topic Re: arcgis.geometry Polygon not converting polyline to polygon in ArcGIS API for Python Questions

arcgis.geometry Polygon not converting polyline to polygon

TylerT — Fri, 11 Mar 2022 15:39:58 GMT

Community,

I have a spatially enabled dataframe (SEDF) with polyline geometry. Ultimately, I need polygon geometry.

My attempt to use arcgis.geometry is shown below. The Polygon() method doesn’t throw any errors but it also doesn’t seem to do anything. The geometry remains a polygon.

Any suggestions would be appreciated. Also, I don’t necessarily need to employ the method I’m trying, but it is desirable to stay within the SEDF.

Regards,

Tyler

Re: arcgis.geometry Polygon not converting polyline to polygon

jcarlson — Fri, 11 Mar 2022 15:50:26 GMT

Any chance you've got a subset of the data to test this on? I've had some luck with accessing the geometry dict and renaming the keys (in this case "paths" → "rings"), then using the Geometry constructor, but I don't have a concrete example handy.

Re: arcgis.geometry Polygon not converting polyline to polygon

DavidAnderson_1701 — Fri, 11 Mar 2022 16:10:05 GMT

The lambda function is doing exactly what you specified - returning the polygon calculation. Recall, this is a function, it returns a value. What is missing is a destination for function return value. The SHAPE field is unchanged because the code does not use that field to store the function results.

Try this

df['SHAPE'] = df['SHAPE'].apply(lambda x: Polygon(x))

Re: arcgis.geometry Polygon not converting polyline to polygon

TylerT — Fri, 11 Mar 2022 17:26:03 GMT

@jcarlson Thanks for looking at this. Here is one polyline:

{'paths': [[[14581, 6618.75], [14583, 6618.75], [14583, 6756.75], [14427, 6756.75], [14427, 6748.75], [14415, 6748.75], [14415, 6618.75], [14581, 6618.75]]], 'spatialReference': {'wkid': None}}

Before posting I had simply tried replacing 'paths' with 'rings' with no luck, but didn't dig any further. Let me know if there are other syntax changes required. Side note, seem like arcgis.geometry would have a simple type caster. The documentation implies polyline to polygon using Polygon should work stating "The same patterna can be repeated for Polygon," at https://developers.arcgis.com/python/api-reference/arcgis.geometry.html, where the example is from line to polyline, so no example of polyline to polygon.

@DavidAnderson_1701, first things first...I need df['SHAPE'].apply(lambda x: Polygon(x)) to return a polygon vice polyline as shown. I'll take care of the assignment later once debugged. Thanks for your input though. Any other thoughts?

Re: arcgis.geometry Polygon not converting polyline to polygon

jcarlson — Fri, 11 Mar 2022 17:47:53 GMT

Seems to work just fine at a single-geometry level.

geom = { 'paths': [[ [14581, 6618.75], [14583, 6618.75], [14583, 6756.75], [14427, 6756.75], [14427, 6748.75], [14415, 6748.75], [14415, 6618.75], [14581, 6618.75] ]], 'spatialReference': {'wkid': None} } geom['rings'] = geom.pop('paths') arcgis.geometry.Geometry(geom).type

Returns

Polygon

But perhaps things operate a bit different at the SEDF level? Could you attach the dataframe itself, or just a subset of it?

sedf.head(10).to_pickle('pline_to_pgon_test.pkl')

Re: arcgis.geometry Polygon not converting polyline to polygon

Anonymous User — Fri, 11 Mar 2022 18:09:11 GMT

@TylerT this is the correct answer. You need an assignment statement.

When you print out the type() in your last line of code, it's not doing what you think it's doing. It's running the apply function, but not saving the results. So unless you assign the results, then the type will not change.

Re: arcgis.geometry Polygon not converting polyline to polygon

TylerT — Fri, 11 Mar 2022 18:15:10 GMT

I wasn't using your technique. I'll give that a go, and let you know. If it fails, I'll post a subset. Thanks @jcarlson.

Re: arcgis.geometry Polygon not converting polyline to polygon

TylerT — Sat, 12 Mar 2022 20:01:58 GMT

@jcarlson; here's what I came up with using your pop suggestion. I'm not thrilled with it, as I'll explain more below.

Test 1) Overwrite 'SHAPE' with apply pl_to_pg. This did not work because the 'SHAPE' column ended up losing it's status as geometry type and therefore the entire df lost status as SEDF. Converting back to SEDF with pd.DataFrame.spatial.from_df(df) would then throw an error.

Test 2) Create new 'SHAPE_PG' (polygon) column with apply pl_to_pg. This didn't work, because oddly, the 'apply' method changed both the 'SHAPE' column values and 'SHAPE_PG' columns values, again losing df status as SEDF. This is not normal behavior for apply. Something to do with geometry type, maybe?

Final) Back to the drawing board. You see finally I went from type_polyline to type JSON to type_polygon, then set the new geometry column to 'SHAPE_PG'. Creating a JSON column allowed apply to behave and SEDF status held, but seems a bit Rube Goldberg. following are the top ten csv polyline records. Let me know if you can come up with something cleaner.

Thank you,

Tyler

idx,FID,SHAPE 2,3,"{'paths': [[[15033.75, 5868.75], [15028.75, 5868.75], [15028.75, 5878.6875], [14842.25, 5878.6875], [14842.25, 5902.6875], [14689.25, 5902.6875], [14689.25, 5652.6875], [14716.75, 5652.6875], [14716.75, 5638.125], [14935.25, 5638.125], [14935.25, 5758.5], [14940.25, 5758.5], [14940.25, 5638.125], [15028.75, 5638.125], [15028.75, 5717.75], [15033.75, 5717.75], [15033.75, 5868.75]]], 'spatialReference': {'wkid': None}}" 3,4,"{'paths': [[[14581, 6618.75], [14583, 6618.75], [14583, 6756.75], [14427, 6756.75], [14427, 6748.75], [14415, 6748.75], [14415, 6618.75], [14581, 6618.75]]], 'spatialReference': {'wkid': None}}" 4,5,"{'paths': [[[14539, 6566.75], [14539, 6614.5], [14415, 6614.5], [14415, 6430.5], [14539, 6430.5], [14539, 6566.75]]], 'spatialReference': {'wkid': None}}" 5,6,"{'paths': [[[14543.25, 6566.75], [14543.25, 6430.5], [14602, 6430.500000000001], [14602, 6378], [14656, 6378], [14656, 6388.5], [14670, 6388.5], [14670, 6378], [14810.75, 6378], [14810.75, 6641.25], [14763, 6641.25], [14763, 6492], [14583, 6492], [14583, 6614.5], [14543.25, 6614.5], [14543.25, 6566.75]]], 'spatialReference': {'wkid': None}}" 6,7,"{'paths': [[[14587.25, 6544.5], [14587.25, 6496.5], [14670.25, 6496.5], [14670.25, 6641.25], [14587.25, 6641.25], [14587.25, 6544.5]]], 'spatialReference': {'wkid': None}}" 7,8,"{'paths': [[[14674.75, 6496.5], [14758.5, 6496.5], [14758.5, 6641.25], [14674.75, 6641.25], [14674.75, 6496.5]]], 'spatialReference': {'wkid': None}}" 8,9,"{'paths': [[[14670.25, 6748.5], [14587.25, 6748.5], [14587.25, 6646.5], [14670.25, 6646.5], [14670.25, 6748.5]]], 'spatialReference': {'wkid': None}}" 9,10,"{'paths': [[[14805.25, 6646.5], [14810.75, 6646.5], [14810.75, 6748.5], [14674.75, 6748.5], [14674.75, 6646.5], [14805.25, 6646.5]]], 'spatialReference': {'wkid': None}}" 10,11,"{'paths': [[[14816, 6639.25], [14816, 6593.5], [14912, 6593.5], [14912, 6748.5], [14816, 6748.5], [14816, 6639.25]]], 'spatialReference': {'wkid': None}}" 11,12,"{'paths': [[[14816, 6473.5], [14912, 6473.5], [14912, 6588.25], [14816, 6588.25], [14816, 6473.5]]], 'spatialReference': {'wkid': None}}"

Re: arcgis.geometry Polygon not converting polyline to polygon

jcarlson — Sun, 13 Mar 2022 03:37:11 GMT

It's hard to say for sure, since the dataframe I create from your text is not necessarily the same dataframe you're working with. I'm able to overwrite the shape column without the issues you're describing.

Any chance you could attach a pickled dataframe? That would ensure that the data I'm testing is the same as what you have.

Re: arcgis.geometry Polygon not converting polyline to polygon

TylerT — Sun, 13 Mar 2022 14:59:55 GMT

Sorry about that. You asked for that before and I wasn't sure what it was. Now I know. Let me know what you think. Thank you.

Kind Regards,

Tyler

Re: arcgis.geometry Polygon not converting polyline to polygon

DavidAnderson_1701 — Mon, 14 Mar 2022 15:32:50 GMT

The above code shows how I was able to get this to work. The SEDF Polygon operator does not quite work as I thought it should. This statement did not wori:

temp_df['new_geom'] = temp_df['coor'].apply(lambda x : arcgis.geometry.Polygon(x))

I dropped back to using arcpy geometry constructors, which seems to work.

Note this is hard coded to just work for a single part polygon with no holes. A more general formalation would be extract the arcpy code into a separate function.

Re: arcgis.geometry Polygon not converting polyline to polygon

TylerT — Wed, 16 Mar 2022 17:34:13 GMT

@jcarlson, any findings with the pickle file?

Re: arcgis.geometry Polygon not converting polyline to polygon

jcarlson — Wed, 16 Mar 2022 17:48:05 GMT

No luck. I get an AttributeError when I attempt to read that file into a new DataFrame:

'DataFrame' object has no attribute '_data'

Re: arcgis.geometry Polygon not converting polyline to polygon

DavidAnderson_1701 — Wed, 16 Mar 2022 21:39:52 GMT

I read in the pickle file, then successfully applied the method I desribed above:

test_df = pd.read_pickle(r"polyline_sample.pkl")

test_df['coor'] = test_df['SHAPE'].apply(lambda x : arcgis.geometry.Polyline.coordinates(x))

test_df['poly_geom'] = test_df['coor'].apply(lambda x : arcpy.Polygon(arcpy.Array([arcpy.Point(*coords) for coords in x[0]])))

test_df.spatial.set_geometry('poly_geom',inplace=True)

Re: arcgis.geometry Polygon not converting polyline to polygon

jcarlson — Wed, 16 Mar 2022 21:43:03 GMT

Excellent! Thanks for doing that! I have no idea why pandas didn't want to read the file in my env, but I had too much going on to really dig into it further.

Re: arcgis.geometry Polygon not converting polyline to polygon

Anonymous User — Thu, 17 Mar 2022 00:09:46 GMT

Worth noting for posterity that this method is dangerous. This is why I couldn't test this data sample.

From the Python docs:

Warning
The pickle module is not secure. Only unpickle data you trust.
It is possible to construct malicious pickle data which will execute arbitrary code during unpickling. Never unpickle data that could have come from an untrusted source, or that could have been tampered with.
Consider signing data with hmac if you need to ensure that it has not been tampered with.
Safer serialization formats such as json may be more appropriate if you are processing untrusted data. See Comparison with json.