From 1885e2fb223d7f262aca24283ef389ac2934d406 Mon Sep 17 00:00:00 2001
From: Alex Wernle <awernle@usgs.gov>
Date: Fri, 4 Oct 2024 10:10:07 -0600
Subject: [PATCH] Deleted field-work attribute and added Field_Work to
categories. Deleted source and added additional validators and class methods
to ArtificialDisturbances.
---
geomagio/metadata/flag/Flag.py | 78 +++++++++++++++++++++-------------
1 file changed, 48 insertions(+), 30 deletions(-)
diff --git a/geomagio/metadata/flag/Flag.py b/geomagio/metadata/flag/Flag.py
index a16c6363..b9d4c98e 100644
--- a/geomagio/metadata/flag/Flag.py
+++ b/geomagio/metadata/flag/Flag.py
@@ -1,7 +1,6 @@
from typing import Dict, Union, List, Optional
from datetime import timedelta
-import numpy as np
from obspy import UTCDateTime
from pydantic import BaseModel, Field, validator
from enum import Enum
@@ -11,6 +10,7 @@ class FlagCategory(str, Enum):
ARTIFICIAL_DISTURBANCE = "ARTIFICIAL_DISTURBANCE"
GAP = "GAP"
EVENT = "EVENT"
+ FIELD_WORK = "FIELD_WORK"
OTHER = "OTHER"
@@ -23,30 +23,26 @@ class Flag(BaseModel):
automatic_flag = Metadata(
created_by='ex_algorithm',
start_time=UTCDateTime('2023-01-01T03:05:10'),
- end_time=UTCDateTime('2023-01-01T03:05:11'),
+ end_time=UTCDateTime('2023-01-01T03:10:11'),
network='NT',
station='BOU',
channel='BEH',
category=MetadataCategory.FLAG,
comment="spike detected",
priority=1,
- data_valid=False,
+ data_valid=True,
metadata= ArtificialDisturbance{
- "description": "Spike in magnetic field strength",
- "field_work": false,
- "corrected": false,
+ "description": "Spikes in magnetic field strength",
"flag_category": ARTIFICIAL_DISTURBANCE,
"artificial_disturbance_type": ArtificialDisturbanceType.SPIKE,
- "source": "Lightning",
"deviation": None,
+ "spikes": ['2023-01-01T03:05:10','2023-01-01T03:07:20','2023-01-01T03:10:11']
}
)
```
"""
description: str = Field(..., description="Description of the flag")
- field_work: bool = Field(..., description="Flag signaling field work")
- corrected: int = Field(..., description="Corrected ID for processing stage")
flag_category: FlagCategory = "OTHER"
@@ -70,24 +66,56 @@ class ArtificialDisturbance(Flag):
----------
artificial_disturbance_type:ArtificialDisturbanceType
The type of artificial disturbance(s).
- source: str
- Source of the disturbance if known or suspected.
deviation: float
Deviation of an offset in nt.
- spikes: np.ndarray
- NumPy array of timestamps as UTCDateTime. Can be a single spike or many spikes.
+ spikes: List[UTCDateTime]
+ Array of timestamps as UTCDateTime. Can be a single spike or many spikes.
"""
artificial_disturbance_type: ArtificialDisturbanceType
deviation: Optional[float] = None
- source: Optional[str] = None
- spikes: Optional[np.ndarray] = None
+ spikes: Optional[List[UTCDateTime]] = None
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.flag_category = "ARTIFICIAL_DISTURBANCE"
+ @validator("spikes", always=True)
+ def validate_spikes(cls, spikes):
+ if len(spikes) == 0:
+ raise ValueError("Spikes list cannot be empty.")
+
+ return spikes
+
+ @validator("spikes", always=True)
+ def check_spikes_duration(cls, spikes):
+ if spikes is None or len(spikes) < 2:
+ return spikes
+
+ duration = spikes[-1] - spikes[0]
+ if duration > timedelta(days=1).total_seconds():
+ raise ValueError(
+ f"The duration between the first and last spike timestamp must not exceed 1 day. Duration: {duration} seconds"
+ )
+
+ return spikes
+
+ @classmethod
+ def check_spikes_match_times(cls, spikes, values):
+ metadata_starttime = values.get("starttime")
+ metadata_endtime = values.get("endtime")
+
+ if spikes[0] != metadata_starttime:
+ raise ValueError(
+ f"The first spike timestamp {spikes[0]} does not match the starttime {metadata_starttime}."
+ )
+
+ if spikes[-1] != metadata_endtime:
+ raise ValueError(
+ f"The last spike timestamp {spikes[-1]} does not match the endtime {metadata_endtime}."
+ )
+
class Gap(Flag):
"""
@@ -145,37 +173,27 @@ class Event(Flag):
# More example usage:
-timestamps_array = np.array(
- [
- UTCDateTime("2023-11-16T12:00:0"),
- UTCDateTime("2023-11-16T12:01:10"),
- UTCDateTime("2023-11-16T12:02:30"),
- ]
-)
+timestamps_array = [
+ UTCDateTime("2023-11-16T12:00:0"),
+ UTCDateTime("2023-11-16T12:01:10"),
+ UTCDateTime("2023-11-16T12:02:30"),
+]
spikes_data = {
"starttime": "2023-11-16 12:00:00",
"endtime": "2023-11-16 12:02:30",
"description": "Spikes description",
- "field_work": False,
- "corrected": 32265,
"disturbance_type": ArtificialDisturbanceType.SPIKES,
- "source": "processing",
"spikes": timestamps_array,
}
offset_data = {
"description": "Offset description",
- "field_work": False,
- "corrected": 47999,
"disturbance_type": ArtificialDisturbanceType.OFFSET,
- "source": "Bin change",
"deviation": 10.0,
}
geomagnetic_storm_data = {
"description": "Geomagnetic storm",
- "field_work": False,
- "corrected": 36999,
"event_type": EventType.GEOMAGNETIC_STORM,
"scale": "G3",
"index": 7,
--
GitLab