צפונות ויקי

יחידה:ParamValidator: הבדלים בין גרסאות בדף

→ העריכה הקודמת
לדרישת הקהל
מ גרסה אחת של הדף wikipedia:he:יחידה:ParamValidator יובאה
 
(54 גרסאות ביניים של 4 משתמשים אינן מוצגות)
שורה 5: שורה 5:
the source of this module is in //he.wikipedia.org/wiki/Module:ParamValidator
the source of this module is in //he.wikipedia.org/wiki/Module:ParamValidator


main purpose: use "templatedata" to verify the parameters passed to a template


Terminology: "numeric parameter" means order-based parameter. e.g. if the template is transcluded like so {{x  | k |  | a = m | b = }}
"a" and "b" are "named" parameters, and there are 2 "numeric", or order based parameters, 1 and 2.
we say that the value of a is "m", the value of 1 is "k", and "b" and 2 are "empty".


This module exports two functions: calculateViolations( frame, subpages ), and validateParams( frame ).


it exports a single function: validateParams, which expects a frame object
calculateViolations( frame, subpages ) finds templatedata, in template page or in one of its subpages in the list, if provided.
it returns a table with the violations. if there are none, the table is empty. otherwise it has the structure
{
violation1 = { param1 = value1, param2 = value2 },
violation2 = { ... },
...
}


it retrieves the template's templatedata, and uses it to validate the parameters passed to the template.
violation1, violation2 etc. are one of the names of specific violations, as described below.  
if any violations found, it returns am appropriate string. this string will typically include some maintenance category, and/or error message.
param1, param2 etc. are either the names of parameter passed to the template, or defined in templatedata.
the operation of the module is controlled by a single parameter passed to it. this string should be valid JSON describing the actions.
value1, value2 etc. are the values passed to the template, or an empty string if no such parameter was passed.
the fields of this structure has the following keys, with string or null value:


the different violations are as follow:
* "no-templatedata": no valid tempaltedata was found in tempalte page, or documentation subpage
* "no-templatedata": no valid tempaltedata was found in tempalte page, or documentation subpage
* "undeclared": named parameters with non-empty value, does not exist in templatedata
* "undeclared": named parameters with non-empty value, does not exist in templatedata
שורה 23: שורה 34:
* "empty-deprecated":  parameters with empty value, marked as "deprecated" in tempaltedata
* "empty-deprecated":  parameters with empty value, marked as "deprecated" in tempaltedata
* "empty-required": missing or empty parameter marked as "required" in tempaltedata
* "empty-required": missing or empty parameter marked as "required" in tempaltedata
* "any": at least one of the above properies has non-empty value, and the corresponding error actually exists.
* "incompatible": a non-empty parameter passed to the template, incompatible with the parameter type defined in templatedata
* "multiple": more than one of the above properies (not including "any") has non-empty value, and the corresponding error actually exists.
* "duplicate": a value is passed for the same parameter (or any of its aliases) more than once


if any of these error-conditions exists, and "options" has non-empty value for the corresponding key, will be appended to the output,
after replacing the follwing tokens:
-- "templatename" is replaced with the actual template name,
-- "tname-naked" is replaced with the naked template name, i.e. without colon and namespace
-- "paramname" is replaced by a comma-separated list of parameters offending the rule.
-- "paramandvalue" is replaced by comma-separated list of "name: value" pairs of parameters and values


if the value of some keys is null, this error condition will be ignored, and not counted when calculating "any" and "multiple" conditions.
The second function, validateParams( frame ), can be called from the tempalte' using #invoke.
it expects a parameter named "options", which contains the definition of the output. typically, it's used by placing something like so:
 
<includeonly>{{#invoke:ParamValidator | validateParams | options = {{PV default options}} }}</includeonly>
 
at the top of the template (be mindful not to add extra spaces and newlines to the template).
to bypass some mediawiki limitation, it is also possible to pass the options as "module", like so (use one of the two, but not both):
<includeonly>{{#invoke:ParamValidator | validateParams | module_options = Module:PV default options}} }}</includeonly>
 
the first form expects a template named "Template:PV default options" which contains the options, and the 2nd form expects a module,
suitable for mw.loadData(), which returns a map of namespace => options (i.e. { [0] = <options>, [2] => <options> } .... )
 
the options parameter should be a JSON-encoded string, defining the output, and some special behaviors.
the example above assumes that a wiki page named [[Template:PV default options]] exists, and contains valid JSON string.
for each of the violations defined above, "options" may define an output string, so basically, "options" looks like so:
{
violation1: outputstring1,
violation2: outputstring2,
.... ,
behavior1: some value,
....
}
 
not all violations have to be defined. a violation not defined in "options" will be ignored.
 
when invoked, it extract "subpages" from the options parameter, and calls:
calculateViolations( frame, subpages )
if the returned table is empty, no violation were found, and an empty string is returned and nothing else happens.
 
otherwise, for each of the violations, i.e., the keys of the returned table, when "options" contains this key,
the corresonding value is appended to the output.
 
some further processing is done:
1) several tokens are replaced with calculated values. these are described below.
2) some "meta" violations are calculated: when any none-ignored violation occured,
the "any" meta-violation is added to the output in the same way,
i.e. the string keyed by "any" in the options is appended to output with appropriate substitutions.
similarly, "multiple" meta-violation is created when more than one type of non-ignored violations occured.
3) if the output is not empty, a prefix and suffix strings are prepended and appended to it.
 
these are the tokens and the replacement.
* "templatename": full template name, including namespace.
* "tname_naked": template name without namespace.
* "paramname":  comma-separated list of parameters
* "paramandvalue": is replaced by comma-separated list of "name: value" pairs of parameters and values
the first two are applied to the whole output, including the suffux and prefix,
and the rest are applied to the individual violations, each with its own list of offending parameters and values.
 
 
the rest of the if the value of some keys is null, this error condition will be ignored, and not counted when calculating "any" and "multiple" conditions.


some other optional fields can be passed via options:
some other optional fields can be passed via options:
* "ignore": list (in square brackets) of parameter names to ignore, both in templatedata and actual parameters. all names should be quoted.
* "doc-subpage": can be either a string, or a list (in square bracktes) of strings, indicating subpages of the template  
* "doc-subpage": can be either a string, or a list (in square bracktes) of strings, indicating subpages of the template  
that may contain templatedata.  
that may contain templatedata.  
* "skip-empty-numeric": if a quoted number, the module will ignore empty numeric parameters up to this number
* "ignore": list of patterns. any parameter whose name matches any pattern, will not considered in violation of any of the rules.
* "skip-empty-numeric": if a quoted number, the module will ignore non-declared empty numeric parameters up to this number
* "wrapper-prefix": openning wrapper element of outpot (defaults to "<div class = 'paramvalidator-wrapper'>")  
* "wrapper-prefix": openning wrapper element of outpot (defaults to "<div class = 'paramvalidator-wrapper'>")  
* "wrapper-suffix": closing wrapper element of output (defaults to "</div>")  
* "wrapper-suffix": closing wrapper element of output (defaults to "</div>")  


additional option parameters, named options1, options2, etc. can be passed. any entry defined in these options will
override the previous value. a typical use may be like so:


typically, this JSON structure will be placed in a special template, and retrieved for the module-use like so:
<includeonly>{{#invoke:ParamValidatoe | validateParams | options = {{PV Default Values}} }}</includeonly>
typically, this JSON structure will be placed in a separate template, and retrieved for the module-use as shown above.
this code is to be placed at the very beginning of the page of the template whose uses are to be validated.
<includeonly>{{#invoke:ParamValidator | validateParams | options = {{PV default options}} | options1 = {"key":"value"} }}</includeonly>
"key" can override any of the options fields described above.


several such options pages may exist, so the user can decide which validation regime to use with each template, by invoking it and passing
the appropriate one for the "options" parameter.


]=]
]=]


function extract_options( frame )
local util = {
local options
empty = function( s )
local args = frame.args
return s == nil  or type( s ) == 'string' and (mw.text.trim( s ) == '' or s == '-') -- compatible with module:תבנית מידע 
end
,
extract_options = function( frame, optionsPrefix )
optionsPrefix = optionsPrefix or 'options'


local options = not empty( args.options ) and mw.text.jsonDecode( args.options ) or {}
local options, n, more = {}
if frame.args['module_options'] then
local module_options = mw.loadData( frame.args['module_options'] )
if type( module_options ) ~= 'table' then return {} end
local title = mw.title.getCurrentTitle()
local local_ptions = module_options[ title.namespace ] or module_options[ title.nsText ] or {}
for k, v in pairs( local_ptions ) do options[k] = v end
end
repeat
ok, more = pcall( mw.text.jsonDecode, frame.args[optionsPrefix .. ( n or '' )] )
if ok and type( more ) == 'table' then
for k, v in pairs( more ) do options[k] = v end
end
n = ( n or 0 ) + 1
until not ok


local n = 1
return options
while not empty( args['options' .. n] ) do
end
local more = mw.text.jsonDecode( args['options' .. n] )
,
for k, v in pairs(more) do options[k] = v end
build_namelist = function ( template_name, sp )
n = n + 1
local res = { template_name }
if sp then
if type( sp ) == 'string' then sp = { sp } end
for _, p in ipairs( sp ) do table.insert( res, template_name .. '/' .. p ) end
end
return res
end
,
table_empty = function( t ) -- normally, test if next(t) is nil, but for some perverse reason, non-empty tables returned by loadData return nil...
if type( t ) ~= 'table' then return true end
for a, b in pairs( t ) do return false end
return true
end
end
,
}


return options
local function _readTemplateData( templateName )
local title = mw.title.makeTitle( 0, templateName ) 
local templateContent = title and title.exists and title:getContent() -- template's raw content
local capture =  templateContent and mw.ustring.match( templateContent, '<templatedata%s*>(.*)</templatedata%s*>' ) -- templatedata as text
-- capture = capture and mw.ustring.gsub( capture, '"(%d+)"', tonumber ) -- convert "1": {} to 1: {}. frame.args uses numerical indexes for order-based params.
local trailingComma = capture and mw.ustring.find( capture, ',%s*[%]%}]' ) -- look for ,] or ,} : jsonDecode allows it, but it's verbotten in json
if capture and not trailingComma then return pcall( mw.text.jsonDecode, capture ) end
return false
end
end


function build_namelist( options, template_name )
local function readTemplateData( templateName )
local res = { template_name }
if type( templateName ) == 'string' then
local sp = options and options['doc-subpage']
templateName = { templateName, templateName .. '/' .. docSubPage }
if sp then
end
if type( sp ) == 'string' then sp = { sp } end
if type( templateName ) == "table" then
for _, p in ipairs( sp ) do table.insert( res, template_name .. '/' .. p ) end
for _, name in ipairs( templateName ) do
local td, result = _readTemplateData( name )  
if td then return result end
end
end
end
return res
return nil
end
end


function empty( s )
return not s or type( s ) == 'string' and mw.text.trim( s ) == ''
end
function validateParams( frame )
local options = extract_options( frame )


-- this is the function to be called by other modules. it expects the frame, and then an optional list of subpages, e.g. { "Documentation" }.
-- if second parameter is nil, only tempalte page will be searched for templatedata.
local function calculateViolations( frame, subpages )
-- used for parameter type validy test. keyed by TD 'type' string. values are function(val) returning bool.
local type_validators = {
['number'] = function( s ) return mw.language.getContentLanguage():parseFormattedNumber( s ) end
}
local function compatible( typ, val )
local func = type_validators[typ]
return type( func ) ~= 'function' or util.empty( val ) or func( val )
end
local function list_empty_or_contains(ar, searched)
if not ar or #ar == 0 then return true end
for _, val in ipairs(ar) do if val == searched then return true end end
return false
end
local t_frame = frame:getParent()
local t_frame = frame:getParent()
local t_args, template_name = t_frame.args, t_frame:getTitle()
local t_args, template_name = t_frame.args, t_frame:getTitle()
local td_source = build_namelist( options, template_name )
local td_source = util.build_namelist( template_name, subpages )
local templatedata = require( 'Module:ReadTd' ).ReadTemplateData( td_source )
local templatedata = readTemplateData( td_source )
local td_p = templatedata and templatedata.params -- now we have options, template args, and td params metadata. we can work now.
local td_params = templatedata and templatedata.params
local report = ''
local all_aliases, all_series = {}, {}
local wrap_report = function()
if empty( report ) then return '' end
if not td_params then return { ['no-templatedata'] = { [''] = '' } } end
report = ( options['wrapper-prefix'] or "<div class = 'paramvalidator-wrapper'>" )
-- from this point on, we know templatedata is valid.
.. report
 
.. ( options['wrapper-suffix'] or "</div>" )
local res = {} -- before returning to caller, we'll prune empty tables
local naked = mw.title.new(template_name)['text']
 
-- allow for aliases
for _, p in pairs( td_params ) do for _, alias in ipairs( p.aliases or {} ) do
all_aliases[alias] = p
if tonumber(alias) then all_aliases[tonumber(alias)] = p end
end end
 
-- handle undeclared and deprecated
local already_seen = {}
local series = frame.args['series']
for p_name, value in pairs( t_args ) do
local tp_param, noval, numeric, table_name = td_params[p_name] or all_aliases[p_name], util.empty( value ), tonumber( p_name )
local hasval = not noval
 
if not tp_param and series then -- 2nd chance. check to see if series
for s_name, p in pairs(td_params) do
if mw.ustring.match( p_name, '^' .. s_name .. '%d+' .. '$') then
-- mw.log('found p_name '.. p_name .. '  s_name:' .. s_name, ' p is:', p) debugging series support
tp_param = p
end -- don't bother breaking. td always correct.
end
end
report = mw.ustring.gsub( raw, 'tname-naked', template_name )
if not tp_param then -- not in TD: this is called undeclared
report = mw.ustring.gsub( report, 'templatename', template_name )
-- calculate the relevant table for this undeclared parameter, based on parameter and value types
return report
table_name =
noval and numeric and 'empty-undeclared-numeric' or
noval and not numeric and 'empty-undeclared' or
hasval and numeric and 'undeclared-numeric' or
'undeclared' -- tzvototi nishar.
else -- in td: test for deprecation and mistype. if deprecated, no further tests
table_name = tp_param.deprecated and hasval and 'deprecated'
or tp_param.deprecated and noval and 'empty-deprecated'
or not compatible( tp_param.type, value ) and 'incompatible'  
or not series and already_seen[tp_param] and hasval and 'duplicate'
or hasval and not list_empty_or_contains(tp_param.suggestedvalues , value) and 'unsuggested-value'
 
already_seen[tp_param] = hasval
end
-- report it.
if table_name then
res[table_name] = res[table_name] or {}
res[table_name][p_name] = value
end
end
-- test for empty/missing paraeters declared "required"
for p_name, param in pairs( td_params ) do
if param.required and util.empty( t_args[p_name] ) then
local is_alias
for _, alias in ipairs( param.aliases or {} ) do is_alias = is_alias or not util.empty( t_args[alias] ) end
if not is_alias then
res['empty-required'] = res['empty-required'] or {}
res['empty-required'][p_name] = ''
end
end
end
end
return res
end
-- wraps report in hidden frame
local function wrapReport(report, template_name, options)
if util.empty( report ) then return '' end
local naked = mw.title.new( template_name )['text']
mw.log(report)
report = ( options['wrapper-prefix'] or "<div class = 'paramvalidator-wrapper'><span class='paramvalidator-error'>" )
.. report
.. ( options['wrapper-suffix'] or "</span></div>" )
report = mw.ustring.gsub( report, 'tname_naked', naked )
report = mw.ustring.gsub( report, 'templatename', template_name )
return report
end
-- this is the "user" version, called with {{#invoke:}} returns a string, as defined by the options parameter
local function validateParams( frame )
-- for purple pages:
if frame:getParent().args['skip parameters validation'] then return '[[ קטגוריה:דפים עם שגיאות פרמטריות שקיבלו חנינה]]' end
local options, report, template_name = util.extract_options( frame ), '', frame:getParent():getTitle()


local ignore = function( p_name )
local ignore = function( p_name )
for _, p in ipairs( options.ignore or {} ) do
for _, pattern in ipairs( options['ignore'] or {} ) do
if tostring( p ) == tostring( p_name ) then return true end
if mw.ustring.match( p_name, '^' .. pattern .. '$' ) then return true end
end
end
return false
return false
שורה 114: שורה 301:


local replace_macros = function( s, param_names )
local replace_macros = function( s, param_names )
local paramstr, paramvaluestr, paramvaluetab = '', '', {}
local function concat_and_escape( t )
local s = table.concat( t, ', ' )
return ( mw.ustring.gsub( s, '%%', '%%%%' ) )
end
if type( param_names ) == 'table' then
if s and ( type( param_names ) == 'table' ) then
paramstr = table.concat( param_names, ', ' )
local k_ar, kv_ar = {}, {}
for _, param_name in ipairs( param_names ) do
for k, v in pairs( param_names ) do
if not empty( t_args[param_name] ) then
table.insert( k_ar, k )
table.insert( paramvaluetab, param_name .. ': ' .. mw.text.trim( t_args[param_name] ) )
table.insert( kv_ar, k .. ': ' .. v)
end
end
end
elseif type( param_names ) == 'string' then
s = mw.ustring.gsub( s, 'paramname', concat_and_escape( k_ar ) )  
paramstr = param_names
s = mw.ustring.gsub( s, 'paramandvalue', concat_and_escape( kv_ar ) )
end
if param_names then
s = mw.ustring.gsub( s, 'paramname', paramstr )
end
if #paramvaluetab then
s = mw.ustring.gsub( s, 'paramandvalue', table.concat( paramvaluetab, ', ' ) )
end
end
return s
return s
שורה 137: שורה 319:


local report_params = function( key, param_names )
local report_params = function( key, param_names )
if options[key] then
local res = replace_macros( options[key], param_names )
report = report .. replace_macros( options[key], param_names )
report = report .. ( res or '' )
end
return res
end
end


local report_table = function( name, tbl )
-- no option no work.
if #tbl ~= 0 then report_params( name, tbl ) end
if util.table_empty( options ) then return '' end
end


if not templatedata then
-- get the errors.
report_params( 'no-templatedata' )
local violations = calculateViolations( frame, options['doc-subpage'] )
report_params( 'any' )
-- special request of bora: use skip_empty_numeric
return wrap_report()
if violations['empty-undeclared-numeric'] then
end
for i = 1, tonumber( options['skip-empty-numeric'] ) or 0 do  
 
violations['empty-undeclared-numeric'][i] = nil
-- from this point on, we know templatedata is valid.
 
-- undeclared and deprecated contains only non-empty parameters.
-- empty_undeclared contains undeclared parameters with no value
local undeclared, empty_undeclared, empty_required, deprecated, empty_deprecated = {}, {}, {}, {}, {}
local undeclared_numeric, empty_undeclared_numeric = {}, {}
 
local lists_and_names = {
[undeclared] = 'undeclared',
[empty_undeclared] = 'empty-undeclared',
[undeclared_numeric] = 'undeclared-numeric',
[empty_undeclared_numeric] = 'empty-undeclared-numeric',
[deprecated] = 'deprecated',
[empty_deprecated] = 'empty-deprecated',
[empty_required] = 'empty-required'
}
 
-- groups to make the loop simpler.
local undec = { [true] = { [true] = empty_undeclared_numeric, [false] = undeclared_numeric }, [false] = { [true] = empty_undeclared, [false] = undeclared } }
local dep = { [true] = empty_deprecated, [false] = deprecated }
local skip_empty_numeric = tonumber( options['skip-empty-numeric'] ) or 0
 
-- DO THE WORK: if paramete does not appear in TD, add it to undeclared or empty-undeclared.
-- if it does appear, check if it's deprecated, and if so, add it to deprecated or empty-deprecated
for p_name, value in pairs( t_args ) do
if not ignore( p_name ) then
local noval = empty( value )
if not td_p[p_name] then -- fill undeclared or undeclared_empty, depends on noval
local numeric = tonumber( p_name ) ~= nil
local skip = numeric and noval and tonumber( p_name ) <=  skip_empty_numeric -- special request of bora
if not skip then
table.insert( undec[numeric][noval], p_name )
end
elseif td_p[p_name].deprecated then
table.insert( dep[noval], p_name )
end
end
end
end
end
 
-- DO MORE WORK: collect "required" parameters that do not exist, or have empty value.
-- handle ignore list, and prune empty violations - in that order!
for p_name, param in pairs( td_p ) do
local offenders = 0
if not ignore( p_name ) and param.required and empty( t_args[p_name] ) then
for name, tab in pairs( violations ) do  
table.insert( empty_required, p_name )
-- remove ignored parameters from all violations
end
for pname in pairs( tab ) do if ignore( pname ) then tab[pname] = nil end end
-- prune empty violations
if util.table_empty( tab ) then violations[name] = nil end
-- WORK IS DONE. report the errors.
-- if report then count it.
if violations[name] and report_params( name, tab ) then offenders = offenders + 1 end  
end
end


-- WORK IS DONE. report the errors.
local offenders = 0
for list, name in pairs( lists_and_names ) do
if #list ~= 0 and not empty( options[name] ) then
offenders = offenders + 1
report_table( name, list )
end
end
if offenders ~= 0 then report_params( 'any' ) end
if offenders > 1 then report_params( 'multiple' ) end
if offenders > 1 then report_params( 'multiple' ) end
 
if offenders ~= 0 then report_params( 'any' ) end -- could have tested for empty( report ), but since we count them anyway...
return wrap_report()
return wrapReport(report, template_name, options)
end
end


return {
return {
['validateparams'] = validateParams
['validateparams'] = validateParams,
['calculateViolations'] = calculateViolations,
['wrapReport'] = wrapReport
}
}
אוחזר מתוך "https://tzefunotwiki.com/index.php/יחידה:ParamValidator"