agent-first-data 0.33.0__tar.gz → 0.35.0__tar.gz

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-first-data
-Version: 0.33.0
+Version: 0.35.0
 Summary: A naming convention that lets AI agents understand your data without being told what it means, plus a CLI and library for reading Markdown structure and safely editing structured JSON, TOML, YAML, dotenv, and INI documents.
 License-Expression: MIT
 Project-URL: Repository, https://github.com/agentfirstkit/agent-first-data
@@ -33,11 +33,11 @@ Useful names use Python casing: `render` (the single value x format x options ->
 ## Behavior Notes
 
 - Default redaction replaces every `_secret` or configured secret-name subtree with `***`, including objects and arrays.
-- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed and internal whitespace redacts the whole field.
+- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed, and internal whitespace or a `://` that is not the start of a clean URL redacts the whole field, where whitespace is the explicit Unicode `White_Space` set the spec enumerates, not the language's native predicate.
 - `OutputOptions.url_names` applies URL treatment to exact legacy field names
   and recurses through collections. `redact_urls_in_text` is explicit and scans
   only complete scheme URLs, never arbitrary prose secrets.
-- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON.
+- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON; it quotes a value only when it contains ASCII space, tab, newline, carriage return, form feed, vertical tab, NBSP, `=`, `"`, or `\`, and escapes `\`, `"`, newline, carriage return, tab, form feed, and vertical tab inside the quotes.
 - Logging records use `kind:"log"` with a nested `log` payload and a separate `level` field, so error-level logs are not terminal protocol errors.
 - `build_cli_error(message, hint?)` returns a strict-ready CLI error with `error.retryable:false` and `trace:{}`.
 - Do not add a raw version/help pre-parser. Until the Python `CliSpec` compiler lands, applications should keep lifecycle parsing in their own parser rather than claim AFDATA closed-world help compatibility.

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/README.md

@@ -24,11 +24,11 @@ Useful names use Python casing: `render` (the single value x format x options ->
 ## Behavior Notes
 
 - Default redaction replaces every `_secret` or configured secret-name subtree with `***`, including objects and arrays.
-- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed and internal whitespace redacts the whole field.
+- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed, and internal whitespace or a `://` that is not the start of a clean URL redacts the whole field, where whitespace is the explicit Unicode `White_Space` set the spec enumerates, not the language's native predicate.
 - `OutputOptions.url_names` applies URL treatment to exact legacy field names
   and recurses through collections. `redact_urls_in_text` is explicit and scans
   only complete scheme URLs, never arbitrary prose secrets.
-- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON.
+- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON; it quotes a value only when it contains ASCII space, tab, newline, carriage return, form feed, vertical tab, NBSP, `=`, `"`, or `\`, and escapes `\`, `"`, newline, carriage return, tab, form feed, and vertical tab inside the quotes.
 - Logging records use `kind:"log"` with a nested `log` payload and a separate `level` field, so error-level logs are not terminal protocol errors.
 - `build_cli_error(message, hint?)` returns a strict-ready CLI error with `error.retryable:false` and `trace:{}`.
 - Do not add a raw version/help pre-parser. Until the Python `CliSpec` compiler lands, applications should keep lifecycle parsing in their own parser rather than claim AFDATA closed-world help compatibility.

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/agent_first_data/assets/cli-spec-v1.schema.json

@@ -100,6 +100,10 @@
         "about": {
           "$ref": "#/$defs/nonEmpty"
         },
+        "reference_note": {
+          "description": "Prose for the generated CLI reference, rendered after this command's argument table. For the paragraph that belongs to no single argument — where a default is read from, what two options mean together — which would otherwise be hand-written into a file whose header says it is generated.",
+          "$ref": "#/$defs/nonEmpty"
+        },
         "arguments": {
           "type": "array",
           "items": {

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/agent_first_data/cli.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import enum
+import os
 import sys
 from typing import Any, Iterator
 
@@ -298,10 +299,27 @@ class CliEmitter:
         # mode (no diagnostic sink) keeps every event on the single writer.
         use_diagnostic = kind != "result" and self._diagnostic is not None
         sink = self._diagnostic if use_diagnostic else self._writer
-        sink.write(render(envelope, self._format, options=self._output_options) + "\n")
-        flush = getattr(sink, "flush", None)
-        if flush is not None:
-            flush()
+        try:
+            sink.write(render(envelope, self._format, options=self._output_options) + "\n")
+            flush = getattr(sink, "flush", None)
+            if flush is not None:
+                flush()
+        except BrokenPipeError:
+            # A failed flush leaves buffered bytes for interpreter shutdown to
+            # retry. Redirect the closed descriptor before finish returns 0.
+            fileno = getattr(sink, "fileno", None)
+            if fileno is not None:
+                try:
+                    descriptor = fileno()
+                except (OSError, ValueError):
+                    descriptor = None
+                if descriptor is not None:
+                    null_descriptor = os.open(os.devnull, os.O_WRONLY)
+                    try:
+                        os.dup2(null_descriptor, descriptor)
+                    finally:
+                        os.close(null_descriptor)
+            raise
         if kind in ("result", "error"):
             self._terminal_emitted = True
 

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/agent_first_data/format.py

@@ -1273,17 +1273,19 @@ def _redact_url_field_value(s: str, context: _RedactionContext) -> str:
     redacted = _redact_url_in_str(s, context)
     if redacted is not None:
         return redacted
-    trimmed = s.strip()
+    trimmed = s.strip("".join(_SPAN_TERMINATOR_WHITESPACE))
     if trimmed != s:
         redacted = _redact_url_in_str(trimmed, context)
         if redacted is not None:
             return redacted
     # Fail closed: a _url value we could not parse as a clean scheme-prefixed
-    # URL, yet which carries a credential sigil ('@' userinfo) or internal
-    # whitespace, is redacted wholesale rather than passed through. A schemeless
-    # connection string like user:pass@host/db has no scheme anchor for the
-    # surgical span logic above, so blanket redaction is the safe default.
-    if any(c.isspace() for c in s) or "@" in s:
+    # URL, yet which carries a credential sigil ('@' userinfo), internal
+    # whitespace, or a '://' scheme separator, is redacted wholesale rather than
+    # passed through. A schemeless connection string like user:pass@host/db
+    # has no scheme anchor for the surgical span logic above, and a URL behind
+    # a non-whitespace prefix (a BOM, a zero-width space, a label) has one the
+    # span logic cannot reach, so blanket redaction is the safe default.
+    if any(_is_span_terminator_whitespace(c) for c in s) or "@" in s or "://" in s:
         return REDACTED_MARKER
     return s
 
@@ -1840,11 +1842,10 @@ def _yaml_key(key: str) -> str:
 def _is_ambiguous_yaml_key(key: str) -> bool:
     if key.lower() in {"true", "false", "null", "~", ".nan", ".inf", "+.inf", "-.inf"}:
         return True
-    try:
-        float(key)
-        return True
-    except ValueError:
-        return False
+    unsigned = key[1:] if key.startswith(("+", "-")) else key
+    return unsigned.lower() in {"inf", "infinity", "nan"} or re.fullmatch(
+        r"[-+]?(?:[0-9]+\.?[0-9]*|\.[0-9]+)(?:[eE][-+]?[0-9]+)?", key
+    ) is not None
 
 
 def _yaml_scalar(value: Any) -> str:
@@ -1945,7 +1946,7 @@ def _plain_scalar_raw(value: Any) -> str:
 def _quote_logfmt_value(value: str) -> str:
     if value == "":
         return ""
-    needs_quote = any(c.isspace() or c in '="\\' for c in value)
+    needs_quote = any(c in ' \t\n\r\f\v\u00a0="\\' for c in value)
     if not needs_quote:
         return value
     escaped = (

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/agent_first_data.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-first-data
-Version: 0.33.0
+Version: 0.35.0
 Summary: A naming convention that lets AI agents understand your data without being told what it means, plus a CLI and library for reading Markdown structure and safely editing structured JSON, TOML, YAML, dotenv, and INI documents.
 License-Expression: MIT
 Project-URL: Repository, https://github.com/agentfirstkit/agent-first-data
@@ -33,11 +33,11 @@ Useful names use Python casing: `render` (the single value x format x options ->
 ## Behavior Notes
 
 - Default redaction replaces every `_secret` or configured secret-name subtree with `***`, including objects and arrays.
-- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed and internal whitespace redacts the whole field.
+- `_url` fields scrub userinfo passwords and secret-named query parameters; surrounding whitespace is trimmed, and internal whitespace or a `://` that is not the start of a clean URL redacts the whole field, where whitespace is the explicit Unicode `White_Space` set the spec enumerates, not the language's native predicate.
 - `OutputOptions.url_names` applies URL treatment to exact legacy field names
   and recurses through collections. `redact_urls_in_text` is explicit and scans
   only complete scheme URLs, never arbitrary prose secrets.
-- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON.
+- YAML keeps original keys and values (structure-preserving, like JSON), sorting keys by UTF-16 code unit order and quoting/escaping unsafe keys and string scalars. Plain strips formatting suffixes, formats values, sorts the same way, and renders nested objects/arrays as canonical JSON; it quotes a value only when it contains ASCII space, tab, newline, carriage return, form feed, vertical tab, NBSP, `=`, `"`, or `\`, and escapes `\`, `"`, newline, carriage return, tab, form feed, and vertical tab inside the quotes.
 - Logging records use `kind:"log"` with a nested `log` payload and a separate `level` field, so error-level logs are not terminal protocol errors.
 - `build_cli_error(message, hint?)` returns a strict-ready CLI error with `error.retryable:false` and `trace:{}`.
 - Do not add a raw version/help pre-parser. Until the Python `CliSpec` compiler lands, applications should keep lifecycle parsing in their own parser rather than claim AFDATA closed-world help compatibility.

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "agent-first-data"
-version = "0.33.0"
+version = "0.35.0"
 description = "A naming convention that lets AI agents understand your data without being told what it means, plus a CLI and library for reading Markdown structure and safely editing structured JSON, TOML, YAML, dotenv, and INI documents."
 license = "MIT"
 readme = "README.md"

{agent_first_data-0.33.0 → agent_first_data-0.35.0}/tests/test_cli.py

@@ -1,5 +1,6 @@
 """Tests for agent_first_data CLI helpers."""
 import json
+import sys
 
 import pytest
 from io import StringIO
@@ -500,3 +501,27 @@ def test_cli_render_version_renders_json():
     assert '"code":"version"' in out
     assert '"name":"agent-cli"' in out
     assert '"version":"1.2.3"' in out
+
+
+@pytest.mark.skipif(
+    sys.platform == "win32",
+    reason="closed-pipe writes surface as OSError(EINVAL) on Windows, not "
+    "BrokenPipeError; the reader-closed contract is asserted on POSIX only",
+)
+def test_real_stdout_closed_pipe_has_clean_process_exit():
+    import os
+    import subprocess
+    import sys
+
+    reader, writer = os.pipe()
+    os.close(reader)
+    try:
+        child = subprocess.run(
+            [sys.executable, "-c", "from agent_first_data import CliEmitter, OutputFormat; "
+             "raise SystemExit(CliEmitter.finite(OutputFormat.JSON).finish_result({'ok': True}))"],
+            stdout=writer, stderr=subprocess.PIPE, timeout=10,
+        )
+    finally:
+        os.close(writer)
+    assert child.returncode == 0, child.stderr
+    assert child.stderr == b""