Add per-repo user-specific diff statistics to graph output

graphs.py

@@ -0,0 +1,146 @@
+import matplotlib.pyplot as plot
+from typing import List
+
+class Graph:
+    def __init__(self,
+                 data: list = [],
+                 minimum: int = 0,
+                 min_count: int = 0,
+                 kind: str = 'pie',
+                 legend: str = 'Default graph legend',
+                 title: str = 'Default graph title',
+                 counter: str = 'total'):
+
+        self.minimum = minimum
+        self.min_count = min_count
+        self.kind = kind
+        self.legend = legend
+        self.title = title
+        self.counter = counter
+
+        # Graph generation code expects a list but callers might specify a single list if they only need one dataset.
+        if isinstance(data, dict):
+            self.data = [data]
+        else:
+            self.data = data
+
+        # Normalize and count the elements
+        self.normalize_data()
+
+        self.count = []
+        for dataset in self.data:
+            self.count.append(len(dataset))
+
+    def normalize_data(self):
+        result = []
+
+        for i,dataset in enumerate(self.data):
+            ds = dict(self.data[i])
+
+            # Remove summatory keys from the dictionary.
+            # The additional 'nope' is there just to avoid having to put everything in a try in case the "total" key does not exist. 
+            ds.pop('total', 'nope')
+            ds.pop('past_year', 'nope')
+
+            other = 0
+
+            for key in list(ds.keys()):
+                if ds[key] < self.minimum:
+                    other += ds.pop(key)
+
+            # Order data dictionary by size of elements
+            ds = {k:v for k,v in sorted(ds.items(), key=lambda x: int(x[1]), reverse=True)}
+
+            if other > 0:
+                ds['other'] = other
+
+            result.append(ds)
+
+        self.data = result
+
+    def is_suitable(self) -> bool:
+        for c in self.count:
+            if c > 0:
+                return True
+
+        return False
+
+
+def generate_figure(graphs: List[Graph], path: str):
+    filtered = sorted([graph for graph in graphs if graph.is_suitable()], key=lambda x: 0 if x.kind == 'pie' else 1)
+    heights = []
+
+    # If we have no suitable graphs, return without doing nothing
+    if len(filtered) == 0:
+        return
+
+    for graph in filtered:
+        if graph.kind == 'pie':
+            heights.append(7)
+        else:
+            heights.append((0.3 * max(graph.count)))
+
+    figure, axis = plot.subplots(len(filtered),
+                                 figsize=(12, sum(heights) + 2.5),
+                                 dpi=600,
+                                 gridspec_kw={'height_ratios': [h / heights[0] for h in heights]})
+
+    # We need a list for the following for loop and if len(filtered) = 1 axis is just an object. Maybe there is a better way to do this?
+    if len(filtered) == 1:
+        axis = [axis]
+
+    for i,graph in enumerate(filtered):
+        generate_chart(graph.data, graph.minimum, graph.kind, graph.legend, graph.counter, graph.title, axis[i])
+
+    plot.tight_layout()
+    plot.savefig(path, bbox_inches='tight')
+    plot.close(figure)
+
+
+def generate_chart(data: list, minimum: int, graph_type: str, legend: str, counter: str, title: str, axis):
+    keys    = [dataset.keys() for dataset in data]
+    values  = [dataset.values() for dataset in data]
+    counts  = [len(v) for v in values]
+    totals  = [sum(dataset[key] for key in dataset) for dataset in data]
+    classes = [len(dataset) if counter == 'classes' else totals[i] for i,dataset in enumerate(data)]
+    labels  = [(f'{key} ({((dataset[key] * 100)/totals[i]):.2f}%)' for key in dataset) for i,dataset in enumerate(data)]
+
+    for i,dataset in enumerate(data):
+        if len(dataset) == 0:
+            continue
+
+        if graph_type == 'pie':
+            # Set the color map and generate a properly sized color cycle
+            colors = []
+            colormaps = {'Pastel1':9, 'Accent':8, 'Set1':9, 'tab20':20, 'tab20b':20}
+
+            for colormap in colormaps:
+                cmap = plot.get_cmap(colormap)
+                colors += [cmap(i/colormaps[colormap]) for i in range(colormaps[colormap])]
+
+            step = int(len(colors)/counts[i])
+            axis.set_prop_cycle('color', [colors[i*step] for i in range(counts[i])])
+
+            wedges, texts = axis.pie(values[i], counterclock=False, startangle=90)
+            legend = axis.legend(wedges, labels[i], title=legend, bbox_to_anchor=(1.01, 1), loc='upper left')
+            axis.set_aspect('equal')
+            axis.set_title(f'{title} (total: {classes[i]})')
+
+        elif graph_type == 'bar':
+            if len(data) == 1:
+                colors = ['C0']
+            else:
+                colors = ['C2', 'C3']
+
+            y = [i for i in range(counts[i])]
+
+            bars = axis.barh(y, values[i], align='center', color=colors[i])
+            axis.set_yticks(y)
+            axis.set_yticklabels(keys[i])
+            axis.invert_yaxis()
+            axis.set_xlabel(legend)
+            axis.set_title(f'{title} (total: {classes[i]})')
+
+            for bar in bars:
+                width = bar.get_width()
+                axis.annotate(str(width), xy=(width, bar.get_y() + bar.get_height() / 2), xytext=(3,0), textcoords='offset points', ha='left', va='center')