Merge pull request #3 from grimlor/fix/windows-utf8-encoding

fix: add explicit encoding="utf-8" to all text file I/O

Author: grimlor

.github/workflows/ci.yml

@@ -15,7 +15,7 @@ jobs:
     if: github.event_name == 'pull_request'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
 
@@ -61,10 +61,10 @@ jobs:
         python-version: ["3.11", "3.12", "3.13"]
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
 
       - name: Install uv
-        uses: astral-sh/setup-uv@v4
+        uses: astral-sh/setup-uv@v8.1.0
 
       - name: Set up Python ${{ matrix.python-version }}
         run: uv python install ${{ matrix.python-version }}

.github/workflows/release.yml

@@ -16,13 +16,13 @@ jobs:
     if: "!startsWith(github.event.head_commit.message, 'chore(release):')"
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0          # full history needed for commit analysis
           fetch-tags: true
 
       - name: Install uv
-        uses: astral-sh/setup-uv@v4
+        uses: astral-sh/setup-uv@v8.1.0
 
       - name: Set up Python
         run: uv python install 3.13

.gitignore

@@ -38,6 +38,7 @@ Thumbs.db
 
 # Coverage
 .coverage
+.coverage.*
 coverage.json
 htmlcov/
 

config/settings.toml

@@ -110,6 +110,9 @@ eval_history_path = "data/eval_history.jsonl"
 [chroma]
 persist_dir = "./data/chroma_db"
 distance_metric = "cosine"
+# Flush HNSW index to disk after every N writes.
+# 1 = immediate durability (required for Windows correctness).
+sync_threshold = 1
 
 [security]
 screen_prompt = """\

pyproject.toml

@@ -45,6 +45,23 @@ dev = [
 [project.scripts]
 jobsearch-rag = "jobsearch_rag.__main__:main"
 
+# ---------------------------------------------------------------------------
+# uv
+# ---------------------------------------------------------------------------
+[tool.uv]
+# Ensure the lock file resolves wheels for every CI platform.
+# Without this, uv may pick a version that lacks wheels for some targets
+# (e.g. onnxruntime 1.26.0 dropped macOS ARM64 wheels for Python 3.12).
+required-environments = [
+    "sys_platform == 'darwin' and platform_machine == 'arm64'",
+    "sys_platform == 'linux' and platform_machine == 'x86_64'",
+    "sys_platform == 'win32' and platform_machine == 'AMD64'",
+]
+
+# onnxruntime 1.26.0 has no macOS ARM64 wheel for Python 3.12.
+# Cap until upstream publishes the missing wheel.
+constraint-dependencies = ["onnxruntime<1.26"]
+
 # ---------------------------------------------------------------------------
 # Taskipy
 # ---------------------------------------------------------------------------
@@ -139,7 +156,7 @@ asyncio_mode = "auto"
 python_files = ["*_test.py", "test_*.py"]
 python_classes = ["Test*"]
 python_functions = ["test_*"]
-addopts = "-v --strict-markers -m 'not integration and not live'"
+addopts = "-v --strict-markers -m 'not integration and not live' -n auto --cov-fail-under=100"
 log_cli = true
 log_cli_level = "INFO"
 log_cli_format = "%(asctime)s %(levelname)-8s %(name)s — %(message)s"

src/jobsearch_rag/adapters/session.py

@@ -417,7 +417,7 @@ async def save_storage_state(self) -> Path:
         path = self._config.storage_state_path
         path.parent.mkdir(parents=True, exist_ok=True)
         state = await self._context.storage_state()
-        path.write_text(json.dumps(state, indent=2))
+        path.write_text(json.dumps(state, indent=2), encoding="utf-8")
         logger.info("Session state saved to %s", path)
         return path
 

src/jobsearch_rag/cli.py

@@ -59,7 +59,7 @@ def _read_jd_text(
     jd_path = jd_dir / filename
     if not jd_path.exists():
         return ""
-    content = jd_path.read_text()
+    content = jd_path.read_text(encoding="utf-8")
     marker = "## Job Description\n"
     idx = content.find(marker)
     if idx == -1:
@@ -79,7 +79,7 @@ def _load_prior_csv(csv_path: Path, *, max_full_text_chars: int) -> list[RankedL
         return []
 
     listings: list[RankedListing] = []
-    with open(csv_path) as f:
+    with open(csv_path, encoding="utf-8") as f:
         reader = csv_mod.DictReader(f)
         for row in reader:
             external_id = row.get("external_id", "")
@@ -218,6 +218,7 @@ def handle_index(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     indexer = Indexer(store=store, embedder=embedder)
 
@@ -376,6 +377,7 @@ def handle_decide(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     recorder = DecisionRecorder(
         store=store, embedder=embedder, decisions_dir=settings.output.decisions_dir
@@ -430,6 +432,7 @@ def handle_decisions(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     recorder = DecisionRecorder(
         store=store, embedder=embedder, decisions_dir=settings.output.decisions_dir
@@ -494,6 +497,7 @@ def handle_review(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     recorder = DecisionRecorder(
         store=store, embedder=embedder, decisions_dir=settings.output.decisions_dir
@@ -509,7 +513,7 @@ def handle_review(args: argparse.Namespace) -> None:
     ranked_listings: list[RankedListing] = []
     jd_dir = settings.output.jd_dir
 
-    with open(csv_path) as f:
+    with open(csv_path, encoding="utf-8") as f:
         reader = csv_mod.DictReader(f)
         for _i, row in enumerate(reader, 1):
             title = row.get("title", "")
@@ -618,6 +622,7 @@ def handle_rescore(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     scorer = Scorer(
         store=store,
@@ -731,6 +736,7 @@ def _build_eval_stack(
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
     scorer = Scorer(
         store=store,
@@ -865,9 +871,9 @@ def handle_export(args: argparse.Namespace) -> None:
 
     fmt = args.format
     if fmt == "markdown" and md_path.exists():
-        print(md_path.read_text())
+        print(md_path.read_text(encoding="utf-8"))
     elif fmt == "csv" and csv_path.exists():
-        print(csv_path.read_text())
+        print(csv_path.read_text(encoding="utf-8"))
     else:
         print(f"No {fmt} export found in {out_dir}. Run 'search' to generate results.")
 
@@ -882,6 +888,7 @@ def handle_reset(args: argparse.Namespace) -> None:
     store = VectorStore(
         persist_dir=settings.chroma.persist_dir,
         distance_metric=settings.chroma.distance_metric,
+        sync_threshold=settings.chroma.sync_threshold,
     )
 
     collections = [args.collection] if args.collection else list(_COLLECTIONS)

src/jobsearch_rag/config.py

@@ -136,6 +136,7 @@ class ChromaConfig:
 
     persist_dir: str
     distance_metric: str
+    sync_threshold: int
 
 
 @dataclass
@@ -461,8 +462,17 @@ def _validate(data: _TOMLDict, filepath: Path) -> Settings:
     chroma = ChromaConfig(
         persist_dir=str(_require_field(chroma_data, "persist_dir", "chroma")),
         distance_metric=str(_require_field(chroma_data, "distance_metric", "chroma")),
+        sync_threshold=int(_require_field(chroma_data, "sync_threshold", "chroma")),
     )
 
+    # Validate sync_threshold
+    if chroma.sync_threshold < 1:
+        raise ActionableError.validation(
+            field_name="chroma.sync_threshold",
+            reason=f"is {chroma.sync_threshold} — must be >= 1",
+            suggestion="Set [chroma].sync_threshold to a positive integer (e.g. 1)",
+        )
+
     # Validate distance_metric
     valid_metrics = {"cosine", "l2", "ip"}
     if chroma.distance_metric not in valid_metrics:

src/jobsearch_rag/export/csv_export.py

@@ -54,7 +54,7 @@ def export(
         qualified = [r for r in listings if not (r.scores.disqualified and r.final_score == 0.0)]
         qualified.sort(key=lambda r: r.final_score, reverse=True)
 
-        with open(output_path, "w", newline="") as f:
+        with open(output_path, "w", newline="", encoding="utf-8") as f:
             writer = csv.writer(f)
             writer.writerow(_COLUMNS)
 

src/jobsearch_rag/export/jd_files.py

@@ -92,7 +92,7 @@ def export(
             filepath = out / filename
 
             content = self._render(r)
-            filepath.write_text(content)
+            filepath.write_text(content, encoding="utf-8")
             paths.append(filepath)
 
         logger.info("Exported %d JD files to %s", len(paths), out)