fix: hide generic citation summaries

Co-authored-by: Codex <noreply@openai.com>

hackathon_advisor/data.py

@@ -11,6 +11,16 @@ import re
 
 
 TOKEN_RE = re.compile(r"[a-z0-9][a-z0-9.+_-]*", re.IGNORECASE)
+GENERIC_PUBLIC_TITLE_RE = re.compile(
+    r"^(?:my\s+)?build\s+small\s+hackathon$",
+    re.IGNORECASE,
+)
+GENERIC_PUBLIC_SUMMARY_RE = re.compile(
+    r"(?:\bthis\s+(?:is\s+)?(?:space\s+is\s+for|my\s+submission)\b.*\b(?:build[-\s]*small|hackathon)\b)"
+    r"|(?:\bhacka?ton\s+project\b)"
+    r"|(?:^\s*todo\s*$)",
+    re.IGNORECASE,
+)
 
 
 @dataclass(frozen=True)
@@ -65,6 +75,23 @@ class Project:
         )
 
     def to_public_dict(self) -> dict:
+        return {
+            "id": self.id,
+            "title": public_project_title(self.title),
+            "summary": public_project_summary(self.summary),
+            "tags": list(self.tags),
+            "models": list(self.models),
+            "datasets": list(self.datasets),
+            "likes": self.likes,
+            "sdk": self.sdk,
+            "license": self.license,
+            "created_at": self.created_at,
+            "last_modified": self.last_modified,
+            "host": self.host,
+            "url": self.url,
+        }
+
+    def to_snapshot_dict(self) -> dict:
         return {
             "id": self.id,
             "title": self.title,
@@ -108,6 +135,24 @@ class WhitespaceItem:
         }
 
 
+def public_project_title(title: str) -> str:
+    cleaned = " ".join(str(title).split())
+    if not cleaned:
+        return "Untitled project"
+    if GENERIC_PUBLIC_TITLE_RE.search(cleaned):
+        return "Untitled project"
+    return cleaned
+
+
+def public_project_summary(summary: str) -> str:
+    cleaned = " ".join(str(summary).split())
+    if not cleaned:
+        return ""
+    if GENERIC_PUBLIC_SUMMARY_RE.search(cleaned):
+        return ""
+    return cleaned
+
+
 @dataclass(frozen=True)
 class WhitespaceSeed:
     label: str
@@ -354,7 +399,7 @@ def project_snapshot_digest(projects: list[Project], generated_at: str, source:
     payload = {
         "generated_at": generated_at,
         "source": source,
-        "projects": [project.to_public_dict() for project in projects],
+        "projects": [project.to_snapshot_dict() for project in projects],
     }
     encoded = json.dumps(payload, sort_keys=True, separators=(",", ":"), ensure_ascii=False).encode("utf-8")
     return sha256(encoded).hexdigest()

static/app.js

@@ -765,10 +765,8 @@ function renderProjects(projects) {
     item.href = project.url;
     item.target = "_blank";
     item.rel = "noreferrer";
-    item.innerHTML = `
-      <strong>${escapeHtml(project.title)}</strong>
-      <p>${escapeHtml(project.summary || project.id)}</p>
-    `;
+    const summary = project.summary ? `<p>${escapeHtml(project.summary)}</p>` : "";
+    item.innerHTML = `<strong>${escapeHtml(project.title || "Untitled project")}</strong>${summary}`;
     projectsEl.append(item);
   }
 }
@@ -788,9 +786,10 @@ function renderCitations(echoes) {
     item.rel = "noreferrer";
     item.title = project.title || project.id || "Project citation";
     const matched = (echo.matched_terms || []).slice(0, 5).join(", ") || "no shared terms";
+    const summary = project.summary ? `<p>${escapeHtml(project.summary)}</p>` : "";
     item.innerHTML = `
-      <strong>Page ${escapeHtml(echo.page_number || "?")} · ${escapeHtml(project.title || project.id || "Untitled")}</strong>
-      <p>${escapeHtml(project.summary || project.id || "")}</p>
+      <strong>Page ${escapeHtml(echo.page_number || "?")} · ${escapeHtml(project.title || "Untitled project")}</strong>
+      ${summary}
       <span class="matched">${Number(echo.score || 0).toFixed(3)} · ${escapeHtml(matched)}</span>
     `;
     projectsEl.append(item);

tests/test_data.py

@@ -1,7 +1,7 @@
 from pathlib import Path
 import json
 
-from hackathon_advisor.data import ProjectIndex
+from hackathon_advisor.data import Project, ProjectIndex, public_project_summary, public_project_title
 
 
 def test_project_index_searches_snapshot() -> None:
@@ -24,6 +24,34 @@ def test_project_index_whitespace() -> None:
     assert all(item.label for item in items)
 
 
+def test_public_project_cards_hide_generic_submission_copy() -> None:
+    assert public_project_title("My Build Small Hackathon") == "Untitled project"
+    assert public_project_summary("This is my submission for the build-small-hackathon") == ""
+    assert public_project_summary("Todo") == ""
+    assert public_project_summary("Local-first personal knowledge agent") == "Local-first personal knowledge agent"
+
+    project = Project(
+        id="build-small-hackathon/my-build-small-hackathon",
+        title="My Build Small Hackathon",
+        summary="This is my submission for the build-small-hackathon",
+        tags=(),
+        models=(),
+        datasets=(),
+        likes=0,
+        sdk="gradio",
+        license="",
+        created_at="",
+        last_modified="",
+        host="",
+        url="https://example.test",
+    )
+
+    public = project.to_public_dict()
+
+    assert public["title"] == "Untitled project"
+    assert public["summary"] == ""
+
+
 def test_project_index_rejects_mismatched_snapshot(tmp_path: Path) -> None:
     payload = json.loads(Path("data/project_index.json").read_text(encoding="utf-8"))
     payload["snapshot_generated_at"] = "2000-01-01T00:00:00+00:00"