Notify user if two datasets with different hashes are compared

shapeout2/gui/pipeline_plot.py

@@ -118,6 +118,16 @@ def update_content_plot(self, plot_state, slot_states, dslist):
         lay = plot_state["layout"]
         sca = plot_state["scatter"]
 
+        # create a hash set for the dcnum hashes
+        hash_set = set()
+        for ds in dslist:
+            pipe_config = ds.config.get("pipeline", {})
+            dcnum_hash = pipe_config.get("dcnum hash", None)
+            if dcnum_hash is not None:
+                hash_set.add(dcnum_hash)
+            else:
+                hash_set.add(None)
+
         # auto range (overrides stored ranges)
         if gen["auto range"]:
             # default range is limits + 5% margin
@@ -168,14 +178,17 @@ def update_content_plot(self, plot_state, slot_states, dslist):
         elif lay["division"] == "each":
             colcount = 0
             for ds, sl in zip(dslist, slot_states):
+                # get the hash flag
+                hash_flag = get_hash_flag(hash_set, ds)
+
                 pp = PipelinePlotItem(parent=linner)
                 self.plot_items.append(pp)
                 linner.addItem(item=pp,
                                row=None,
                                col=None,
                                rowspan=1,
                                colspan=1)
-                pp.redraw([ds], [sl], plot_state)
+                pp.redraw([ds], [sl], plot_state, hash_flag)
                 colcount += 1
                 if colcount % lay["column count"] == 0:
                     linner.nextRow()
@@ -185,14 +198,17 @@ def update_content_plot(self, plot_state, slot_states, dslist):
             plot_state_scatter = copy.deepcopy(plot_state)
             plot_state_scatter["contour"]["enabled"] = False
             for ds, sl in zip(dslist, slot_states):
+                # get the hash flag
+                hash_flag = get_hash_flag(hash_set, ds)
+
                 pp = PipelinePlotItem(parent=linner)
                 self.plot_items.append(pp)
                 linner.addItem(item=pp,
                                row=None,
                                col=None,
                                rowspan=1,
                                colspan=1)
-                pp.redraw([ds], [sl], plot_state_scatter)
+                pp.redraw([ds], [sl], plot_state_scatter, hash_flag)
                 colcount += 1
                 if colcount % lay["column count"] == 0:
                     linner.nextRow()
@@ -294,7 +310,7 @@ def perform_export(self, file):
             exp = exporters.SVGExporter(win.scene())
         exp.export(file)
 
-    def redraw(self, dslist, slot_states, plot_state):
+    def redraw(self, dslist, slot_states, plot_state, hash_flag=None):
         # Remove everything
         for el in self._plot_elements:
             self.removeItem(el)
@@ -333,7 +349,8 @@ def redraw(self, dslist, slot_states, plot_state):
                 sct = add_scatter(plot_item=self,
                                   rtdc_ds=rtdc_ds,
                                   plot_state=plot_state,
-                                  slot_state=ss
+                                  slot_state=ss,
+                                  hash_flag=hash_flag
                                   )
                 self._plot_elements += sct
         # Contour data
@@ -525,7 +542,7 @@ def add_isoelastics(plot_item, axis_x, axis_y, channel_width, pixel_size,
     return elements
 
 
-def add_scatter(plot_item, plot_state, rtdc_ds, slot_state):
+def add_scatter(plot_item, plot_state, rtdc_ds, slot_state, hash_flag):
     gen = plot_state["general"]
     sca = plot_state["scatter"]
     scatter = pg.ScatterPlotItem(size=sca["marker size"],
@@ -592,6 +609,17 @@ def add_scatter(plot_item, plot_state, rtdc_ds, slot_state):
     if gen["scale y"] == "log":
         y = np.log10(y)
 
+    # add dcnum hash label
+    if hash_flag:
+        add_label(
+            hash_flag,
+            anchor_parent=plot_item.axes["top"]["item"],
+            font_size_diff=-1,
+            color="red",
+            text_halign="left",
+            text_valign="top",
+        )
+
     scatter.setData(x=x, y=y, brush=brush)
     scatter.setZValue(-50)
     return [scatter]
@@ -733,6 +761,27 @@ def set_viewbox(plot, range_x, range_y, scale_x="linear", scale_y="linear",
                   )
 
 
+def get_hash_flag(hash_set, rtdc_ds):
+    """Helper function to determine the hash flag based on the dataset and
+    hash set."""
+    short_hash_set = set(h[:4] if h is not None else None for h in hash_set)
+
+    req_hash_len = 4
+    if len(short_hash_set) != len(hash_set):
+        req_hash_len = 5
+
+    if len(hash_set) == 1:
+        # only one hash, no need to show it
+        return None
+    else:
+        # get the pipeline hash
+        pipe_config = rtdc_ds.config.get("pipeline", {})
+        dcnum_hash = pipe_config.get("dcnum hash", None)
+        # use the first `req_hash_len` characters of the hash
+        short_hash = dcnum_hash[:req_hash_len] if dcnum_hash else None
+        return f"Pipeline: {short_hash}" if short_hash else None
+
+
 linestyles = {
     "solid": QtCore.Qt.PenStyle.SolidLine,
     "dashed": QtCore.Qt.PenStyle.DashLine,

tests/test_gui_pipeline_plot.py

@@ -1,7 +1,12 @@
 import numpy as np
+import pathlib
+
+import dclab
 
 from shapeout2.gui import pipeline_plot
 
+datapath = pathlib.Path(__file__).parent / "data"
+
 
 def test_compute_contour_opening_angles():
     contour = [
@@ -71,3 +76,26 @@ def test_compute_contour_opening_angles_log_scale():
     angles = pipeline_plot.compute_contour_opening_angles(
         plot_state=plot_state, contour=contour)
     assert np.allclose(angles, np.pi/3)
+
+
+def test_get_hash_flag():
+    rtdc_paths = datapath.glob("*.rtdc")
+
+    hash_set = set()
+    rtdc_ds_list = []
+    expected = []
+    for path in rtdc_paths:
+        ds = dclab.new_dataset(path)
+        # get the hash flag
+        pipe_config = ds.config.get("pipeline", {})
+        dcnum_hash = pipe_config.get("dcnum hash", None)
+        hash_set.add(dcnum_hash)
+        expected.append(f"Pipeline: {dcnum_hash[:4]}" if dcnum_hash else None)
+        rtdc_ds_list.append(ds)
+
+    assert len(hash_set) == 2
+
+    for ds, exp in zip(rtdc_ds_list, expected):
+        result = pipeline_plot.get_hash_flag(hash_set, ds)
+        print(result)
+        assert result == exp