etlplus 0.12.13__py3-none-any.whl → 0.13.0__py3-none-any.whl

etlplus/config/jobs.py

@@ -34,10 +34,7 @@ __all__ = [
 ]
 
 
-# SECTION: TYPE ALIASES ===================================================== #
-
-
-# SECTION: CLASSES ========================================================== #
+# SECTION: DATA CLASSES ===================================================== #
 
 
 @dataclass(kw_only=True, slots=True)
@@ -100,6 +97,8 @@ class JobConfig:
         Unique job name.
     description : str | None
         Optional human-friendly description.
+    depends_on : list[str]
+        Optional job dependency list. Dependencies must refer to other jobs.
     extract : ExtractRef | None
         Extraction reference.
     validate : ValidationRef | None
@@ -114,6 +113,7 @@ class JobConfig:
 
     name: str
     description: str | None = None
+    depends_on: list[str] = field(default_factory=list)
     extract: ExtractRef | None = None
     validate: ValidationRef | None = None
     transform: TransformRef | None = None
@@ -149,9 +149,19 @@ class JobConfig:
         if description is not None and not isinstance(description, str):
             description = str(description)
 
+        depends_raw = data.get('depends_on')
+        depends_on: list[str] = []
+        if isinstance(depends_raw, str):
+            depends_on = [depends_raw]
+        elif isinstance(depends_raw, list):
+            for entry in depends_raw:
+                if isinstance(entry, str):
+                    depends_on.append(entry)
+
         return cls(
             name=name,
             description=description,
+            depends_on=depends_on,
             extract=ExtractRef.from_obj(data.get('extract')),
             validate=ValidationRef.from_obj(data.get('validate')),
             transform=TransformRef.from_obj(data.get('transform')),

etlplus/dag.py

@@ -0,0 +1,103 @@
+"""
+:mod:`etlplus.dag` module.
+
+Lightweight directed acyclic graph (DAG) helpers for ordering jobs based on
+``depends_on``.
+"""
+
+from __future__ import annotations
+
+from collections import deque
+from dataclasses import dataclass
+
+from .config.jobs import JobConfig
+
+# SECTION: EXPORTS ========================================================== #
+
+
+__all__ = [
+    'DagError',
+    'topological_sort_jobs',
+]
+
+
+# SECTION: ERRORS =========================================================== #
+
+
+@dataclass(slots=True)
+class DagError(ValueError):
+    """
+    Raised when the job dependency graph is invalid.
+
+    Attributes
+    ----------
+    message : str
+        Error message.
+    """
+
+    # -- Attributes -- #
+
+    message: str
+
+    # -- Magic Methods (Object Representation) -- #
+
+    def __str__(self) -> str:
+        return self.message
+
+
+# SECTION: FUNCTIONS ======================================================== #
+
+
+def topological_sort_jobs(
+    jobs: list[JobConfig],
+) -> list[JobConfig]:
+    """
+    Return jobs in topological order based on ``depends_on``.
+
+    Parameters
+    ----------
+    jobs : list[JobConfig]
+        List of job configurations to sort.
+
+    Returns
+    -------
+    list[JobConfig]
+        Jobs sorted in topological order.
+
+    Raises
+    ------
+    DagError
+        If a dependency is missing, self-referential, or when a cycle is
+        detected.
+    """
+    index = {job.name: job for job in jobs}
+    edges: dict[str, set[str]] = {name: set() for name in index}
+    indegree: dict[str, int] = {name: 0 for name in index}
+
+    for job in jobs:
+        for dep in job.depends_on:
+            if dep not in index:
+                raise DagError(
+                    f'Unknown dependency "{dep}" in job "{job.name}"',
+                )
+            if dep == job.name:
+                raise DagError(f'Job "{job.name}" depends on itself')
+            if job.name not in edges[dep]:
+                edges[dep].add(job.name)
+                indegree[job.name] += 1
+
+    queue = deque(sorted(name for name, deg in indegree.items() if deg == 0))
+    ordered: list[str] = []
+
+    while queue:
+        name = queue.popleft()
+        ordered.append(name)
+        for child in sorted(edges[name]):
+            indegree[child] -= 1
+            if indegree[child] == 0:
+                queue.append(child)
+
+    if len(ordered) != len(jobs):
+        raise DagError('Dependency cycle detected')
+
+    return [index[name] for name in ordered]

{etlplus-0.12.13.dist-info → etlplus-0.13.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: etlplus
-Version: 0.12.13
+Version: 0.13.0
 Summary: A Swiss Army knife for simple ETL operations
 Home-page: https://github.com/Dagitali/ETLPlus
 Author: ETLPlus Team

{etlplus-0.12.13.dist-info → etlplus-0.13.0.dist-info}/RECORD

@@ -2,6 +2,7 @@ etlplus/README.md,sha256=5jNes37UIy_THNmUr5HSAyS5mTCTa5tqRfEPnvsgV4s,1455
 etlplus/__init__.py,sha256=M2gScnyir6WOMAh_EuoQIiAzdcTls0_5hbd_Q6of8I0,1021
 etlplus/__main__.py,sha256=btoROneNiigyfBU7BSzPKZ1R9gzBMpxcpsbPwmuHwTM,479
 etlplus/__version__.py,sha256=1E0GMK_yUWCMQFKxXjTvyMwofi0qT2k4CDNiHWiymWE,327
+etlplus/dag.py,sha256=4EYmBsJax3y4clHv10jjdp3GrBBD_WblvtxUb_JxGCQ,2464
 etlplus/enums.py,sha256=WyxpUEUPdYdXlueKDXGaSEo7o9OqCXyzjDOOPqmW8tw,8326
 etlplus/extract.py,sha256=LOyL8_KCOaIGemTxSnKbN_ttfLWUljqT4OQxANe7G3k,6089
 etlplus/load.py,sha256=aufl-2CpuI_J1hKBY1uFsoVf9Gfl9bKQjs233dYFf00,8631
@@ -43,7 +44,7 @@ etlplus/cli/types.py,sha256=tclhKVJXDqHzlTQBYKARfqMgDOcuBJ-Zej2pvFy96WM,652
 etlplus/config/README.md,sha256=ot6oFZxTz4x83mj1_FrQ13dO0z2QkRFDnkCkx7NPsSs,1636
 etlplus/config/__init__.py,sha256=VZWzOg7d2YR9NT6UwKTv44yf2FRUMjTHynkm1Dl5Qzo,1486
 etlplus/config/connector.py,sha256=0-TIwevHbKRHVmucvyGpPd-3tB1dKHB-dj0yJ6kq5eY,9809
-etlplus/config/jobs.py,sha256=hmzRCqt0OvCEZZR4ONKrd3lvSv0OmayjLc4yOBk3ug8,7399
+etlplus/config/jobs.py,sha256=oa2rNwacy2b_1HP_iFDLarGnn812ZVP2k5cHt4eqBIg,7843
 etlplus/config/pipeline.py,sha256=m4Jh0ctFcKrIx6zR7LEC0sYY5wq0o8NqOruWPlz6qmA,9494
 etlplus/config/profile.py,sha256=Ss2zedQGjkaGSpvBLTD4SZaWViMJ7TJPLB8Q2_BTpPg,1898
 etlplus/config/types.py,sha256=a0epJ3z16HQ5bY3Ctf8s_cQPa3f0HHcwdOcjCP2xoG4,4954
@@ -125,9 +126,9 @@ etlplus/templates/view.sql.j2,sha256=Iy8DHfhq5yyvrUKDxqp_aHIEXY4Tm6j4wT7YDEFWAhk
 etlplus/validation/README.md,sha256=qusyiyJu2DsaK80jlwfXVZ0iDgeuTPOX2EL3a_fcFiw,1401
 etlplus/validation/__init__.py,sha256=Pe5Xg1_EA4uiNZGYu5WTF3j7odjmyxnAJ8rcioaplSQ,1254
 etlplus/validation/utils.py,sha256=Mtqg449VIke0ziy_wd2r6yrwJzQkA1iulZC87FzXMjo,10201
-etlplus-0.12.13.dist-info/licenses/LICENSE,sha256=MuNO63i6kWmgnV2pbP2SLqP54mk1BGmu7CmbtxMmT-U,1069
-etlplus-0.12.13.dist-info/METADATA,sha256=ERZKncgR9agi3oVefJwnDRjPvokwZUlzYu701I6TMUg,28105
-etlplus-0.12.13.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-etlplus-0.12.13.dist-info/entry_points.txt,sha256=6w-2-jzuPa55spzK34h-UKh2JTEShh38adFRONNP9QE,45
-etlplus-0.12.13.dist-info/top_level.txt,sha256=aWWF-udn_sLGuHTM6W6MLh99ArS9ROkUWO8Mi8y1_2U,8
-etlplus-0.12.13.dist-info/RECORD,,
+etlplus-0.13.0.dist-info/licenses/LICENSE,sha256=MuNO63i6kWmgnV2pbP2SLqP54mk1BGmu7CmbtxMmT-U,1069
+etlplus-0.13.0.dist-info/METADATA,sha256=_L0jck50nGtiKn2XwWnpUwHd9ylP3grWBZhATo9ibLM,28104
+etlplus-0.13.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+etlplus-0.13.0.dist-info/entry_points.txt,sha256=6w-2-jzuPa55spzK34h-UKh2JTEShh38adFRONNP9QE,45
+etlplus-0.13.0.dist-info/top_level.txt,sha256=aWWF-udn_sLGuHTM6W6MLh99ArS9ROkUWO8Mi8y1_2U,8
+etlplus-0.13.0.dist-info/RECORD,,