marvisx-cli 0.1.0__py3-none-any.whl

core/api/services/inbox_taxonomy.py

@@ -0,0 +1,195 @@
+from __future__ import annotations
+
+from typing import Final
+
+
+VALID_INBOX_TOPICS: Final[tuple[str, ...]] = (
+    "ai-news",
+    "ai-products",
+    "tooling",
+    "security-devtools",
+    "pv-energy",
+    "strategy-business",
+    "policy-politics",
+    "general",
+)
+
+VALID_INBOX_TREATMENTS: Final[tuple[str, ...]] = (
+    "read",
+    "save",
+    "read_save",
+    "ignore",
+)
+
+_TOPIC_SET = set(VALID_INBOX_TOPICS)
+_TREATMENT_SET = set(VALID_INBOX_TREATMENTS)
+
+
+def normalize_inbox_topic(value: str | None) -> str:
+    if not value:
+        return "general"
+    topic = value.strip().lower()
+    return topic if topic in _TOPIC_SET else "general"
+
+
+def normalize_inbox_treatment(value: str | None) -> str:
+    if not value:
+        return "read"
+    treatment = value.strip().lower()
+    return treatment if treatment in _TREATMENT_SET else "read"
+
+
+def infer_topic_from_metadata(
+    metadata: dict[str, object],
+    *,
+    title: str | None = None,
+    content: str | None = None,
+) -> str:
+    hint = metadata.get("topicHint")
+    if isinstance(hint, str):
+        return normalize_inbox_topic(hint)
+
+    feed_url = metadata.get("feedUrl")
+    if isinstance(feed_url, str):
+        url = feed_url.lower()
+        if any(
+            token in url
+            for token in (
+                "arxiv",
+                "importai",
+                "latent.space",
+                "deeplearning.ai",
+                "alignmentfeed",
+                "latentsignal",
+            )
+        ):
+            return "ai-news"
+        if any(
+            token in url
+            for token in (
+                "anthropics/claude-code",
+                "anomalyco/opencode",
+                "simonwillison.net/atom/everything",
+            )
+        ):
+            return "tooling"
+        if any(token in url for token in ("pv-magazine", "renewablesnow", "pveurope")):
+            return "pv-energy"
+        if any(
+            token in url
+            for token in ("stratechery", "newcomer", "a16z", "notboring", "thediff")
+        ):
+            return "strategy-business"
+        if any(token in url for token in ("ilpost", "linkiesta", "lavoce", "politico")):
+            return "policy-politics"
+
+    feed_name = metadata.get("feedName")
+    if isinstance(feed_name, str):
+        name = feed_name.lower()
+        if any(
+            token in name
+            for token in ("claude", "opencode", "datasette", "tool", "simon willison")
+        ):
+            return "tooling"
+        if any(token in name for token in ("security", "secret", "vuln")):
+            return "security-devtools"
+        if any(token in name for token in ("pv", "renewable", "energy")):
+            return "pv-energy"
+        if any(
+            token in name for token in ("politico", "ilpost", "linkiesta", "lavoce")
+        ):
+            return "policy-politics"
+
+    corpus = " ".join(filter(None, [title, content])).lower()
+    if any(
+        token in corpus
+        for token in (
+            "anthropic",
+            "project glasswing",
+            "mythos preview",
+            "model preview",
+            "model release",
+            "llm release",
+            "openai",
+            "google ai",
+        )
+    ):
+        return "ai-news"
+    if any(
+        token in corpus
+        for token in (
+            "claude code",
+            "opencode",
+            "datasette",
+            "sqlite wal",
+            "scan-for-secrets",
+        )
+    ):
+        return "tooling"
+    if any(
+        token in corpus
+        for token in ("supply chain", "vulnerability", "secret", "security", "axios")
+    ):
+        return "security-devtools"
+    if any(
+        token in corpus
+        for token in (
+            "glm",
+            "gemma",
+            "research",
+            "paper",
+            "benchmark",
+            "model",
+            "agentic",
+        )
+    ):
+        return "ai-news"
+    if any(
+        token in corpus
+        for token in ("product", "use case", "customer", "community", "feedback")
+    ):
+        return "ai-products"
+    if any(
+        token in corpus for token in ("solar", "pv", "renewable", "battery", "energy")
+    ):
+        return "pv-energy"
+    if any(
+        token in corpus
+        for token in (
+            "market",
+            "startup",
+            "funding",
+            "strategy",
+            "venture",
+            "acquisition",
+        )
+    ):
+        return "strategy-business"
+    if any(
+        token in corpus
+        for token in (
+            "politics",
+            "regulation",
+            "policy",
+            "europe",
+            "government",
+            "election",
+        )
+    ):
+        return "policy-politics"
+    return "general"
+
+
+def infer_treatment(topic: str, metadata: dict[str, object]) -> str:
+    source_class = metadata.get("sourceClass")
+    if source_class == "operational_email":
+        return "save"
+    if topic in {"ai-products", "strategy-business", "policy-politics"}:
+        return "read_save"
+    if topic in {
+        "tooling",
+        "security-devtools",
+        "pv-energy",
+    }:
+        return "save"
+    return "read"