Merge pull request #127 from mhearne-usgs/magfix

Moved get all mags functionality into utils module.

Author: Mike Hearne

bin/getmags

@@ -3,59 +3,16 @@ import argparse
 import sys
 
 from libcomcat.search import search, count
-from libcomcat.utils import get_detail_data_frame, get_summary_data_frame, maketime
-from impactutils.time.ancient_time import HistoricTime
+from libcomcat.utils import (maketime, get_all_mags)
 from obspy.clients.fdsn import Client
 import pandas as pd
 
 
-class CustomFormatter(argparse.ArgumentDefaultsHelpFormatter, argparse.RawDescriptionHelpFormatter):
+class CustomFormatter(argparse.ArgumentDefaultsHelpFormatter,
+                      argparse.RawDescriptionHelpFormatter):
     pass
 
 
-def get_mag_src(mag):
-    """Try to find the best magnitude source from a Magnitude object.
-
-    Note: This can be difficult, as there is a great deal of variance
-    in how magnitude information is submitted in QuakeML within ComCat.
-
-    Args:
-        mag (obspy Magnitude): Magnitude object from obspy.
-    Returns:
-        str: String indicating the most likely source of the magnitude solution.
-
-    """
-    if mag.creation_info is not None and mag.creation_info.agency_id is not None:
-        magsrc = mag.creation_info.agency_id.lower()
-    else:
-        has_gcmt = mag.resource_id.id.lower().find('gcmt') > -1
-        has_at = mag.resource_id.id.lower().find('at') > -1
-        has_pt = mag.resource_id.id.lower().find('pt') > -1
-        has_ak = (mag.resource_id.id.lower().find('ak') > -1 or
-                  mag.resource_id.id.lower().find('alaska') > -1)
-        has_pr = mag.resource_id.id.lower().find('pr') > -1
-        has_dup = mag.resource_id.id.lower().find('duputel') > -1
-        has_us = mag.resource_id.id.lower().find('us') > -1
-        if has_gcmt:
-            magsrc = 'gcmt'
-        elif has_dup:
-            magsrc = 'duputel'
-        elif has_at:
-            magsrc = 'at'
-        elif has_pt:
-            magsrc = 'pt'
-        elif has_ak:
-            magsrc = 'ak'
-        elif has_pr:
-            magsrc = 'pr'
-        elif has_us:
-            magsrc = 'us'
-        else:
-            magsrc = 'unknown'
-
-    return magsrc
-
-
 def get_parser():
     desc = '''Download epicenter and all contributed magnitudes in line format (csv, tab, etc.).
 
@@ -186,7 +143,6 @@ def main():
     # create a dataframe with these columns - we'll add more later
     df = pd.DataFrame(columns=['id', 'time', 'lat', 'lon', 'depth',
                                'location', 'url', 'hypo_src'])
-    client = Client('USGS')
     ievent = 1
 
     for event in events:
@@ -208,24 +164,14 @@ def main():
             print('Parsing event %s (%i of %i) - %i origins' % tpl)
         ievent += 1
         errors = []
+        mags = {}
         for eid in id_list:
-            try:
-                obsevent = client.get_events(eventid=eid).events[0]
-            except Exception as e:
-                msg = 'Failed to download data for event %s.  Skipping.' % eid
-                errors.append(msg)
-            for mag in obsevent.magnitudes:
-                magvalue = mag.mag
-                magtype = mag.magnitude_type
-                magsrc = get_mag_src(mag)
-                if magsrc == 'unknown' and args.verbose:
-                    print('Unable to determine magnitude source from %s' % eid)
-                colname = '%s-%s' % (magsrc, magtype)
-                if colname in row:
-                    # print('Duplicate column %s for event %s - skipping' % (colname,event.id))
-                    continue
-                row[colname] = magvalue
-                imag += 1
+            magtypes, msg = get_all_mags(eid)
+            if args.verbose and len(msg):
+                print(msg)
+            mags.update(magtypes)
+            imag += 1
+        row = pd.concat([row, pd.Series(mags)])
         df = df.append(row, ignore_index=True)
 
     if len(errors):
@@ -234,9 +180,9 @@ def main():
             print('\t%s' % error)
 
     if args.format == 'excel':
-        df.to_excel(args.filename)
+        df.to_excel(args.filename, index=False)
     else:
-        df.to_csv(args.filename)
+        df.to_csv(args.filename, index=False)
     print('%i records saved to %s.' % (len(df), args.filename))
     sys.exit(0)
 

libcomcat/utils.py

@@ -10,6 +10,7 @@
 import numpy as np
 import pandas as pd
 from obspy.io.quakeml.core import Unpickler
+from obspy.clients.fdsn import Client
 from impactutils.time.ancient_time import HistoricTime
 from openpyxl import load_workbook
 
@@ -22,6 +23,83 @@
 DATEFMT = '%Y-%m-%d'
 
 
+def get_mag_src(mag):
+    """Try to find the best magnitude source from a Magnitude object.
+
+    Note: This can be difficult, as there is a great deal of variance
+    in how magnitude information is submitted in QuakeML within ComCat.
+
+    Args:
+        mag (obspy Magnitude): Magnitude object from obspy.
+    Returns:
+        str: String indicating the most likely source of the
+             magnitude solution.
+
+    """
+    c1 = mag.creation_info is not None
+    if c1:
+        c2 = mag.creation_info.agency_id is not None
+    else:
+        c2 = False
+    if c2:
+        magsrc = mag.creation_info.agency_id.lower()
+    else:
+        has_gcmt = mag.resource_id.id.lower().find('gcmt') > -1
+        has_at = mag.resource_id.id.lower().find('at') > -1
+        has_pt = mag.resource_id.id.lower().find('pt') > -1
+        has_ak = (mag.resource_id.id.lower().find('ak') > -1 or
+                  mag.resource_id.id.lower().find('alaska') > -1)
+        has_pr = mag.resource_id.id.lower().find('pr') > -1
+        has_dup = mag.resource_id.id.lower().find('duputel') > -1
+        has_us = mag.resource_id.id.lower().find('us') > -1
+        if has_gcmt:
+            magsrc = 'gcmt'
+        elif has_dup:
+            magsrc = 'duputel'
+        elif has_at:
+            magsrc = 'at'
+        elif has_pt:
+            magsrc = 'pt'
+        elif has_ak:
+            magsrc = 'ak'
+        elif has_pr:
+            magsrc = 'pr'
+        elif has_us:
+            magsrc = 'us'
+        else:
+            magsrc = 'unknown'
+
+    return magsrc
+
+
+def get_all_mags(eventid):
+    """Get all magnitudes for a given event ID.
+
+    Args:
+        eventid (str): ComCat Event ID.
+    Returns:
+        dict: Dictionary where keys are "magsrc-magtype" and values
+              are magnitude value.
+
+    """
+    row = {}
+    msg = ''
+    client = Client('USGS')
+    try:
+        obsevent = client.get_events(eventid=eventid).events[0]
+    except Exception as e:
+        msg = 'Failed to download event %s, error "%s".' % (eventid, str(e))
+    for mag in obsevent.magnitudes:
+        magvalue = mag.mag
+        magtype = mag.magnitude_type
+        magsrc = get_mag_src(mag)
+        colname = '%s-%s' % (magsrc, magtype)
+        if colname in row:
+            continue
+        row[colname] = magvalue
+    return (row, msg)
+
+
 def read_phases(filename):
     """Read a phase file CSV or Excel file into data structures.