fix: #295 transcluding source file with search-option

[nobiot]

fix #295
We really should have unit-test.
@AndreasMatthias , I'd appreciate it if you ran your unit-test.el for no-web. On my end, I don't think I am running it correctly but it looks like the test fails.

[AndreasMatthias]

I see that the indentation is wrong for transclusions without a :src-block.
But as far as I can see this issue is not related to :noweb-chunk.

Here is a simple test case where the indentation is wrong, too.
This is broken since 6e5e6eb.

File foo.py:

def foo():
    print("foo")

File test.org:

#+transclude: [[./foo.py]]

Another issue that you might hit when running test-noweb-chunk.org is the
global setting of org-edit-src-content-indentation that was 0 when I created
the detached file.

[AndreasMatthias]

Since the indentation inside a src-block is correct, the question is: Do we need to support code transclusion without a src-block at all?

[AndreasMatthias]

Regarding unit-tests: What kind of unit-tests do you have in mind?

Some time ago I started with an ERT testsuite. It's not a strict unit testing, but a top-level white-box testing that does comparisions with detached files, similar to the noweb test file. This testsuite displays diff buffers for all tests that failed, and it can check simple live-editing as well.

If you are interested I can provide the code. But I need to refactor the code first.

[gggion]

Was just about to post an issue for this, I was fixing #304 related to thing-at-point (PR at #305 ) and also noticed the search-option issue.

Regarding more tests, I've added several thing-at-point org tests here:
https://raw.githubusercontent.com/gggion/org-transclusion/61f76beb509b21d26cb8f0f21f19b46b88bb6eda/test/things-at-point.org

I noticed after fixing 'defun' thing-at-point, even if the match to the defun was succesful, the transcluded region began at the exact position of the match instead of ht beginning of the line/indentation like it used to.

In this org test file I went over:

• thing-at-point

  • elisp related, tested these with and without :src

    • sexp and sexp :end "N"
    • list
    • defun and defun :end "N"
    • symbol

  • prose related (tested :end with most of these too)

    • word
    • sentence
    • paragraph
    • line

I think I could make most of these into ert tests without much trouble since I've delineated expected results from multiple types of transclusions for all thing-at-point transclusions I could think of, we could add them to any other ert test suite for src-lines.el, since my tests focus mainly on thing-at-point.

I have also have several tests for org-transclusion-blocks.el which could be ported over to org-transclusion.el as ert tests, for :lines and :src, also have some for text properties and fringe creation.

[gggion]

I've ran the tests mentioned over this branch and it seems to have fixed the issues I was describing before.

[AndreasMatthias]

@gggion: Looks good to me.

While looking into this, I noticed a few things that are not related to this fix, but are somehow related to this topic. Here's an example:

File code.el:

(defun foo ()
  (message "aaa") ;; This is important [id-123]
  (message "bbb"))

(message "this is not part of foo()")

File test.org:

Comment is stripped here. Is this intended?
#+transclude: [[./code.el::id-123]]  :src elisp :thing-at-point sexp

Comment not stripped in this case
#+transclude: [[./code.el::id-123]]  :src elisp

What's the supposed transclusion here?
I was expecting function `foo()` to be transcluded.
#+transclude: [[./code.el::id-123]]  :src emacs-lisp :thing-at-point defun