scriptoria 0.6.3__tar.gz → 0.6.4__tar.gz

{scriptoria-0.6.3 → scriptoria-0.6.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptoria
-Version: 0.6.3
+Version: 0.6.4
 Summary: Deterministic scriptorium-keeper (the `scrip` CLI): staleness, provenance integrity, and fact queries for an agent-compiled knowledge base
 Project-URL: Homepage, https://github.com/coredipper/scriptorium
 Project-URL: Changelog, https://github.com/coredipper/scriptorium/blob/main/CHANGELOG.md

{scriptoria-0.6.3 → scriptoria-0.6.4}/pyproject.toml

@@ -2,7 +2,7 @@
 # Distribution name on PyPI is `scriptoria` (scrip/scriptorium were taken); the
 # CLI command and the import package both remain `scrip`.
 name = "scriptoria"
-version = "0.6.3"
+version = "0.6.4"
 description = "Deterministic scriptorium-keeper (the `scrip` CLI): staleness, provenance integrity, and fact queries for an agent-compiled knowledge base"
 readme = "README.md"
 requires-python = ">=3.10"

{scriptoria-0.6.3 → scriptoria-0.6.4}/src/scrip/__init__.py

@@ -13,7 +13,7 @@ from __future__ import annotations
 
 from pathlib import Path
 
-__version__ = "0.6.3"
+__version__ = "0.6.4"
 
 # --- canonical vault layout ------------------------------------------------
 # ``root`` is the repo/instance root: the directory containing ``vault/``.

{scriptoria-0.6.3 → scriptoria-0.6.4}/src/scrip/blocks.py

@@ -35,7 +35,9 @@ def _is_blank(line: str) -> bool:
 
 def split_blocks(text: str) -> list[dict]:
     """Return a list of ``{"block_id", "span": [start, end], "hash"}``."""
-    # ⚡ Bolt: Single pass block segmentation to avoid intermediate O(N) allocation
+    # Use splitlines(keepends=True): its boundary set (form feed, NEL, U+2028/9,
+    # …) is part of the deterministic segmentation contract — io.StringIO only
+    # splits on \r/\n and would re-hash existing blocks that contain those chars.
     ranges: list[list[int]] = []  # [start, end] per block
     cur: list[int] | None = None
     start = 0

{scriptoria-0.6.3 → scriptoria-0.6.4}/src/scrip/facts.py

@@ -77,7 +77,11 @@ def parse_ndjson(text: str) -> list[dict]:
     """Parse proposed records (one JSON object per line). Malformed input is a
     :class:`DataError` with its line number; an empty input is a usage error."""
     records: list[dict] = []
-    for lineno, raw_line in enumerate(text.splitlines(), start=1):
+    # Split on "\n" only, not str.splitlines(): NDJSON records are newline-
+    # delimited and may legally contain U+2028/U+2029/NEL inside a JSON string,
+    # which splitlines() would wrongly treat as record breaks. trailing \r (from
+    # \r\n) and the trailing empty element from a final newline are dropped below.
+    for lineno, raw_line in enumerate(text.split("\n"), start=1):
         line = raw_line.strip()
         if not line:
             continue

{scriptoria-0.6.3 → scriptoria-0.6.4}/tests/test_blocks.py

@@ -151,3 +151,14 @@ def test_blank_runs_are_boundaries_not_blocks():
     assert len(bs) == 1
     s, e = bs[0]["span"]
     assert "only paragraph" in "\n\nonly paragraph\n\n"[s:e]
+
+
+def test_form_feed_is_a_line_boundary():
+    """Segmentation uses str.splitlines()'s full boundary set (form feed here,
+    also NEL/U+2028/U+2029). Splitting only on \\r/\\n — e.g. via io.StringIO —
+    would merge these into one block and change every affected block's id/hash,
+    breaking the deterministic-segmentation contract (SPEC §7.2)."""
+    text = "# Heading\x0c## Sub\n"
+    spans = [tuple(b["span"]) for b in blocks.split_blocks(text)]
+    # two separate heading blocks, split at the form feed
+    assert spans == [(0, 10), (10, 17)]

scriptoria-0.6.4/tests/test_facts.py

@@ -0,0 +1,34 @@
+import pytest
+from scrip.errors import DataError, UsageError
+
+from scrip import facts
+
+
+def test_parse_ndjson_basic():
+    text = '{"a": 1}\n{"b": 2}\n'
+    assert facts.parse_ndjson(text) == [{"a": 1}, {"b": 2}]
+
+
+def test_parse_ndjson_skips_blank_lines_and_strips_crlf():
+    text = '{"a": 1}\r\n\r\n{"b": 2}\r\n'
+    assert facts.parse_ndjson(text) == [{"a": 1}, {"b": 2}]
+
+
+def test_parse_ndjson_keeps_u2028_inside_json_string():
+    # U+2028 / U+2029 are legal *inside* a JSON string value. str.splitlines()
+    # treats them as line breaks and would split this record into invalid
+    # fragments; splitting on "\n" only keeps the record intact.
+    text = '{"a": "x
y
z"}\n{"b": 2}\n'
+    assert facts.parse_ndjson(text) == [{"a": "x
y
z"}, {"b": 2}]
+
+
+def test_parse_ndjson_reports_line_number_on_bad_json():
+    text = '{"a": 1}\nnot json\n'
+    with pytest.raises(DataError) as exc:
+        facts.parse_ndjson(text)
+    assert "line 2" in str(exc.value)
+
+
+def test_parse_ndjson_empty_is_usage_error():
+    with pytest.raises(UsageError):
+        facts.parse_ndjson("  \n  \n")

{scriptoria-0.6.3 → scriptoria-0.6.4}/uv.lock

@@ -1141,7 +1141,7 @@ wheels = [
 
 [[package]]
 name = "scriptoria"
-version = "0.6.2"
+version = "0.6.4"
 source = { editable = "." }
 dependencies = [
     { name = "duckdb" },