head 1.3; access; symbols; locks; strict; comment @# @; 1.3 date 2007.01.16.04.12.01; author TWikiContributor; state Exp; branches; next 1.2; 1.2 date 2006.06.25.16.26.33; author TWikiContributor; state Exp; branches; next 1.1; 1.1 date 2006.02.01.12.01.24; author TWikiContributor; state Exp; branches; next ; desc @new-topic @ 1.3 log @buildrelease @ text @---+ Package =TWiki= TWiki operates by creating a singleton object (known as the Session object) that acts as a point of reference for all the different modules in the system. This package is the class for this singleton, and also contains the vast bulk of the basic constants and the per- site configuration mechanisms. Global variables are avoided wherever possible to avoid problems with CGI accelerators such as mod_perl. %TOC% ---++ StaticMethod *getTWikiLibDir* () -> $path STATIC method. Returns the full path of the directory containing TWiki.pm ---++ ObjectMethod *UTF82SiteCharSet* ($utf8) -> $ascii Auto-detect UTF-8 vs. site charset in string, and convert UTF-8 into site charset. ---++ ObjectMethod *writeCompletePage* ($text,$pageType,$contentType) Write a complete HTML page with basic header to the browser. * =$text= is the text of the page body (<html> to </html> if it's HTML) * =$pageType= - May be "edit", which will cause headers to be generated that force caching for 24 hours, to prevent Codev.BackFromPreviewLosesText bug, which caused data loss with IE5 and IE6. * =$contentType= - page content type | text/html This method removes noautolink and nop tags before outputting the page unless $contentType is text/plain. ---++ ObjectMethod *writePageHeader* ($query,$pageType,$contentType,$contentLength) All parameters are optional. * =$query= CGI query object | Session CGI query (there is no good reason to set this) * =$pageType= - May be "edit", which will cause headers to be generated that force caching for 24 hours, to prevent Codev.BackFromPreviewLosesText bug, which caused data loss with IE5 and IE6. * =$contentType= - page content type | text/html * =$contentLength= - content-length | no content-length will be set if this is undefined, as required by HTTP1.1 Implements the post-Dec2001 release plugin API, which requires the writeHeaderHandler in plugin to return a string of HTTP headers, CR/LF delimited. Filters any illegal headers. Plugin headers will override core settings. ---++ ObjectMethod *redirect* ($url,$passthrough) Redirects the request to =$url=, *unless* 1 It is overridden by a plugin declaring a =redirectCgiQueryHandler=. 1 =$session->{cgiQuery}= is =undef= or 1 $query->param('noredirect') is set to a true value. Thus a redirect is only generated when in a CGI context. Normally this method will ignore parameters to the current query. If $passthrough is set, then it will pass all parameters that were passed to the current query on to the redirect target. If the request_method was GET, then all parameters can be passed in the URL. If the request_method was POST then it caches the form data and passes over a cache reference in the redirect GET. Passthrough is only meaningful if the redirect target is on the same server. ---++ ObjectMethod *cacheQuery* () -> $queryString Caches the current query in the params cache, and returns a rewritten query string for the cache to be picked up again on the other side of a redirect. We can't encode post params into a redirect, because they may exceed the size of the GET request. So we cache the params, and reload them when the redirect target is reached. ---++ StaticMethod *isValidWikiWord* ($name) -> $boolean Check for a valid WikiWord or WikiName ---++ StaticMethod *isValidTopicName* ($name) -> $boolean Check for a valid topic name ---++ StaticMethod *isValidAbbrev* ($name) -> $boolean Check for a valid ABBREV (acronym) ---++ StaticMethod *isValidWebName* ($name,$system) -> $boolean STATIC Check for a valid web name. If $system is true, then system web names are considered valid (names starting with _) otherwise only user web names are valid ---++ ObjectMethod *readOnlyMirrorWeb* ($theWeb) -> ($mirrorSiteName,$mirrorViewURL,$mirrorLink,$mirrorNote) If this is a mirrored web, return information about the mirror. The info is returned in a quadruple: | site name | URL | link | note | ---++ ObjectMethod *getSkin* () -> $string Get the currently requested skin path ---++ ObjectMethod *getScriptUrl* ($absolute,$script,$web,$topic,...) -> $scriptURL Returns the URL to a TWiki script, providing the web and topic as "path info" parameters. The result looks something like this: "http://host/twiki/bin/$script/$web/$topic". * =...= - an arbitrary number of name,value parameter pairs that will be url-encoded and added to the url. The special parameter name '#' is reserved for specifying an anchor. e.g. getScriptUrl('x','y','view','#'=>'XXX',a=>1,b=>2) will give .../view/x/y?a=1&b=2#XXX If $absolute is set, generates an absolute URL. $absolute is advisory only; TWiki can decide to generate absolute URLs (for example when run from the command-line) even when relative URLs have been requested. The default script url is taken from {ScriptUrlPath}, unless there is an exception defined for the given script in {ScriptUrlPaths}. Both {ScriptUrlPath} and {ScriptUrlPaths} may be absolute or relative URIs. If they are absolute, then they will always generate absolute URLs. if they are relative, then they will be converted to absolute when required (e.g. when running from the command line, or when generating rss). If $script is not given, absolute URLs will always be generated. If either the web or the topic is defined, will generate a full url (including web and topic). Otherwise will generate only up to the script name. An undefined web will default to the main web name. ---++ ObjectMethod *getPubUrl* ($absolute,$web,$topic,$attachment) -> $url Composes a pub url. If $absolute is set, returns an absolute URL. If $absolute is set, generates an absolute URL. $absolute is advisory only; TWiki can decide to generate absolute URLs (for example when run from the command-line) even when relative URLs have been requested. $web, $topic and $attachment are optional. A partial URL path will be generated if one or all is not given. ---++ ObjectMethod *getIconUrl* ($absolute,$iconName) -> $iconURL Map an icon name to a URL path. ---++ ObjectMethod *mapToIconFileName* ($fileName,$default) -> $fileName Maps from a filename (or just the extension) to the name of the file that contains the image for that file type. ---++ ObjectMethod *getOopsUrl* ($template,@@options) -> $absoluteOopsURL Composes a URL for an "oops" error page. The @@options consists of a list of key => value pairs. The following keys are used: * =-web= - web name * =-topic= - topic name * =-def= - optional template def within the main template file * =-params= - a single parameter, or a reference to an array of parameters These are passed in the URL as '¶m1=' etc. Do _not_ include the "oops" part in front of the template name. Alternatively you can pass a reference to an OopsException in place of the template. All other parameters will be ignored. The returned URL ends up looking something like this: "http://host/twiki/bin/oops/$web/$topic?template=$template¶m1=$scriptParams[0]..." Note: if {keep} is true in the params, then they will also be pushed into the current query. ---++ ObjectMethod *normalizeWebTopicName* ($theWeb,$theTopic) -> ($theWeb,$theTopic) Normalize a Web.TopicName See TWikiFuncDotPm for a full specification of the expansion (not duplicated here) *WARNING* if there is no web specification (in the web or topic parameters) the web defaults to $TWiki::cfg{UsersWebName}. If there is no topic specification, or the topic is '0', the topic defaults to the web home topic name. ---++ ClassMethod *new* ($loginName,$query,\%initialContext) Constructs a new TWiki object. Parameters are taken from the query object. * =$loginName= is the username of the user you want to be logged-in if none is available from a session or browser. Used mainly for side scripts and debugging. * =$query= the CGI query (may be undef, in which case an empty query is used) * =\%initialContext= - reference to a hash containing context name=value pairs to be pre-installed in the context hash ---++ ObjectMethod *finish* Complete processing after the client's HTTP request has been responded to. Right now this does two things: 1 calling TWiki::Client to flushing the user's session (if any) to disk, 2 breaking circular references to allow garbage collection in persistent environments ---++ ObjectMethod *writeLog* ($action,$webTopic,$extra,$user) * =$action= - what happened, e.g. view, save, rename * =$wbTopic= - what it happened to * =$extra= - extra info, such as minor flag * =$user= - user who did the saving (user object or string user name) Write the log for an event to the logfile ---++ ObjectMethod *writeWarning* ($text) Prints date, time, and contents $text to $TWiki::cfg{WarningFileName}, typically 'warnings.txt'. Use for warnings and errors that may require admin intervention. Use this for defensive programming warnings (e.g. assertions). ---++ ObjectMethod *writeDebug* ($text) Prints date, time, and contents of $text to $TWiki::cfg{DebugFileName}, typically 'debug.txt'. Use for debugging messages. ---++ StaticMethod *applyPatternToIncludedText* ($text,$pattern) -> $text Apply a pattern on included text to extract a subset ---++ ObjectMethod *inlineAlert* ($template,$def,...) -> $string Format an error for inline inclusion in rendered output. The message string is obtained from the template 'oops'.$template, and the DEF $def is selected. The parameters (...) are used to populate %PARAM1%..%PARAMn% ---++ StaticMethod *parseSections* ($text) -> ($string,$sectionlistref) Generic parser for sections within a topic. Sections are delimited by STARTSECTION and ENDSECTION, which may be nested, overlapped or otherwise abused. The parser builds an array of sections, which is ordered by the order of the STARTSECTION within the topic. It also removes all the SECTION tags from the text, and returns the text and the array of sections. Each section is a =TWiki::Attrs= object, which contains the attributes {type, name, start, end} where start and end are character offsets in the string *after all section tags have been removed*. All sections are required to be uniquely named; if a section is unnamed, it will be given a generated name. Sections may overlap or nest. See test/unit/Fn_SECTION.pm for detailed testcases that round out the spec. ---++ ObjectMethod *expandVariablesOnTopicCreation* ($text,$user) -> $text * =$text= - text to expand * =$user= - reference to user object. This is the user expanded in e.g. %USERNAME. Optional, defaults to logged-in user. Expand limited set of variables during topic creation. These are variables expected in templates that must be statically expanded in new content. # SMELL: no plugin handler ---++ StaticMethod *entityEncode* ($text,$extras) -> $encodedText Escape special characters to HTML numeric entities. This is *not* a generic encoding, it is tuned specifically for use in TWiki. HTML4.0 spec: "Certain characters in HTML are reserved for use as markup and must be escaped to appear literally. The "<" character may be represented with an entity, &lt;. Similarly, ">" is escaped as &gt;, and "&" is escaped as &amp;. If an attribute value contains a double quotation mark and is delimited by double quotation marks, then the quote should be escaped as &quot;.

Other entities exist for special characters that cannot easily be entered with some keyboards..." This method encodes HTML special and any non-printable ascii characters (except for \n and \r) using numeric entities. FURTHER this method also encodes characters that are special in TWiki meta-language. $extras is an optional param that may be used to include *additional* characters in the set of encoded characters. It should be a string containing the additional chars. ---++ StaticMethod *entityDecode* ($encodedText) -> $text Decodes all numeric entities (e.g. &#123;). _Does not_ decode named entities such as &amp; (use HTML::Entities for that) ---++ StaticMethod *urlEncode* ($string) -> encodedstring Encode by converting characters that are illegal in URLs to their %NN equivalents. This method is used for encoding strings that must be embedded _verbatim_ in URLs; it cannot be applied to URLs themselves, as it escapes reserved characters such as = and ?. RFC 1738, Dec. '94: > ...Only alphanumerics [0-9a-zA-Z], the special characters $-_.+!*'(), and reserved characters used for their reserved purposes may be used unencoded within a URL. Reserved characters are $&+,/:;=?@@ - these are _also_ encoded by this method. SMELL: For non-ISO-8859-1 $TWiki::cfg{Site}{CharSet}, need to convert to UTF-8 before URL encoding. This encoding only supports 8-bit character codes. ---++ StaticMethod *urlDecode* ($string) -> decodedstring Reverses the encoding done in urlEncode. ---++ StaticMethod *isTrue* ($value,$default) -> $boolean Returns 1 if =$value= is true, and 0 otherwise. "true" means set to something with a Perl true value, with the special cases that "off", "false" and "no" (case insensitive) are forced to false. Leading and trailing spaces in =$value= are ignored. If the value is undef, then =$default= is returned. If =$default= is not specified it is taken as 0. ---++ StaticMethod *spaceOutWikiWord* ($word,$sep) -> $string Spaces out a wiki word by inserting a string (default: one space) between each word component. With parameter $sep any string may be used as separator between the word components; if $sep is undefined it defaults to a space. ---++ ObjectMethod *enterContext* ($id,$val) Add the context id $id into the set of active contexts. The $val can be anything you like, but should always evaluate to boolean TRUE. An example of the use of contexts is in the use of tag expansion. The commonTagsHandler in plugins is called every time tags need to be expanded, and the context of that expansion is signalled by the expanding module using a context id. So the forms module adds the context id "form" before invoking common tags expansion. Contexts are not just useful for tag expansion; they are also relevant when rendering. Contexts are intended for use mainly by plugins. Core modules can use $session->inContext( $id ) to determine if a context is active. ---++ ObjectMethod *leaveContext* ($id) Remove the context id $id from the set of active contexts. (see =enterContext= for more information on contexts) ---++ ObjectMethod *inContext* ($id) Return the value for the given context id (see =enterContext= for more information on contexts) ---++ StaticMethod *registerTagHandler* ($tag,$fnref) STATIC Add a tag handler to the function tag handlers. * =$tag= name of the tag e.g. MYTAG * =$fnref= Function to execute. Will be passed ($session, \%params, $web, $topic ) ---++ StaticMethod *registerRESTHandler* ($subject,$verb,\&fn) Adds a function to the dispatch table of the REST interface for a given subject. See TWikiScripts#rest for more info. * =$subject= - The subject under which the function will be registered. * =$verb= - The verb under which the function will be registered. * =\&fn= - Reference to the function. The handler function must be of the form: sub handler(\%session,$subject,$verb) -> $text where: * =\%session= - a reference to the TWiki session object (may be ignored) * =$subject= - The invoked subject (may be ignored) * =$verb= - The invoked verb (may be ignored) *Since:* TWiki::Plugins::VERSION 1.1 ---++ StaticMethod *restDispatch* ($subject,$verb)=>\&fn Returns the handler function associated to the given $subject and $werb, or undef if none is found. *Since:* TWiki::Plugins::VERSION 1.1 ---++ ObjectMethod *handleCommonTags* ($text,$web,$topic) -> $text Processes %VARIABLE%, and %TOC% syntax; also includes 'commonTagsHandler' plugin hook. Returns the text of the topic, after file inclusion, variable substitution, table-of-contents generation, and any plugin changes from commonTagsHandler. ---++ ObjectMethod *addToHEAD* ($id,$html) Add =$html= to the HEAD tag of the page currently being generated. Note that TWiki variables may be used in the HEAD. They will be expanded according to normal variable expansion rules. The 'id' is used to ensure that multiple adds of the same block of HTML don't result in it being added many times. ---++ StaticMethod *initialize* ($pathInfo,$remoteUser,$topic,$url,$query) -> ($topicName,$webName,$scriptUrlPath,$userName,$dataDir) Return value: ( $topicName, $webName, $TWiki::cfg{ScriptUrlPath}, $userName, $TWiki::cfg{DataDir} ) Static method to construct a new singleton session instance. It creates a new TWiki and sets the Plugins $SESSION variable to point to it, so that TWiki::Func methods will work. This method is *DEPRECATED* but is maintained for script compatibility. Note that $theUrl, if specified, must be identical to $query->url() ---++ StaticMethod *readFile* ($filename) -> $text Returns the entire contents of the given file, which can be specified in any format acceptable to the Perl open() function. Fast, but inherently unsafe. WARNING: Never, ever use this for accessing topics or attachments! Use the Store API for that. This is for global control files only, and should be used *only* if there is *absolutely no alternative*. ---++ StaticMethod *expandStandardEscapes* ($str) -> $unescapedStr Expands standard escapes used in parameter values to block evaluation. The following escapes are handled: | *Escape:* | *Expands To:* | | =$n= or =$n()= | New line. Use =$n()= if followed by alphanumeric character, e.g. write =Foo$n()Bar= instead of =Foo$nBar= | | =$nop= or =$nop()= | Is a "no operation". | | =$quot= | Double quote (="=) | | =$percnt= | Percent sign (=%=) | | =$dollar= | Dollar sign (=$=) | @ 1.2 log @buildrelease @ text @d24 1 d33 5 a37 1 $text is the HTML of the page body (<html> to </html>) d39 2 a40 1 This method removes noautolink and nop tags before outputting the page. d60 14 a73 1 ---++ ObjectMethod *redirect* ($url,...) d75 1 a75 3 Generate a CGI redirect to $url unless (1) $session->{cgiQuery} is undef or (2) $query->param('noredirect') is set to a true value. Thus a redirect is only generated when in a CGI context. a76 2 The ... parameters are concatenated to the message written when printing to STDOUT, and are ignored for a redirect. d78 10 a87 2 Redirects the request to $url, via the CGI module object $query unless overridden by a plugin declaring a =redirectCgiQueryHandler=. d137 1 a137 1 * =...= - an arbitrary number of name,value parameter pairs that will be url-encoded and added to the url. The special parameter name '#' is reserved for specifying an anchor. e.g. getScriptUrl('x','y','view','#'=>'XXX',a=>1,b=>2) will give .../view/x/y#XXX?a=1&b=2 d196 3 a203 12 
Input:                      Return:
  ( 'Web',  'Topic' )         ( 'Web',  'Topic' )
  ( '',     'Topic' )         ( 'Main', 'Topic' )
  ( '',     '' )              ( 'Main', 'WebHome' )
  ( '',     'Web/Topic' )     ( 'Web',  'Topic' )
  ( '',     'Web.Topic' )     ( 'Web',  'Topic' )
  ( 'Web1', 'Web2.Topic' )    ( 'Web2', 'Topic' )
  ( '%MAINWEB%', 'Web2.Topic' ) ( 'Main', 'Topic' )
  ( '%TWIKIWEB%', 'Web2.Topic' ) ( 'TWiki', 'Topic' )
Note: Function renamed from getWebTopic d205 1 a205 2 SMELL: WARNING: this function defaults the web and topic names. Be very careful where you use it! d207 3 d212 3 a214 1 ---++ ClassMethod *new* ($remoteUser,$query) d217 5 a221 2 * =$remoteUser= the logged-in user (login name) * =$query= the query d226 1 d236 1 d267 1 d274 21 d296 1 d433 2 a434 1 ---+++ registerRESTHandler( $subject, $verb, \&fn ) d453 4 a456 1 ---+++ restDispatch( $subject, $verb) => \&fn d462 2 d465 1 d475 1 d510 14 @ 1.1 log @buildrelease @ text @d206 4 a209 3 to. Right now this only entails one activity: calling TWiki::Client to flushing the user's session (if any) to disk. @