@danhachuel/thunderbolt 0.6.60 → 0.6.62

package/app/social_networks_ui.py

@@ -15,7 +15,7 @@ from hermes_ui.influencers import InfluencerBackendError, STANDALONE_CONTENT_INF
 from hermes_ui.storage import read_json, write_json
 from hermes_ui.countries import COUNTRY_OPTIONS
 from hermes_ui.languages import LANGUAGE_CODES, language_code, language_label
-from integrations.instagram_public import fetch_public_instagram_posts, fetch_public_instagram_profile, normalize_instagram_bio, normalize_instagram_metric
+from integrations.instagram_public import fetch_public_instagram_posts, fetch_public_instagram_profile, normalize_instagram_metric
 from integrations.meta_social import test_facebook_pages_api_card, test_instagram_api_card
 
 
@@ -46,17 +46,13 @@ def _metric_input(value: Any) -> int | None:
 
 
 def _render_instagram_bio(value: Any) -> None:
-    bio = normalize_instagram_bio(value)
+    bio = str(value or "")
     if bio:
         st.text(bio)
 
 
 def _profile_bio(profile: Mapping[str, Any]) -> str:
-    for key in ("bio", "bio_raw", "biography", "description"):
-        bio = normalize_instagram_bio(profile.get(key))
-        if bio:
-            return bio
-    return ""
+    return str(profile.get("bio") or profile.get("biography", ""))
 
 
 def _country_key(value: Any) -> str:
@@ -539,7 +535,7 @@ def _render_instagram_card(profile: dict[str, Any], characters: list[dict[str, A
                     following = st.number_input("seguindo", min_value=0, value=_metric_input(_profile_metric(profile, "following_count", "following", "follows", "followingCount", "edge_follow")), placeholder="Não encontrado", key=f"instagram_edit_following_{profile_id}")
                 save_edit = st.form_submit_button("Guardar alterações", type="primary", use_container_width=True)
             if save_edit:
-                update_channel(profile_id, {"name": name.strip(), "handle": handle.strip(), "bio": normalize_instagram_bio(bio), "bio_raw": bio, "country": _normalise_country(country), "language": language, "post_count": _metric_input(posts), "subscriber_count": _metric_input(followers), "following_count": _metric_input(following)})
+                update_channel(profile_id, {"name": name.strip(), "handle": handle.strip(), "bio": bio, "bio_raw": bio, "country": _normalise_country(country), "language": language, "post_count": _metric_input(posts), "subscriber_count": _metric_input(followers), "following_count": _metric_input(following)})
                 st.session_state.pop(edit_key, None)
                 st.success("Conta Instagram actualizada.")
                 st.rerun()
@@ -625,7 +621,7 @@ def render_social_networks(settings: dict[str, Any]) -> None:
                     save_profile = st.form_submit_button("Cadastrar conta Instagram", type="primary", use_container_width=True)
                 if save_profile:
                     profile_data = dict(data)
-                    profile_data.update({"name": name, "bio": normalize_instagram_bio(bio), "bio_raw": bio, "post_count": _metric_input(posts), "subscriber_count": _metric_input(followers), "following_count": _metric_input(following)})
+                    profile_data.update({"name": name, "bio": bio, "bio_raw": bio, "post_count": _metric_input(posts), "subscriber_count": _metric_input(followers), "following_count": _metric_input(following)})
                     profile_data["country"] = _normalise_country(country)
                     _save_public_profile(profile_data, country=profile_data["country"], language=language, character_id=selected_character)
                     st.success("Conta Instagram cadastrada em Redes Sociais.")

package/integrations/instagram_public.py

@@ -78,8 +78,8 @@ class IntegrationResult:
 
 
 def normalize_instagram_bio(value: Any) -> str:
-    """Preserve the complete biography returned by Instagram."""
-    return str(value or '').strip()
+    """Deprecated compatibility helper; Bio values are now passed through unchanged."""
+    return str(value or '')
 
 
 def normalize_instagram_reference(source: str) -> dict[str, str]:
@@ -185,23 +185,45 @@ def _metric_text_value(value: Any) -> int | None:
     return int(digits) * multiplier
 
 
+def _extract_bio_from_html(document: str) -> str:
+    """Extract only the JSON biography value; never use an SEO description as Bio."""
+    patterns = (
+        r'"biography"\s*:\s*"((?:\\.|[^"\\])*)"',
+        r"'biography'\s*:\s*'((?:\\.|[^'\\])*)'",
+    )
+    for pattern in patterns:
+        for match in re.finditer(pattern, document, flags=re.IGNORECASE | re.DOTALL):
+            raw_value = match.group(1)
+            try:
+                decoded = json.loads(f'"{raw_value}"')
+            except (TypeError, ValueError, json.JSONDecodeError):
+                decoded = bytes(raw_value, 'utf-8').decode('unicode_escape')
+            if str(decoded or '').strip():
+                return str(decoded).strip()
+    return ''
+
+
 def _embedded_profile_user(document: str, username: str) -> Mapping[str, Any]:
-    wanted = username.casefold()
+    wanted = str(username or '').strip().lstrip('@').casefold()
     for payload in _embedded_json_documents(document):
         for node in _walk_json(payload):
             if not isinstance(node, Mapping):
                 continue
-            node_username = str(node.get('username') or node.get('handle') or '').lstrip('@').casefold()
-            if node_username == wanted and any(key in node for key in ('biography', 'bio', 'edge_followed_by', 'edge_follow', 'followers', 'following')):
+            node_username = str(node.get('username') or node.get('handle') or '').strip().lstrip('@').casefold()
+            has_profile_fields = any(key in node for key in ('biography', 'bio', 'edge_followed_by', 'edge_follow', 'followers', 'following'))
+            if node_username == wanted and has_profile_fields:
                 return node
     return {}
 
 
 def _extract_profile_user_from_html(document: str, username: str) -> dict[str, Any]:
     embedded_user = _embedded_profile_user(document, username)
+    regex_biography = _extract_bio_from_html(document)
     seo_description = _meta(document, 'og:description')
-    biography = str(embedded_user.get('biography') or embedded_user.get('bio') or '').strip()
-    description = biography or seo_description
+    biography = str(embedded_user.get('biography') or embedded_user.get('bio') or regex_biography or '').strip()
+    print(f'embedded_user for @{username}: {embedded_user}')
+    print(f'biography extracted for @{username}: {biography!r}')
+    description = biography
     title = str(embedded_user.get('full_name') or embedded_user.get('name') or _meta(document, 'og:title') or username).split('(')[0].strip() or username
     avatar_url = str(embedded_user.get('profile_pic_url_hd') or embedded_user.get('profile_pic_url') or _meta(document, 'og:image')).strip()
 
@@ -235,7 +257,7 @@ def _extract_profile_user_from_html(document: str, username: str) -> dict[str, A
         'profile_pic_url': avatar_url,
         '_html_posts': posts,
     }
-    return user if any((description, followers, following, post_count, country, posts)) else {}
+    return user if any((biography, followers, following, post_count, country, posts)) else {}
 
 
 def _fetch_web_profile_user(username: str) -> dict[str, Any] | None:
@@ -382,7 +404,7 @@ def _profile_data_from_api(user: Mapping[str, Any], reference: Mapping[str, str]
         'id': f"instagram_{reference['username']}",
         **reference,
         'name': str(user.get('full_name') or user.get('username') or reference['username']).strip(),
-        'bio': normalize_instagram_bio(biography),
+        'bio': biography,
         'bio_raw': biography,
         'country': country,
         'avatar_url': str(user.get('profile_pic_url_hd') or user.get('profile_pic_url') or '').strip(),
@@ -492,7 +514,7 @@ def fetch_public_instagram_profile(source: str) -> IntegrationResult:
         embedded_country = extract_public_instagram_country(document)
         if embedded_country:
             break
-    data = {'id': f"instagram_{reference['username']}", **reference, 'name': title.split('(')[0].strip() or reference['username'], 'bio': normalize_instagram_bio(description), 'bio_raw': description, 'country': embedded_country, 'avatar_url': avatar_url, 'subscriber_count': followers, 'following_count': following, 'post_count': posts, 'public_lookup': True, 'metrics_source': 'instagram_public_page', 'last_public_lookup_at': datetime.now(timezone.utc).isoformat()}
+    data = {'id': f"instagram_{reference['username']}", **reference, 'name': title.split('(')[0].strip() or reference['username'], 'bio': description, 'bio_raw': description, 'country': embedded_country, 'avatar_url': avatar_url, 'subscriber_count': followers, 'following_count': following, 'post_count': posts, 'public_lookup': True, 'metrics_source': 'instagram_public_page', 'last_public_lookup_at': datetime.now(timezone.utc).isoformat()}
     return IntegrationResult(True, 'Perfil Instagram encontrado publicamente.', data)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@danhachuel/thunderbolt",
-  "version": "0.6.60",
+  "version": "0.6.62",
   "description": "Thunderbolt — interface local para operação de canais faceless e motor MoneyPrinterTurbo",
   "license": "MIT",
   "main": "scripts/cli.mjs",