engramkeep 0.5.1__py3-none-any.whl

engramkeep/concepts.py

@@ -0,0 +1,198 @@
+"""Deterministic bilingual (EN/TR) concept expansion for keyword search.
+
+Engramkeep stores decisions in plain Markdown. People write memory cards in
+the language they think in: a card saved as "Capacitor Android keystore
+sorunu" must still surface for the query "mobile" or "keystore". This module
+maps common engineering concepts to their EN and TR spellings so a search in
+one language finds cards written in the other.
+
+The map is deliberately small, generic, and free of project-specific terms.
+It is a scoring aid, not an ontology: matched expanded terms carry a lower
+weight than the literal query tokens.
+"""
+
+from __future__ import annotations
+
+import re
+
+SEMANTIC_CONCEPTS: dict[str, tuple[str, ...]] = {
+    "mobile": (
+        "mobil",
+        "android",
+        "ios",
+        "iphone",
+        "apk",
+        "capacitor",
+        "react-native",
+        "flutter",
+        "keystore",
+        "hybrid",
+    ),
+    "database": (
+        "veritabanı",
+        "sqlite",
+        "postgres",
+        "postgresql",
+        "mysql",
+        "mariadb",
+        "mongodb",
+        "redis",
+        "prisma",
+        "sql",
+        "db",
+        "storage",
+    ),
+    "frontend": (
+        "ön yüz",
+        "arayüz",
+        "react",
+        "next.js",
+        "nextjs",
+        "vue",
+        "svelte",
+        "pwa",
+        "tailwind",
+        "css",
+        "component",
+        "ui",
+        "ux",
+    ),
+    "backend": (
+        "api",
+        "rest",
+        "graphql",
+        "fastapi",
+        "flask",
+        "django",
+        "express",
+        "node",
+        "microservice",
+        "endpoint",
+        "arka yüz",
+    ),
+    "auth": (
+        "authentication",
+        "authorization",
+        "jwt",
+        "oauth",
+        "session",
+        "login",
+        "token",
+        "kimlik",
+        "yetki",
+        "şifre",
+    ),
+    "devops": (
+        "docker",
+        "kubernetes",
+        "k8s",
+        "ci/cd",
+        "github actions",
+        "vercel",
+        "aws",
+        "cloudflare",
+        "nginx",
+        "deploy",
+        "dağıtım",
+    ),
+    "automation": (
+        "otomasyon",
+        "webhook",
+        "workflow",
+        "cron",
+        "schedule",
+        "bot",
+        "scraper",
+        "pipeline",
+    ),
+    "testing": (
+        "test",
+        "sınama",
+        "pytest",
+        "unittest",
+        "e2e",
+        "integration",
+        "mock",
+        "fixture",
+    ),
+    "security": (
+        "güvenlik",
+        "vulnerability",
+        "xss",
+        "csrf",
+        "injection",
+        "redact",
+        "secret",
+        "encrypt",
+        "şifreleme",
+    ),
+    "search": (
+        "arama",
+        "query",
+        "index",
+        "indeks",
+        "full-text",
+        "fts",
+        "rank",
+        "snippet",
+    ),
+    "sync": (
+        "senkronizasyon",
+        "senkron",
+        "backup",
+        "yedek",
+        "restore",
+        "cloud",
+        "drive",
+        "sync",
+    ),
+    "performance": (
+        "performans",
+        "latency",
+        "cache",
+        "önbellek",
+        "profiling",
+        "optimizasyon",
+        "memory",
+        "bellek",
+    ),
+    "media": (
+        "görsel",
+        "image",
+        "video",
+        "audio",
+        "ffmpeg",
+        "logo",
+        "svg",
+        "banner",
+        "thumbnail",
+    ),
+}
+
+_TOKEN_SPLIT = re.compile(r"[^\w/.@-]+", re.UNICODE)
+
+
+def expand_query_terms(query: str) -> set[str]:
+    """Return the normalized query terms plus matching EN/TR concept terms."""
+    normalized = " ".join(query.lower().split())
+    if not normalized:
+        return set()
+
+    tokens = {token for token in _TOKEN_SPLIT.split(normalized) if token}
+    expanded = set(tokens)
+    if not tokens:
+        expanded.add(normalized)
+
+    for primary, related in SEMANTIC_CONCEPTS.items():
+        family = (primary, *related)
+        if any(_term_matches(term, normalized, tokens) for term in family):
+            expanded.update(family)
+    return expanded
+
+
+def _term_matches(term: str, normalized: str, tokens: set[str]) -> bool:
+    if term in tokens:
+        return True
+    # Substring match in both directions so "mobil" finds "mobile"-family
+    # entries and multi-word terms like "github actions" still match.
+    return term in normalized or normalized in term

engramkeep/config.py

@@ -0,0 +1,79 @@
+"""Resolve the local Engramkeep data directory.
+
+Precedence is intentionally simple and explicit:
+
+1. the ``--home`` command-line value,
+2. the ``ENGRAMKEEP_HOME`` environment variable,
+3. a platform-specific default.
+"""
+
+from __future__ import annotations
+
+import os
+import platform
+from collections.abc import Mapping
+from pathlib import Path
+
+from .errors import ConfigError
+
+APP_NAME = "engramkeep"
+DISPLAY_NAME = "Engramkeep"
+ENV_HOME = "ENGRAMKEEP_HOME"
+SCHEMA_VERSION = 1
+
+
+def platform_default_home(
+    *,
+    system: str | None = None,
+    env: Mapping[str, str] | None = None,
+    user_home: Path | None = None,
+) -> Path:
+    """Return the platform-specific default data directory."""
+    system = system or platform.system()
+    env = os.environ if env is None else env
+    user_home = Path.home() if user_home is None else Path(user_home)
+
+    if system == "Windows":
+        base = env.get("LOCALAPPDATA")
+        root = Path(base) if base else user_home / "AppData" / "Local"
+        return (root / DISPLAY_NAME).resolve()
+    if system == "Darwin":
+        return (user_home / "Library" / "Application Support" / DISPLAY_NAME).resolve()
+    base = env.get("XDG_DATA_HOME")
+    root = Path(base) if base else user_home / ".local" / "share"
+    return (root / APP_NAME).resolve()
+
+
+def resolve_home(
+    cli_home: str | os.PathLike[str] | None = None,
+    *,
+    env: Mapping[str, str] | None = None,
+    system: str | None = None,
+    user_home: Path | None = None,
+) -> Path:
+    """Resolve the active memory home.
+
+    An explicit command-line value wins over ``ENGRAMKEEP_HOME``, which wins
+    over the platform default. Empty values are ignored.
+    """
+    env = os.environ if env is None else env
+    raw: str | os.PathLike[str] | None = cli_home
+    source = "command line"
+
+    if raw is None or str(raw).strip() == "":
+        raw = env.get(ENV_HOME)
+        source = ENV_HOME
+    if raw is None or str(raw).strip() == "":
+        return platform_default_home(system=system, env=env, user_home=user_home)
+
+    path = Path(raw).expanduser()
+    try:
+        resolved = path.resolve()
+    except OSError as exc:
+        raise ConfigError(f"cannot resolve {source} path: {path}") from exc
+
+    if resolved == Path(resolved.anchor):
+        raise ConfigError(f"{source} must not point at a filesystem root: {resolved}")
+    if resolved.exists() and not resolved.is_dir():
+        raise ConfigError(f"{source} is not a directory: {resolved}")
+    return resolved