NAME
CPAN::Testers::Data::Generator - Download and summarize CPAN Testers data
SYNOPSIS
% cpanstats
# ... wait patiently, very patiently
# ... then use cpanstats.db, an SQLite database
# ... or the MySQL databaseDESCRIPTION
This distribution was originally written by Leon Brocard to download and summarize CPAN Testers data. However, all of the original code has been rewritten to use the CPAN Testers Statistics database generation code. This now means that all the CPAN Testers sites including the Reports site, the Statistics site and the CPAN Dependencies site, can use the same database.
This module downloads articles from the cpan-testers newsgroup, generating or updating an SQLite database containing all the most important information. You can then query this database, or use CPAN::WWW::Testers to present it over the web.
A good example query for Acme-Colour would be:
SELECT version, status, count(*) FROM cpanstats WHERE
distribution = "Acme-Colour" group by version, state;To create a database from scratch can take several days, as there are now over 2 million articles in the newgroup. As such updating from a known copy of the database is much more advisable. If you don't want to generate the database yourself, you can obtain the latest official copy (compressed with gzip) at http://devel.cpantesters.org/cpanstats.db.gz
With over 6 million articles in the archive, if you do plan to run this software to generate the databases it is recommended you utilise a high-end processor machine. Even with a reasonable processor it can take a week!
SQLite DATABASE SCHEMA
The cpanstats database schema is very straightforward, one main table with several index tables to speed up searches. The main table is as below:
+--------------------------------+
| cpanstats |
+----------+---------------------+
| id | INTEGER PRIMARY KEY |
| state | TEXT |
| postdate | TEXT |
| tester | TEXT |
| dist | TEXT |
| version | TEXT |
| platform | TEXT |
| perl | TEXT |
| osname | TEXT |
| osvers | TEXT |
| date | TEXT |
| guid | TEXT |
| type | INTEGER |
+----------+---------------------+It should be noted that 'postdate' refers to the YYYYMM formatted date, whereas the 'date' field refers to the YYYYMMDDhhmm formatted date and time.
The metabase database schema is again very straightforward, and consists of one table, as below:
+--------------------------------+
| metabase |
+----------+---------------------+
| guid | TEXT PRIMARY KEY |
| report | TEXT |
+----------+---------------------+The report field is JSON encoded, and is a cached version of the one extracted from Metabase::Librarian.
SIGNIFICANT CHANGES
v0.31 CHANGES
With the release of v0.31, a number of changes to the codebase were made as a further move towards CPAN Testers 2.0. The first change is the name for this distribution. Now titled 'CPAN-Testers-Data-Generator', this now fits more appropriately within the CPAN-Testers namespace on CPAN.
The second significant change is to now reference a MySQL cpanstats database. The SQLite version is still updated as before, as a number of other websites and toolsets still rely on that database file format. However, in order to make the CPAN Testers Reports website more dynamic, an SQLite database is not really appropriate for a high demand website.
The database creation code is now available as a standalone program, in the examples directory, and all the database communication is now handled by the new distribution CPAN-Testers-Common-DBUtils.
v0.41 CHANGES
In the next stage of development of CPAN Testers 2.0, the id field used within the database schema above for the cpanstats table no longer matches the NNTP ID value, although the id in the articles does still reference the NNTP ID.
In order to correctly reference the id in the articles table, you will need to use the function guid_to_nntp() with CPAN::Testers::Common::Utils, using the new guid field in the cpanstats table.
As of this release the cpanstats id field is a unique auto incrementing field.
The next release of this distribution will be focused on generation of stats using the Metabase storage API.
v1.00 CHANGES
Moved to Metabase API. The change to a definite major version number hopefully indicates that this is a major interface change. All previous NNTP access has been dropped and is no longer relavent. All report updates are now fed from the Metabase API.
INTERFACE
The Constructor
new
Instatiates the object CPAN::Testers::Data::Generator. Accepts a hash containing values to prepare the object. These are described as:
my $obj = CPAN::Testers::Data::Generator->new( logfile => './here/logfile', config => './here/config.ini' );Where 'logfile' is the location to write log messages. Log messages are only written if a logfile entry is specified, and will always append to any existing file. The 'config' should contain the path to the configuration file, used to define the database access and general operation settings.
Public Methods
generate
Starting from the last cached report, retrieves all the more recent reports from the Metabase Report Submission server, parsing each and recording each report in both the cpanstats databases (MySQL & SQLite) and the metabase cache database.
regenerate
For a given date range, retrieves all the reports from the Metabase Report Submission server, parsing each and recording each report in both the cpanstats databases (MySQL & SQLite) and the metabase cache database.
Note that as only 2500 can be returned at any one time due to Amazon SimpleDB restrictions, this method will only process the guids returned from a given start data, up to a maxiumu of 2500 guids.
This method will return the guid of the last report processed.
rebuild
In the event that the cpanstats database needs regenerating, either in part or for the whole database, this method allow you to do so. You may supply parameters as to the 'start' and 'end' values (inclusive), where all records are assumed by default. Records are rebuilt using the local metabase cache database.
reparse
Rather than a complete rebuild the option to selective reparse selected entries is useful if there are reports which were previously unable to correctly supply a particular field, which now has supporting parsing code within the codebase.
In addition there is the option to exclude fields from parsing checks, where they may be corrupted, and can be later amended using the 'cpanstats-update' tool.
Private Methods
commit
To speed up the transaction process, a commit is performed every 500 inserts. This method is used as part of the clean up process to ensure all transactions are completed.
get_next_guids
Get the list of GUIDs for the reports that have been submitted since the last cached report.
retrieve_reports
Abstracted loop of requesting GUIDs, then parsing, storing and caching each report as appropriate. Updates Recent JSON & RSS files.
already_saved
Given a guid, determines whether it has already been saved in the local metabase cache.
get_fact
Get a specific report factfor a given GUID.
parse_report
Parses a report extracting the metadata required for the cpanstats database.
reparse_report
Parses a report (from a local metabase cache) extracting the metadata required for the stats database.
retrieve_report
Given a guid will attempt to return the report metadata from the cpanstats database.
store_report
Inserts the components of a parsed report into the cpanstats database.
cache_report
Inserts a serialised report into a local metabase cache database.
cache_update
For the current report will update the local metabase cache with the id used within the cpanstats database.
HISTORY
The CPAN testers was conceived back in May 1998 by Graham Barr and Chris Nandor as a way to provide multi-platform testing for modules. Today there are over 2 million tester reports and more than 100 testers each month giving valuable feedback for users and authors alike.
BECOME A TESTER
Whether you have a common platform or a very unusual one, you can help by testing modules you install and submitting reports. There are plenty of module authors who could use test reports and helpful feedback on their modules and distributions.
If you'd like to get involved, please take a look at the CPAN Testers Wiki, where you can learn how to install and configure one of the recommended smoke tools.
For further help and advice, please subscribe to the the CPAN Testers discussion mailing list.
CPAN Testers Wiki
- http://wiki.cpantesters.org
CPAN Testers Discuss mailing list
- http://lists.cpan.org/showlist.cgi?name=cpan-testers-discussBUGS, PATCHES & FIXES
There are no known bugs at the time of this release. However, if you spot a bug or are experiencing difficulties, that is not explained within the POD documentation, please send bug reports and patches to the RT Queue (see below).
Fixes are dependant upon their severity and my availablity. Should a fix not be forthcoming, please feel free to (politely) remind me.
RT Queue - http://rt.cpan.org/Public/Dist/Display.html?Name=CPAN-Testers-Data-Generator
SEE ALSO
CPAN::Testers::Report, Metabase, Metabase::Fact, CPAN::Testers::Fact::LegacyReport, CPAN::Testers::Fact::TestSummary, CPAN::Testers::Metabase::AWS
CPAN::Testers::WWW::Statistics
http://www.cpantesters.org/, http://stats.cpantesters.org/, http://wiki.cpantesters.org/
AUTHOR
It should be noted that the original code for this distribution began life under another name. The original distribution generated data for the original CPAN Testers website. However, in 2008 the code was reworked to generate data in the format for the statistics data analysis, which in turn was reworked to drive the redesign of the all the CPAN Testers websites. To reflect the code changes, a new name was given to the distribution.
CPAN-WWW-Testers-Generator
Original author: Leon Brocard <acme@astray.com> (C) 2002-2008
Current maintainer: Barbie <barbie@cpan.org> (C) 2008-2010CPAN-Testers-Data-Generator
Original author: Barbie <barbie@cpan.org> (C) 2008-2011LICENSE
This code is distributed under the Artistic License 2.0.