track number of changes in patch stage (resolve #138)

seal5/flow.py

@@ -764,6 +764,7 @@ def patch(self, verbose: bool = False, stages: List[PatchStage] = None, force: b
             stages = list(map(PatchStage, range(PatchStage.PHASE_5 + 1)))
         assert len(stages) > 0
         patches_per_stage = self.collect_patches()
+        stages_metrics = {}
         for stage in stages:
             logger.info("Current stage: %s", stage)
             patches = patches_per_stage.get(stage, [])
@@ -779,11 +780,23 @@ def patch(self, verbose: bool = False, stages: List[PatchStage] = None, force: b
             # author = git_.get_author(self.settings.git)
             # self.repo.create_tag(tag_name, message=tag_msg, force=True, author=author)
             self.repo.create_tag(tag_name, message=tag_msg, force=True)
+            base_tag = f"seal5-{self.name}-base"
+            if int(stage) > 0:
+                prev_tag = f"seal5-{self.name}-stage{int(stage)-1}"
+            else:
+                prev_tag = base_tag
+            n_files_changed, n_insertions, n_deletions = inject_patches.analyze_diff(self.repo, cur=tag_name, base=prev_tag)
+            stage_metrics = {}
+            stage_metrics["n_files_changed"] = n_files_changed
+            stage_metrics["n_insertions"] = n_insertions
+            stage_metrics["n_deletions"] = n_deletions
+            stages_metrics[PatchStage(stage).name] = stage_metrics
         end = time.time()
         diff = end - start
         metrics["start"] = start
         metrics["end"] = end
         metrics["time_s"] = diff
+        metrics["stages"] = stages_metrics
         self.settings.metrics.append({"patch": metrics})
         self.settings.save()
         logger.info("Completed application of Seal5 patches")

seal5/tools/inject_patches.py

@@ -20,10 +20,12 @@
 import os
 import argparse
 from pathlib import Path
+from typing import Optional
 
 from email.utils import formatdate
 import yaml
 
+from seal5 import utils
 from seal5.logging import get_logger
 
 logger = get_logger()
@@ -230,6 +232,35 @@ def process_arguments():
     return args
 
 
+def analyze_diff(repo, base: str, cur: Optional[str] = None):
+    args = ["git", "diff", "--shortstat"]
+    args += [base]
+    if cur is not None:
+        args += [cur]
+    out = utils.exec_getout(
+        *args,
+        cwd=repo.working_tree_dir,
+        print_func=lambda *args, **kwargs: None,
+        live=False,
+    )
+    n_files_changed = 0
+    n_insertions = 0
+    n_deletions = 0
+    for x in out.split(","):
+        x = x.strip()
+        if len(x) == 0:
+            continue
+        val, key = x.split(" ", 1)
+        val = int(val)
+        if "files changed" in key:
+            n_files_changed = val
+        elif "insertions" in key:
+            n_insertions = val
+        elif "deletions" in key:
+            n_deletions = val
+    return n_files_changed, n_insertions, n_deletions
+
+
 def main():
     """Main entry point."""
     args = process_arguments()