file-observer 1.2.2__tar.gz → 1.2.3__tar.gz

{file_observer-1.2.2/src/file_observer.egg-info → file_observer-1.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: file-observer
-Version: 1.2.2
+Version: 1.2.3
 Summary: Not a file watcher — a one-shot, read-only scan that emits a deterministic JSON manifest of what's in a directory before you ingest it.
 Author-email: Russell Pfister <russalo@russalo.com>
 License: AGPL-3.0-or-later

{file_observer-1.2.2 → file_observer-1.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "file-observer"
-version = "1.2.2"
+version = "1.2.3"
 description = "Not a file watcher — a one-shot, read-only scan that emits a deterministic JSON manifest of what's in a directory before you ingest it."
 readme = "docs/README.md"
 requires-python = ">=3.12"

{file_observer-1.2.2 → file_observer-1.2.3}/src/file_observer/scanner.py

@@ -71,8 +71,8 @@ except ImportError:
     _defusedxml_available = False
 
 
-SCANNER_VERSION = "1.2.2"
-LOGIC_VERSION = "1.1.2"
+SCANNER_VERSION = "1.2.3"
+LOGIC_VERSION = "1.1.3"
 SCHEMA_VERSION = "1.2"
 
 
@@ -546,7 +546,7 @@ CHATLOG_TOOL = "chatlog_signals"
 # any detection regex or extraction algorithm requires bumping METHOD_VERSION
 # and updating the rules definition string.
 CHATLOG_VECTOR_ID = "chatlog"
-CHATLOG_METHOD_VERSION = 6  # v1.2.2: prose Rule 1 requires a recurring speaker; expanded stop-list (FP fixes)
+CHATLOG_METHOD_VERSION = 7  # v1.2.3: + Question/Answer stop-list (FAQ FP stopgap)
 CHATLOG_RULES_DEFINITION = (
     "detect:speaker_label_re(distinct>=2,total>=3,recurring>=1,stop_list),"
     "h3_header_re(5+)|section_divider_re(3+)[require speaker_cosignal_distinct>=2],"
@@ -556,10 +556,10 @@ CHATLOG_RULES_DEFINITION = (
     "turn_char_stats,speaker_turn_counts,speaker_turn_chars,alternation,"
     "reference_tokens(at_mentions,wiki_links,code_fence_blocks,url_count),"
     "top_capitalized_tokens(freq>=3,top20),vocabulary_size_estimate;"
-    "stop_list:Allow,Arguments,Authorization,Bcc,CAUTION,Cc,Command,Commands,"
+    "stop_list:ANSWER,Allow,Answer,Arguments,Authorization,Bcc,CAUTION,Cc,Command,Commands,"
     "Copyright,Date,Description,Disallow,Distribution,Documentation,Error,Example,"
     "Examples,FIXME,Format,From,IMPORTANT,License,Lines,Message,NOTE,Newsgroups,"
-    "Note,Options,Organization,Parameters,Password,Path,References,Result,Returns,"
+    "Note,Options,Organization,Parameters,Password,Path,QUESTION,Question,References,Result,Returns,"
     "Sender,Subject,Summary,Synopsis,TIP,TODO,To,Usage,Version,Warning"
 )
 CHATLOG_STATIC_TUNING = {
@@ -624,6 +624,13 @@ CHATLOG_SPEAKER_STOP_LIST: set[str] = {
     "Version", "Usage", "Options", "Command", "Commands", "Format", "Synopsis",
     "Description", "Arguments", "Parameters", "Returns", "Summary",
     "Authorization", "Documentation", "License", "Copyright", "Password",
+    # v1.2.3: FAQ labels — `Question:`/`Answer:` recur like speakers but aren't
+    # (stopgap for the most common recurring-Key:value FP; the real fix is a
+    # non-count signal — see scratch/review/v1_2_2_fp_findings.md). Deliberately
+    # NOT adding single letters `Q`/`A`: `A:`/`B:` is legitimate anonymized dialogue.
+    # Membership is case-sensitive and the label regex matches all-caps, so the
+    # common ALL-CAPS FAQ style needs its own entries (PR #30 review).
+    "Question", "Answer", "QUESTION", "ANSWER",
 }
 
 # v0.9: Reference tokens vector identity constants.

{file_observer-1.2.2 → file_observer-1.2.3/src/file_observer.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: file-observer
-Version: 1.2.2
+Version: 1.2.3
 Summary: Not a file watcher — a one-shot, read-only scan that emits a deterministic JSON manifest of what's in a directory before you ingest it.
 Author-email: Russell Pfister <russalo@russalo.com>
 License: AGPL-3.0-or-later

{file_observer-1.2.2 → file_observer-1.2.3}/src/file_observer.egg-info/SOURCES.txt

@@ -20,4 +20,5 @@ tests/test_unit.py
 tests/test_v1_1.py
 tests/test_v1_2.py
 tests/test_v1_2_1.py
-tests/test_v1_2_2.py
+tests/test_v1_2_2.py
+tests/test_v1_2_3.py

{file_observer-1.2.2 → file_observer-1.2.3}/tests/test_packaging.py

@@ -17,7 +17,7 @@ def test_canonical_top_level_api():
 
     import file_observer
 
-    assert file_observer.__version__ == "1.2.2"
+    assert file_observer.__version__ == "1.2.3"
     assert "Scanner" in file_observer.__all__
 
 
@@ -49,9 +49,9 @@ def test_canonical_submodule_constants_unchanged():
     """Constants and the manifest field they feed stay stable across the 1.0.x → 1.1 line (the 1.0.1 import-package rename left them intact)."""
     from file_observer.scanner import SCANNER_VERSION, LOGIC_VERSION, SCHEMA_VERSION
 
-    assert SCANNER_VERSION == "1.2.2"
+    assert SCANNER_VERSION == "1.2.3"
     assert SCHEMA_VERSION == "1.2"  # additive minor bump (v1.1 corpus-intelligence fields)
-    assert LOGIC_VERSION == "1.1.2"  # v1.2.2: prose chatlog detection tightened (recurrence + stop-list)
+    assert LOGIC_VERSION == "1.1.3"  # v1.2.3: + Question/Answer stop-list (FAQ FP stopgap)
 
 
 def test_legacy_scanner_import_warns():

{file_observer-1.2.2 → file_observer-1.2.3}/tests/test_unit.py

@@ -811,8 +811,8 @@ class TestScanContext:
         (tmp_path / "a.txt").write_text("hello")
         manifest = Scanner(source_dir=tmp_path).scan()
         ctx = manifest.context
-        assert ctx.scanner_version == "1.2.2"
-        assert ctx.logic_version == "1.1.2"
+        assert ctx.scanner_version == "1.2.3"
+        assert ctx.logic_version == "1.1.3"
         assert ctx.python_version  # non-empty
         assert ctx.platform  # non-empty
 
@@ -850,7 +850,7 @@ class TestScanContext:
         manifest = Scanner(source_dir=tmp_path).scan()
         data = json_mod.loads(manifest_to_json(manifest))
         assert "context" in data
-        assert data["context"]["scanner_version"] == "1.2.2"
+        assert data["context"]["scanner_version"] == "1.2.3"
 
 
 # ---------------------------------------------------------------------------
@@ -2042,8 +2042,8 @@ class TestSemanticToolNames:
 
     def test_version_is_current(self) -> None:
         from file_observer.scanner import SCANNER_VERSION, LOGIC_VERSION
-        assert SCANNER_VERSION == "1.2.2"
-        assert LOGIC_VERSION == "1.1.2"
+        assert SCANNER_VERSION == "1.2.3"
+        assert LOGIC_VERSION == "1.1.3"
 
 
 # ---------------------------------------------------------------------------
@@ -3530,4 +3530,4 @@ class TestMarkdownReport:
         from file_observer.scanner import manifest_to_markdown
         (tmp_path / "a.txt").write_text("hello")
         md = manifest_to_markdown(Scanner(source_dir=tmp_path).scan())
-        assert "1.2.2" in md
+        assert "1.2.3" in md

file_observer-1.2.3/tests/test_v1_2_3.py

@@ -0,0 +1,41 @@
+"""v1.2.3 — FAQ false-positive stopgap.
+
+Pro + flash both flagged that recurring `Question:`/`Answer:` labels (a FAQ doc)
+satisfy the v1.2.2 prose rule (≥2 distinct, ≥3 total, ≥1 recurring) and
+false-positive as a chatlog. v1.2.3 adds `Question`/`Answer` to the stop-list.
+
+This is a STOPGAP for the most common single case — NOT the real fix. The root
+issue (prose `Key:value` is ambiguous with dialogue) needs a non-count signal;
+see scratch/review/v1_2_2_fp_findings.md. We deliberately do NOT stop-list single
+letters like `Q`/`A` — `A:`/`B:` is a legitimate anonymized-dialogue pattern.
+
+Falsify-first: the FAQ test FAILS on v1.2.2, passes after the stop-list add.
+"""
+
+from file_observer.scanner import Scanner, ScannerConfig
+
+
+def _is_chatlog(tmp_path, name, content):
+    (tmp_path / name).write_text(content)
+    return Scanner(source_dir=tmp_path, config=ScannerConfig()).scan().files[0].is_chatlog
+
+
+def test_faq_question_answer_not_flagged(tmp_path):
+    c = ("Question: How do I install?\nAnswer: Use pip.\n"
+         "Question: Where are the docs?\nAnswer: In docs/.\n")
+    assert _is_chatlog(tmp_path, "FAQ.md", c) is False
+
+
+def test_faq_allcaps_not_flagged(tmp_path):
+    # case-sensitive stop-list + the label regex matches all-caps → ALL-CAPS FAQ
+    # style needs explicit coverage (PR #30 review: flash/codex/copilot flagged it).
+    c = ("QUESTION: How do I install?\nANSWER: Use pip.\n"
+         "QUESTION: Where are the docs?\nANSWER: In docs/.\n")
+    assert _is_chatlog(tmp_path, "FAQ.md", c) is False
+
+
+def test_anonymized_ab_dialogue_still_detected(tmp_path):
+    # `A:`/`B:` anonymized speakers are a LEGIT conversation — must stay detected
+    # (guards against over-stop-listing single letters while fixing FAQ).
+    c = "A: hey did you see this\nB: yeah just now\nA: what do you think\n"
+    assert _is_chatlog(tmp_path, "dialogue.txt", c) is True