I want to modify the network analysis library because th graph builder does something I don't want. When there is an intersection between to lines, it creates a node in it. I want to avoid this behaviour, so I did something like :
Calculate the coordinates of every beginning and end of a line in my .shp file
Create a shp that has every node as a point
Associate the first shp containg the lines the id of the beginning point and end point in new attributes called "from" and "to"
define a new function that will each time I want to calculate the graph. It initialises a new Qgsgrapg(), then I loop over the node shp and use the addVertex() method, when it ends, I loop over my line shp and use the addEdge() method. in the end I return the graph.
My problem is it that this last function consumes a bit of time. I got over something like 300k lines, and I guess there is some python overhead in the loop. (I don't know much about these kind of things, I just assume from what I read)
So I thought that now that I made a working project, I could try to optimize it. I'm trying to understand how this function works :
http://qgis.org/api/classQgsLineVectorLayerDirector.html#af14ff34e31a2860edd156090c0f746ef
I suspect it's the one responsible of my problem. I don't know how it really works, but I guess I should find the lines responsible of the new nodes in intersections, delete them (maybe it would need some code adjustment), find a way to recompile this (There are some small c++ plugin guides in github), and tell my module from where to import the new dll (I guess that compiling this would lead to a .dll, I don't know much about c++, I have enough troubles to write std::string)
Ps: I tried to use the networkx library and it's read_shp() method to see if it's fast enough, I got a memory after 2min of processing.
أكثر...
Calculate the coordinates of every beginning and end of a line in my .shp file
Create a shp that has every node as a point
Associate the first shp containg the lines the id of the beginning point and end point in new attributes called "from" and "to"
define a new function that will each time I want to calculate the graph. It initialises a new Qgsgrapg(), then I loop over the node shp and use the addVertex() method, when it ends, I loop over my line shp and use the addEdge() method. in the end I return the graph.
My problem is it that this last function consumes a bit of time. I got over something like 300k lines, and I guess there is some python overhead in the loop. (I don't know much about these kind of things, I just assume from what I read)
So I thought that now that I made a working project, I could try to optimize it. I'm trying to understand how this function works :
http://qgis.org/api/classQgsLineVectorLayerDirector.html#af14ff34e31a2860edd156090c0f746ef
I suspect it's the one responsible of my problem. I don't know how it really works, but I guess I should find the lines responsible of the new nodes in intersections, delete them (maybe it would need some code adjustment), find a way to recompile this (There are some small c++ plugin guides in github), and tell my module from where to import the new dll (I guess that compiling this would lead to a .dll, I don't know much about c++, I have enough troubles to write std::string)
Ps: I tried to use the networkx library and it's read_shp() method to see if it's fast enough, I got a memory after 2min of processing.
أكثر...