sondalex / cliriculum Goto Github PK

Add optional location field in contact section (with fontawesome location logo if provided).

When viewing the PDF (created with chromium print) in firefox, the sidebar (whitesmoke) color is displayed differently than in chromium based pdf viewers or Evince or Acrobat Reader.

More precisely the linear gradient does not follow the same pattern.

The font of the resume png image in production User Guide is not found thus defaults to the container (github action's ubuntu-latest) browser font.

In some cases a KeyError is raised by cliriculum.markdown.ParseMd
when using add_location.

Potential cause:

• The lack of correction of self.heading_index in method add_contact

cliriculum.pdf.chromium_print fails more than before.
After analysis this seems to come from either chromium update or Ubuntu update.

I ran chromium headless print from the command line to verify, it did not work (empty pdf). On the other side, print to pdf from the browser (GUI) works correctly.

Deserializers should be rigourously tested for null values and clearly define necessary field and unnecessary field for each level.

Currently doc gets uploaded on Pull Request, this should not be the case.
However, it should be partially disabled, building the doc on PR could still be useful as a test.

Feature Request:

• Adding metadata about job application for later use in chrome_print(filename, ...)

Add kwargs in some/all deserializers with instance dynamic attribute creation such that they can be easily extended

Currently, conftest.py is located in repository root.
As a result pytest scope is aware of cliriculum source directory. As a result tests are run against source and not against installed version. See pytest/fixture for more explanation on scope.

conftest.py should be moved -> tests/conftest.py

pip install on PyPi test server does not work

Error:

ERROR: Could not find a version that satisfies the requirement mistletoe (from cliriculum) (from versions: none)
ERROR: No matching distribution found for mistletoe

Currently, nth level list are aligned horizontally with first level.

This should be corrected, however alignment of bullet point from second degree and above level with previous level text is complicated a priori.

Temporary solution for users:

Adding custom css rule to the custom css file (i.e the one specified with --stylesheet):

ul li ul > li {
    padding-left: 1.6em;
}

Note that the value should be adjusted based on browser specificities.

See https://github.com/sondalex/cliriculum/actions/runs/4026885655/jobs/6921941089#step:5:76

To recapitulate cliriculum.utils.copy_resources test fails.
_isnot_pycache_or_init may return False positive

Currently there is no support for location information to map to Second level headers by matching ID.

One solution not breaking current implementation:

Add --location pointing towards a file containing location information with keys corresponding to --main file section keys.

If provided insert node ontop of eventual date information.

Use Markdown specification or reduced markdown (bold and italic) for this section with potential font awesome icon addition

Chromium headless print does not work correctly, a blank page is generated.

I.e the following command

chromium --headless http://127.0.0.1:8000 --print-to-pdf --disable-gpu

generates a blank page.

Solution:

Setting --virtual-time-budget appropriately, 20000 works good on my setup.

Previous commit introduced a print statement by accident while debugging an issue.
It should be removed, more precisely:
https://github.com/sondalex/cliriculum/blob/master/cliriculum/utils.py#L41

By looking how to render HTMLBlock in mistletoe, I came across miyuchina/mistletoe#106

Maybe I should refactor the code such that the parser parses in the context of
the cliriculum.renderers.Renderer (without calling the render method).
This avoids fiddling with mistletoe.base_token._token_types (used by mistletoe.Document)

command line flag --overwrite set 1 does not overwrite every files.
Indeed after I installed new version with updated style and ran overwrite, style did not get updated in the resume directory.

Currently, sometimes pdf conversion creates a pdf with no content. Although the flag --virtual-time-budget is set to 20000, making this scenario quite rare there is no insurance mechanism to warn the user when such case happen.

A warning parameter could be added to chrome_print which would use under the hood pypdf, and load it dynamically rather to avoid adding a new dependency as requirement.

Another methodology could just parse the raw pdf and search for

<</Length1 123
/Filter /FlateDecode
/Length 126>> stream

...some content...
endstream

patterns, if only one of zero length then the document is empty -> raise warning

Currently, if contact key "profession" is associated with value null, the html renderer renders None instead of empty string.
This behaviour is potentially generalized to other keys

Text which overlap page limits
does not get pushed to the next page.

This behaviour is probably due to the absolute positioning of
sidebar and main.