also simplify a bit definition as ZODB is common in all versions

----

kirr:

- cover only last 2 py3 releases: 3.6 and 3.7 for now (3.8 is not yet
  released)
- separate ZODB3 as it supports only python2.

Py3 tests are failing for now and we'll be getting them to pass
incrementally - step by step.

Using dateparser to support absolute and relative dates in natural language.

/reviewed-by @kirr
/reviewed-on nexedi/zodbtools!8

Fixes to get all tests passing with all ZODB versions we care about.  

Organize testing coverage for all cases via tox, similarly to wendelin.core .

/reviewed-on nexedi/zodbtools!11

Use tox to test with all kinds of ZODB.
With preceding 3 patches tests pass with all versions of upstream ZODB.

TODO: test coverage for both py2 and py3.

maxtid is in ZODB.utils starting only from ZODB5.

ZODB{3,4} want txn._extension, while ZODB5 deprecate it in favour of
txn.extension.

IStorageTransactionMetaData is ZODB5-only interface. Bug introduced in
dd959b28 (zodbdump += DumpReader - to read/parse zodbdump stream).

Currently we exercise zodbdump and zodbcommit+zodbdump with non-empty
extensions, which works if ZODB is patched for txn.extension_bytes
support, but fails on pristine ZODB.

Support for txn.extension_bytes cannot get into upstream ZODB for
more than a year:

https://github.com/zopefoundation/ZODB/pull/183
https://github.com/zopefoundation/ZODB/pull/207

and  even if it somehow will make it, it will likely be only in ZODB5,
while we still care to support ZODB4 and ZODB3.

Skipping zodbdump / zodbcommit tests, if a ZODB does not have
txn.extension_bytes support, would result in significant reduction of
zodbtools test coverage, because practically that is the current
situation with all upstream ZODB{3,4,5}. Dropping test coverage for
non-empty extensions is neither a good option.

For those reason, let's rework the tests and test both zodbdump and
zodbcommit with two scenarios:

1. on a test database where transactions extensions are always empty.
   This should work on all ZODB irregardless of whether
   txn.extension_bytes patch is there or not.

2. on a test database where transactions extensions are present.
   This should work if ZODB has txn.extension_bytes support, but if not,
   we can mark this case as xfail, since the failure is expected.

This way we make the testsuite pass irregardless of whether
txn.extension_bytes support is there, and we don't abandon dump/commit
testing coverage.

/helped-by Jérome Perrin <jerome@nexedi.com>

Fix for this kind of errors:

```
(env)$ zodb analyze demo.fs ffffffffffffffff..
# ø
Processed 0 records in 0 transactions
Traceback (most recent call last):
  File "/srv/slapgrid/slappart8/srv/runner/project/zodbtools/env/bin/zodb", line 11, in <module>
    load_entry_point('zodbtools', 'console_scripts', 'zodb')()
  File "/srv/slapgrid/slappart8/srv/runner/project/zodbtools/zodbtools/zodb.py", line 130, in main
    return command_module.main(argv)
  File "/srv/slapgrid/slappart8/srv/runner/project/zodbtools/zodbtools/zodbanalyze.py", line 305, in main
    report(analyze(path, use_dbm, delta_fs, tidmin, tidmax), csv)
  File "/srv/slapgrid/slappart8/srv/runner/project/zodbtools/zodbtools/zodbanalyze.py", line 102, in report
    print "Average record size is %7.2f bytes" % (rep.DBYTES * 1.0 / rep.OIDS)
ZeroDivisionError: float division by zero
```

and also small fixes for python3 compatibility

/reviewed-on !9

To keep a consistent output.

and use six.moves for python3 compatibility.

Previously we were using "anydbm" which selects dbhash, gdbm or dbm, but
opening the db with the f flag that's only valid for gdm, so de-facto we
were supporting only gdbm.

this also solves the following error on python3:

 AttributeError: 'dict_keys' object has no attribute 'sort'

/reviewed-on !10

- add help tidrange topic.
- change all commands to refer to it.
- add TODO to parse tid from absolute and relative dates (e.g.
  1.month.ago, similarly to how git can do). Dateparser

	https://dateparser.readthedocs.io/

  will probably be of help here.

/reviewed-on nexedi/zodbtools!7

Currently zodbanalyze analyzes whole storage. However it becomes
non practical to make a full zodbanalyze run on whole storage because
usually there are many transactions and objects and the time to run full
zodbanalyze is huge.

However, similarly to zodbdump, we can teach zodbanalyze to analyze a
particular range of transactions. This should help to analyze a range of
changes for e.g.  yesterday, or for last week or similar.

/reviewed-on nexedi/zodbtools!7

Analyze uses regular ZODB storage API: .iterator() & friends. This way
it should be possible apply it not only to FileStorage, but to other
type of storages as well - for example to NEO and ZEO.

Use zodbtools.util.storageFromURL to open a storage by knowing its URL.
Preserve support to directly apply zodbanalyze to FileStorage deltas.

/reviewed-on !7

Storages need to be closed to indicate a clean access end. If a storage is
not closed cleanly it might require to spend time and resources on next
open. For example FileStorage might need to recompute the index.

/reviewed-on nexedi/zodbtools!6

This should be in 7f0bbf7e (zodbdump: Add golden test).

Zodbcommit reads transaction description from stdin and commits read data into
ZODB. The transaction to be committed is read in zodbdump format, but without
first 'txn' header line. For example:

    user "author"
    description "change 123"
    extension ""
    obj 0000000000000001 4 null:00
    ZZZZ

This tool could be useful for testing and for low-level database
maintenance. Please see zodbcommit.py docstring for more details.

We will likely need this reader for `zodb restore` in the future.
We will also use this reader for `zodb commit` in the next patch.

pygolang dependency v↑ becuase we use recently introduced
golang.strconv to unquote user/desc/extension strings.

Python2 works. Python3 support is only minimal and incomplete.

hashRegistry will be needed for zodbdump reader to create a hasher by
its name. Since std hashlib does not have adler32/crc32 nor null - let's
add our own implementation for those hashers too.

The code is based on

	neo@a60c472c, and
	neo@3f578560

Before now we were emitting extra LF only in between transactions as a
separator. However the dump format states LF always goes after
transaction, and there is a reason for it:

	without LF in the end, it becomes ambiguous at EOF - whether it
	is a proper transaction end, or the transaction was cut.

So avoid the ambiguity by always emitting trailing LF after transaction
record.

I originally added escapeqq as part of 75c03368 (zodbdump: Start to
stabilize output format) with the task for this utility to quote string
into valid "..." string always quoted with ".

This utility was later copied to pygolang:

	kirr/pygolang@afa46cf5

and then further improved there to work under both Python2 and Python3
and to not escape printable UTF-8 characters:

	kirr/pygolang@02dddb97

So stop the duplication and simply switch to the better version.

We start to stabilize output format of `zodb dump`. It is actually now robust and the only thing I would contemplate to potentially change is to also cover transaction metadata by hash checksum. So please take a look at updated format (details in patch 1) to provide feedback because it is likely close to what it  will be in its final form.

We also add a program to generate test database which uses various fancy ZODB features and check `zodb dump` output on it to golden one (patch 3).

To be able to dump transaction metadata in raw form ZODB is patched a bit:

https://github.com/zopefoundation/ZODB/pull/183

and we try to detect whether appropriate support is there at runtime and if yes use it to streamline obtaining transaction extension as raw (patch 2).
Pleae see patch 1 (second half of `zodbdump.py` about what has to be taken on without such support and that it still can't work fully reliably).

/cc @nexedi
/reviewed-on nexedi/zodbtools!3

We add a program to generate a test database with all fancy features and
then check `zodb dump` output on it to golden on.

The test database itself is commited to git because we want to make sure
zodbdump works ok for particular exact data and because transaction
extension dict is potentially saved differently on various runs.

Quoting

  https://docs.python.org/2.7/library/stdtypes.html#dict.items    and
  https://docs.python.org/3.7/library/stdtypes.html#dictionary-view-objects

""" CPython implementation detail: Keys and values are listed in an arbitrary
    order which is non-random, varies across Python implementations, and depends
    on the dictionary’s history of insertions and deletions. """

This way on test/gen_testdata.py changes it has to be run manually, and
then the output result of the run committed back together with
gen_testdata.py changes.

This patch test at runtime whether used storage can provide transaction
metadata in raw form and if so used thi form directly without going
through long fragile way of stable pickling extension back to bytes.

Corresponding ZODB support is at

https://github.com/zopefoundation/ZODB/pull/183

Since zodbdump start (c0a6299f "zodbdump - Tool to dump content of a
ZODB database   (draft)") and up till now zodbdump output format was not
good. For example user and description transaction properties were
output without proper quoting, which in situation when there would be
fancy characters in there would break the output.

So start the format stabilization:

- user and description are output as quoted, so now they are guaranteed
  to be on one line. The quoting character is always " (instead of e.g.
  smartly quoting either by ' or " as python does) for easier
  compatibility with ZODB implementations in other languages.

- transaction extension is now printed as raw bytes, not as dict.
  The idea here is that `zodb dump`

  * should perform dump of raw data as stored inside ZODB so that later
    `zodb restore` could restore the database identically to the same state.

  * we should dump raw data instead of unpickled ones because generally
    on-disk extension's format can be any raw bytes and this information
    should be preserved.

- transaction status is now also output as quoted to preserve line
  breakage on fancy status codes.

- it is documented that sha1 is not the only allowed hash function that
  might be used.

- in hashonly mode we add trailing " -" to obj string so that it is
  possible to distinguish outputs of `zodb dump` and `zodb dump -hashonly`
  without knowing a-priory the way it was produced.

  The reason to do so is that it would be not good to e.g. by accident
  feed hashonly output to (future) `zodb restore`, which, without having
  a way to see it should not consume object data would read following
  transaction information as raw object data with confusing later
  errors (and a small chance to restore completely different database
  without reporting error at all).

Because ZODB iteration API gives us already unpickled extension and only
that, for now to dump it as raw we get a long road to pickle it back
also caring to try to pickle in stable order.

Hopefully this will be only a fallback because of

https://github.com/zopefoundation/ZODB/pull/183

and next zodbtools patch.

~~~~

For testing purposes (currently only quoting function unit test) py.test
usage is introduced.

The code is also generally polished here and there.

Relicense to GPLv3+ with wide exception for all Free Software / Open Source projects + Business options.

Nexedi stack is licensed under Free Software licenses with various exceptions
that cover three business cases:

- Free Software
- Proprietary Software
- Rebranding

As long as one intends to develop Free Software based on Nexedi stack, no
license cost is involved. Developing proprietary software based on Nexedi stack
may require a proprietary exception license. Rebranding Nexedi stack is
prohibited unless rebranding license is acquired.

Through this licensing approach, Nexedi expects to encourage Free Software
development without restrictions and at the same time create a framework for
proprietary software to contribute to the long term sustainability of the
Nexedi stack.

Please see https://www.nexedi.com/licensing for details, rationale and options.

Hello up there.

Recently with Ivan we needed a way to obtain last transaction ID of a ZODB storage for resiliency checking. For this `zodb info` utility to print general information about a ZODB database is introduced. Along the way, as it was already planned, all zodbtools utilities are now covered by only one `zodb` command which can invoke other subcommands and show help topics. I also used the occassion to switch how storages are specified from being  ZConfig-file-always to be specified by URL e.g.

     neo://neo1@127.0.0.1:24573
     zeo://...
     file://...

without loosing generality because zconfig:// scheme is also supported.

Please find more details about all changes in individal commit messages.

Thanks for feedback,  
Kirill

/cc @kazuhiko, @jm, @vpelletier, @jerome, @Tyagov, @klaus, @alain.takoudjou, @rafael
/reviewed-on nexedi/zodbtools!2

Either all general parameters at once:

    $ zodb info neo://neo1@127.0.0.1:24573
    name=NEOStorage(neo1)
    size=0
    last_tid=03be7484ddc7f6ee

or one particular parameter:

    $ zodb info neo://neo1@127.0.0.1:24573 last_tid
    03be7484ddc7f6ee

Most of our tools need only read access for working. However e.g.
FileStorage, when opened in read-write mode, automatically creates
database file and index.

This way if database is opened in read-write mode a simple typo in path,
e.g. to `zodb dump path` would lead to:

- new database at path will be created
- the dump will print nothing (empty database)
- exit status will be 0 (ok) and no error will be reported.

For this reason it is better tools declare access level they need so for
read-only access request we can catch it with an error from storage.

This, however, requires quite recent ZODB to work:

https://github.com/zopefoundation/ZODB/pull/153

P.S.

We don't want to force users to always specify read-only in URLs or
zconf files because:

- this is error prone
- URL or zconf can be though as of file
- when a program opens a file the program, not file, declares which type
  of access it wants.

That's why access mode declaration has to be internal.