Add geospatial/temporal bounds to Processrun's output

src/data/lrauv_deployment_plots.py

@@ -313,6 +313,7 @@ def plot_deployment(  # noqa: C901, PLR0912, PLR0913, PLR0915
                 nc_files,
                 verbose=verbose,
                 update_ssds_provenance=update_ssds_provenance,
+                force=force,
                 notify=notify,
             )
 
@@ -327,6 +328,7 @@ def _build_and_write_html(  # noqa: PLR0913
         nc_files: list[str],
         verbose: int = 0,
         update_ssds_provenance: bool = False,  # noqa: FBT001, FBT002
+        force: bool = False,  # noqa: FBT001, FBT002
         notify: str | None = None,
     ) -> None:
         """Fetch STOQS permalink and write per-PNG HTML pages."""
@@ -363,7 +365,7 @@ def _build_and_write_html(  # noqa: PLR0913
         html_paths = [
             Path(p).with_suffix(".html") for p in png_paths if Path(p).with_suffix(".html").exists()
         ]
-        self._notify(notify or "", raw_name or plot_name_stem, html_paths, stoqs_url)
+        self._notify(notify or "", raw_name or plot_name_stem, html_paths, stoqs_url, force=force)
         if update_ssds_provenance:
             self._submit_provenance(
                 deployment_dir=deployment_dir,
@@ -379,6 +381,7 @@ def _send_notify_email(
         recipient: str,
         deployment_name: str,
         html_paths: list[Path],
+        force: bool = False,  # noqa: FBT001, FBT002
     ) -> None:
         """Send a plain-HTML email with the standard inline PNG and a single web link."""
         import smtplib  # noqa: PLC0415
@@ -395,9 +398,10 @@ def _send_notify_email(
             std_png = None
         web_url = get_web_url(str(std_html)) if std_html else ""
 
+        prefix = "" if force else "New "
         sent_on = datetime.now(tz=UTC).astimezone().strftime("%Y-%m-%d %H:%M:%S %Z")
         plain = (
-            f"New LRAUV deployment plots: {deployment_name}\n\n"
+            f"{prefix}LRAUV deployment plots: {deployment_name}\n\n"
             f"  View this and related information on the web:\n  {web_url}\n\n"
             f"Sent on: {sent_on}"
         )
@@ -413,7 +417,7 @@ def _send_notify_email(
         )
 
         outer = MIMEMultipart("related")
-        outer["Subject"] = f"New LRAUV deployment plots: {deployment_name}"
+        outer["Subject"] = f"{prefix}LRAUV deployment plots: {deployment_name}"
         outer["From"] = "auv-python@mbari.org"
         outer["To"] = recipient
         alt = MIMEMultipart("alternative")
@@ -435,12 +439,13 @@ def _send_notify_email(
         except Exception as exc:  # noqa: BLE001
             self.logger.warning("Email notification failed: %s", exc)
 
-    def _notify(
+    def _notify(  # noqa: PLR0913
         self,
         target: str,
         deployment_name: str,
         html_paths: list[Path],
         stoqs_url: str | None,
+        force: bool = False,  # noqa: FBT001, FBT002
     ) -> None:
         """Send an email or Slack notification with links to the new deployment HTML pages.
 
@@ -460,7 +465,8 @@ def _notify(
             import requests  # noqa: PLC0415
 
             plot_links = [(get_web_url(str(p)), self._plot_label(str(p))) for p in html_paths]
-            lines = [f"New LRAUV deployment plots available: {deployment_name}", ""]
+            prefix = "" if force else "New "
+            lines = [f"{prefix}LRAUV deployment plots available: {deployment_name}", ""]
             for url, label in plot_links:
                 lines.append(f"  {label}: {url}")
             if stoqs_url:
@@ -475,7 +481,7 @@ def _notify(
             except Exception as exc:  # noqa: BLE001
                 self.logger.warning("Slack notification failed: %s", exc)
         else:
-            self._send_notify_email(resolved, deployment_name, html_paths)
+            self._send_notify_email(resolved, deployment_name, html_paths, force=force)
 
     def _submit_provenance(  # noqa: PLR0913
         self,

src/data/process.py

@@ -75,7 +75,7 @@ class data are: download_process and calibrate, while for LRAUV class data
 from logs2netcdfs import BASE_PATH, MISSIONLOGS, MISSIONNETCDFS, AUV_NetCDF
 from lopcToNetCDF import LOPC_Processor, UnexpectedAreaOfCode
 from nc42netcdfs import BASE_LRAUV_PATH, BASE_LRAUV_WEB, GROUP, Extract
-from provenance import get_dods_url, get_web_url, submit_process_run
+from provenance import ds_geobounds, get_dods_url, get_web_url, submit_process_run
 from resample import (
     AUVCTD_OPENDAP_BASE,
     FLASH_THRESHOLD,
@@ -811,6 +811,15 @@ def _submit_provenance(  # noqa: PLR0913
                 self.logger.debug("Output %s not found, skipping provenance", full_nc)
                 return
             log_url = get_dods_url(log_file) if log_file else None
+            import xarray as xr  # noqa: PLC0415
+
+            try:
+                _ds = xr.open_dataset(full_nc)
+                _geobounds = ds_geobounds(_ds)
+                _ds.close()
+            except Exception:  # noqa: BLE001
+                self.logger.debug("Could not extract geobounds from %s", full_nc, exc_info=True)
+                _geobounds = {}
             submit_process_run(
                 producer_name=(
                     f"auv-python - Execution of {Path(script_name).name}"
@@ -826,6 +835,7 @@ def _submit_provenance(  # noqa: PLR0913
                 log_file_url=log_url,
                 additional_resources=additional_resources,
                 log=self.logger,
+                **_geobounds,
             )
         except Exception:  # noqa: BLE001
             self.logger.warning("Provenance submission failed", exc_info=True)

src/data/provenance.py

@@ -127,6 +127,37 @@ def get_script_github_url(script_name: str) -> str:
     return get_git_url(script_name, _get_git_version())
 
 
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+def ds_geobounds(ds: object) -> dict:
+    """Return output_* kwargs for submit_process_run extracted from an xarray Dataset.
+
+    Silently skips any variable that is missing or raises on min/max.
+    """
+    import pandas as pd  # noqa: PLC0415
+
+    bounds: dict = {}
+    for var, out_min, out_max in (
+        ("latitude", "output_minlatitude", "output_maxlatitude"),
+        ("longitude", "output_minlongitude", "output_maxlongitude"),
+        ("depth", "output_mindepth", "output_maxdepth"),
+    ):
+        if var in ds:  # type: ignore[operator]
+            try:
+                bounds[out_min] = float(ds[var].min())  # type: ignore[index]
+                bounds[out_max] = float(ds[var].max())  # type: ignore[index]
+            except Exception:  # noqa: BLE001
+                logger.debug("Could not extract %s/%s bounds", out_min, out_max, exc_info=True)
+    try:
+        t = ds["time"]  # type: ignore[index]
+        bounds["output_startdate"] = pd.Timestamp(t.min().values).isoformat()
+        bounds["output_enddate"] = pd.Timestamp(t.max().values).isoformat()
+    except Exception:  # noqa: BLE001
+        logger.debug("Could not extract temporal bounds", exc_info=True)
+    return bounds
+
+
 # ---------------------------------------------------------------------------
 # Core function
 # ---------------------------------------------------------------------------
@@ -139,6 +170,14 @@ def submit_process_run(  # noqa: PLR0913
     poc_email: str = "mccann@mbari.org",
     pr_start: str | None = None,
     pr_end: str | None = None,
+    output_startdate: str | None = None,
+    output_enddate: str | None = None,
+    output_minlatitude: float | None = None,
+    output_maxlatitude: float | None = None,
+    output_minlongitude: float | None = None,
+    output_maxlongitude: float | None = None,
+    output_mindepth: float | None = None,
+    output_maxdepth: float | None = None,
     software_name: str = "auv-python",
     software_version: str | None = None,
     script_name: str = "src/data/process.py",
@@ -212,6 +251,14 @@ def submit_process_run(  # noqa: PLR0913
         "person_email": poc_email,
         "startdate": pr_start,
         "enddate": pr_end,
+        "output_startdate": output_startdate,
+        "output_enddate": output_enddate,
+        "output_minlatitude": output_minlatitude,
+        "output_maxlatitude": output_maxlatitude,
+        "output_minlongitude": output_minlongitude,
+        "output_maxlongitude": output_maxlongitude,
+        "output_mindepth": output_mindepth,
+        "output_maxdepth": output_maxdepth,
         "resources": resources,
     }
     if nc_file_path is not None: