datoon 1.7.0__tar.gz → 1.7.2__tar.gz

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/package-installer-smoke.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup Python
         uses: actions/setup-python@v6

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/pages.yml

@@ -26,7 +26,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup Pages
         id: setup_pages

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/pre-commit.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup uv
         uses: astral-sh/setup-uv@v7

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/publish-mcp.yml

@@ -1,15 +1,18 @@
 name: Publish to MCP Registry
 
-# Publishes server.json to the official MCP Registry after a GitHub Release.
+# Publishes server.json to the official MCP Registry after a release.
 # Ownership is proven via GitHub OIDC (namespace io.github.andrii-su/*) and the
 # `mcp-name: io.github.andrii-su/datoon` token in the PyPI package README.
 #
-# Runs after publish.yml has pushed the tagged version to PyPI, so the registry
-# can validate the package + ownership token for the released version.
+# Keys off the Release workflow's completion (GITHUB_TOKEN-created releases do
+# not fire `release: published`). Runs only when main HEAD carries a new version
+# tag, and waits for that version to appear on PyPI before publishing, so it
+# never races the parallel Publish to PyPI workflow.
 
 on:
-  release:
-    types: [published]
+  workflow_run:
+    workflows: ["Release"]
+    types: [completed]
   workflow_dispatch:
 
 permissions:
@@ -18,16 +21,32 @@ permissions:
 
 jobs:
   publish:
+    if: >-
+      github.event_name == 'workflow_dispatch'
+      || github.event.workflow_run.conclusion == 'success'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
+        with:
+          ref: main
+          fetch-depth: 0
 
-      - name: Resolve release version
+      - name: Resolve version to publish
         id: version
-        run: echo "version=${GITHUB_REF_NAME#v}" >> "$GITHUB_OUTPUT"
+        run: |
+          version="$(git tag --points-at HEAD | sed 's/^v//' | head -1)"
+          if [ -z "$version" ] && [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            version="$(git describe --tags --abbrev=0 | sed 's/^v//')"
+          fi
+          echo "version=${version}" >> "$GITHUB_OUTPUT"
+          if [ -z "$version" ]; then
+            echo "No release tag at HEAD - nothing to publish."
+          else
+            echo "Publishing datoon ${version} to the MCP Registry."
+          fi
 
-      - name: Sync server.json version to release tag
-        if: github.event_name == 'release'
+      - name: Sync server.json version
+        if: steps.version.outputs.version != ''
         run: |
           version="${{ steps.version.outputs.version }}"
           tmp="$(mktemp)"
@@ -38,16 +57,18 @@ jobs:
           cat server.json
 
       - name: Install mcp-publisher
+        if: steps.version.outputs.version != ''
         run: |
           curl -sSL "https://github.com/modelcontextprotocol/registry/releases/latest/download/mcp-publisher_linux_amd64.tar.gz" \
             | tar -xz mcp-publisher
           sudo mv mcp-publisher /usr/local/bin/
 
       - name: Validate server.json
+        if: steps.version.outputs.version != ''
         run: mcp-publisher validate
 
       - name: Wait for the released version to appear on PyPI
-        if: github.event_name == 'release'
+        if: steps.version.outputs.version != ''
         run: |
           version="${{ steps.version.outputs.version }}"
           url="https://pypi.org/pypi/datoon/${version}/json"
@@ -64,7 +85,9 @@ jobs:
           exit 1
 
       - name: Login via GitHub OIDC
+        if: steps.version.outputs.version != ''
         run: mcp-publisher login github-oidc
 
       - name: Publish to MCP Registry
+        if: steps.version.outputs.version != ''
         run: mcp-publisher publish

datoon-1.7.2/.github/workflows/publish.yml

@@ -0,0 +1,82 @@
+name: Publish to PyPI
+
+# Releases are cut by the Release workflow using GITHUB_TOKEN, and GitHub does
+# not fire `release: published` for events created by GITHUB_TOKEN. So this
+# workflow keys off the Release workflow's completion instead, and publishes
+# only when the current main HEAD carries a new version tag.
+
+on:
+  workflow_run:
+    workflows: ["Release"]
+    types: [completed]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+  id-token: write
+
+jobs:
+  build:
+    if: >-
+      github.event_name == 'workflow_dispatch'
+      || github.event.workflow_run.conclusion == 'success'
+    runs-on: ubuntu-latest
+    outputs:
+      version: ${{ steps.ver.outputs.version }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          ref: main
+          fetch-depth: 0
+
+      - name: Resolve version to publish
+        id: ver
+        run: |
+          version="$(git tag --points-at HEAD | sed 's/^v//' | head -1)"
+          if [ -z "$version" ] && [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            version="$(git describe --tags --abbrev=0 | sed 's/^v//')"
+          fi
+          echo "version=${version}" >> "$GITHUB_OUTPUT"
+          if [ -z "$version" ]; then
+            echo "No release tag at HEAD - nothing to publish."
+          else
+            echo "Publishing datoon ${version}."
+          fi
+
+      - name: Check out the exact release tag
+        if: steps.ver.outputs.version != ''
+        run: git checkout "v${{ steps.ver.outputs.version }}"
+
+      - name: Install uv
+        if: steps.ver.outputs.version != ''
+        uses: astral-sh/setup-uv@v7
+
+      - name: Build distributions
+        if: steps.ver.outputs.version != ''
+        run: uv build
+
+      - name: Upload distributions
+        if: steps.ver.outputs.version != ''
+        uses: actions/upload-artifact@v7
+        with:
+          name: dist
+          path: dist/
+
+  publish:
+    needs: build
+    if: needs.build.outputs.version != ''
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/project/datoon/
+    steps:
+      - name: Download distributions
+        uses: actions/download-artifact@v8
+        with:
+          name: dist
+          path: dist/
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          skip-existing: true

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/release.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
 

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/tests.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup uv
         uses: astral-sh/setup-uv@v7
@@ -27,7 +27,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup uv
         uses: astral-sh/setup-uv@v7

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/validate-artifacts.yml

@@ -18,7 +18,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Setup Python
         uses: actions/setup-python@v6

{datoon-1.7.0 → datoon-1.7.2}/.github/workflows/workflow-lint.yml

@@ -22,7 +22,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v6
         with:
           persist-credentials: false
 
@@ -35,7 +35,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v6
         with:
           persist-credentials: false
 

{datoon-1.7.0 → datoon-1.7.2}/CHANGELOG.md

@@ -1,3 +1,17 @@
+## [1.7.2](https://github.com/andrii-su/datoon/compare/v1.7.1...v1.7.2) (2026-07-07)
+
+
+### Bug Fixes
+
+* **readers:** preserve semantic values during scalar and row coercion ([#36](https://github.com/andrii-su/datoon/issues/36)) ([5b24c00](https://github.com/andrii-su/datoon/commit/5b24c0098da620e14158b457a04d639663ba33f0))
+
+## [1.7.1](https://github.com/andrii-su/datoon/compare/v1.7.0...v1.7.1) (2026-06-22)
+
+
+### Bug Fixes
+
+* **mcp:** report datoon version in the MCP initialize handshake ([22ba5fd](https://github.com/andrii-su/datoon/commit/22ba5fd23071cf2fa6a2d2c51d306d761dce9dc4))
+
 # [1.7.0](https://github.com/andrii-su/datoon/compare/v1.6.0...v1.7.0) (2026-06-22)
 
 

{datoon-1.7.0 → datoon-1.7.2}/MARKETPLACES.md

@@ -22,17 +22,25 @@ ______________________________________________________________________
   [`README.md`](./README.md) (PyPI renders the README as the package description,
   which the registry validates).
 - **Publishing:** [`.github/workflows/publish-mcp.yml`](./.github/workflows/publish-mcp.yml)
-  runs on every GitHub Release. It syncs `server.json`'s version to the release
-  tag, validates, authenticates via **GitHub OIDC** (no stored secret), and runs
-  `mcp-publisher publish`.
+  runs when the **Release** workflow completes (via `workflow_run`), and only
+  when main HEAD carries a new version tag. It syncs `server.json`'s version to
+  that tag, validates, waits for the version to appear on PyPI, authenticates via
+  **GitHub OIDC** (no stored secret), and runs `mcp-publisher publish`.
 
 > [!NOTE]
-> The ownership token first reaches PyPI with the release that includes this
-> change. The publish workflow validates against the released version on PyPI,
-> so the first successful registry publish happens on the **next** release after
-> this is merged.
+> **Trigger note.** Releases are cut by semantic-release using `GITHUB_TOKEN`,
+> and GitHub does not fire `release: published` for `GITHUB_TOKEN`-created
+> events. Both [`publish.yml`](./.github/workflows/publish.yml) (PyPI) and
+> publish-mcp.yml therefore key off the **Release** workflow's completion
+> instead, gated on a new tag at HEAD.
 
-Manual publish (if needed):
+> [!NOTE]
+> The ownership token first reaches PyPI with the release that includes the
+> marketplace change. The first successful publish happens on the next release
+> after that merge.
+
+Manual publish — **Publish to MCP Registry** workflow → **Run workflow**
+(`workflow_dispatch`); it publishes the latest tag. Or locally:
 
 ```bash
 # Install mcp-publisher (see modelcontextprotocol/registry releases)
@@ -41,8 +49,6 @@ mcp-publisher login github        # interactive OAuth, namespace io.github.andri
 mcp-publisher publish
 ```
 
-Re-run via the **Publish to MCP Registry** workflow's `workflow_dispatch`.
-
 ______________________________________________________________________
 
 ## 2. Smithery — one-time connect

{datoon-1.7.0 → datoon-1.7.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datoon
-Version: 1.7.0
+Version: 1.7.2
 Summary: Smart structured-data-to-TOON gateway with pragmatic auto-gating for LLM prompts.
 Author-email: Andrii Suruhov <andrii.suruhov@gmail.com>
 License: MIT License

{datoon-1.7.0 → datoon-1.7.2}/src/datoon/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '1.7.0'
-__version_tuple__ = version_tuple = (1, 7, 0)
+__version__ = version = '1.7.2'
+__version_tuple__ = version_tuple = (1, 7, 2)
 
-__commit_id__ = commit_id = 'g50cf0e107'
+__commit_id__ = commit_id = 'ga011eb92d'

{datoon-1.7.0 → datoon-1.7.2}/src/datoon/mcp_server.py

@@ -12,6 +12,7 @@ except ImportError as exc:
         "MCP support requires the 'mcp' extra: pip install datoon[mcp]"
     ) from exc
 
+from datoon import __version__
 from datoon.analyzer import analyze_payload
 from datoon.converter import DatoonError, convert_json_for_llm
 from datoon.models import ConversionConfig
@@ -25,6 +26,10 @@ mcp = FastMCP(
         "Use analyze_json to check if a payload is a good TOON candidate without converting."
     ),
 )
+# FastMCP has no `version` parameter; set it on the underlying server so the
+# MCP `initialize` handshake reports datoon's version instead of the mcp
+# library's default.
+mcp._mcp_server.version = __version__
 
 
 @mcp.tool()

datoon-1.7.2/src/datoon/readers/_coerce.py

@@ -0,0 +1,48 @@
+"""Shared scalar coercion for text readers (CSV, XML).
+
+Coercion must never change semantic values. Two traps are guarded here:
+
+* Leading-zero integers (ZIP codes, phone numbers, IDs) stay strings — only
+  values that round-trip through ``int`` are coerced.
+* ``inf``/``nan`` and overflowing exponents would serialize to ``Infinity``/
+  ``NaN`` (invalid JSON), so non-finite floats stay strings.
+"""
+
+from __future__ import annotations
+
+import math
+import re
+
+# Canonical decimal/scientific literal: no leading zeros, no underscores. Guards
+# against float("00755") -> 755.0 and float("1_000") -> 1000.0 quietly rewriting
+# IDs and formatted numbers.
+_FLOAT_RE = re.compile(r"^[+-]?(?:0|[1-9]\d*)(?:\.\d+)?(?:[eE][+-]?\d+)?$")
+
+
+def coerce_scalar(value: str) -> int | float | bool | str | None:
+    """Coerce a raw string cell to a JSON-safe scalar without losing meaning."""
+    stripped = value.strip()
+    if stripped == "":
+        return None
+
+    low = stripped.lower()
+    if low == "true":
+        return True
+    if low == "false":
+        return False
+
+    # Only accept integers that round-trip: rejects "00123", "+5", "1_000".
+    try:
+        as_int = int(stripped)
+    except ValueError:
+        pass
+    else:
+        if str(as_int) == stripped:
+            return as_int
+
+    if _FLOAT_RE.match(stripped):
+        as_float = float(stripped)
+        # Reject overflowing exponents (1e400 -> inf) that break JSON serialization.
+        if math.isfinite(as_float):
+            return as_float
+    return stripped

datoon-1.7.2/src/datoon/readers/_tabular.py

@@ -0,0 +1,46 @@
+"""Shared header/data-row normalization for spreadsheet readers (Excel, Numbers)."""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+def header_rows_to_dicts(rows: list[Any]) -> list[dict[str, Any]]:
+    """Turn ``[header, *data]`` cell tuples into row dicts with stable keys.
+
+    * A ``None`` header cell becomes ``col{index}``.
+    * Data rows wider than the header keep their overflow cells under synthetic
+      ``col{index}`` keys instead of being silently dropped.
+    * Data rows narrower than the header get ``None`` for missing trailing cells,
+      so every emitted row shares an identical key set.
+    * Rows whose cells are all ``None`` are skipped.
+    """
+    if not rows:
+        return []
+
+    data_rows = rows[1:]
+    if not data_rows:
+        return []
+
+    widest = max((len(row) for row in data_rows), default=0)
+    header_row = rows[0]
+    column_count = max(len(header_row), widest)
+
+    headers: list[str] = []
+    for i in range(column_count):
+        raw = header_row[i] if i < len(header_row) else None
+        headers.append(str(raw) if raw is not None else f"col{i}")
+
+    result: list[dict[str, Any]] = []
+    for row in data_rows:
+        if not any(cell is not None for cell in row):
+            continue
+        result.append(dict(zip(headers, _pad(row, column_count))))
+    return result
+
+
+def _pad(row: Any, width: int) -> list[Any]:
+    cells = list(row)
+    if len(cells) < width:
+        cells.extend([None] * (width - len(cells)))
+    return cells

datoon-1.7.2/src/datoon/readers/csv.py

@@ -0,0 +1,17 @@
+"""CSV reader with basic type coercion."""
+
+from __future__ import annotations
+
+import csv
+import io
+from typing import Any
+
+from datoon.readers._coerce import coerce_scalar
+
+
+def read_csv(text: str, *, coerce_types: bool = True) -> list[dict[str, Any]]:
+    """Parse CSV text into a list of row dicts."""
+    reader = csv.DictReader(io.StringIO(text))
+    if coerce_types:
+        return [{k: coerce_scalar(str(v)) for k, v in row.items()} for row in reader]
+    return [dict(row) for row in reader]

{datoon-1.7.0 → datoon-1.7.2}/src/datoon/readers/excel.py

@@ -5,6 +5,8 @@ from __future__ import annotations
 from pathlib import Path
 from typing import Any
 
+from datoon.readers._tabular import header_rows_to_dicts
+
 
 def read_excel(path: Path, *, sheet: int = 0) -> list[dict[str, Any]]:
     """Read the first sheet of an Excel file into a list of row dicts."""
@@ -20,12 +22,4 @@ def read_excel(path: Path, *, sheet: int = 0) -> list[dict[str, Any]]:
     rows = list(ws.iter_rows(values_only=True))
     wb.close()
 
-    if not rows:
-        return []
-
-    headers = [str(h) if h is not None else f"col{i}" for i, h in enumerate(rows[0])]
-    return [
-        {h: v for h, v in zip(headers, row)}
-        for row in rows[1:]
-        if any(v is not None for v in row)
-    ]
+    return header_rows_to_dicts(rows)

{datoon-1.7.0 → datoon-1.7.2}/src/datoon/readers/numbers.py

@@ -5,6 +5,8 @@ from __future__ import annotations
 from pathlib import Path
 from typing import Any
 
+from datoon.readers._tabular import header_rows_to_dicts
+
 
 def read_numbers(path: Path, *, sheet: int = 0, table: int = 0) -> list[dict[str, Any]]:
     """Read an Apple Numbers file into a list of row dicts."""
@@ -19,12 +21,4 @@ def read_numbers(path: Path, *, sheet: int = 0, table: int = 0) -> list[dict[str
     tbl = doc.sheets[sheet].tables[table]
     rows = list(tbl.rows(values_only=True))
 
-    if not rows:
-        return []
-
-    headers = [str(h) if h is not None else f"col{i}" for i, h in enumerate(rows[0])]
-    return [
-        {h: v for h, v in zip(headers, row)}
-        for row in rows[1:]
-        if any(v is not None for v in row)
-    ]
+    return header_rows_to_dicts(rows)

{datoon-1.7.0 → datoon-1.7.2}/src/datoon/readers/xml.py

@@ -5,6 +5,8 @@ from __future__ import annotations
 import xml.etree.ElementTree as ET
 from typing import Any
 
+from datoon.readers._coerce import coerce_scalar as _coerce
+
 
 def read_xml(text: str) -> list[dict[str, Any]]:
     """Parse XML into a list of row dicts from the dominant repeated child element."""
@@ -36,23 +38,3 @@ def _element_to_dict(element: ET.Element) -> dict[str, Any]:
     if not list(element) and element.text and element.text.strip():
         result["_text"] = element.text.strip()
     return result
-
-
-def _coerce(value: str) -> int | float | bool | str | None:
-    stripped = value.strip()
-    if not stripped:
-        return None
-    low = stripped.lower()
-    if low == "true":
-        return True
-    if low == "false":
-        return False
-    try:
-        return int(stripped)
-    except ValueError:
-        pass
-    try:
-        return float(stripped)
-    except ValueError:
-        pass
-    return stripped

{datoon-1.7.0 → datoon-1.7.2}/src/datoon.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datoon
-Version: 1.7.0
+Version: 1.7.2
 Summary: Smart structured-data-to-TOON gateway with pragmatic auto-gating for LLM prompts.
 Author-email: Andrii Suruhov <andrii.suruhov@gmail.com>
 License: MIT License

{datoon-1.7.0 → datoon-1.7.2}/src/datoon.egg-info/SOURCES.txt

@@ -103,6 +103,8 @@ src/datoon.egg-info/scm_file_list.json
 src/datoon.egg-info/scm_version.json
 src/datoon.egg-info/top_level.txt
 src/datoon/readers/__init__.py
+src/datoon/readers/_coerce.py
+src/datoon/readers/_tabular.py
 src/datoon/readers/columnar.py
 src/datoon/readers/csv.py
 src/datoon/readers/excel.py
@@ -119,4 +121,5 @@ tests/unit/test_converter_policy.py
 tests/unit/test_installer.py
 tests/unit/test_mcp_server.py
 tests/unit/test_models.py
-tests/unit/test_readers.py
+tests/unit/test_readers.py
+tests/unit/test_tabular_rows.py

{datoon-1.7.0 → datoon-1.7.2}/src/datoon.egg-info/scm_file_list.json

@@ -40,11 +40,13 @@
     "src/datoon/converter.py",
     "src/datoon/cli.py",
     "src/datoon/readers/xml.py",
+    "src/datoon/readers/_coerce.py",
     "src/datoon/readers/__init__.py",
     "src/datoon/readers/jsonl.py",
     "src/datoon/readers/numbers.py",
     "src/datoon/readers/columnar.py",
     "src/datoon/readers/csv.py",
+    "src/datoon/readers/_tabular.py",
     "src/datoon/readers/excel.py",
     "src/datoon/readers/yaml.py",
     "assets/datoon-banner.svg",
@@ -59,6 +61,7 @@
     "tests/unit/test_models.py",
     "tests/unit/test_converter_policy.py",
     "tests/unit/test_mcp_server.py",
+    "tests/unit/test_tabular_rows.py",
     "tests/unit/test_analyzer.py",
     "tests/unit/test_installer.py",
     "tests/unit/test_cli.py",

datoon-1.7.2/src/datoon.egg-info/scm_version.json

@@ -0,0 +1,8 @@
+{
+  "tag": "1.7.2",
+  "distance": 0,
+  "node": "ga011eb92d6c550e4d87db4ebdd3b5efe9cb4f533",
+  "dirty": false,
+  "branch": "HEAD",
+  "node_date": "2026-07-07"
+}