reboost 0.4.3__tar.gz → 0.5.1__tar.gz

{reboost-0.4.3 → reboost-0.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: reboost
-Version: 0.4.3
+Version: 0.5.1
 Summary: New LEGEND Monte-Carlo simulation post-processing
 Author-email: Manuel Huber <info@manuelhu.de>, Toby Dixon <toby.dixon.23@ucl.ac.uk>, Luigi Pertoldi <gipert@pm.me>
 Maintainer: The LEGEND Collaboration
@@ -700,7 +700,7 @@ Requires-Dist: colorlog
 Requires-Dist: numpy
 Requires-Dist: scipy
 Requires-Dist: numba
-Requires-Dist: legend-pydataobj>=1.11.10
+Requires-Dist: legend-pydataobj>=1.12.0
 Requires-Dist: legend-pygeom-optics>=0.6.5
 Requires-Dist: hist
 Requires-Dist: dbetto

{reboost-0.4.3 → reboost-0.5.1}/pyproject.toml

@@ -36,7 +36,7 @@ dependencies = [
     "numpy",
     "scipy",
     "numba",
-    "legend-pydataobj>=1.11.10",
+    "legend-pydataobj>=1.12.0",
     "legend-pygeom-optics>=0.6.5",
     "hist",
     "dbetto",

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/__init__.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+from lgdo import lh5
+
 from . import build_hit, core, iterator, math, shape
 from ._version import version as __version__
 
@@ -15,3 +17,5 @@ __all__ = [
     "optmap",
     "shape",
 ]
+
+lh5.settings.DEFAULT_HDF5_SETTINGS = {"shuffle": True, "compression": "lzf"}

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/_version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '0.4.3'
-__version_tuple__ = version_tuple = (0, 4, 3)
+__version__ = version = '0.5.1'
+__version_tuple__ = version_tuple = (0, 5, 1)

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/build_evt.py

@@ -102,9 +102,9 @@ def build_evt(
         elif isinstance(info, list):
             channels[group] = info
 
-    for tcm_lh5, _, n_rows_read in LH5Iterator(tcm_file, "tcm", buffer_len=buffer):
+    for tcm_lh5 in LH5Iterator(tcm_file, "tcm", buffer_len=buffer):
         tcm_lh5_sel = tcm_lh5
-        tcm_ak = tcm_lh5_sel.view_as("ak")[:n_rows_read]
+        tcm_ak = tcm_lh5_sel.view_as("ak")
 
         tcm = TCMData(
             id=np.array(ak.flatten(tcm_ak.array_id)),

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/build_glm.py

@@ -146,13 +146,14 @@ def get_stp_evtids(
 
     while start_row < n_rows_tot:
         # read the file
-        lh5_obj, n_read = store.read(
+        lh5_obj = store.read(
             f"{lh5_table}/{id_name}",
             stp_file,
             start_row=start_row,
             n_rows=stp_buffer,
         )
         evtids = lh5_obj.view_as("ak")
+        n_read = len(evtids)
 
         # pick the first evtid
         if evtids.ndim > 1:
@@ -247,9 +248,9 @@ def build_glm(
         vfield = f"vtx/{id_name}"
 
         # iterate over the vertex table
-        for vert_obj, vidx, n_evtid in LH5Iterator(stp_file, vfield, buffer_len=evtid_buffer):
+        for vidx, vert_obj in enumerate(LH5Iterator(stp_file, vfield, buffer_len=evtid_buffer)):
             # range of vertices
-            vert_ak = vert_obj.view_as("ak")[:n_evtid]
+            vert_ak = vert_obj.view_as("ak")
 
             msg = f"... read chunk {vidx}"
             log.debug(msg)

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/build_hit.py

@@ -41,10 +41,6 @@ A :func:`build_hit` to parse the following configuration file:
              phyvol: OBJECTS.geometry.physical_volume_dict[DETECTOR]
              drift_time_map: lgdo.lh5.read(DETECTOR, ARGS.dtmap_file)
 
-          # this defines "hits", i.e. layout of the output hit table
-          # group steps by time and evtid with 10us window
-          hit_table_layout: reboost.shape.group_by_time(STEPS, window=10)
-
           # finally, the processing chain
           operations:
 
@@ -190,7 +186,6 @@ def build_hit(
     *,
     start_evtid: int = 0,
     n_evtid: int | None = None,
-    in_field: str = "stp",
     out_field: str = "hit",
     buffer: int = int(5e6),
     overwrite: bool = False,
@@ -214,8 +209,6 @@ def build_hit(
         first evtid to read.
     n_evtid
         number of evtid to read, if `None` read all.
-    in_field
-        name of the input field in the remage output.
     out_field
         name of the output field
     buffer
@@ -280,13 +273,17 @@ def build_hit(
                     time_dict=time_dict[proc_name],
                 )
 
+                lh5_group = proc_group.get("lh5_group", "stp")
+                if lh5_group is None:
+                    lh5_group = "/"
+
                 # begin iterating over the glm
                 iterator = GLMIterator(
                     glm_file,
                     stp_file,
                     lh5_group=in_detector,
                     start_row=start_evtid,
-                    stp_field=in_field,
+                    stp_field=lh5_group,
                     n_rows=n_evtid,
                     buffer=buffer,
                     time_dict=time_dict[proc_name],
@@ -396,7 +393,7 @@ def build_hit(
                         lh5.read(obj, stp_file),
                         obj,
                         files.hit[file_idx],
-                        wo_mode="write_safe",
+                        wo_mode="write_safe" if file_idx == 0 else "append",
                     )
                 except LH5EncodeError as e:
                     msg = f"cannot forward object {obj} as it has been already processed by reboost"

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/cli.py

@@ -103,7 +103,6 @@ def cli(args=None) -> None:
     hit_parser.add_argument(
         "--n-evtid", type=int, default=None, help="Number of event id to process."
     )
-    hit_parser.add_argument("--in-field", type=str, default="stp", help="Input field name.")
     hit_parser.add_argument("--out-field", type=str, default="hit", help="Output field name.")
     hit_parser.add_argument("--buffer", type=int, default=int(5e6), help="Buffer size.")
 
@@ -174,7 +173,6 @@ def cli(args=None) -> None:
         msg += f"    hit files:      {hit_files}\n"
         msg += f"    start_evtid:    {args.start_evtid}\n"
         msg += f"    n_evtid:        {args.n_evtid}\n"
-        msg += f"    in_field:       {args.in_field}\n"
         msg += f"    out_field:      {args.out_field}\n"
         msg += f"    buffer:         {args.buffer} \n"
         msg += f"    overwrite:      {args.overwrite} \n"
@@ -189,7 +187,6 @@ def cli(args=None) -> None:
             hit_files=hit_files,
             start_evtid=args.start_evtid,
             n_evtid=args.n_evtid,
-            in_field=args.in_field,
             out_field=args.out_field,
             buffer=args.buffer,
             overwrite=args.overwrite,

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/iterator.py

@@ -139,12 +139,14 @@ class GLMIterator:
 
         if self.use_glm:
             if self.glm_file is not None:
-                glm_rows, n_rows_read = self.sto.read(
+                glm_rows = self.sto.read(
                     f"glm/{self.lh5_group}",
                     self.glm_file,
                     start_row=self.start_row_tmp,
                     n_rows=n_rows,
                 )
+                n_rows_read = len(glm_rows.view_as("ak"))
+
             else:
                 # get the maximum row to read
                 max_row = self.start_row_tmp + n_rows
@@ -198,12 +200,14 @@ class GLMIterator:
             time_start = time.time()
 
         try:
-            stp_rows, n_steps = self.sto.read(
+            stp_rows = self.sto.read(
                 f"{self.stp_field}/{self.lh5_group}",
                 self.stp_file,
                 start_row=int(start),
                 n_rows=int(n),
             )
+            n_steps = len(stp_rows.view_as("ak"))
+
         except OverflowError:
             raise StopIteration from None
 

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/optmap/convolve.py

@@ -356,9 +356,9 @@ def convolve(
 
     log.info("opening energy deposition hit output %s", edep_file)
     it = LH5Iterator(edep_file, edep_path, buffer_len=buffer_len)
-    for it_count, (edep_lgdo, edep_events, edep_n_rows) in enumerate(it):
-        assert (it_count == 0) == (edep_events == 0)
-        edep_df = edep_lgdo.view_as("pd").iloc[0:edep_n_rows].to_records()
+
+    for it_count, edep_lgdo in enumerate(it):
+        edep_df = edep_lgdo.view_as("pd").iloc.to_records()
 
         log.info("start event processing (%d)", it_count)
         output_map = iterate_stepwise_depositions(

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/optmap/create.py

@@ -130,9 +130,8 @@ def _create_optical_maps_chunk(
 
     hits_per_primary = np.zeros(10, dtype=np.int64)
     hits_per_primary_len = 0
-    for it_count, (events_lgdo, events_entry, event_n_rows) in enumerate(optmap_events_it):
-        assert (it_count == 0) == (events_entry == 0)
-        optmap_events = events_lgdo.view_as("pd").iloc[0:event_n_rows]
+    for it_count, events_lgdo in enumerate(optmap_events_it):
+        optmap_events = events_lgdo.view_as("pd").iloc
         hitcounts = optmap_events[all_det_ids].to_numpy()
         loc = optmap_events[["xloc", "yloc", "zloc"]].to_numpy()
 

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/optmap/evt.py

@@ -80,8 +80,8 @@ def build_optmap_evt(
             _store_vert_df()
 
             # read the next vertex chunk into memory.
-            (vert_lgdo, vert_entry, vert_n_rows) = next(vert_it)
-            vert_df = vert_lgdo.view_as("pd").iloc[0:vert_n_rows]
+            (vert_lgdo, vert_entry) = next(vert_it)
+            vert_df = vert_lgdo.view_as("pd").iloc
 
             # prepare vertex coordinates.
             vert_df = vert_df.set_index("evtid", drop=True).drop(["n_part", "time"], axis=1)
@@ -94,9 +94,8 @@ def build_optmap_evt(
     log.info("prepare evt table")
     # use smaller integer type uint8 to spare RAM when storing types.
     hit_count_type = np.uint8
-    for opti_it_count, (opti_lgdo, opti_entry, opti_n_rows) in enumerate(opti_it):
-        assert (opti_it_count == 0) == (opti_entry == 0)
-        opti_df = opti_lgdo.view_as("pd").iloc[0:opti_n_rows]
+    for opti_it_count, opti_lgdo in enumerate(opti_it):
+        opti_df = opti_lgdo.view_as("pd").iloc
 
         log.info("build evt table (%d)", opti_it_count)
         for t in opti_df[["evtid", "det_uid"]].itertuples(name=None, index=False):

{reboost-0.4.3 → reboost-0.5.1}/src/reboost/utils.py

@@ -30,7 +30,9 @@ def get_wo_mode(
         return "overwrite_file" if overwrite else "write_safe"
 
     # if we have a detector not the first and chunk 0 append column
-    if ((in_det > 0) or (out_det > 0)) & (chunk == 0):
+    is_append = ((in_det > 0) or (out_det > 0)) & (chunk == 0)
+    is_append = is_append or (in_det == 0 and out_det == 0 and chunk == 0 and group > 0)
+    if is_append:
         return "append_column"
     return "append"
 

{reboost-0.4.3 → reboost-0.5.1}/src/reboost.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: reboost
-Version: 0.4.3
+Version: 0.5.1
 Summary: New LEGEND Monte-Carlo simulation post-processing
 Author-email: Manuel Huber <info@manuelhu.de>, Toby Dixon <toby.dixon.23@ucl.ac.uk>, Luigi Pertoldi <gipert@pm.me>
 Maintainer: The LEGEND Collaboration
@@ -700,7 +700,7 @@ Requires-Dist: colorlog
 Requires-Dist: numpy
 Requires-Dist: scipy
 Requires-Dist: numba
-Requires-Dist: legend-pydataobj>=1.11.10
+Requires-Dist: legend-pydataobj>=1.12.0
 Requires-Dist: legend-pygeom-optics>=0.6.5
 Requires-Dist: hist
 Requires-Dist: dbetto

{reboost-0.4.3 → reboost-0.5.1}/src/reboost.egg-info/SOURCES.txt

@@ -6,7 +6,6 @@ src/reboost/_version.py
 src/reboost/build_evt.py
 src/reboost/build_glm.py
 src/reboost/build_hit.py
-src/reboost/build_tcm.py
 src/reboost/cli.py
 src/reboost/core.py
 src/reboost/iterator.py
@@ -56,8 +55,7 @@ tests/hit/configs/basic.yaml
 tests/hit/configs/geom.gdml
 tests/hit/configs/hit_config.yaml
 tests/hit/configs/pars.yaml
-tests/hit/test_files/beta_small.lh5
-tests/hit/test_files/beta_small_hit.lh5
+tests/hit/configs/reshape.yaml
 tests/hpge/test_current.py
 tests/hpge/test_dt_heuristic.py
 tests/hpge/test_hpge_map.py

{reboost-0.4.3 → reboost-0.5.1}/src/reboost.egg-info/requires.txt

@@ -2,7 +2,7 @@ colorlog
 numpy
 scipy
 numba
-legend-pydataobj>=1.11.10
+legend-pydataobj>=1.12.0
 legend-pygeom-optics>=0.6.5
 hist
 dbetto

{reboost-0.4.3 → reboost-0.5.1}/tests/conftest.py

@@ -12,6 +12,8 @@ from legendtestdata import LegendTestData
 
 _tmptestdir = Path(gettempdir()) / f"reboost-tests-{getuser()}-{uuid.uuid4()!s}"
 
+collect_ignore = ["test_optmap.py"]
+
 
 @pytest.fixture(scope="session")
 def tmptestdir_global():

{reboost-0.4.3 → reboost-0.5.1}/tests/hit/configs/basic.yaml

@@ -9,8 +9,6 @@ processing_groups:
       - evtid
       - energy
 
-    hit_table_layout: reboost.shape.group.group_by_time(STEPS, window=10)
-
     operations:
       t0: ak.fill_none(ak.firsts(HITS.time, axis=-1), np.nan)
       first_evtid: ak.fill_none(ak.firsts(HITS.evtid, axis=-1), np.nan)

{reboost-0.4.3 → reboost-0.5.1}/tests/hit/configs/hit_config.yaml

@@ -7,8 +7,8 @@ processing_groups:
   - name: geds
 
     detector_mapping:
-      - output: det001
-      - output: det002
+      - output: det1
+      - output: det2
     detector_objects:
       meta: dbetto.AttrsDict(pygeomtools.get_sensvol_metadata(OBJECTS.geometry,
         OBJECTS.user_pars[DETECTOR].name ))
@@ -24,8 +24,6 @@ processing_groups:
       - active_energy
       - smeared_energy
 
-    hit_table_layout: reboost.shape.group.group_by_evtid(STEPS)
-
     operations:
       t0: ak.fill_none(ak.firsts(HITS.time, axis=-1), np.nan)
       first_evtid: ak.fill_none(ak.firsts(HITS.evtid, axis=-1), np.nan)
@@ -44,3 +42,9 @@ processing_groups:
 
       active_energy: ak.sum(HITS.edep*HITS.activeness, axis=-1)
       smeared_energy: reboost.math.stats.gaussian_sample(HITS.active_energy,DETECTOR_OBJECTS.det_pars.reso_in_sigma)
+
+  - name: vtx
+    detector_mapping:
+      - output: vtx
+
+    lh5_group: null

{reboost-0.4.3 → reboost-0.5.1}/tests/hit/configs/pars.yaml

@@ -1,9 +1,9 @@
-det001:
+det1:
   fccd_in_mm: 0.71
   dlf: 0.7
   reso_in_sigma: 1.1
   name: BEGe
-det002:
+det2:
   fccd_in_mm: 2.2
   dlf: 0.4
   reso_in_sigma: 2.32

reboost-0.5.1/tests/hit/configs/reshape.yaml

@@ -0,0 +1,11 @@
+processing_groups:
+  - name: all
+
+    detector_mapping:
+      - output: det1
+      - output: det2
+
+    hit_table_layout: "reboost.shape.group.group_by_time(STEPS, 10)"
+
+forward:
+  - /vtx

reboost-0.5.1/tests/hit/test_build_hit.py

@@ -0,0 +1,219 @@
+from __future__ import annotations
+
+import copy
+from pathlib import Path
+
+import awkward as ak
+import dbetto
+import h5py
+import pytest
+from lgdo import Array, Struct, Table, VectorOfVectors, lh5
+
+import reboost
+
+
+@pytest.fixture(scope="module")
+def test_gen_lh5(tmptestdir):
+    # write a basic lh5 file
+
+    stp_path = str(tmptestdir / "basic.lh5")
+
+    data = {}
+    data["evtid"] = VectorOfVectors([[0, 0], [1, 1, 1]])
+    data["edep"] = VectorOfVectors([[100, 200], [10, 20, 300]])  # keV
+    data["time"] = VectorOfVectors([[0, 1.5], [0.1, 2.1, 3.7]])  # ns
+
+    data["xloc"] = VectorOfVectors([[0.01, 0.02], [0.001, 0.003, 0.005]])  # m
+    data["yloc"] = VectorOfVectors([[0.01, 0.02], [0.001, 0.003, 0.005]])  # m
+    data["zloc"] = VectorOfVectors([[0.04, 0.02], [0.001, 0.023, 0.005]])  # m
+    data["dist_to_surf"] = VectorOfVectors([[0.04, 0.02], [0.011, 0.003, 0.051]])  # m
+
+    vertices = [0, 1]
+    tab = Table(data)
+    tab2 = copy.deepcopy(tab)
+
+    lh5.write(tab, "stp/det1", stp_path, wo_mode="of")
+    lh5.write(tab2, "stp/det2", stp_path, wo_mode="append")
+    lh5.write(
+        Table({"evtid": Array(vertices)}),
+        "vtx",
+        stp_path,
+        wo_mode="append",
+    )
+
+    return stp_path
+
+
+@pytest.fixture(scope="module")
+def test_gen_lh5_flat(tmptestdir):
+    # write a basic lh5 file
+
+    stp_path = str(tmptestdir / "basic_flat.lh5")
+
+    data = {}
+    data["evtid"] = Array([0, 0, 1, 1, 1])
+    data["edep"] = Array([100, 200, 10, 20, 300])  # keV
+    data["time"] = Array([0, 1.5, 0.1, 2.1, 3.7])  # ns
+
+    data["xloc"] = Array([0.01, 0.02, 0.001, 0.003, 0.005])  # m
+    data["yloc"] = Array([0.01, 0.02, 0.001, 0.003, 0.005])  # m
+    data["zloc"] = Array([0.04, 0.02, 0.001, 0.023, 0.005])  # m
+    data["dist_to_surf"] = Array([0.04, 0.02, 0.011, 0.003, 0.051])  # m
+
+    vertices = [0, 1]
+    tab = Table(data)
+    tab2 = copy.deepcopy(tab)
+
+    lh5.write(tab, "stp/det1", stp_path, wo_mode="of")
+    lh5.write(tab2, "stp/det2", stp_path, wo_mode="append")
+    lh5.write(
+        Table({"evtid": Array(vertices)}),
+        "vtx",
+        stp_path,
+        wo_mode="append",
+    )
+
+    return stp_path
+
+
+def test_reshape(test_gen_lh5_flat, tmptestdir):
+    outfile = f"{tmptestdir}/basic_hit_reshaped.lh5"
+
+    reboost.build_hit.build_hit(
+        f"{Path(__file__).parent}/configs/reshape.yaml",
+        args={},
+        stp_files=test_gen_lh5_flat,
+        glm_files=None,
+        hit_files=outfile,
+        out_field="stp",
+        overwrite=True,
+    )
+
+    data = {}
+    data["evtid"] = VectorOfVectors([[0, 0], [1, 1, 1]])
+    data["edep"] = VectorOfVectors([[100, 200], [10, 20, 300]])  # keV
+    data["time"] = VectorOfVectors([[0, 1.5], [0.1, 2.1, 3.7]])  # ns
+
+    data["xloc"] = VectorOfVectors([[0.01, 0.02], [0.001, 0.003, 0.005]])  # m
+    data["yloc"] = VectorOfVectors([[0.01, 0.02], [0.001, 0.003, 0.005]])  # m
+    data["zloc"] = VectorOfVectors([[0.04, 0.02], [0.001, 0.023, 0.005]])  # m
+    data["dist_to_surf"] = VectorOfVectors([[0.04, 0.02], [0.011, 0.003, 0.051]])  # m
+    tab = Table(data)
+
+    output = lh5.read("stp/det1", outfile)
+
+    # check the outputs
+    assert output == tab
+    assert lh5.read("vtx", outfile) == Table({"evtid": Array([0, 1])})
+
+
+def test_basic(test_gen_lh5, tmptestdir):
+    outfile = f"{tmptestdir}/basic_hit.lh5"
+
+    reboost.build_hit.build_hit(
+        f"{Path(__file__).parent}/configs/basic.yaml",
+        args={},
+        stp_files=test_gen_lh5,
+        glm_files=None,
+        hit_files=outfile,
+        overwrite=True,
+    )
+
+    assert lh5.ls(outfile) == ["hit", "vtx"]
+
+    with h5py.File(outfile) as h5f:
+        assert h5f["/hit/det1/energy"].shuffle is True
+        assert h5f["/hit/det1/energy"].compression == "lzf"
+
+    hits = lh5.read("hit/det1", outfile).view_as("ak")
+
+    assert ak.all(hits.energy == [300, 330])
+    assert ak.all(hits.t0 == [0, 0.1])
+    assert ak.all(hits.evtid[0] == [0, 0])
+    assert ak.all(hits.evtid[1] == [1, 1, 1])
+
+    assert len(hits) == 2
+
+    # test in memory
+
+    hits, time_dict = reboost.build_hit.build_hit(
+        f"{Path(__file__).parent}/configs/basic.yaml",
+        args={},
+        stp_files=test_gen_lh5,
+        glm_files=None,
+        hit_files=None,
+    )
+
+    assert ak.all(hits["det1"].energy == [300, 330])
+    assert ak.all(hits["det1"].t0 == [0, 0.1])
+    assert ak.all(hits["det1"].evtid[0] == [0, 0])
+    assert ak.all(hits["det1"].evtid[1] == [1, 1, 1])
+
+    assert set(time_dict.keys()) == {"global_objects", "geds"}
+    assert set(time_dict["geds"].keys()) == {
+        "detector_objects",
+        "read",
+        "conv",
+        "expressions",
+    }
+    assert set(time_dict["geds"]["read"].keys()) == {"stp"}
+    assert set(time_dict["geds"]["expressions"].keys()) == {"t0", "first_evtid", "energy"}
+
+
+def test_file_merging(test_gen_lh5, tmptestdir):
+    outfile = f"{tmptestdir}/basic_hit_merged.lh5"
+
+    reboost.build_hit.build_hit(
+        f"{Path(__file__).parent}/configs/basic.yaml",
+        args={},
+        stp_files=[test_gen_lh5, test_gen_lh5],
+        glm_files=None,
+        hit_files=outfile,
+        overwrite=True,
+    )
+
+    assert lh5.ls(outfile) == ["hit", "vtx"]
+
+    hits = lh5.read("hit/det1", outfile).view_as("ak")
+
+    assert len(hits) == 4
+
+
+def test_full_chain(test_gen_lh5, tmptestdir):
+    args = dbetto.AttrsDict(
+        {
+            "gdml": f"{Path(__file__).parent}/configs/geom.gdml",
+            "pars": f"{Path(__file__).parent}/configs/pars.yaml",
+        }
+    )
+
+    _, time_dict = reboost.build_hit.build_hit(
+        f"{Path(__file__).parent}/configs/hit_config.yaml",
+        args=args,
+        stp_files=test_gen_lh5,
+        glm_files=None,
+        hit_files=str(tmptestdir / "beta_small_hit.lh5"),
+        overwrite=True,
+    )
+    hits = lh5.read("hit", str(tmptestdir / "beta_small_hit.lh5"))
+
+    assert isinstance(hits, Struct)
+
+    assert set(hits["det1"].view_as("ak").fields) == {
+        "evtid",
+        "t0",
+        "truth_energy",
+        "active_energy",
+        "smeared_energy",
+    }
+    assert set(hits["det2"].view_as("ak").fields) == {
+        "evtid",
+        "t0",
+        "truth_energy",
+        "active_energy",
+        "smeared_energy",
+    }
+
+    # also check the processing of the vtx table
+
+    assert hits["vtx"] == Table({"evtid": Array([0, 1])})

reboost-0.5.1/tests/test_cli.py

@@ -0,0 +1,81 @@
+from __future__ import annotations
+
+import copy
+from pathlib import Path
+
+import pytest
+from lgdo import Array, Table, lh5
+
+from reboost.cli import cli
+
+
+@pytest.fixture(scope="module")
+def test_gen_lh5_flat(tmptestdir):
+    # write a basic lh5 file
+
+    stp_path = str(tmptestdir / "basic_flat.lh5")
+
+    data = {}
+    data["evtid"] = Array([0, 0, 1, 1, 1])
+    data["edep"] = Array([100, 200, 10, 20, 300])  # keV
+    data["time"] = Array([0, 1.5, 0.1, 2.1, 3.7])  # ns
+
+    data["xloc"] = Array([0.01, 0.02, 0.001, 0.003, 0.005])  # m
+    data["yloc"] = Array([0.01, 0.02, 0.001, 0.003, 0.005])  # m
+    data["zloc"] = Array([0.04, 0.02, 0.001, 0.023, 0.005])  # m
+    data["dist_to_surf"] = Array([0.04, 0.02, 0.011, 0.003, 0.051])  # m
+
+    vertices = [0, 1]
+    tab = Table(data)
+    tab2 = copy.deepcopy(tab)
+
+    lh5.write(tab, "stp/det1", stp_path, wo_mode="of")
+    lh5.write(tab2, "stp/det2", stp_path, wo_mode="append")
+    lh5.write(
+        Table({"evtid": Array(vertices)}),
+        "vtx",
+        stp_path,
+        wo_mode="append",
+    )
+
+    return stp_path
+
+
+def test_cli(tmptestdir, test_gen_lh5_flat):
+    test_file_dir = Path(__file__).parent / "hit"
+
+    # test cli for build_glm
+    cli(
+        [
+            "build-glm",
+            "--id-name",
+            "evtid",
+            "-w",
+            "--glm-file",
+            f"{tmptestdir}/glm.lh5",
+            "--stp-file",
+            test_gen_lh5_flat,
+        ]
+    )
+
+    glm = lh5.read("glm/det1", f"{tmptestdir}/glm.lh5").view_as("ak")
+    assert glm.fields == ["evtid", "n_rows", "start_row"]
+
+    # test cli for build_hit
+    cli(
+        [
+            "build-hit",
+            "--config",
+            f"{test_file_dir}/configs/reshape.yaml",
+            "-w",
+            "--stp-file",
+            test_gen_lh5_flat,
+            "--hit-file",
+            f"{tmptestdir}/hit.lh5",
+            "--args",
+            f"{test_file_dir}/configs/args.yaml",
+        ]
+    )
+
+    hit1 = lh5.read("hit/det1", f"{tmptestdir}/hit.lh5").view_as("ak")
+    assert set(hit1.fields) == {"xloc", "yloc", "zloc", "dist_to_surf", "time", "edep", "evtid"}

{reboost-0.4.3 → reboost-0.5.1}/tests/test_utils.py

@@ -118,10 +118,10 @@ def test_save_dict(tmp_path):
 
 
 def test_wo_mode():
-    assert get_wo_mode(0, 0, 0, 0, True, overwrite=True) == "of"
-    assert get_wo_mode(0, 0, 0, 0, True, overwrite=False) == "w"
-    assert get_wo_mode(0, 0, 1, 0, True, overwrite=False) == "ac"
-    assert get_wo_mode(0, 0, 1, 1, True, overwrite=False) == "a"
+    assert get_wo_mode(0, 0, 0, 0, True, overwrite=True) == "overwrite_file"
+    assert get_wo_mode(0, 0, 0, 0, True, overwrite=False) == "write_safe"
+    assert get_wo_mode(0, 0, 1, 0, True, overwrite=False) == "append_column"
+    assert get_wo_mode(0, 0, 1, 1, True, overwrite=False) == "append"
 
 
 def test_get_files_dict():

reboost-0.4.3/src/reboost/build_tcm.py

@@ -1,111 +0,0 @@
-from __future__ import annotations
-
-import logging
-import re
-
-import awkward as ak
-from lgdo import Table, lh5
-
-from .shape import group
-
-log = logging.getLogger(__name__)
-
-
-def build_tcm(
-    hit_file: str,
-    out_file: str,
-    channels: list[str],
-    time_name: str = "t0",
-    idx_name: str = "global_evtid",
-    time_window_in_us: float = 10,
-) -> None:
-    """Build the (Time Coincidence Map) TCM from the hit tier.
-
-    Parameters
-    ----------
-    hit_file
-        path to hit tier file.
-    out_file
-        output path for tcm.
-    channels
-        list of channel names to include.
-    time_name
-        name of the hit tier field used for time grouping.
-    idx_name
-        name of the hit tier field used for index grouping.
-    time_window_in_us
-        time window used to define the grouping.
-    """
-    hash_func = r"\d+"
-
-    msg = "start building time-coincidence map"
-    log.info(msg)
-
-    chan_ids = [re.search(hash_func, channel).group() for channel in channels]
-
-    hit_data = []
-    for channel in channels:
-        hit_data.append(
-            lh5.read(f"{channel}/hit", hit_file, field_mask=[idx_name, time_name]).view_as("ak")
-        )
-    tcm = get_tcm_from_ak(
-        hit_data, chan_ids, window=time_window_in_us, time_name=time_name, idx_name=idx_name
-    )
-
-    if tcm is not None:
-        lh5.write(tcm, "tcm", out_file, wo_mode="overwrite_file")
-
-
-def get_tcm_from_ak(
-    hit_data: list[ak.Array],
-    channels: list[int],
-    *,
-    window: float = 10,
-    time_name: str = "t0",
-    idx_name: str = "global_evtid",
-) -> Table:
-    """Builds a time-coincidence map from a hit of hit data Tables.
-
-    - build an ak.Array of the data merging channels with fields base on "time_name", and "idx_name" and adding a field `rawid` from the channel idx, also add the row (`hit_idx`)
-    - sorts this array by "idx_name" then "time_name" fields
-    - group by "idx_name" and "time_name" based on the window parameter
-
-    Parameters
-    ----------
-    hit_data
-        list of hit tier data for each channel
-    channels
-        list of channel indices
-    window
-        time window for selecting coincidences (in us)
-    time_name
-        name of the field for time information
-    idx_name
-        name of the decay index field
-
-    Returns
-    -------
-    an LGDO.VectorOfVectors containing the time-coincidence map
-    """
-    # build ak_obj for sorting
-    sort_objs = []
-
-    for ch_idx, data_tmp in zip(channels, hit_data):
-        obj_tmp = ak.copy(data_tmp)
-        obj_tmp = obj_tmp[[time_name, idx_name]]
-        hit_idx = ak.local_index(obj_tmp)
-
-        obj_tmp = ak.with_field(obj_tmp, hit_idx, "array_idx")
-
-        obj_tmp["array_id"] = int(ch_idx)
-        sort_objs.append(obj_tmp)
-
-    obj_tot = ak.concatenate(sort_objs)
-
-    return group.group_by_time(
-        obj_tot,
-        time_name=time_name,
-        evtid_name=idx_name,
-        window=window,
-        fields=["array_id", "array_idx"],
-    )

reboost-0.4.3/tests/hit/test_build_hit.py

@@ -1,126 +0,0 @@
-from __future__ import annotations
-
-from pathlib import Path
-
-import awkward as ak
-import dbetto
-import pytest
-from lgdo import Array, Struct, Table, lh5
-
-import reboost
-from reboost.build_glm import build_glm
-
-
-@pytest.fixture(scope="module")
-def test_gen_lh5(tmptestdir):
-    # write a basic lh5 file
-
-    stp_path = str(tmptestdir / "basic.lh5")
-    glm_path = str(tmptestdir / "basic_glm.lh5")
-
-    evtid = [0, 0, 1, 1, 1]
-    edep = [100, 200, 10, 20, 300]  # keV
-    time = [0, 1.5, 0.1, 2.1, 3.7]  # ns
-    vertices = [0, 1]
-    tab = Table({"evtid": Array(evtid), "edep": Array(edep), "time": Array(time)})
-    lh5.write(tab, "stp/det1", stp_path, wo_mode="of")
-    lh5.write(
-        Table({"evtid": Array(vertices)}),
-        "vtx",
-        stp_path,
-        wo_mode="append",
-    )
-
-    build_glm(stp_path, glm_path, id_name="evtid")
-
-    return stp_path, glm_path
-
-
-def test_basic(test_gen_lh5, tmptestdir):
-    stp_path, glm_path = test_gen_lh5
-
-    reboost.build_hit.build_hit(
-        f"{Path(__file__).parent}/configs/basic.yaml",
-        args={},
-        stp_files=stp_path,
-        glm_files=glm_path,
-        hit_files=f"{tmptestdir}/basic_hit.lh5",
-        overwrite=True,
-    )
-
-    assert lh5.ls(f"{tmptestdir}/basic_hit.lh5") == ["hit", "vtx"]
-
-    hits = lh5.read("hit/det1", f"{tmptestdir}/basic_hit.lh5").view_as("ak")
-
-    assert ak.all(hits.energy == [300, 330])
-    assert ak.all(hits.t0 == [0, 0.1])
-    assert ak.all(hits.evtid[0] == [0, 0])
-    assert ak.all(hits.evtid[1] == [1, 1, 1])
-
-    # test in memory
-
-    hits, time_dict = reboost.build_hit.build_hit(
-        f"{Path(__file__).parent}/configs/basic.yaml",
-        args={},
-        stp_files=stp_path,
-        glm_files=glm_path,
-        hit_files=None,
-    )
-
-    assert ak.all(hits["det1"].energy == [300, 330])
-    assert ak.all(hits["det1"].t0 == [0, 0.1])
-    assert ak.all(hits["det1"].evtid[0] == [0, 0])
-    assert ak.all(hits["det1"].evtid[1] == [1, 1, 1])
-
-    assert list(time_dict.keys()) == ["global_objects", "geds"]
-    assert list(time_dict["geds"].keys()) == [
-        "detector_objects",
-        "read",
-        "conv",
-        "hit_layout",
-        "expressions",
-    ]
-    assert list(time_dict["geds"]["read"].keys()) == ["glm", "stp"]
-    assert list(time_dict["geds"]["expressions"].keys()) == ["t0", "first_evtid", "energy"]
-
-
-def test_full_chain(tmptestdir):
-    build_glm(
-        f"{Path(__file__).parent}/test_files/beta_small.lh5",
-        str(tmptestdir / "beta_small_glm.lh5"),
-        id_name="evtid",
-    )
-
-    args = dbetto.AttrsDict(
-        {
-            "gdml": f"{Path(__file__).parent}/configs/geom.gdml",
-            "pars": f"{Path(__file__).parent}/configs/pars.yaml",
-        }
-    )
-
-    _, time_dict = reboost.build_hit.build_hit(
-        f"{Path(__file__).parent}/configs/hit_config.yaml",
-        args=args,
-        stp_files=f"{Path(__file__).parent}/test_files/beta_small.lh5",
-        glm_files=str(tmptestdir / "beta_small_glm.lh5"),
-        hit_files=str(tmptestdir / "beta_small_hit.lh5"),
-        overwrite=True,
-    )
-    hits = lh5.read("hit", str(tmptestdir / "beta_small_hit.lh5"))
-
-    assert isinstance(hits, Struct)
-
-    assert set(hits["det001"].view_as("ak").fields) == {
-        "evtid",
-        "t0",
-        "truth_energy",
-        "active_energy",
-        "smeared_energy",
-    }
-    assert set(hits["det002"].view_as("ak").fields) == {
-        "evtid",
-        "t0",
-        "truth_energy",
-        "active_energy",
-        "smeared_energy",
-    }

reboost-0.4.3/tests/test_cli.py

@@ -1,49 +0,0 @@
-from __future__ import annotations
-
-from pathlib import Path
-
-from lgdo import lh5
-
-from reboost.cli import cli
-
-
-def test_cli(tmptestdir):
-    test_file_dir = Path(__file__).parent / "hit"
-
-    # test cli for build_glm
-    cli(
-        [
-            "build-glm",
-            "--id-name",
-            "evtid",
-            "-w",
-            "--glm-file",
-            f"{tmptestdir}/glm.lh5",
-            "--stp-file",
-            f"{test_file_dir}/test_files/beta_small.lh5",
-        ]
-    )
-
-    glm = lh5.read("glm/det001", f"{tmptestdir}/glm.lh5").view_as("ak")
-    assert glm.fields == ["evtid", "n_rows", "start_row"]
-
-    # test cli for build_hit
-    cli(
-        [
-            "build-hit",
-            "--config",
-            f"{test_file_dir}/configs/hit_config.yaml",
-            "-w",
-            "--glm-file",
-            f"{tmptestdir}/glm.lh5",
-            "--stp-file",
-            f"{test_file_dir}/test_files/beta_small.lh5",
-            "--hit-file",
-            f"{tmptestdir}/hit.lh5",
-            "--args",
-            f"{test_file_dir}/configs/args.yaml",
-        ]
-    )
-
-    hit1 = lh5.read("hit/det001", f"{tmptestdir}/hit.lh5").view_as("ak")
-    assert set(hit1.fields) == {"evtid", "t0", "truth_energy", "active_energy", "smeared_energy"}