotation(self):
+        import sitesources
+
+        (_label, items), = sitesources.source_ref_items(
+            {}, "demul.exe 0x6b4d80-0x6b4ea3 (dc table), 0x53ca4d-0x53ca7e (load, fatal)")
+        self.assertEqual([text for text, _ in items], [
+            "offset `0x6b4d80-0x6b4ea3` in `demul.exe` (dc table)",
+            "offset `0x53ca4d-0x53ca7e` in `demul.exe` (load, fatal)",
+        ])
+
+
 class ForgesUpstreamDoesNotRead(unittest.TestCase):
     """Profiles pinned on SourceForge, voidpoint.io, Bitbucket or Software
     Heritage showed no revision, and their citations were plain code."""