Relax NG Schema for CSTUG Bulletin v. 1.0 This is a Relax NG schema for CSTUG bulletin table of contents, abstracts and keywords. The elements will have its own namespace , attributes will have empty namespace. Some elements will belong to other namespaces. The types will be defined using XML Schema Datatype Library. The schema contains Schematron 1.3 assertions intended for use with relames.jar plugin for msv. If you run validation from a command line, you must use relames.jar explicitely, otherwise the assertions will be ignored just as comments and pure Relax NG validation will be performed. Similarly, if you run validation from the JARVTask in Ant, you must specify schemaLanguage = 'http://relaxng.org/ns/structure/1.0+http://www.ascc.net/xml/schematron'. As you will se later, information is often spread accross several files. The assertions verify only data items which must always link to the same file. Some conditions are therefore not validated but there is not a great risk that they will be wrong. The assertions do not verify that the author and keyword id's refer to existing information in the respective file although it could be implemented. Use the stylesheet unknown.xsl for this purpose. The schema is used for the whole group of documents. The start element is defined at the end of this file (see the table of contents above). The schema is documented via elements from a special namespace. A processing instruction links a stylesheet designed for pretty-printing in Mozilla. It works partially even in MS Internet Explorer but nesting of the code is lost. The name attributes of the ref elements are transformed to hyperlinks to the define elements where the pattern is defined. The description of each pattern contains a list of links to patterns where it is used. Please do not save this file alone. It requires a stylesheet for pretty-printing. You can find all files zipped into a single archive at http://bulletin.cstug.cz/xml. The information will always be spread to several XML files. Their home location is, as written above, http://bulletin.cstug.cz/xml. However, we would like to allow their placement in private local copies, redistribution on a CD or DVD etc. In order to make it easier, there will always be an index document containing the links to other files. The first four attributes mean: toc: table of contents keywords: list of keywords authors: list of authors acronyms: list of acronyms The links may be either relative or absolute. It is always preferable to use relative links. If you really need absolute links, put all files to the same directory and define the absolute part in the base attribute. The templates in the supplied stylesheets obtain the URL as xsl:value-of select='concat(@base, @toc)'/. The element optionally contains the date and time of last modification. The contents is empty. All linked files must point back to the index document by an indexhref attribute. The contents of some elements may be either present in the same file or may be included from another file. Inclusion will use standard XML mechanisms. We will therefore need a subset of XInclude. xml Some elements need information on language. We will make advantage of the lang attribute. The texts in the bulletin are in Czech, Slovak, English and German only. We will therefore limit the values to the ISO 639-2 codes of these languages only but the repertoire may be extended in the future if it becomes necessary. cze slo eng ger Some elements require a token as its ID. Each principal element (the one which can appear in start) must point to the index document. The link may be either relative or absolute. Some elements may optionally have an attribute containing the date (and time) of the last modification according to ISO 8601. It is allowed to specify either the date or the full time. The time is used to distinguis several releases within the same day. The attribute represents the date/time when the document was physically modified. It may differ from the file timestamp. The list of acronyms is used to store associations of the ID's and real texts. The root element must contain a link to the index file. Each acronym assigns an ID's to the HTML and TEX representation of logos. The text version is used in the plain text output and as a sort key. Optional meaning can be used for explanation. The acronym ID must be unique. The "id" attribute of the "acronym" element must be unique. The TEX part may contain text and references to other acronyms. The HTML part can contain any markup from namespaces different from this one. References to other acronyms can also be used. The content model for the html element must be defined separately because it can be used recursively. The text part may only contain plain text. The meaning may only contain plain text. The contents of the reference to an acronym will be the acronym's ID. We decided to put the ID to the contents instead of the attribute so that users, who do not process the document with our stylesheets which expadn references, see at least something. The following pattern will be used repeatedly. The keywords are stored in their own file. The root element must contain a link to the index file. Each keyword has its (human readable) identifier and text which may refer to an acronym. The Czech text is mandatory, the English and Slovak translations are optional. The name of the child element is taken from the 3-letter language code according to ISO 639-2. However, the acronyms or product names have the same form in all languages. In such a case we use the alement named ALL. No other language variants are then allowed. You can make use of one of the supplied stylesheets: kwd.xsl for XSLT 1.0 or kwd2.xsl for XSLT 2.0. They expand the keyword in the following way: If the element of the required language exists, it is expanded by applying the templates (mainly acro). If the Czech version exists (it must exist in each valid document) it is used. The first child element is used. However, this might happen only if the document is not valid because all valid documents must have the Czech version. The keyword ID must be unique. The "id" attribute of the "keyword" element must be unique. The list of authors is used to replace author names with ID's. The root element must contain a link to the index file. The author's ID must be a NMTOKEN which can be used for sorting. The surname, first name nad initials may contain references to acronyms. The full name can sometimes be expressed as an acronym. In such cases this optional child is preferred to concatenation of the initials and he surname. Again the author ID must be unique. The "id" attribute of the "author" element must be unique. The main table of contents will be in its own file. The volumes may be directly present in the element or included via XInclude. The ISSN* elements have fixed values. 1211-6661 1213-8185 Each volume can be stored in its own file, therefore it must contain the link to the index file. This link is usually unused if the elements are contained within the toc file as children of the toc element. The element needs an ID which is identical to the year of publication. Another attribute is the volume. The issues can either appear as immediate children or included via XInclude. The "year" and "id" attributes of the "volume" element must be equal. The "volume" attributes of the "volume" element is incorrect. The year and the ID of the volume will share the same type. 1991 Each issue can appear in its own file or can be present in the file of volume. The link to the index file is therefore mandatory (but see the comment for volumes). The ID contains both the year and issue number so that it is unique. Generally you are not able to ask the year and the volume number if the issue resides in its own file. The year can thus be obtained as the part of the ID before the slash and the volume number is calculated by subtracting 1990 from the year. The first and last page of the whole issue are optional but they must either both be given or both omitted. The href attribute contains the link to the web page with contents of the issue, the optional pdfref attribute links to the PDF version of the issue provided it is already available. The element contains a few Schematron assertions. Notice that sometimes the last article may end at the cover page which is traditionally reserved just for the colophon. Thus the number of pages need not be strictly divisible by 4. The "id" and "num" attributes have conflicting values. The number of a multiple issue is invalid. The last page must be greater than the first page. The number of pages must be divisible by 4. Sometimes several issues are published together, thus the number is defined by the following pattern. [1-4](-[2-4])? This is the pattern for the issue ID as discussed above. [0-9]{4}/[1-4](-[2-4])? The articles are usually included in the issue element. The only mandatory attributes are therefore its serial number and ID and the language of the main body of the article. The number of the first page is mandatory, the number of the last page is optional. For reasons explained later we allow to insert optionally also the year, volume and issue number. However, it is mandatory either to include all of them or omit all of them. Another attribute, which can optionally be included for the same reason, is the link to the web page. The article elements may also appear in the bibliography listings. In such a case their order will be random and some assertions thus make no sense. Moreover, some material (mainly photos) may be placed between articles and they do not belong to any of them. Thus sometimes firstPage[n] - lastPage[n-1] > 1. The "serial" and "id" attributes are in conflict. The "serial" attribute is not an increment. The first "article" element must have serial="1". The number of a multiple issue within "article" is invalid. The "year", "num", "serial", and "id" attributes of the "article" element are in conflict. The last page of the "article" must not be less than the first page. Page numbering conflict (first page). Page numbering conflict (last page). The article ID consists of the issue ID followed by the hash mark and the serial number. [0-9]{4}/[1-4](-[2-4])?#[0-9]+ The article must always have a title in the main language. Author(s), keyword(s) and abstract(s) are optional. If the titles and abstracts are present in several languages, they must always be given as adjacent elements. The title must contain the language specification (it is the language of the title). Its text may contain references to acronyms. If the language of the article is not English, we often provide an English translation of the title. In the future we wish to translate most titles into English, even if the translation does not appear in the bulletin. The titles in a particular language must not be duplicated. The "title" language must not be duplicated. Article authors will be defined by authors' ID's in the body of the element. Remember that their order is important. The same author must not be entered twice. The author tokens must not be duplicated. Similarly the keywords are also given as keyword ID's in the element body but their order is irrelevant. The keywords must not be duplicated. The keyword tokens must not be duplicated. Abstracts have similar semantics as title with the only exception that they need not be present. However, in the future we would like to translate all existing abstracts into English and possibly even provide abstracts to other articles. The abstract contains usually a few lines of text which are best represented with a plain text and possibly acronyms. Sometimes it may have a complex structure. In such a case the markup must be enclosed within one or more para elements. The abstract in a particular language must not be duplicated. The "abstract" language must not be duplicated. As written above, the abstract contents is a simple text or one or more paragraphs. The paragraphs are enclosed in the para elements. The paragraphs may contain simple text with acronyms or a few markup elements borrowed from HTML. The paragraph elements consists of lists and inline elements. The list elements contain one or more li elements. The inline elements are basically font changing commands and, of course, plain text. Element cmd represents the \cmd macro. The inline elements may contain a text with acronyms and other inline elements. The main purpose of the list of references is to retrieve information on articles into the TEX file. It is, however, allowed to store the retrieved information also as an XML file. The root element must always contain the link to the index file. Its contents is either a list of requested bibliography entries or retrieved articles (usually with the full information as discussed in the description of the article element). The two types of elements must not be mixed. The cite element needs an ID which is identical to the ID of the cited article. Start of the grammar.