xlcnd / isbntools Goto Github PK

Together with #41.

Using PIL?

• expose package path & version
• functions to write to conf file

Before fully solve #27, do this substitution.

One more test for return None.

You should use an appropriate framework if you need throttling in service calls!

1. Divide it in subsections.
2. Write about helper functions.

People are importing private modules... and are getting strange errors! So rename private modules according with pep8 to signal that those modules MUST NOT be imported directly. They functionality is exposed in another way, namely in the namespaces:

• isbntools
• isbntools.conf
• isbntools.dev
• isbntools.dev.lab

Check setconf.py

Like description, language, ...

... with a simplified data structure (data4mask.py is a bit complicated to compile!)

Starting with v3.1.5 the sphinx's build of docs should be tested on Travis.

This feature already exists in the Windows version.

Periodically new blocks of ISBNs are issued to agencies. The range of these blocks is on a database that mask uses. So it could happen, if you have a version of isbntools that is too old, mask doesn't work for valid (recent) issued ISBNs.

The solution? Update isbntools often!

The meta method and the isbn_meta script sometimes give a wrong result (this is due to errors on the provider service) or don't find the data.

Solution 1? Try another service (wcat, goob, openl, isbndb, ...) or (better) use the provider merge (it merges wcat with goob in order to maximize the quality of data).

Solution 2? Calls to metadata services are cached by default. Sometimes this cache could get corrupted. To mitigate that, just enter into a command line isbn_conf delcache. If by any reason you don't want the cache, just enter isbn_conf setopt cache no.

Plugins with API-KEYS or external dependencies should go to contrib`.

goom is an example of a contrib's module with its own script isbn_goom

This is a feature very helpful in conjunction with renaming the file by its ISBN.

Discussion about Workflow and Dev Tools for the project.

Start by reading the Contributing Guidelines.

Move many of the functionality in isbn_ren to dev.rename.

It is OK to annotate code with no cover in the following cases:

1. pure defensive code
2. code only for py2 or only for py3
3. code covered by tests that are only executed locally and not on Travis

See http://www.doi.org/factsheets/ISBN-A.html

Fuzzy matching (just like git) make it easy the use of commands...

The rule should be:

Everything that is not essential should be moved from the core.

crossref.org, arxiv.org, PubMed, ...

Use pep257.py

All options after DEFAULT_SERVICE are not set!

It seems that this happens because there are cases (remember the call to google is NOT deterministic) where industryIdentifiers is missing! And the _mapper in goom was assuming that it exist.

In addition, we can create:

• a new metadata provider that pulls metadata from that cache.
• an option to only go to the web if the data is not available in the cache.
• an option to go local if internet connection is not available.

The fmt is very simple and not very flexible... we need something better!

The bulk of For Devs section must go to readthedocs.

isbntools.conf should not be overwritten but must be updated.

In researching this bug, was discovered that there is no guarantee that isbntools.conf would be installed in ~/.isbntools!

Despite you can access and use isbntools as a library, in some Windows installations, the packaged scripts will not run! At least in the * anaconda distribution* of python for Windows ...

Solution?

See some suggestions below, but can you help?

Useful for the heavy user!

rename (an ebook file) using metadata from the ISBN of the book (more difficult than it seems!)

Now they can be put in conf file or in the command line, but would be nice to have them encrypted!

The only way to the user to input preferences is by the command line.

Starting with the next version (2.1.2), it will be possible the end user write a .conf file to express some options (isbndb api key, default metadata provider,...).

This is a discussion to collect and evaluate new features.

Start by reading the issues with label enhancement.

If you are in a slow internet connection, you could receive timeout errors.

Solution? Starting in version 2.1.2 (and if you are in a UNIX machine) you will be able to change some parameters to fix that.

However, you could do it now by increasing the values of SOCKETS_TIMEOUT
and THREADS_TIMEOUT in setup.py.

The present one works fine for providers and things like that... but it would be nice to split the lib between core functionality (that shouldn't have external dependencies) and addins that could use other libraries.

Change vias.serial so that it doesn't stop if one of the providers fail.

By default merge (the default provider for meta) uses vias.parallel that is slower (than vias.serial) for fast internet connections.

Solution?
If you have a fast internet connection write the following in the [SERVICES] section of .isbntools.conf:

VIAS_MERGE=serial

Like the one that exists for Windows https://sourceforge.net/projects/isbntools/.

This would open the door to some 'legacy' providers that use the MARC format

The isbntools works internally with unicode, however this doesn't solve errors of lost information due to bad encode/decode at the origin!

There are some 'concerns' that should be taken into account:

• There will be many situations where proposed new filename is equal to an existing file (this could lead to data loss) DON'T ALLOW RENAMING IN THIS CASE.
• If we rename files to another file system, we will get in trouble (even in UNIX systems), so keep renaming inside the SAME directory.
• THERE IS NO SAFE WAY to deal with unicode in filenames. "The solution" of restricting to ASCII characters is NOT applicable here, so allow unicode encoded in UTF-8 but filter some characters.
• Limit the length of the filename to avoid serious problems with Windows
• Let other things (permissions, ...) to the operating system and catch them with OSError.

The library can be installed in (python) virtual environments but the configuration files are installed outside!