Fail to convert lines to polygons correctly,

Hi Jamal,

 

 

I think segment snapping before the Intersector would be effective. And stroking arcs (and ellipses if exist) before the snapping might also be effective.

 

 

In addition, found these exceptional cases in the data.

 

1) three polygons that contain multiple texts for each.

 

2) one polygon that contains no text.

 

3) one text that does not correspond to any polygon.

 

4) a few lines that don't become polygon edges.

 

 

Takashi

5) two texts that duplicate about both geometry and text string.

I agree with Takashi, serious data integrity checks and cleaning should be done before transformation to polygons.

Thanks Takashi and ETLS-Itay,

 

Sure the CAD data is really “dirty”. This is why we need to clean it as much as possible.

 

1. I have applied the ArcStroker

 

2. But couldn’t find the “segment snapping” tool.

 

 

However, I think that the number of polygons derived from lines should be CLOSE to the number of texts. Otherwise, this will indicate that the lines are not yet cleaned.

 

 

By the way, there is no effect neither for the “ArcStroker” nor for the “matcher”.

 

 

What other tools should be applied to make the number of polygons quite close to the number of texts?

 

 

 

 

Take a look at the "Snapping Type" parameter of the Snapper transformer.

You are correct Takashi,

 

The data is dirty to a degree that all tools will never clean it. I check the resulted polygons and found out many slivers.

 

Better if the data is correct at the level of AutoCAD and then applying FME.

 

 

 

 

Why not try Snapper (Snapping Type: Segment Snapping) ?

 

I inspected your data with this workspace, FYI.

 

 

Many thanks Takashi. It is much better now with the snapper. But we still need some manual work on the original CAD data.

 

 

 

 

 

 

 

 

Best

 

 

Jamal

Please check again my first post in this thread.

 

"I think segment snapping before the Intersector would be effective. And stroking arcs (and ellipses if exist) before the snapping might also be effective."

 

 

This is the result of my inspection about your data. Red polygon contains no text; Green polygons contain multiple texts for each; Red text (left-top corner) does not correspond to any polygon.

 

 

Very much appreciated Takashi for the massive efforts. Your valuable advices introduce better results

 

 

 

 

 

 

I tried to build same workbench as yours but

 

 

1. I fail to build the testfilter

 

2. To enlarge the text to be visible in the “fme inspector”

 

 

 

 

 

 

 

Could you please share your workbench?

Find this link.

 

https://www.dropbox.com/s/drgpym10j6ty7sp/140911_acad2none.fmw?dl=0

 

 

To enlarge text size in FME Data Inspector, modify "fme_text_size" attribute value.

Thank you very much Takashi,

 

 

I have applied the same settings for the “testfilter” but with no luck.

 

 

CAD file:

 

http://www.mediafire.com/download/or2juzglb27ja7g/Parcels_Al_Samou_04.dwg

 

 

Workbench:

 

http://www.mediafire.com/view/6prk49av2uk621k/Parcels_CAD_To_GIS_05.fmw

 

 

 

 

 

 

 

I didn't get the same result for:

 

 

Single text

 

Multiple text

 

No text

The SpatialRelator outputs the Requestor features after collecting attributes of related Supplier(s).

 

And, which of the ArcStroker and the Snapper should be used previously?

 

See more carefully my workspace, posts, and related help docs., please.

Perfect. This is really great.

 

 

Very much appreciated for the massive efforts and help.

 

 

 

 

 

ponceta wrote:

The LineCloser transformer has done the job in my case.

I didn't had the snapping needs btw.