Branch: Tag:

2000-07-19

2000-07-19 16:21:23 by Johan Sundström <oyasumi@gmail.com>

MODULE_LOCATION refdoc.

Rev: server/modules/examples/location.pike:1.1

1: + inherit "module"; + // All roxen modules must inherit module.pike    -  + constant cvs_version = "$Id: location.pike,v 1.1 2000/07/19 16:21:23 jhs Exp $"; + constant module_type = MODULE_LOCATION; + constant module_name = "RefDoc for MODULE_LOCATION"; + constant module_doc = "This module does nothing, but its inlined " +  "documentation gets imported into the roxen " +  "programmer manual. You really don't want to " +  "add this module to your virtual server, promise!"; +  + // see common_api.pike for further explanations of the above constants +  + void create() + { +  defvar("mountpoint", +  Variable.Location("/dev/urandom/sort/of/", 0, +  "Mount point", +  "This is where the module will be inserted " +  "in the namespace of your server.")); + } +  + mapping|Stdio.File|void find_file( string path, RequestID id ) + //! The return value is either a <ref>result mapping</ref>, an + //! open file descriptor or 0, signifying that your module did not + //! handle the request. + //! + //! This is the fundamental method of all location modules and is, as + //! such, required. It will be called to handle all accesses below the + //! module's mount point. The path argument contains the path to the + //! resource, in the module's own name space, and the id argument + //! contains the request information object. + //! + //! That the path is in the modules name space means that the path + //! will only contain the part of the URL after the module's mount + //! point. If a module is mounted on <tt>/here/</tt> and a user + //! requests <tt>/here/it/is.jpg</tt>, the module will be called with + //! a path of <tt>it/is.jpg</tt>. That way, the administrator can set + //! the mount point to anything she wants, and the module will keep + //! working. Note that changing the mount point to <tt>/here</tt> + //! would give the module the path <tt>/it/is.jpg</tt> for that + //! request. + //! + //! A zero return value means that the module could not find the + //! requested resource. In that case roxen will move on and try to + //! find the resource in in other location modules. Returning -1 means + //! that the requested resource is a directory, in which case the + //! request will be handled by a directory type module. + //! + //! If the module could handle the request, the return value is either + //! a <ref>response mapping</ref> or a Stdio.File object containing + //! the requested file. + { +  return Roxen.http_string_answer(make_random_string( 17, 17 )); + } +  + string make_random_string(int min_len, int|void max_len) + { +  int length = min_len + (max_len ? random(max_len - min_len) +  : random(17)); +  return (string)allocate(length, lambda(){ return random(256); })(); + } +  + string query_location() + //! Returns the location in the virtual server's where your location + //! module is mounted. If you make a location module and leave out + //! this method, the default behaviour inherited from module.pike + //! will return the value of the module variable 'location'. + { +  return query( "mountpoint" ); + } +  + array(int) stat_file( string path, RequestID id ) + //! The stat_file() method emulates Pike's <pi>file_stat()</pi> + //! method, returning information about a file or directory. path + //! is the path to the file or directory in the module's name space, + //! id is the request information object. + //! + //! stat_file() is most commonly used by directory type modules to + //! provide informative directory listings, or by the ftp protocol + //! module to create directory listings. + //! + //! The return value it is expected to be an array of integers in the + //! following format: + //! + //! ({ mode, size, atime, mtime, ctime, uid, gid }) + //! + //! <tt>mode</tt> is an integer containing the unix file permissions + //! of the file. It can be ignored. + //! + //! <tt>size</tt> is an integer containing the size of the file, or a + //! special value in case the object is not actually a file. Minus two + //! means that it is a directory, minus three that it is a symbolic + //! link and minus four that it is a device special file. This value + //! must be given. + //! + //! <tt>atime</tt> is a unixtime integer containing the last time the + //! file was accessed (seconds since 1970). It can be ignored. + //! + //! <tt>mtime</tt> is a unixtime integer containing the last time the + //! file was modified (seconds since 1970). It will be used to handle + //! Last-Modified-Since requests and should be supplied if possible. + //! + //! <tt>ctime</tt> is a unixtime integer containing the time the file + //! was created (seconds since 1970). It can be ignored. + //! + //! <tt>uid</tt> is an integer containing the user id of this file. It + //! will be correlated with the information from the current + //! authentification type module, and used by the CGI executable + //! support module to start CGI scripts as the correct user. It is + //! only necessary for location modules that provide access to a real + //! file system and that implement the <ref>real_file()</ref> method. + //! + //! <tt>gid</tt> is an integer containing the group id of the file. It + //! is needed when uid is needed. + { +  return ({ 0775, // mode +  ({ 17, -2 })[random(2)], // size/special +  963331858, // atime +  963331858, // mtime +  963331858, // ctime +  0, // uid +  0 /* gid */ }); + } // Of course, it's typically silly to return something like this. +  + string|void real_file( string path, RequestID id ) + //! This method translates the path of a file in the module's name + //! space to the path to the file in the real file system. path + //! is the path to the file in the module's name space, id is the + //! request information object. + //! + //! If the file could not be found, or the file doesn't exist on a + //! real file system, zero should be returned. Only location modules + //! that access server files from a real file system need implement + //! this method. See also the <ref>stat_file()</ref> method. + {} +  + array(string)|void find_dir( string path, RequestID id ) + //! The find_dir() returns a directory listing; an array of strings + //! containing the names of all files and directories in this + //! directory. The path parameter is the path to the directory, in the + //! module's name space, id is the request information object. + //! + //! This method is usually called because a previous call to + //! find_file() returned that this path contained a directory and a + //! directory type module is right now trying to create a directory + //! listing of this directory. Note that it is possible that the + //! find_dir() is called in several location modules, and that the + //! actual directory listing shown to the user will be the + //! concatenated result of all those calls. + //! + //! To find information about each entry in the returned array the + //! <ref>stat_file()</ref> is used. + { +  return allocate(random(47)+11, make_random_string)(4, 71); + }   Newline at end of file added.