Hi, I found the Q&A; Entry for the "Keep Clipper attributes" concerning raster-data,
but I have the same problem with a vectorbased source (dxf):
I have a line/point-file (dxf) and and a polygon dataset (shp). I have used the Clipper transformer to clip the dxf, but would also need to keep the attribute ID from the polygon dataset.
I have tried "Merge Attributes" on the clipper, but I can't get the Clipper-ID in the output.
Any suggestions?
Best answer by redgeographics
For starters, I'm not sure whether it's really necessary to test which tiles are used. I suspect it may add overhead. I would also recommend testing this on a small sample of the data first, if you limit the DGN reader to say the first 1000 features you can test whether this is working before letting it run through the entire set of 1M features.
If you make sure the shapefile gets read first and set the Clipper's 'Clipper Type' parameter to Clippers First you'll be able to more quickly work through the data, that way it doesn't need to chache all data before it can start clipping.
In your sample you've set the accumulation mode to 'Only use Clipper', that means you'll lose all original DGN attributes. If that's what you want then fine, otherwise maybe change it to one of the other options.
The attribute KACHEL is coming out of the Clipper, so you should be able to fan out on that. You just need to send the original tiles to the same writer, which should have that same attribute. Hope this helps.
+49If you set the accumulation mode to "Merge Clipper" or "Prefix Clipper" it will pass the Clipper attributes on to the Clippees that are inside.
+43A couple of things occur to me.
1) If the clipper ID attribute is the same name as the that on the clippee (dxf) then it might not overwrite. As @redgeographics suggests, try setting a prefix
2) How are you inspecting the output? Is it being directed to an Inspector direct from the Clipper? Or are you writing to a dataset and inspecting that? You want to be sure that it's the Clipper that is the problem and not the writer (maybe that attribute isn't defined on the writer, or is defined with the wrong data type?)
3) I would set a Breakpoint on the clippee input to the transformer and run this in debug mode to check each feature step-by-step to make sure you know exactly what is happening and to confirm that it's the Clipper that is the problem. Or maybe connect a Logger to the Inside port of the Clipper to log what is being output. It *should* have the attributes from the Clipper
Hope something there helps
Mark
Hi,
thank you both for the tips - but it still doesn't work how I expect...
let me summorize my task:
I need the following output:
What have I done so far is the reckognition whether a tile is used, or not:
this works... I have all used tiles - single files in a folder - and cut together in a single file.
what I want to do after that, is clipping the original vector file with the used tiles.
And that doesn't work. It is now possible to select the ID (in my case "KACHEL") in the Clipper
(thanks to @redgeographics) but in no case (differnt Accumulation Modes, different grouping...)
I get the output I need...
< Image of the workflow attached >
the attached workflow was stopped, because it will run for days with 100% RAM-usage (and I have
a workstation with 16 kernels and 32GB RAM)
+49For starters, I'm not sure whether it's really necessary to test which tiles are used. I suspect it may add overhead. I would also recommend testing this on a small sample of the data first, if you limit the DGN reader to say the first 1000 features you can test whether this is working before letting it run through the entire set of 1M features.
If you make sure the shapefile gets read first and set the Clipper's 'Clipper Type' parameter to Clippers First you'll be able to more quickly work through the data, that way it doesn't need to chache all data before it can start clipping.
In your sample you've set the accumulation mode to 'Only use Clipper', that means you'll lose all original DGN attributes. If that's what you want then fine, otherwise maybe change it to one of the other options.
The attribute KACHEL is coming out of the Clipper, so you should be able to fan out on that. You just need to send the original tiles to the same writer, which should have that same attribute. Hope this helps.
Hi,
thanks for the help - it works :-)
I tried it with less objects, but for 14 Tiles and 4.500 lines and points it lasted nearly 3 hours.
Is there a trick to speed it up?
fme-2.png
Hi,
about the performance-problems, I found out that this seems to be 'clipper-specific':
https://knowledge.safe.com/questions/3857/clipper-speed-comments-from-safe.html
Has anyone an idea?
+15I don't think 14 tiles and 4500 should take 3 hours of processing.
Why not go straight for a LineOnArea overlayer (instead of chopping them to pints first) and then test for overlaps? (and check fme_feature_type if needed).
Seems to save quite a lot of comparisons.
+17Hi,
thanks for the help - it works :-)
I tried it with less objects, but for 14 Tiles and 4.500 lines and points it lasted nearly 3 hours.
Is there a trick to speed it up?
fme-2.png
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
