Refactor Graph.__init__

[nathanncohen]

This (very unpleasant) ticket attempts to clean a bit more the content of Graph.__init__.

To this end, it creates a graph_input modules that gathers into individual functions the different pieces of code that build a graph from .

Admittedly, this is just one step toward a cleaner graph constructor, but we are not there yet.

Nothing smart is done in this branch: the pieces of code from Graph.__init__ are moved to new individual functions (some variables/parameters are renamed). One from DiGraph.__init__ is moved too.

In particular, some pieces of code from DiGraph.__init__ may eventually be merged with the functions that this branch creates, but this will be done later.

This branch is already sufficiently long and sufficiently unpleasant to write/review :-/

Nathann

CC: @sagetrac-borassi @dcoudert @dimpase @videlec

Component: graph theory

Author: Nathann Cohen

Branch/Commit: 9bdf527

Reviewer: David Coudert

Issue created by migration from https://trac.sagemath.org/ticket/19381

[nathanncohen]

Last 10 new commits:

daff6d5	trac #19382: graph6
69cf259	trac #19382: sparse6
cb6af84	trac #19382: disclaimer
d230783	trac #19382: seidel adjacency matrix
eebc428	trac #19382: adjacency matrix
102e977	trac #19382: Fix seidel adjacency
8d50d1f	trac #19382: Incidence matrix
28ecfce	trac #19382: dict_of_dicts
fa5d01c	trac #19382: dict of lists
a74b0ff	trac #19382: dig6

[nathanncohen]

Commit: a74b0ff

[nathanncohen]

Branch: u/ncohen/19381

[tscrim]

comment:2

Would it be possible in this ticket to add support for constructing a graph by [V, E] like this:

sage: V = [1,2,3,4]
sage: E = [[1,2]]
sage: Graph([V, E])

I know I can do it like this

sage: G = Graph({1:[2], 2:[], 3:[], 4:[]})
sage: G.vertices()
[1, 2, 3, 4]

but this is somewhat cumbersome (and doesn't work for graphs with labeled (multiple) edges). The other way I know how to do this is I have to pass the edge set, then add the vertices.

The reason I ask is because I have wanted to create immutable graphs with larger vertex sets than the connectivity of the edges and I don't want to have to waste time copying a mutable graph backend into the static backend. (As an additional reason, it also follows the basic intro to combinatorics definition of a graph.)

I should also be able to review this ticket too (and I definitely will if you make the above change).

[nathanncohen]

comment:3

Would it be possible in this ticket to add support for constructing a graph by [V, E] like this:

Certainly not in this ticket. It refactors code, and it is already (as you can see) a lot to review. I guess it can be done in another ticket, though, I don't thing that there is any problem. With 5-6 lines any ambiguity with the other kind of inputs can be cleared ([V,f], list_of_edges, ...)

I know I can do it like this

sage: G = Graph({1:[2], 2:[], 3:[], 4:[]})
sage: G.vertices()
[1, 2, 3, 4]

but this is somewhat cumbersome (and doesn't work for graphs with labeled (multiple) edges).

sage: Graph({1:{2:['l1','l2']}},multiedges=True).edges()
[(1, 2, 'l1'), (1, 2, 'l2')]

See also Graph.to_dictionary which generates this kind of output.

Nathann

[dcoudert]

comment:4

• Note: This is a internal... -> 'an internal`?

Preliminary tests are ok and the doc build and display properly.
I'm currently running tests on sage/graphs/. I'll let you know soon.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

dc36f02

trac #19381: Typo

[sagetrac-git]

Changed commit from a74b0ff to dc36f02

[dimpase]

comment:6

hey, your commit messages have the ticket number wrong...

[dcoudert]

comment:7

For me the patch is good to go, but I don't know if Nathann has something to do about the commit ticket numbers...

Note: the (possible) segfault in the test src/sage/graphs is independent from this patch and is resolved in #19337.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. Last 10 new commits:

7a33fad	trac #19381: sparse6
efb842b	trac #19381: disclaimer
d9522bb	trac #19381: seidel adjacency matrix
f0f8ee4	trac #19381: adjacency matrix
7a51fde	trac #19381: Fix seidel adjacency
5abbd4b	trac #19381: Incidence matrix
cf7f515	trac #19381: dict_of_dicts
f9fc7a8	trac #19381: dict of lists
dd28d93	trac #19381: dig6
9bdf527	trac #19381: Typo

[sagetrac-git]

Changed commit from dc36f02 to 9bdf527

[dimpase]

comment:9

Replying to @dcoudert:

For me the patch is good to go, but I don't know if Nathann has something to do about the commit ticket numbers...

well, he adds these numbers by hand.
If you think it's OK, then it's OK with me, too.

Note: the (possible) segfault in the test src/sage/graphs is independent from this patch and is resolved in #19337.

[dcoudert]

Reviewer: David Coudert

[dcoudert]

comment:10

Then let's go!

[nathanncohen]

comment:11

Thaaaaaaaaanks for this quick review O_O

Nathann

[vbraun]

Changed branch from u/ncohen/19381 to 9bdf527