feat: Ghost CMS integration — auto-publish blog + /publish command

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/generator/generator.py

@@ -2,6 +2,12 @@
 ResearchOwl Generators
 Produces structured outputs from processed research using Claude or Ollama
 """
+import base64
+import hashlib
+import hmac
+import json
+import time
+
 import structlog
 
 from src.config import settings
@@ -196,6 +202,92 @@ Material disponible (resumen):
 """
 
 
+# ─── Ghost CMS ────────────────────────────────────────────────────────────────
+
+def _b64url(data: bytes | str) -> str:
+    if isinstance(data, str):
+        data = data.encode()
+    return base64.urlsafe_b64encode(data).rstrip(b"=").decode()
+
+
+def _extract_title(content: str) -> str:
+    """Return first H1 heading from markdown, skipping the ResearchOwl header block."""
+    in_header = False
+    for line in content.splitlines():
+        stripped = line.strip()
+        if stripped == "---":
+            in_header = not in_header
+            continue
+        if in_header:
+            continue
+        if stripped.startswith("# ") and not stripped.startswith("## "):
+            return stripped[2:].strip()
+    return ""
+
+
+def _strip_researchowl_header(content: str) -> str:
+    """Remove the ---...--- metadata block that ResearchOwl prepends to outputs."""
+    lines = content.splitlines(keepends=True)
+    dashes_seen = 0
+    for i, line in enumerate(lines):
+        if line.strip() == "---":
+            dashes_seen += 1
+            if dashes_seen == 2:
+                return "".join(lines[i + 1:]).lstrip("\n")
+    return content
+
+
+class GhostPublisher:
+    def __init__(self):
+        self.url = (settings.ghost_url or "").rstrip("/")
+        self.api_key = settings.ghost_api_key or ""
+
+    def is_configured(self) -> bool:
+        return bool(self.url and self.api_key)
+
+    def _make_token(self) -> str:
+        key_id, secret = self.api_key.split(":", 1)
+        now = int(time.time())
+        header = _b64url(json.dumps({"alg": "HS256", "typ": "JWT", "kid": key_id}))
+        payload = _b64url(json.dumps({"iat": now, "exp": now + 300, "aud": "/admin/"}))
+        signing = f"{header}.{payload}"
+        sig = _b64url(
+            hmac.new(bytes.fromhex(secret), signing.encode(), hashlib.sha256).digest()
+        )
+        return f"{signing}.{sig}"
+
+    async def publish_draft(self, title: str, markdown_content: str,
+                            tags: list[str] | None = None) -> dict:
+        import aiohttp as _aio
+        import markdown as _md
+        clean = _strip_researchowl_header(markdown_content)
+        html = _md.markdown(clean, extensions=["extra"])
+        token = self._make_token()
+        body = {
+            "posts": [{
+                "title": title,
+                "html": html,
+                "status": "draft",
+                "tags": [{"name": t} for t in (tags or ["investigacion"])],
+            }]
+        }
+        async with _aio.ClientSession() as sess:
+            async with sess.post(
+                f"{self.url}/ghost/api/admin/posts/",
+                json=body,
+                headers={
+                    "Authorization": f"Ghost {token}",
+                    "Accept-Version": "v5.0",
+                },
+            ) as resp:
+                if resp.status not in (200, 201):
+                    text = await resp.text()
+                    raise ValueError(f"Ghost API {resp.status}: {text[:300]}")
+                return await resp.json()
+
+
+# ─── Output generation ────────────────────────────────────────────────────────
+
 class OutputGenerator:
     def __init__(self, db: ResearchDB, ollama: OllamaClient, processor: ContentProcessor):
         self.db = db
@@ -250,8 +342,26 @@ class OutputGenerator:
         # Save to DB
         await self.db.save_output(session_id, output_type, full_output)
 
+        # Auto-publish to Ghost for blog outputs
+        ghost_notice = ""
+        if output_type in (OutputType.BLOG, OutputType.BLOG_EXTENDED):
+            ghost = GhostPublisher()
+            if ghost.is_configured():
+                try:
+                    title = _extract_title(full_output) or topic
+                    result = await ghost.publish_draft(title, full_output)
+                    post = result["posts"][0]
+                    ghost_notice = (
+                        f"\n\n---\n"
+                        f"📤 *Borrador publicado en Ghost*\n"
+                        f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}"
+                    )
+                    logger.info("Auto-published blog to Ghost", post_id=post["id"])
+                except Exception as e:
+                    logger.warning("Auto-publish to Ghost failed", error=str(e))
+
         logger.info("Output generated", type=output_type, length=len(full_output))
-        return full_output
+        return full_output + ghost_notice
 
     async def _generate(self, prompt: str, system: str, output_type: OutputType,
                          session_id: int | None = None) -> str:
@@ -403,9 +513,28 @@ class OutputGenerator:
         full_output = header + "\n\n" + full_content
 
         await self.db.save_output(session_id, output_type, full_output)
+
+        # Auto-publish to Ghost for extended blog outputs
+        ghost_notice = ""
+        if output_type == OutputType.BLOG_EXTENDED:
+            ghost = GhostPublisher()
+            if ghost.is_configured():
+                try:
+                    title = _extract_title(full_output) or topic
+                    result = await ghost.publish_draft(title, full_output)
+                    post = result["posts"][0]
+                    ghost_notice = (
+                        f"\n\n---\n"
+                        f"📤 *Borrador publicado en Ghost*\n"
+                        f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}"
+                    )
+                    logger.info("Auto-published extended blog to Ghost", post_id=post["id"])
+                except Exception as e:
+                    logger.warning("Auto-publish to Ghost failed (extended)", error=str(e))
+
         logger.info("Extended output generated", type=output_type,
                     sections=len(sections), length=len(full_output))
-        return full_output
+        return full_output + ghost_notice
 
     async def _generate_raw(self, prompt: str,
                              session_id: int | None = None) -> str: