Introduce basic github action for the box files

Author: Jakub Sliacan

.github/workflows/box.yml

@@ -0,0 +1,36 @@
+name: Process Box
+
+on:
+  push:
+    branches: [ main ] 
+  pull_request: {}
+
+jobs:
+  read-data:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+      
+      - name: Set up python 3.x
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.x'
+          cache: 'pip3'
+        run: pip3 install -r requirements.txt
+     
+      - name: Process CSV files
+        run: |
+          python gha/process-box.py
+
+      - name: Upload OT figures
+        uses: actions/upload-artifact@v4
+        with:
+          name: OT-figures
+          path: gha/out/ot-figures/*.zip
+
+      - name: Upload OT events file
+        uses: actions/upload-artifact@v4
+        with:
+          name: OT-events
+          path: gha/out/ot-events.csv

gha/process-box.py

@@ -0,0 +1,93 @@
+import os
+import matplotlib.pyplot as plt
+import networkx as nx
+from src import box, constants, util
+from src import modularity as mod
+
+bikeLogs = "samples/BikeLogs"
+dflist = util.get_box_files(bikeLogs)
+util.ensure_date_in_filenames(dflist)
+dflist = util.get_box_files(bikeLogs)
+
+for csv_file in dflist:
+    print(csv_file, flush=True)
+    csvr = util.read_csv(csv_file)
+    ldata = box.make_ldata(csvr)  # CSV data as a list
+    # presses
+    press_starts, press_lengths = box.get_press_lengths_and_starts(ldata)
+    date_string = csv_file.split("/")[-1][:8]
+
+    """
+    b_partitions, a_partitions, b_modularities, a_modularities = mod.get_partitions(ldata, press_starts, press_lengths)
+    
+    for j, b_parts in enumerate(b_partitions):
+        
+        if b_modularities[j] == -1: # it's an oncoming case
+            continue
+
+        ps = press_starts[j]
+        pmod = b_modularities[j]
+        part_and_size_pairs = [(len(x), x) for x in b_parts]
+        part_and_size_pairs.sort(reverse=True) # sorts pairs by 1st item, i.e. size
+        press_gap = sum([s for s, p in part_and_size_pairs])
+        lat_dists = [ldata[ps-x][4] for x in range(press_gap)]
+        dispersion_score = mod.dispersion_score(lat_dists)
+                     
+        plt.scatter(range(len(lat_dists)), lat_dists, c='b')
+
+        # ---- process the partition -----
+        for s, p in part_and_size_pairs:
+
+            lp = list(p)
+            lp.sort() # should already be sorted
+
+            # skip the maxed-out readings
+            if min([lat_dists[x] for x in p]) > 450:
+                continue
+            
+            # skip the readings stuck too low
+            if max([lat_dists[x] for x in p]) < 50: 
+                continue
+
+            # Split part into subparts >8 lines apart
+            # Discard tiny subparts
+            subparts = mod.strip_and_split(lp, lat_dists, 8)
+            if len(subparts) == 0:
+                continue
+            lp = subparts[0]
+            s = len(lp)
+            
+                
+            # --- max clique method ----
+            G = nx.Graph()
+            G.add_nodes_from(lp)
+
+            G.add_edges_from([(lp[x],lp[y]) for x in range(s-1) for y in range(x+1, s) if abs(lat_dists[lp[x]]-lat_dists[lp[y]])<40])
+            mc = nx.approximation.max_clique(G)
+            additional_vertices = set()
+            for v in mc:
+                v_ball = [lat_dists[v]-30, lat_dists[v]+30]
+                for u in G.nodes:
+                    if u in mc:
+                        continue
+                    # make sure the vertex we include with the clique doesn't have a very small degree
+                    if lat_dists[u] > v_ball[0] and lat_dists[u] < v_ball[1] and G.degree(u) > 0.7*len(mc):
+                        additional_vertices.add(u)
+                
+            ot = list(mc)+list(additional_vertices)
+            ot_event = [date_string, ps, dispersion_score, pmod, len(ot), ot]
+            ot_events.append(ot_event)
+            
+            plt.scatter(ot, [lat_dists[x] for x in ot], c='r')
+            
+            # --- end of max clique ----
+
+            break # after getting to the OT event
+            
+        plt.ylim([0,700])
+        plt.savefig(os.path.join("out", "mod", date_string+"_ld_"+str(press_starts[j])+"_"+"{:.6f}".format(pmod)+"_disp="+str(dispersion_score)+"_clique.png"))
+        plt.clf()
+
+util.write_to_csv_file(os.path.join("data", "ot_events.csv"), ot_events)
+
+        """