How to determine number of records using arcpy.da.SearchCursor

[BruceBacia]

Does anyone know of a quick way to return the number of records in a query constrained Search Cursor using the da module?

rows = arcpy.da.SearchCursor(someFC,someFields,someQuery)  print (len(rows))  #I'm sure this used to work with the old cursors, but doesn't work with da.SearchCursor

[JamesCrandall]

This works (and should honor any selections):

myFeatClass = r'C:\MyGDB\myFeatureClass'
result = int(arcpy.GetCount_management(fc).getOutput(0))

print result

This also works:

rowcount = 0
myFeatClass = r'C:\MyGDB\myFeatureClass'
with arcpy.da.SearchCursor(myFeatClass, "*") as cursor:
    for row in cursor:
        count = count + 1

print rowcount

Edit:  You can also apply an sql conditional statement too

rowcount = 0
sql = "OBJECTID = 29821"
myFeatClass = r'C:\MyGDB\myFeatureClass'
with arcpy.da.SearchCursor(myFeatClass, "*", sql) as cursor:
    for row in cursor:
        count = count + 1

print rowcount

[BruceBacia]

Thanks for the suggestions, but I'm looking to avoid a counter.

[JamesCrandall]

Thanks for the suggestions, but I'm looking to avoid a counter.

arcpy.GetCount_management doesn't work for you?

[MattEiben · Accepted solution]

You were actually very close in your original entry.  Try this:

rows = [row for row in arcpy.da.SearchCursor(someFC,someFields,someQuery)]  print len(rows)

The cursor creates an iterator, which you can't find the length of.  The key is to use list comprehension for a nice one liner with the cursor to make your list.

Of course, you can always just this as an alternative:

int(arcpy.GetCount_management(someFC).getOutput(0))

Hope this helps!

[JasonScheirer]

Please, please, please use James' GetCount solution:

result = int(arcpy.GetCount_management(fc).getOutput(0))

It will be the fastest.

[FilipKrál]

Indeed, use arcpy.GetCount_management. In order to limit the number of record by a where clause, create a feature layer (or table view) first.

fc = r'c:\path\to\fc'
w = '"OBJECTID" < 10'
lr = arcpy.MakeFeatureLayer_management(fc, 'tmp_layer', w).getOutput(0)
result = int(arcpy.GetCount_management(lr).getOutput(0))
arcpy.Delete_management(lr) # deletes the layer, not the feature class data
# result is 9

[benberman]

I actually just built one like this today.

lst = [row.getvalue() for row in arcpy.SearchCursor(dataset,expression...enter other parameters you want to include)]
print len(lst)

[benberman]

I would err on the side of the SearchCursor rather than GetCount in case if a request may arise when he may need to include queries on the dataset.

[ChrisSnyder]

The number of records determines what method (GetCount or da.SearchCursor) is faster.

If the number of records is quite large, GetCount is faster, otherwise the searchcursor one liner is faster. I use the latter method (cursor) in some code that recursively (and quickly) needs to determine selected set counts.

Jason - Curious why you indicate the cursor method is a poor choice (locks?, memory?, ???).

import arcpy, time
fc = r"C:\csny490\overlay_20130620\ldo_20130620\ldo_database.gdb\crew_code"
time1 = time.clock()
result = int(arcpy.GetCount_management(fc).getOutput(0))
time2 = time.clock()
print "GetCount (" + str(result) + "), took " + str(time2-time1) + " seconds..."
time1 = time.clock()
result = len([r[0] for r in arcpy.da.SearchCursor(fc, ["OID@"])])
time2 = time.clock()
print "SearchCursor (" + str(result) + "), took " + str(time2-time1) + " seconds..." 

Some results:

Table with 8 records:
GetCount (8), took 0.755854384327 seconds...
SearchCursor (8), took 0.136624540949 seconds...

Table with 81k records:
GetCount (81327), took 0.996590826688 seconds...
SearchCursor (81327), took 0.648470391747 seconds...

Table with 2 mm records:
GetCount (2133404), took 4.37459715564 seconds...
SearchCursor (2133404), took 11.5886194669 seconds...