Rework open PRs to look at PRs from a week prior to the given date

metrics/github/api.py

@@ -1,12 +1,12 @@
 import json
 import os
 import textwrap
-from datetime import date, timedelta
+from datetime import date
 
 import requests
 import structlog
 
-from ..tools.dates import datetime_from_iso
+from ..tools.dates import date_from_iso
 
 
 log = structlog.get_logger()
@@ -136,18 +136,17 @@ def _iter_pull_requests(org, date_range):
     results = list(_iter_query_results(query, searchQuery=search_query))
     for pr in results:
         yield {
-            "created": datetime_from_iso(pr["createdAt"]).date(),
-            "closed": datetime_from_iso(pr["closedAt"]).date(),
+            "created": date_from_iso(pr["createdAt"]),
+            "closed": date_from_iso(pr["closedAt"]),
             "author": pr["author"]["login"],
             "repo": pr["repository"]["name"],
             "org": pr["repository"]["owner"]["login"],
         }
 
 
-def prs_open_on_date(org, date):
-    start = date.isoformat()
-    end = (date + timedelta(days=1)).isoformat()
-
+def prs_open_in_range(org, start, end):
+    start = start.isoformat()
+    end = end.isoformat()
     date_range = f"created:<={start} closed:>={end}"
 
     return list(_iter_pull_requests(org, date_range))

metrics/github/backfill.py

@@ -9,7 +9,7 @@
 from ..logs import setup_logging
 from ..timescaledb import TimescaleDBWriter
 from ..timescaledb.tables import GitHubPullRequests
-from ..tools.dates import date_from_iso, iter_days
+from ..tools.dates import date_from_iso, iter_days, previous_weekday
 from .prs import process_prs
 
 
@@ -57,13 +57,15 @@ def get_prs(db):
     return list(cur.execute(sql))
 
 
-def open_prs(prs, org, start, days_threshold):
-    dates = list(iter_days(start, date.today()))
+def open_prs(prs, org, days_threshold):
+    earliest = date_from_iso(min([pr["created"] for pr in prs]))
+    start = previous_weekday(earliest, 0)  # Monday
+    mondays = list(iter_days(start, date.today(), step=timedelta(days=7)))
 
     today = date.today()
     threshold = timedelta(days=days_threshold)
 
-    def open_on_day(date, pr, today):
+    def open_on_day(pr, start, end):
         """
         Filter function for PRs
 
@@ -73,25 +75,31 @@ def open_on_day(date, pr, today):
         closed = date_from_iso(pr["closed"]) or today
         opened = date_from_iso(pr["created"])
 
-        open_today = (opened <= date) and (closed >= day)
+        open_today = (opened <= start) and (closed >= end)
         if not open_today:
             return False
 
         return (closed - opened) >= threshold
 
     with TimescaleDBWriter(GitHubPullRequests) as writer:
-        for day in dates:
-            prs_on_day = [pr for pr in prs if open_on_day(day, pr, today)]
+        for start in mondays:
+            end = start + timedelta(days=6)
+            prs_on_day = [pr for pr in prs if open_on_day(pr, start, end)]
 
             name = f"queue_older_than_{days_threshold}_days"
 
             log.info(
-                "%s | %s | %s | Processing %s PRs", name, day, org, len(prs_on_day)
+                "%s | %s | Processing %s PRs from week starting %s",
+                name,
+                org,
+                len(prs_on_day),
+                start,
             )
-            process_prs(writer, prs_on_day, day, name=name)
+            process_prs(writer, prs_on_day, start, name=name)
 
 
-def pr_throughput(prs, org, start):
+def pr_throughput(prs, org):
+    start = date_from_iso(min([pr["created"] for pr in prs]))
     days = list(iter_days(start, date.today()))
 
     with TimescaleDBWriter(GitHubPullRequests) as writer:
@@ -125,9 +133,6 @@ def backfill(ctx, org, pull_data, db_path):
 
     org_prs = [pr for pr in prs if pr["org"] == org]
     log.info("Backfilling with %s PRs for %s", len(org_prs), org)
-    start_date = date_from_iso(min([pr["created"] for pr in org_prs]))
 
-    for day in [2, 7, 10, 30, 60]:
-        open_prs(org_prs, org, start_date, days_threshold=day)
-
-    pr_throughput(org_prs, org, start_date)
+    open_prs(org_prs, org, days_threshold=7)
+    pr_throughput(org_prs, org)

metrics/github/cli.py

@@ -5,6 +5,7 @@
 
 from ..timescaledb import TimescaleDBWriter
 from ..timescaledb.tables import GitHubPullRequests
+from ..tools.dates import previous_weekday
 from . import api
 from .backfill import backfill
 from .prs import process_prs
@@ -25,12 +26,24 @@ def github(ctx, token):
 @github.command()
 @click.argument("org")
 @click.argument("date", type=click.DateTime())
-@click.argument("days-threshold", type=int)
+@click.argument("--days-threshold", type=int, default=7)
 @click.pass_context
 def open_prs(ctx, org, date, days_threshold):
-    """The number of PRs open for DAYS_THRESHOLD or longer on the given date"""
+    """
+    How many open PRs were there this week?
+
+    The number of PRs open for DAYS_THRESHOLD (defaults to 7 days) in the
+    previous week to the given date.
+
+    Week here is defined as the dates covering the most recent Monday to Sunday
+    (inclusive) before the given date, eg if the given date is a Tuesday this
+    command will step back a week+1 day to collect a full weeks worth of data.
+    """
     date = date.date()
-    prs = api.prs_open_on_date(org, date)
+
+    end = previous_weekday(date, 6)  # Most recent Sunday
+    start = end - timedelta(days=6)  # Monday before that Sunday
+    prs = api.prs_open_in_range(org, start, end)
 
     # remove PRs which have been open <days_threshold days
     open_prs = [

metrics/tools/dates.py

@@ -22,3 +22,21 @@ def iter_days(start, end, step=DELTA):
     while start <= end:
         yield start
         start += step
+
+
+def previous_weekday(d, weekday):
+    """
+    Get the date for a previous week day
+
+    Starting at the given date, walk backwards through days until the given
+    weekday is found, returning the date for that weekday.
+
+    For example, when giving the date 2023-11-16 and asking for the previous
+    Sunday, the returned date would be 2023-11-12.
+    """
+    output = d
+
+    while output.weekday() != weekday:
+        output -= timedelta(days=1)
+
+    return output

tests/metrics/tools/test_dates.py

@@ -2,7 +2,23 @@
 
 import pytest
 
-from metrics.tools.dates import date_from_iso, datetime_from_iso, iter_days
+from metrics.tools.dates import (
+    date_from_iso,
+    datetime_from_iso,
+    iter_days,
+    previous_weekday,
+)
+
+
+# TODO: remove when we switch to 3.12, this has been added to the calendar
+# module in stdlib
+MONDAY = 0
+TUESDAY = 1
+WEDNESDAY = 2
+THURSDAY = 3
+FRIDAY = 4
+SATURDAY = 5
+SUNDAY = 6
 
 
 @pytest.mark.parametrize(
@@ -48,3 +64,20 @@ def test_iter_days_with_empty_values():
 
     with pytest.raises(TypeError):
         list(iter_days(date(2020, 7, 8), date(2022, 7, 8), None))
+
+
+@pytest.mark.parametrize(
+    "d,weekday,expected",
+    [
+        (date(2023, 11, 16), MONDAY, date(2023, 11, 13)),
+        (date(2023, 11, 16), TUESDAY, date(2023, 11, 14)),
+        (date(2023, 11, 16), WEDNESDAY, date(2023, 11, 15)),
+        (date(2023, 11, 16), THURSDAY, date(2023, 11, 16)),
+        (date(2023, 11, 16), FRIDAY, date(2023, 11, 10)),
+        (date(2023, 11, 16), SATURDAY, date(2023, 11, 11)),
+        (date(2023, 11, 16), SUNDAY, date(2023, 11, 12)),
+    ],
+    ids=["monday", "tuesday", "wednesday", "thursday", "friday", "saturday", "sunday"],
+)
+def test_previous_weekday(d, weekday, expected):
+    assert previous_weekday(d, weekday) == expected