WIP: inserting tables with \tableinput command

[cserteGT3]

As discussed in #195 this PR adds the \tableinput{header}{path} command. The current implementation works in the sense that the markdown strings are generated and included in the page. As I see, the followings are need to be done:

• conversion to html
• error handling: check if the provided header's length==number of columns
• tests
• docs

I need help with the first two tasks (and will do the other two later on):

1. I just don't know which function to call on the result string to convert it to html.
2. If the header's and the table's length doesn't match: error should be thrown or an @assert is fine, or should just ignore it and write an error message?

[codecov]

Codecov Report

Merging #197 into master will decrease coverage by 2.55%.
The diff coverage is 2.94%.

@@            Coverage Diff             @@
##           master     #197      +/-   ##
==========================================
- Coverage   87.21%   84.65%   -2.56%     
==========================================
  Files          26       26              
  Lines        1087     1121      +34     
==========================================
+ Hits          948      949       +1     
- Misses        139      172      +33

Impacted Files	Coverage Δ
src/JuDoc.jl	100% <ø> (ø)	⬆️
src/converter/lx_simple.jl	45% <0%> (-55%)	⬇️
src/jd_vars.jl	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a7572f5...12ab362. Read the comment docs.

[codecov]

Codecov Report

Merging #197 into master will decrease coverage by 0.18%.
The diff coverage is 80.64%.

@@            Coverage Diff             @@
##           master     #197      +/-   ##
==========================================
- Coverage   87.21%   87.03%   -0.19%     
==========================================
  Files          26       26              
  Lines        1087     1118      +31     
==========================================
+ Hits          948      973      +25     
- Misses        139      145       +6

Impacted Files	Coverage Δ
src/JuDoc.jl	100% <ø> (ø)	⬆️
src/jd_vars.jl	100% <100%> (ø)	⬆️
src/converter/lx_simple.jl	89.47% <80%> (-10.53%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a7572f5...cbd9700. Read the comment docs.

[tlienart]

Ok I think your solution is a bit too complicated:

Let's say that the file is test.csv

a, b, c, d
0, 1, 2, 3
4, 5, 6, 7

Then

using DelimitedFiles
content = readdlm("test.csv", ',', String)
open("out.md", "w") do fout
    for row in eachrow(content)
        write(fout, "| " * prod(e * " | " for e in row))
        write(fout, '\n')
    end
end

Does most of the job (header part missing but it's easy to add). Btw, you may have to write rowidx in 1:size(content, 1) because eachrow doesn't work on 1.0 I think)

For your question (1) see review comment, just declare the command in LX_SIMPLE_REPROCESS, nothing else.

For (2), an html_err is probably best so that it doesn't stop the engine but just writes a message on the HTML page.

[cserteGT3]

Only reason for this (from the docs):

As illustrated in the above example each column of | characters must be aligned vertically.

If the cells in a column doesn't have the same length (like in the followings), | will not be aligned:

a, b, c, d
string1, 1.123, 2, 3
string126, 0.1, 6, 7

Result markdown:

| a |  b |  c |  d | 
| string1 |  1.123 |  2 |  3 | 
| string126 |  0.1 |  6 |  7 | 

I didn't tested how the parser handles this kind of table, but in the docs it's stated that they must align, so I would rather align them.

[tlienart]

Here's a bit more code which may help with testing etc:

using Markdown

test = """
    This is a table:

    | a |  b |  c |  d |
    | :---: | :---: | :---: | :---: |
    | string1 |  1.123 |  2 |  3 |
    | string126 |  0.1 |  6 |  7 |

    Done.
    """

result = test |> Markdown.parse |> Markdown.html

result = """<style>table, th, td {
  border: 1px solid black;
}</style>\n""" * result

write("/Users/tlienart/Desktop/index.html", result)

if you open that in a browser you'll get:

which is what we want (separators etc are styled in CSS), so you don't need to have things aligned.

Note: everything is left-aligned even though there's a specification :---: which should produce centered columns; all this can be styled in CSS though afaik so not too much of a problem for now.

By the way, JuDoc also uses this Markdown.parse |> Markdown.html chain though after a bit of processing.

[cserteGT3]

I'll try to answer everything in this comment.
My two questions:
As discussed html_err() is used to handle DimensionMismatch of the header and if the file is not present. Therefore I left the command in LX_SIMPLE and used the md2html function to convert the generated markdown to html. Hope that is a good solution.

I experienced too, that alignment doesn't matter for the parser, moreover if | are aligned, the parser can't parse the string into the table, which I don't really understand. I may open an issue, because currently the behaviour is totally different than the documentation 😄

I'll soon figure out the testing, and write the docs.

[tlienart]

Hey thanks a lot for this, I'll try to look at this tomorrow and get back to you with, hopefully helpful, comments.

PS: given that you're one of the first to actually have a look at the code, don't hesitate to also let me know the things that don't make sense and I'll try to add comments (or docs) as appropriate to guide future contributions

[cserteGT3]

A quick note:
While I was writing the docs I realized that Documenter needs to have the |-s aligned.
And also the alignment (left-middle-right) works.

[cserteGT3]

I've finished the first version of documentation and tests. Let me know if they serve their purpose.

PS: This is my first larger PR, so I'm not sure what else should be done (after approving the changes). Rebasing and/or squashing commits? If so, please do let me know, I'd like to do them to practice open source development 😃 .

[cserteGT3]

Answering to my own issue:

I experienced too, that alignment doesn't matter for the parser, moreover if | are aligned, the parser can't parse the string into the table, which I don't really understand. I may open an issue, because currently the behaviour is totally different than the documentation 😄

I started writing an MWE and realized that the parser works totally fine. Also I misunderstood what you've written. Now I understand that CSS is dominant.

[tlienart]

Ok, I've added some feedback on "just the code", will now pull your branch, play a bit with it and see if it works as expected, assuming that + fixing of minor comments above, I'll merge. Thanks for contributing!

[tlienart]

Ok it mostly looks good, one more thing though: this test:

# header specified
    h = raw"""
        A table:
        \tableinput{A,B,C}{/assets/testcsv.csv}
        Done.
        """ |> seval
    shouldbe = """<p>A table: <table><tr><th>A</th><th>B</th><th>C</th></tr>
            <tr><td>string1</td><td>1.567</td><td>0</td></tr>
            <tr><td></td><td></td><td></td></tr>
            <tr><td>l i n e</td><td>.158</td><td>99999999</td></tr></table>
            Done.</p>"""
    @test isapproxstr(h, shouldbe)

looks incorrect; when the header is specified you should not over-write the first line of the CSV file, so the first row after A, B, C should be h1, h2, h3 which is present in the file

[cserteGT3]

Sorry for the silence, I had a busy week. I've updated the code, tests and documentation (and resolved every review). Please review them particularly the documentation

[tlienart]

you're -> your

[tlienart]

with -> by a

[tlienart]

of the file -> in the file

[tlienart]

looks good to me minus minor nitpicking, thanks a lot for this Tamás! and actually given it's minor, I'll just merge and do it :-)