MondoReport Documentation  
Version 0.01 alpha 24-Nov-2001.  iron@mso.oz.net or mso@oz.net.
Copyright (c) 2001 Mike Orr.  License: same as Python or Cheetah.

* * * * *
STATUS:  previous/next batches and query string are not implemented yet.
Sorting not designed yet.  Considering "click on this column header to sort by
this field" and multiple ascending/descending sort fields for a future version.

Tested with Python 2.2b1.  May work with Python 2.1 or 2.0.

* * * * *
OVERVIEW

MondoReport -- provide information about a list that is useful in generating
any kind of report.  The module consists of one main public class, and some
generic functions you may find useful in other programs.  This file contains an
overview, syntax reference and examples.  The module is designed both for
standalone use and for integration with the Cheetah template system
(http://www.cheetahtemplate.org/), so the examples are in both Python and
Cheetah.  The main uses of MondoReport are: 

(A) to iterate through a list.  In this sense MR is a for-loop enhancer,
providing information that would be verbose to calculate otherwise.

(B) to separate a list into equal-size "pages" (or "batches"--the two terms are
interchangeable) and only display the current page, plus limited information
about the previous and next pages.

(C) to extract summary statistics about a certain column ("field") in the list.

* * * * *
MAIN PUBLIC CLASS

To create a MondoReport instance, supply a list to operate on.

    mr = MondoReport(origList)

The list may be a list of anything, but if you use the 'field' argument in any
of the methods below, the elements must be instances or dictionaries.

MondoReport assumes it's operating on an unchanging list.  Do not modify the
list or any of its elements until you are completely finished with the
ModoReport object and its sub-objects.  Otherwise, you may get an exception or
incorrect results.

MondoReport instances have three methods:

    .page(size, start, overlap=0, orphan=0
        sort=None, reverse=False)                => list of (r, a, b).

'size' is an integer >= 1.  'start', 'overlap' and 'orphan' are integers >= 0.
The list returned contains one triple for each record in the current page.  'r'
is the original record.  'a' is a BatchRecord instance for the current record
in relation to all records in the origList.  'b' is a BatchRecord instance for
the current record in relation to all the records in that batch/page.  (There
is a .batch method that's identical to .page.)

The other options aren't implemented yet, but 'overlap' duplicates this many
records on adjacent batches.  'orphan' moves this many records or fewer, if
they are on a page alone, onto the neighboring page.  'sort' (string) specifies
a field to sort the records by.  It may be suffixed by ":desc" to sort in
descending order.  'reverse' (boolean) reverses the sort order.  If both
":desc" and 'reverse' are specified, they will cancel each other out.  This
sorting/reversal happens on a copy of the origList, and all objects returned
by this method use the sorted list, except when resorting the next time.
To do more complicated sorting, such as a hierarchy of columns, do it to the
original list before creating the ModoReport object.

    .all(sort=None, reverse=False)              => list of (r, a).

Same, but the current page spans the entire origList.

    .summary()                                  => Summary instance.

Summary statistics for the entire origList.

In Python, use .page or .all in a for loop:

    from Cheetah.Tools.MondoReport import MondoReport
    mr = MondoReport(myList)
    for r, a, b in mr.page(20, 40):
        # Do something with r, a and b.  The current page is the third page,
        # with twenty records corresponding to origList[40:60].
    if not myList:
        # Warn the user there are no records in the list.

It works the same way in Cheetah, just convert to Cheetah syntax.  This example
assumes the template doubles as a Webware servlet, so we use the servlet's
'$request' method to look up the CGI parameter 'start'.  The default value is 0
for the first page.

    #from Cheetah.Tools.MondoReport import MondoReport
    #set $mr = $MondoReport($bigList)
    #set $start = $request.field("start", 0)
    #for $o, $a, $b in $mr.page(20, $start)
        ... do something with $o, $a and $b ...
    #end for
    #unless $bigList
        This is displayed if the original list has no elements.
        It's equivalent to the "else" part Zope DTML's <dtml-in>.
    #end unless

* * * * *
USING 'r' RECORDS

Use 'r' just as you would the original element.  For instance:

    print r.attribute     # If r is an instance.
    print r['key']        # If r is a dictionary.
    print r               # If r is numeric or a string.

In Cheetah, you can take advantage of Universal Dotted Notation and autocalling:

    $r.name        ## 'name' may be an attribute or key of 'r'.  If 'r' and/or
                   ## 'name' is a function or method, it will be called without
                   ## arguments.
    $r.attribute
    $r['key']
    $r
    $r().attribute()['key']()

If origList is a list of name/value pairs (2-tuples or 2-lists), you may
prefer to do this:

    for (key, value), a, b in mr.page(20, 40):
        print key, "=>", value
    
    #for ($key, $value), $a, $b in $mr.page(20, $start)
        $key =&gt; $value
    #end for

* * * * *
STATISTICS METHODS AND FIELD VALUES

Certain methods below have an optional argument 'field'.  If specified, 
MondoReport will look up that field in each affected record and use its value
in the calculation.  MondoReport uses Cheetah's NameMapper if available,
otherwise it uses a minimal NameMapper substitute that looks for an attribute
or dictionary key called "field".  You'll get an exception if any record is a
type without attributes or keys, or if one or more records is missing that
attribute/key.

If 'field' is None, MondoReport will use the entire record in its
calculation.  This makes sense mainly if the records are a numeric type.

All statistics methods filter out None values from their calculations, and
reduce the number of records accordingly.  Most filter out non-numeric fields
(or records).  Some raise NegativeError if a numeric field (or record) is
negative.


* * * * *
BatchRecord METHODS

The 'a' and 'b' objects of MondoReport.page() and MondoReport.all() provide
these methods.

    .index()

The current subscript.  For 'a', this is the true subscript into origList.
For 'b', this is relative to the current page, so the first record will be 0.
Hint: In Cheetah, use autocalling to skip the parentheses: '$b.index'.

    .number()

The record's position starting from 1.  This is always '.index() + 1'.

    .Letter()

The letter ("A", "B", "C") corresponding to .number().  Undefined if .number()
> 26.  The current implementation just adds the offset to 'a' and returns
whatever character it happens to be.  

To make a less dumb implementation (e.g., "Z, AA, BB" or "Z, A1, B1"):
1) Subclass BatchRecord and override the .Letter method. 
2) Subclass MondoReport and set the class variable .BatchRecordClass to your
new improved class.

    .letter()

Same but lower case.

    .Roman()

The Roman numeral corresponding to .number().

    .roman()

Same but lower case.

    .even()

True if .number() is even.

    .odd()

True if .number() is odd.

    .even_i()

True if .index() is even.

    .odd_i()

True if .index() is odd.

    .length()

For 'a', number of records in origList.  For 'b', number of records on this
page.

    .item()

The record itself.  You don't need this in the normal case since it's the same
as 'r', but it's useful for previous/next batches.

    .size()

The 'size' argument used when this BatchRecord was created.  
'a.size() == b.size()'.

    .first()

True if this is the first record.

    .last()

True if this is the last record.

    .firstValue(field=None)

True if there is no previous record, or if the previous field/record has a 
different value.  Used for to print section headers.  For instance, if you
are printing addresses by country, this will be true at the first occurrance
of each country.  Or for indexes, you can have a non-printing field showing
which letter of the alphablet this entry starts with, and then print a "B"
header before printing the first record starting with "B".

    .lastValue(field=None)

True if this is the last record containing the current value in the
field/record.

    .percentOfTotal(field=None, suffix="%", default="N/A", decimals=2)

Returns the percent that the current field/record is of all fields/records.
If 'suffix' is None, returns a number; otherwise it returns a string with
'suffix' suffixed.  If the current value is non-numeric, returns 'default'
instead (without 'suffix').  'decimals' tells the number of decimal places to
return; if 0, there will be no decimal point.

    .prev()

Returns a PrevNextBatch instance for the previous page.  If there is no
previous page, returns None.  [Not implemented yet.]

    .next()

Returns a PrevNextBatch instance for the next page.  If there is no next page,
returns None.  [Not implemented yet.]

    .prevPages()

Returns a list of PrevNextPage instances for every previous page, or [] if no
previous pages.  [Not implemented yet.]

    .nextPages()

Returns a list of PrevNextPage instances for every next page, or [] if no next
pages.  [Not implemented yet.]

    .query(start=None, label=None, attribName="start", attribs=[])

[Not implemented yet.]

With no arguments, returns the HTML query string with start value removed (so
you can append a new start value in your hyperlink).  The query string is taken
from the 'QUERY_STRING' environmental variable, or "" if missing.  (This is
Webware compatible.)  

With 'start' (an integer >= 0), returns the query string with an updated start
value, normally for the next or previous batch.

With 'label' (a string), returns a complete HTML hyperlink:
'<A HREF="?new_query_string">label</A>'.  You'll get a TypeError if you specify
'label' but not 'start'.

With 'attribName' (a string), uses this attribute name rather than "start".
Useful if you have another CGI parameter "start" that's used for something
else.

With 'attribs' (a dictionary), adds these attributes to the hyperlink.
For instance, 'attribs={"target": "_blank"}'.  Ignored unless 'label' is 
specified too.

This method assumes the start parameter is a GET variable, not a POST variable.

    .summary()

Returns a Summary instance.  'a.summary()' refers to all records in the
origList, so it's the same as MondoReport.summary().  'b.summary()' refers only
to the records on the current page.  [Not implemented yet.]

* * * * *
PrevNextPage INSTANCES

[Not implemented yet.]

PrevNextPage instances have the following methods:

    .start()

The index (true index of origList) that that page starts at.  You may also use
'.start().index()', '.start().number()', etc.  Also
'.start().item(field=None)'.  (Oh, so *that*'s what .item is for!)

    .end()

The index (true index of origList) that that page ends at.  You may also use
'.end().index()', '.end().number()', etc.   Also
'.end().item(field=None)'.

    .length()

Number of records on that page.

    .query(label=None, attribName="start", attribs={}, before="", after="")

[Not implemented yet.]

Similar to 'a.query()' and 'b.query()', but automatically calculates the start
value for the appropriate page.  

For fancy HTML formatting, 'before' is prepended to the returned text and 
'after' is appended.  (There was an argument 'else_' for if there is no such
batch, but it was removed because you can't even get to this method at all in
that case.)

* * * * * *
SUMMARY STATISTICS

These methods are supported by the Summary instances returned by
MondoReport.Summary():

    .sum(field=None)

Sum of all numeric values in a field, or sum of all records.

    .total(field=None)

Same.

    .count(field=None)

Number of fields/records with non-None values.

    .min(field=None)

Minimum value in that field/record.  Ignores None values.

    .max(field=None)

Maximum value in that field/record.  Ignores None values.

    .mean(field=None)

The mean (=average) of all numeric values in that field/record.

    .average(field=None)

Same.

    .median(field=None)

The median of all numeric values in that field/record.  This is done by sorting
the values and taking the middle value.

    .variance(field=None), .variance_n(field=None)
    .standardDeviation(field=None), .standardDeviation_n(field=None)

[Not implemented yet.]


* * * * *
To run the regression tests (requires unittest.py, which is standard with
Python 2.2), run MondoReportTest.py from the command line.  The regression test
double as usage examples.


# vim: shiftwidth=4 tabstop=4 expandtab textwidth=79