Merge pull request #192 from andrewm4894/precommit-improve

Precommit improve

Author: andrewm4894

.pre-commit-config.yaml

@@ -14,20 +14,10 @@ repos:
       #   args: ["--branch", "main"]
       - id: trailing-whitespace
         args: ["--markdown-linebreak-ext=md"]
-  - repo: https://github.com/pycqa/isort
-    rev: 5.12.0
-    hooks:
-      - id: isort
-        args:
-          - "--profile"
-          - "black"
-  - repo: https://github.com/charliermarsh/ruff-pre-commit
-    rev: v0.0.241
+# isort removed - ruff handles import sorting now
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.8.4
     hooks:
       - id: ruff
-        args:
-          - "--fix"
-          - "--exclude"
-          - "metrics/defaults/python/prompt.py"
-          - "--exclude"
-          - "anomstack/llm/completion.py"
+        args: ["--fix"]
+      - id: ruff-format

anomstack/alerts/asciiart.py

@@ -91,9 +91,7 @@ def __init__(
         else:
             self.graphsymbol = graphsymbol
         if self._len_noansi(self.graphsymbol) != 1:
-            raise Exception(
-                "Bad graphsymbol length, must be 1", self._len_noansi(self.graphsymbol)
-            )
+            raise Exception("Bad graphsymbol length, must be 1", self._len_noansi(self.graphsymbol))
         self.multivalue = multivalue
         self.hsymbols = [
             self._u(""),
@@ -181,9 +179,7 @@ def _get_thresholds(self, data):
                     if self.multivalue:
                         totalvalue_len += len("," + self._trans_hr(ivalue))
                     else:
-                        totalvalue_len = max(
-                            totalvalue_len, len(self._trans_hr(ivalue))
-                        )
+                        totalvalue_len = max(totalvalue_len, len(self._trans_hr(ivalue)))
 
                 if self.multivalue:
                     # remove one comma if multivalues
@@ -214,17 +210,13 @@ def _gen_graph_string(
     ):
         """Generate the bar + its paddings (left and right)"""
 
-        def _gen_graph_string_part(
-            value, max_value, min_neg_value, graph_length, color
-        ):
+        def _gen_graph_string_part(value, max_value, min_neg_value, graph_length, color):
             all_width = max_value + abs(min_neg_value)
 
             if all_width == 0:
                 bar_width = 0
             else:
-                bar_width = int(
-                    abs(float(value)) * float(graph_length) / float(all_width)
-                )
+                bar_width = int(abs(float(value)) * float(graph_length) / float(all_width))
 
             return (
                 Pyasciigraph._color_string(self.graphsymbol * bar_width, color),
@@ -236,9 +228,7 @@ def _gen_graph_string_part(
         if all_width == 0:
             neg_width = 0
         else:
-            neg_width = int(
-                abs(float(min_neg_value)) * float(graph_length) / float(all_width)
-            )
+            neg_width = int(abs(float(min_neg_value)) * float(graph_length) / float(all_width))
             int(abs(max_value) * graph_length / all_width)
 
         if isinstance(value, Iterable):
@@ -264,9 +254,7 @@ def _gen_graph_string_part(
                 accuvalue += scaled_value
 
             # left padding
-            totalstring = (
-                Pyasciigraph._u(" ") * (neg_width - abs(totalsquares)) + totalstring
-            )
+            totalstring = Pyasciigraph._u(" ") * (neg_width - abs(totalsquares)) + totalstring
 
             # reset some counters
             accuvalue = 0
@@ -285,9 +273,7 @@ def _gen_graph_string_part(
                 accuvalue += scaled_value
 
             # right padding
-            totalstring += Pyasciigraph._u(" ") * (
-                start_value_pos - neg_width - abs(totalsquares)
-            )
+            totalstring += Pyasciigraph._u(" ") * (start_value_pos - neg_width - abs(totalsquares))
             return totalstring
         else:
             # handling for single value item
@@ -312,9 +298,7 @@ def _gen_info_string(self, info, start_info_pos, line_length):
         number_of_space = line_length - start_info_pos - self._len_noansi(info)
         return info + Pyasciigraph._u(" ") * number_of_space
 
-    def _gen_value_string(
-        self, value, min_neg_value, color, start_value_pos, start_info_pos
-    ):
+    def _gen_value_string(self, value, min_neg_value, color, start_value_pos, start_info_pos):
         """Generate the value string + padding"""
         icount = 0
         if isinstance(value, Iterable) and self.multivalue:
@@ -324,14 +308,10 @@ def _gen_value_string(
                     # with the len() function even when they are not printed to
                     # the screen.
                     totalvalue_len = len(self._trans_hr(ivalue))
-                    totalvalue = Pyasciigraph._color_string(
-                        self._trans_hr(ivalue), icolor
-                    )
+                    totalvalue = Pyasciigraph._color_string(self._trans_hr(ivalue), icolor)
                 else:
                     totalvalue_len += len("," + self._trans_hr(ivalue))
-                    totalvalue += "," + Pyasciigraph._color_string(
-                        self._trans_hr(ivalue), icolor
-                    )
+                    totalvalue += "," + Pyasciigraph._color_string(self._trans_hr(ivalue), icolor)
                 icount += 1
         elif isinstance(value, Iterable):
             max_value = min_neg_value
@@ -347,9 +327,7 @@ def _gen_value_string(
             totalvalue_len = len(self._trans_hr(value))
             totalvalue = Pyasciigraph._color_string(self._trans_hr(value), color)
 
-        number_space = (
-            start_info_pos - start_value_pos - totalvalue_len - self.separator_length
-        )
+        number_space = start_info_pos - start_value_pos - totalvalue_len - self.separator_length
 
         # This must not be negitive, this happens when the string length is
         # larger than the separator length
@@ -457,18 +435,14 @@ def graph(self, label=None, data=[]):
             # calcul of where to start info
             start_info_pos = self.line_length - all_thre["info_max_length"]
             # calcul of where to start value
-            start_value_pos = (
-                start_info_pos - self.separator_length - all_thre["value_max_length"]
-            )
+            start_value_pos = start_info_pos - self.separator_length - all_thre["value_max_length"]
             # calcul of where to end graph
             graph_length = start_value_pos - self.separator_length
         else:
             # calcul of where to start value
             start_value_pos = self.min_graph_length + self.separator_length
             # calcul of where to start info
-            start_info_pos = (
-                start_value_pos + all_thre["value_max_length"] + self.separator_length
-            )
+            start_info_pos = start_value_pos + all_thre["value_max_length"] + self.separator_length
             # calcul of where to end graph
             graph_length = start_value_pos - self.separator_length
             # calcul of the real line length
@@ -514,32 +488,18 @@ def make_alert_message(
     score_col="metric_score_smooth",
     ascii_graph=False,
 ):
-    df_alert_metric = df_alert_metric.sort_values(
-        by="metric_timestamp", ascending=False
-    ).dropna()
-    df_alert_metric["metric_timestamp"] = pd.to_datetime(
-        df_alert_metric["metric_timestamp"]
-    )
+    df_alert_metric = df_alert_metric.sort_values(by="metric_timestamp", ascending=False).dropna()
+    df_alert_metric["metric_timestamp"] = pd.to_datetime(df_alert_metric["metric_timestamp"])
     x = df_alert_metric["metric_value"].round(2).values.tolist()
     metric_name = df_alert_metric["metric_name"].unique()[0]
-    metric_timestamp_from = (
-        df_alert_metric["metric_timestamp"].min().strftime("%Y-%m-%d %H:%M")
-    )
-    metric_timestamp_to = (
-        df_alert_metric["metric_timestamp"].max().strftime("%Y-%m-%d %H:%M")
-    )
+    metric_timestamp_from = df_alert_metric["metric_timestamp"].min().strftime("%Y-%m-%d %H:%M")
+    metric_timestamp_to = df_alert_metric["metric_timestamp"].max().strftime("%Y-%m-%d %H:%M")
     graph_title = f"{metric_name} ({metric_timestamp_from} to {metric_timestamp_to})"
     message = ""
     if ascii_graph:
         labels = (
-            np.where(
-                df_alert_metric["metric_alert"] == 1,
-                anomaly_symbol,
-                normal_symbol
-            )
-            + (df_alert_metric[score_col].round(2) * 100)
-            .astype("int")
-            .astype("str")
+            np.where(df_alert_metric["metric_alert"] == 1, anomaly_symbol, normal_symbol)
+            + (df_alert_metric[score_col].round(2) * 100).astype("int").astype("str")
             + "% "
         )
         data = zip(labels, x)

anomstack/alerts/email.py

@@ -2,14 +2,14 @@
 Helper functions for sending alerts via email.
 """
 
-import os
-import smtplib
-import ssl
-import tempfile
 from email import encoders
 from email.mime.base import MIMEBase
 from email.mime.multipart import MIMEMultipart
 from email.mime.text import MIMEText
+import os
+import smtplib
+import ssl
+import tempfile
 
 from dagster import get_dagster_logger
 
@@ -55,12 +55,8 @@ def send_email_with_plot(
     host = os.getenv("ANOMSTACK_ALERT_EMAIL_SMTP_HOST")
     port = os.getenv("ANOMSTACK_ALERT_EMAIL_SMTP_PORT")
 
-    with tempfile.NamedTemporaryFile(
-        prefix=attachment_name, suffix=".png", delete=False
-    ) as temp:
-        fig = make_alert_plot(
-            df, metric_name, threshold, score_col, score_title, tags=tags
-        )
+    with tempfile.NamedTemporaryFile(prefix=attachment_name, suffix=".png", delete=False) as temp:
+        fig = make_alert_plot(df, metric_name, threshold, score_col, score_title, tags=tags)
         fig.savefig(temp.name)
 
         msg = MIMEMultipart()
@@ -70,14 +66,10 @@ def send_email_with_plot(
 
         msg.attach(MIMEText(body, "html"))
         binary_file = open(temp.name, "rb")
-        payload = MIMEBase(
-            "application", "octate-stream", Name=f"{attachment_name}.png"
-        )
+        payload = MIMEBase("application", "octate-stream", Name=f"{attachment_name}.png")
         payload.set_payload((binary_file).read())
         encoders.encode_base64(payload)
-        payload.add_header(
-            "Content-Decomposition", "attachment", filename=f"{attachment_name}.png"
-        )
+        payload.add_header("Content-Decomposition", "attachment", filename=f"{attachment_name}.png")
         msg.attach(payload)
 
         context = ssl.create_default_context()
@@ -126,7 +118,7 @@ def send_email(
     with smtplib.SMTP(host, port) as server:
         server.connect(host, port)
         server.starttls(context=context)
-        server.login(sender, password) # type: ignore
+        server.login(sender, password)  # type: ignore
         text = msg.as_string()
         server.sendmail(sender, to, text)
         server.quit()

anomstack/alerts/send.py

@@ -2,8 +2,8 @@
 Helper functions to send alerts.
 """
 
-import pandas as pd
 from dagster import get_dagster_logger
+import pandas as pd
 
 from anomstack.alerts.asciiart import make_alert_message
 from anomstack.alerts.email import send_email, send_email_with_plot
@@ -46,9 +46,7 @@ def send_alert(
     """
     logger = get_dagster_logger()
     logger.debug(f"alerts to send: \n{df}")
-    message = make_alert_message(
-        df, description=description, tags=tags, score_col=score_col
-    )
+    message = make_alert_message(df, description=description, tags=tags, score_col=score_col)
     if "slack" in alert_methods:
         send_alert_slack_with_plot(
             df=df,

anomstack/alerts/slack.py

@@ -41,8 +41,8 @@
 import os
 import tempfile
 
-import matplotlib.pyplot as plt
 from dagster import get_dagster_logger
+import matplotlib.pyplot as plt
 from slack_sdk import WebClient
 from slack_sdk.errors import SlackApiError
 
@@ -73,10 +73,7 @@ def send_alert_slack(
     slack_token = os.environ.get("ANOMSTACK_SLACK_BOT_TOKEN")
     if not slack_token:
         raise ValueError(
-            (
-                "Slack bot token not found in environment variable "
-                "ANOMSTACK_SLACK_BOT_TOKEN"
-            )
+            ("Slack bot token not found in environment variable " "ANOMSTACK_SLACK_BOT_TOKEN")
         )
 
     client = WebClient(token=slack_token)
@@ -119,9 +116,7 @@ def send_alert_slack(
                     filename=os.path.basename(image_file_path),
                 )
         else:
-            response = client.chat_postMessage(
-                channel=channel_id, text=f"*{title}*\n{message}"
-            )
+            response = client.chat_postMessage(channel=channel_id, text=f"*{title}*\n{message}")
     except SlackApiError as e:
         logger.error(f"Error sending message to Slack channel {channel_name, channel_id}: {e}")
 
@@ -146,13 +141,9 @@ def send_alert_slack_with_plot(
         channel_name = os.environ.get("ANOMSTACK_SLACK_CHANNEL")
 
     with tempfile.NamedTemporaryFile(
-        prefix=f"{metric_name}_{metric_timestamp}_",
-        suffix=".png",
-        delete=False
+        prefix=f"{metric_name}_{metric_timestamp}_", suffix=".png", delete=False
     ) as temp:
-        fig = make_alert_plot(
-            df, metric_name, threshold, score_col, score_title, tags=tags
-        )
+        fig = make_alert_plot(df, metric_name, threshold, score_col, score_title, tags=tags)
         fig.savefig(temp.name)
         plt.close(fig)