@r1z4x/tezgah 1.9.2 → 1.11.0

package/hooks/tezgah_extension.py

@@ -0,0 +1,337 @@
+#!/usr/bin/env python3
+"""Statusline extensions: a repository's own metrics, on tezgah's line.
+
+A repository adds chips to the status line without touching tezgah. It drops one
+manifest at `<repo>/.tezgah/extension.json`:
+
+    {"statusline": {"command": ["bin/engagement-statusline", "--segments"],
+                    "timeout_ms": 300}}
+
+and that command prints structured segments on stdout:
+
+    {"segments": [{"key": "eng", "state": "info", "text": "pentest-001",
+                   "glyph": "", "group": 0}]}
+
+The command is handed its own call on stdin as one JSON object -
+`{"api": 1, "event": "statusline", "cwd": ..., "repo_root": ...,
+"session_id": ...}` (`call_envelope()`) - so it knows which repository and
+session asked, rather than guessing from its environment.
+
+tezgah discovers the manifest by walking up from the working directory, runs the
+command once per render (bounded, fail-open) and appends what it returns after
+its own marks. The core owns rendering: an extension produces data, never a
+formatted line, so color, the narrow-surface drop and the group separators stay
+identical for every extension, and the extension never has to know tezgah's
+gutter.
+
+Why a subprocess and not an import: the extension lives in the user's repo, in
+whatever language, and may be slow or broken. Importing it runs its code in the
+harness's own process; a bounded child is the seam that keeps a broken extension
+costing one chip on one line, never the gate that carries every turn.
+
+The line is drawn on every redraw, so the cost is bounded on purpose: the walk
+stops at the first manifest, the child gets a hard timeout, and a repo with no
+manifest pays a single `os.path.isfile` per level and nothing else.
+
+The same manifest carries four more keys. Each is read here, consumed by one
+caller and validated the way `statusline` is, so a broken value costs that key
+and never the line or the gate (docs/status-line.md, "The manifest's other
+keys"):
+
+    {"checks": ["./bin/run-tool"],                          # repo_checks
+     "untrusted": ["nmap", "sqlmap"],                       # untrusted_programs
+     "agents": [{"match": "(?i)recon", "agent": "recon"}],  # repo_agent
+     "statusline": {"command": [...], "override": true}}    # override
+"""
+import json
+import os
+import re
+import shlex
+import subprocess
+
+EXT_DIR = ".tezgah"
+MANIFEST_NAME = "extension.json"
+# The variable a domain repository names its engagement root with: a session
+# working anywhere under `$TEZGAH_ENGAGEMENTS_ROOT/<slug>` reads that
+# engagement's own manifest even when the walk up ran out first.
+ENGAGEMENTS_ENV = "TEZGAH_ENGAGEMENTS_ROOT"
+# How far the walk climbs before giving up. A repo root is at most a few levels
+# above any working directory an editor or a host opens; a bound keeps a
+# manifest-less path from stat-ing every ancestor up to `/`.
+MAX_WALK = 12
+# A healthy extension costs one interpreter start (~25 ms measured 2026-10-19),
+# but a cold start under a loaded machine spiked past 500 ms, so the default
+# carries headroom: the budget is a hang detector, not a latency budget, and a
+# manifest that knows its own cost lowers it.
+DEFAULT_TIMEOUT_MS = 1000
+# A chip budget, not a data budget: the status line is one row, and an extension
+# that floods it is a bug in the extension. The per-extension cap keeps one
+# broken manifest from pushing the built-in marks off the line.
+MAX_SEGMENTS = 8
+STATES = ("on", "ready", "off", "info")
+# The call envelope's version, not this module's: it is what tells an extension
+# which keys to expect on its stdin. Adding a key leaves it alone; changing what
+# a key means bumps it.
+CALL_API = 1
+# The other keys' bounds. A list past its cap is cut, not refused, so a long
+# manifest still contributes its first entries. A pattern is bounded in length
+# and the brief it reads in characters, so a repo's regex cannot hold route() up.
+MAX_CHECKS = 32
+# A `checks` prefix that is one of these programs alone checks nothing, and a
+# run of it would license a pass; with a word beyond the program (`echo tested`)
+# it names one command and stays.
+NO_OP_PROGRAMS = frozenset(("echo", "true", "cat", "printf", "ls"))
+MAX_UNTRUSTED = 64
+MAX_AGENTS = 32
+MAX_PATTERN = 512
+MAX_BRIEF = 4000
+# An agent name lands in a ledger row's `agent=` word and in a spawn call, so it
+# is one word of a file-name alphabet; a program name is one basename.
+NAME = re.compile(r"^[A-Za-z0-9][\w.+-]{0,63}$")
+
+
+def call_envelope(repo_root, cwd, session_id=None):
+    """The one JSON object every extension call carries on stdin.
+
+    The child is told which repository it was called for, from which directory
+    and in which session, instead of inferring that from its own environment. A
+    key an extension does not know it ignores; `api` is the key it can gate on,
+    so a future envelope reads as a named refusal rather than a wrong chip."""
+    return {"api": CALL_API, "event": "statusline",
+            "cwd": os.path.realpath(cwd), "repo_root": repo_root,
+            "session_id": session_id or None}
+
+
+def manifest_path(cwd):
+    """(repo_root, manifest_path) for the nearest `.tezgah/extension.json` above
+    `cwd`, or (None, None) when there is none within MAX_WALK levels.
+
+    The first manifest wins: nesting is a mistake, not a feature, and stopping
+    early is what makes the walk cheap. When the walk finds none, a cwd under
+    `$TEZGAH_ENGAGEMENTS_ROOT/<slug>` reads that engagement's manifest however
+    deep it sits (`_engagement_manifest`)."""
+    path = os.path.realpath(cwd)
+    for _ in range(MAX_WALK):
+        candidate = os.path.join(path, EXT_DIR, MANIFEST_NAME)
+        if os.path.isfile(candidate):
+            return path, candidate
+        parent = os.path.dirname(path)
+        if parent == path:
+            break
+        path = parent
+    return _engagement_manifest(cwd)
+
+
+def _engagement_manifest(cwd):
+    """(engagement_root, manifest_path) for a cwd under
+    `$TEZGAH_ENGAGEMENTS_ROOT/<slug>`, or (None, None): no root set, a cwd
+    outside it or at the root itself, or no manifest in that engagement. One
+    `isfile`, so an unset or unrelated root costs nothing."""
+    root = os.environ.get(ENGAGEMENTS_ENV)
+    if not root:
+        return None, None
+    root = os.path.realpath(os.path.expanduser(root))
+    rel = os.path.relpath(os.path.realpath(cwd), root)
+    slug = rel.split(os.sep, 1)[0]
+    if slug in (os.curdir, os.pardir):
+        return None, None
+    engagement = os.path.join(root, slug)
+    candidate = os.path.join(engagement, EXT_DIR, MANIFEST_NAME)
+    return (engagement, candidate) if os.path.isfile(candidate) else (None, None)
+
+
+def given_manifest(manifest_file):
+    """(repo_root, manifest_path) for a manifest named outright (the renderer's
+    `--manifest`): the repo root is the directory holding its `.tezgah`, else
+    the directory holding the file. (None, None) when it is no file."""
+    path = os.path.realpath(os.path.expanduser(manifest_file))
+    if not os.path.isfile(path):
+        return None, None
+    folder = os.path.dirname(path)
+    return (os.path.dirname(folder) if os.path.basename(folder) == EXT_DIR
+            else folder), path
+
+
+def _load_manifest(manifest_file):
+    """The manifest dict, or None when it is missing or malformed. A bad manifest
+    costs the extension, never the line."""
+    try:
+        with open(manifest_file, encoding="utf-8") as fh:
+            data = json.load(fh)
+    except (OSError, ValueError, UnicodeDecodeError):
+        return None
+    return data if isinstance(data, dict) else None
+
+
+def load(cwd, manifest_file=None):
+    """(repo_root, manifest dict) for `cwd` - or for `manifest_file` when one
+    is named - else (None, None). Never raises: every key reader below goes
+    through it, and a reader that raised would cost its caller the call."""
+    try:
+        repo_root, found = (given_manifest(manifest_file) if manifest_file
+                            else manifest_path(cwd))
+    except (OSError, ValueError, TypeError):
+        return None, None
+    manifest = _load_manifest(found) if found else None
+    return (repo_root, manifest) if manifest else (None, None)
+
+
+def repo_checks(cwd):
+    """[argv prefix] the manifest's `checks` declares, each a list of words.
+
+    An entry is a string split the shell's way (`"bin/audit --check"` is two
+    words) or a list of strings; anything else - a number, an empty string, a
+    line shlex rejects - is dropped alone. The first word is normalised
+    (`./bin/x` and `bin/x` are one prefix), the way a command word is read. A
+    lone NO_OP_PROGRAMS word (`echo`, `/bin/true`) is dropped too."""
+    _root, manifest = load(cwd)
+    raw = (manifest or {}).get("checks")
+    if not isinstance(raw, list):
+        return []
+    out = []
+    for entry in raw[:MAX_CHECKS]:
+        try:
+            words = (shlex.split(entry) if isinstance(entry, str)
+                     else list(entry) if isinstance(entry, list) else [])
+        except ValueError:
+            continue
+        if words and all(isinstance(w, str) and w for w in words) and not (
+                len(words) == 1 and os.path.basename(words[0]) in NO_OP_PROGRAMS):
+            out.append([os.path.normpath(words[0])] + words[1:])
+    return out
+
+
+def untrusted_programs(cwd):
+    """frozenset of the program names the manifest's `untrusted` declares: one
+    basename each, case kept (a shell's program word is case-sensitive). A
+    non-list, or an entry that is no name, is dropped."""
+    _root, manifest = load(cwd)
+    raw = (manifest or {}).get("untrusted")
+    if not isinstance(raw, list):
+        return frozenset()
+    return frozenset(os.path.basename(p) for p in raw[:MAX_UNTRUSTED]
+                     if isinstance(p, str) and NAME.match(os.path.basename(p)))
+
+
+def repo_agent(brief, cwd):
+    """(agent, pattern) of the first `agents` entry whose `match` regex finds
+    the brief, or None. An entry with no string pattern, a pattern longer than
+    MAX_PATTERN or one `re` rejects, or an agent that is no one-word name, is
+    dropped alone; the brief is read on its first MAX_BRIEF characters."""
+    _root, manifest = load(cwd)
+    raw = (manifest or {}).get("agents")
+    if not isinstance(raw, list):
+        return None
+    text = str(brief or "")[:MAX_BRIEF]
+    for entry in raw[:MAX_AGENTS]:
+        if not isinstance(entry, dict):
+            continue
+        pattern, agent = entry.get("match"), entry.get("agent")
+        if not (isinstance(pattern, str) and pattern and len(pattern) <= MAX_PATTERN
+                and isinstance(agent, str) and NAME.match(agent)):
+            continue
+        try:
+            if re.search(pattern, text):
+                return agent, pattern
+        except re.error:
+            continue
+    return None
+
+
+def override(cwd, manifest_file=None):
+    """True when the manifest's `statusline.override` is literally `true`: the
+    project owns its status line and points it at tezgah's renderer, so that
+    renderer is the only line. Any other value is False."""
+    _root, manifest = load(cwd, manifest_file)
+    spec = (manifest or {}).get("statusline")
+    return isinstance(spec, dict) and spec.get("override") is True
+
+
+def _valid_segment(seg, index):
+    """One segment dict, normalized, or None when it cannot be rendered.
+
+    `text` and `state` are the only required fields: text is what the reader
+    sees, state is what the glyph and the color mean. A missing or unknown state
+    is `info` (the state-less mark), never a guess at "on"."""
+    if not isinstance(seg, dict):
+        return None
+    text = seg.get("text")
+    if not isinstance(text, str) or not text.strip():
+        return None
+    state = seg.get("state")
+    if state not in STATES:
+        state = "info"
+    key = seg.get("key")
+    if not isinstance(key, str) or not key:
+        key = "ext-%d" % index
+    glyph = seg.get("glyph")
+    if not isinstance(glyph, str):
+        glyph = ""
+    return {"key": key, "state": state, "glyph": glyph, "text": text.strip()}
+
+
+def run_manifest(repo_root, manifest, envelope=None):
+    """The segments a manifest's command returns, or [] on any failure.
+
+    Failure is a closed door, not a raised one: a missing command, a non-zero
+    exit, a timeout, malformed JSON, a wrong shape and an over-budget list all
+    answer [] - the line loses this extension's chips and keeps its own.
+
+    `envelope` reaches the child as one JSON object on stdin. A command that
+    never reads its stdin sees the same closed, empty input it always did."""
+    spec = manifest.get("statusline")
+    if not isinstance(spec, dict):
+        return []
+    command = spec.get("command")
+    if (not isinstance(command, list) or not command
+            or not all(isinstance(a, str) for a in command)):
+        return []
+    timeout_ms = spec.get("timeout_ms", DEFAULT_TIMEOUT_MS)
+    if not isinstance(timeout_ms, (int, float)) or timeout_ms <= 0:
+        timeout_ms = DEFAULT_TIMEOUT_MS
+    argv = list(command)
+    # A bare program name resolves on PATH; a path is relative to the repo root,
+    # so a manifest travels with its clone instead of naming an absolute path.
+    if os.sep in argv[0] and not os.path.isabs(argv[0]):
+        argv[0] = os.path.join(repo_root, argv[0])
+    if os.sep in argv[0] and not os.path.isfile(argv[0]):
+        return []
+    try:
+        proc = subprocess.run(argv, cwd=repo_root, capture_output=True,
+                              text=True, encoding="utf-8",
+                              input=(json.dumps(envelope) if envelope is not None
+                                     else ""),
+                              timeout=timeout_ms / 1000.0)
+    except (OSError, subprocess.SubprocessError, ValueError):
+        return []
+    if proc.returncode != 0:
+        return []
+    try:
+        payload = json.loads(proc.stdout or "")
+    except ValueError:
+        return []
+    if not isinstance(payload, dict):
+        return []
+    raw = payload.get("segments")
+    if not isinstance(raw, list):
+        return []
+    out = []
+    for index, seg in enumerate(raw[:MAX_SEGMENTS]):
+        clean = _valid_segment(seg, index)
+        if clean is not None:
+            out.append(clean)
+    return out
+
+
+def segments(cwd, session_id=None, manifest_file=None):
+    """Every extension segment for a render of `cwd`, in manifest order.
+    `manifest_file` names the manifest outright (the renderer's `--manifest`)
+    instead of discovering it from `cwd`.
+
+    One manifest today (the first found); the return is a list so a second
+    extension is a discovery change, not a caller change."""
+    repo_root, manifest = load(cwd, manifest_file)
+    if not manifest:
+        return []
+    return run_manifest(repo_root, manifest,
+                        call_envelope(repo_root, cwd, session_id))