Add capability for new constraintDecl element: * Require scheme * Require a sch:ns prefix="tei" uri="" child * Create pop-up set of other sch:ns elements to insert * Require other children be in Schematron namespace Added sch:rule elements PRN to avoid new warning about contextless Schematron. Remove the altIdent-only-NCName constraint, as it is no longer needed — the content of altIdent in P5 is now just xs:NCName. update to use version 4.6.0 of P5 Per TEI ticket #2285 disallow altIdent as a direct child of classSpec, constraintSpec, dataSpec, macroSpec; thus leaving it as only available as a child of attDef, elementSpec, and valItem. In order to do this, remove altIdent from model.identSynonyms (leaving that class with only gloss and equiv, i.e. the same as model.identEquiv, sigh) and add it back to attDef and valItem. We will soon not allow more than 1 child of content, so updated the content model of schemaSpec to have only 1 child (in this case, sequence). Since TEI ticket #1735 has now addressed the ( content | valList )* portion of the content models of dataSpec and macroSpec (changing them to ( content | valList )?), we no longer need to do that here. Thus undo most of the changes of 2018-07-19. Motivated by #136, restrict docLang of schemaSpec to only 1 value. Per #2050, replace nonfatal with warning or information on role attributes. Turns out there is only one such case (on tei-source), which I changed to information. Also fix the comment that precedes that rule, as P5 no longer has the att.readFrom class. As part of working through Stylesheets ticket #402, Martin Holmes and I thought that an elementSpec addressing a TEI element should not be in mode add. This is intended to enforce that rule. Further addressing TEI ticket #1735: constrain content of both dataSpec and macroSpec so that at most 1 content or 1 valList child is permitted. While there give the ident attribute a semi-controlled vocabulary. update to use version 3.3.0 of P5 format $versionDate so it does not have the timezone appended use $versionDate in date in TEI Header, too (instead of calling current-date() again) Per #1735, add constraints on content child of elementSpec: if mode of elementSpec is add or replace, require a content if mode of elementSpec is delete, require said elementSpec to be empty Add constraint to check that xml:id values are unique. We do this because this file (and the RELAX NG produced from it) needs to be validated with ID/IDREF checking turned off. Since TEI does not use IDREFs you might think that it is not a problem to turn it off, but that is also the feature that checks for ID uniqueness. So we test for it separately. Add xml:lang and a somewhat bogus versionDate (set from a parameter) to various gloss, desc, and remarks elements as required by tei_odds schema. Change altIdent-only-NCName to test value for a colon if not a child of taxonomy or valItem (rather than test that value is castable as NCName if it is a child of an ODD element other than valItem, because some processors don't like the NCName test). Bug fix: check for specGrpRef was checking that it pointed to elementSpec, not specGrp Remove our definition of content, as TEI’s now does what we want. Remove constraints testing for both url and key on moduleRef, and for both include and except on either moduleRef or classRef, as the TEI mechanism for doing so (an attList with org=choice) seems to be working now. Change generation system — create odd4odds.odd from a single XSLT transform run with TEI P5 is input Update driver file to P5 3.1.0 (from 2.7.0) Remove declaration of content of schemaSpec, as P5 now allows constraintSpec. Convert to Pure ODD restore deleted model.resourceLike allow funder use dataRef instead of macroRef inside datatype test the source attr of dataRef, too lots of whitespace tweaks just to make more readable delete two remaining references to non-existent allowText loosen the mode-child-sanity constraint to consider an attribute other than mode or ident to be a child. This may be a bit too loose, may want to expand list of attrs that are excepted in the test, and thus flag an error Fix bug in only-1-per contraint (which was added 674 days ago — why did this bug last even 1 day?) Changes for Pure ODD: constrain content of content delete att.global.rendition and att.global.responsibility delete allowText (as I think TEI has decided to go with textNode instead) require key of elementRef change expression of all datatypes (which happen to all be data.enumerated) from RNG to Pure ODD change expression of most of our content models from RNG to Pure ODD. Changes for Pure ODD: constrain key of classRef, elementRef, and macroRef constrain include and except of classRef, also flagging an error if both are present Re-generate based on current TEI (in preperation for conversion to Pure ODD) Allow editionStmt (and edition) in the teiHeader Remove our check for colon in prefix of schemaSpec and elementSpec, as TEI no longer allows it. Removed our alterations to type of list, as TEI has improved it. Add constraint only-1-per, which warns user iff there are more than 1 elementSpec with the same ident warn if no schemaSpec or greater than 1 of 'em warn against 2 moduleRefs with same key check that the 7 required elements are included when include is used (we already check that they are not deleted or excepted) warn when an element with a mode of delete has children warn when an element with a mode of change, add, or replace has no children Change name Add health warning about using Check ID/IDREF in oXygen Added sch:pattern to list of elements allowed inside constraint. Remove the new notatedMusic and gb elments. neaten up a bit for readers of this XML file allow xi:fallback to have XML mixed content check that moduleRef that uses url has a (new!) prefix Test that required elements are not removed. changed check for colons in prefix of schemaSpec from a Schematron warning to a RELAX NG constraint warn if the value of prefix of moduleRef matches the prefix of anything else Beefed up prose, then corrections per Julia, including changing name of language (and thus prefix). remove addition of att.identifiable to att.combineable, as it's there already in current P5 remove (extraneous?) rule elements that have no context retro-write several change log entries make mode of classes required issue warning if we find a classSpec, elementSpec, or a macroSpec outside of schemaSpec, and not referenced from within schemaSpec bug fix: include of moduleRef had been a semi list — changed to closed require ident of classSpec require ident of elementSpec Constrain ident of classSpec. During workshop: constrain start of schemaSpec constrain key of elmentRef improve remakrs of constraint so that it correctly reflects that we only permit ISO Schematron constrain content of altIdent, unless it is a child of valItem Now that stylesheets support sucking in multiple external schemas that happen to have pattern name collision, and references to patterns within them, take advantage of it by using real RELAX NG and ISO Schematron schemas. Add persName, placeName, and orgName. add this revisionDesc add body as a starting root element made lists into XIncludes remove scriptNote add check for required modules delete model.entryPart.top delete model.msItemPart delete model.personPart delete model.titlepagePart put att.identifiable into att.combinable (reflect a change in P5, maybe?) various first crack One or more of the elements included on the '' ＜moduleRef＞ are not actually available in that module One or more of the elements excepted on the '' ＜moduleRef＞ are not actually available in that module class: module: ************************ WARNING ************************************* WARNING: This is a derived file. Edit the source XSLT program and run it against P5 again to make changes. See <sourceDesc> and <appInfo> for info on the XSLT program that was used. ********************************************************************** TEI ODD Customization for writing TEI ODD Customizations Syd Bauman Women Writers Project Text Encoding Initiative Consortium

Copyright 2017 Syd Bauman and Northeastern WWP; some rights reserved.

This TEI-encoded ODD file is available under the terms of the OSI BSD 2-clause License, to wit:

Redistribution and use in source and derived forms, with or without modification, are permitted provided that the following conditions are met: Redistributions of source code must retain the above copyright notice, this list of conditions, and the following disclaimer. Redistributions in binary or other derived form must reproduce the above copyright notice, this list of conditions, and the following disclaimer in the documentation or other materials provided with the distribution. Note, however, that although modification and redistribution of derivatives of this ODD (e.g., the schemas that result from processing with Roma) is legal, the resulting schema is (definitionally) not TEI-conformant.

This software is provided by the copyright holders and contributors as is and any express or implied warranties, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose are disclaimed. In no event shall the copyright holder or contributors be liable for any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited to, procurement of substitute goods or services; loss of use, data, or profits; or business interruption) however caused and on any theory of liability, whether in contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use of this software, even if advised of the possibility of such damage.

Generated from TEI-to-tei_customization.xslt, which is based on the Women Writers Project’s odd4odds_driver.odd, itself based on the TEI Consoritum’s tei_odds.odd by Sebastian Rahtz, which can be found on GitHub

 This ODD file was generated by running against TEI P5 “”, found at on
Introduction

The primary intent of this schema is to help you and your XML editor write a customization ODD; it in no way pretends to be the last arbiter of what is or is not allowed in an ODD file in general, nor even in a TEI customization ODD. Thus it makes lots of assumptions (read: enforces lots of constraints) that probably make sense when writing a customization schema, but may not in your particular case. Thus this schema may well flag as invalid things that are perfectly reasonable to have in your ODD.

Background

The TEI ODD language was designed both for the creation and customization of the TEI Guidelines, and also for the creation (and perhaps customization) of other, non-TEI, markup languages. Thus the TEI ODD langauge is, by default, much more flexible than needed for writing TEI customization ODDs. For example, the Guidelines define the key attribute of the moduleRef element as any XML name (without a namespace prefix, i.e. an xsd:NCName), even though the only possible values when used to customize TEI are the 20 or so module names defined in the TEI Guidelines.

Of course, when using RomaThe canonical installation of Roma is available at the TEI-C site, but it is an open source tool available on GitHub, which may be installed on any GNU/Linux system., a web-based front-end editor for ODD files, this is not a problem. The web form gives the user only the appropriate TEI values to choose from. However, when editing ODD files by hand, and thus when teaching TEI customization, it is much more efficient to catch errors like mis-spelled module names before handing the ODD file to an ODD processor (e.g. the aforementioned Roma or teitorelaxng).

Thus the Women Writers Project has developed this TEI customization for the purpose of having a schema to use that deliberately makes it easier to write a TEI customization, at the expense of the complete flexibility ODD provides.

This schema permits a valid document to use a variety of elements as the root element. This is for debugging and file maintenance convenience. As always, any conforming TEI ODD must have either TEI or teiCorpus as the root element.

REMINDER: This language is not canonical — it is intended to be helpful, not definitive

Technical Information

Of course the lists of elements, classes, and on rare occasion even modules change as TEI P5 matures. Rather than requiring that this customization be manually edited with each release of TEI P5, the ODD for this customization is itself generated from a source file. To generate this customization, an XSLT stylesheet reads in the source to TEI P5, and from it generates the lists of elements, classes, and modules needed. The output of the stylesheet is the ODD file for this customization.

We call this TEI language the TEI ODD Customization for writing TEI ODD Customizations language (or TOCTOC for short). It is instantited in the TEI exemplar tei_customization, which differs from tei_oddsAvailable in the main TEI P5 git repository at GitHub in a variety of ways, including: constraining the elements available (e.g., gap and imprimatur are not available) requiring some otherwise optional attributes (e.g., mode of classes, or module on elementSpec unless the mode is specified as add) limiting the content of constraint to ISO Schematron, and thus permitting validation of the Schematron therein constraining the values of several attributes requiring that the four required modules be imported requiring that altIdent be an NCName (i.e., not have a prefix) unless it is the child of valItem (or taxonomy) flagging *Spec elements that are neither in the schemaSpec nor referred to from a specGrpRef within it warning if required elements (e.g., teiHeader) are deleted using, and encouraging the use of, the prefix attribute of moduleRef

If editing an XML document associated with this schema in oXygen, you will need to uncheck the Check ID/IDREF preference box. It can be found in Options > Preferences > XML > XML Parser > RELAX NG.

Schema Specification a schema to help writing TEI customization ODDs; will (incorrectly) flag extensions as invalid required modules minus lots of elements that probably don't make sense in an ODD for an ODD, tagdocs is required too access to the RELAX NG and ISO Schematron languages Note: end user may need to change paths (Here they are relative paths to where they would be found from TEIC/TEI/P5/Exmplars/.) allow use of the specialized naming elements, as many people are used to them allow tables, figures, and formulæ The specification element ＜＞ has both a mode= of "delete" and child elements, which is incongruous The specification element ＜＞ has a mode= of "", but does not have any child elements or schema-changing attributes, which is incongruous required modules missing one or more of the required modules (tei, core, header, textstructure). A *Spec element should either be within schemaSpec, or be in a specGrp referred to by a specGrpRef wihin schemaSpec. ＜＞ should be in ＜schemaSpec＞ or referred to from within ＜schemaSpec＞ TEI permits schemaSpec as a repeatable child of a variety of elements (including front, body, back, encodingDesc, div, or any numbered division element). But roma only processes the first schemaSpec found (in document order). So we just limit ourselves to 1 and only 1. There's no ＜schemaSpec＞, so this isn't much of a TEI customization You have more than one ＜schemaSpec＞; current ODD processors will only look at the first one specifies which language to use when creating documentation if the description for an element, attribute, class, or macro is available in more than one language

The current TEI ODD processor only handles 1 value as the value of docLang.

 specifies entry points to the schema, i.e. which elements may be used as the root of documents conforming to it.

The content of the tei_customization schemaSpec is somewhat more restrictive than TEI. To wit, neither altIdent nor equiv are permitted there can be at most one gloss, which (if present) must precede the desc there must be one and only one desc the four core modules must be included the value of start requires values that are valid TEI element names

** The following set of <elementSpec> elements exists to allow us to ** refer to them from within the content model for <constraintDecl>. ** This allows for completion assistance from XML tools such as oXygen. ** To change the set of <sch:ns> elements that are defined, change ** the set of namespaces bound to the root <xsl:stylesheet> of this ** XSLT program. ns ISO Schematron

The tei_customization version of constraintDecl is more strict than the vanilla TEI version: it requires that the scheme be schematron and that the contents be only Schematron elements, the first of which must be sch:ns prefix="tei" uri="".

This is not strictly necessary. The TEI patterns have a default prefix (the value of ident of schemaSpec), so if only one external module is imported, it does not need a prefix — there will not be any collisions of pattern names. But it is good practice to use a prefix anyway, and it is required if more than one external module that use patterns with the same name are included. This is the case, in fact, for the RELAX NG and ISO Schematron schemas that are included in this schema. a ＜moduleRef＞ that uses the url= attribute should have a prefix= attribute, too The '' module is included (by reference) more than once We could get away with using a single test, i.e. "('TEI','teiHeader','fileDesc','titleStmt','title','publicationStmt','sourceDesc')=$exceptions", but then we wouldn't get individualized error msgs. In either case, if & when TEI changes the list of elements which are *required*, we need to change this code (in ) manually. Removing ＜TEI＞ from your schema guarantees it is not TEI conformant, and will will likely be outright invalid Removing ＜teiHeader＞ from your schema guarantees it is not TEI conformant Removing ＜fileDesc＞ from your schema guarantees it is not TEI conformant Removing ＜titleStmt＞ from your schema guarantees it is not TEI conformant Removing ＜title＞ from your schema guarantees it is not TEI conformant Removing ＜publicationStmt＞ from your schema guarantees it is not TEI conformant Removing ＜sourceDesc＞ from your schema guarantees it is not TEI conformant Not including ＜TEI＞ in your schema guarantees it is not TEI conformant, and will likely be outright invalid Not including ＜teiHeader＞ in your schema guarantees it is not TEI conformant Not including ＜fileDesc＞ in your schema guarantees it is not TEI conformant Not including ＜titleStmt＞ in your schema guarantees it is not TEI conformant Not including ＜publicationStmt＞ in your schema guarantees it is not TEI conformant Not including ＜sourceDesc＞ in your schema guarantees it is not TEI conformant Not including ＜title＞ in your schema guarantees it is not TEI conformant supplies a list of the elements which are to be copied from the specified module into the schema being defined. supplies a list of the elements which are not to be copied from the specified module into the schema being defined. supplies a list of class members which are to be included in the schema being defined. supplies a list of class members which are to be excluded from the schema being defined.

The content of the tei_customization classSpec is somewhat more restrictive than TEI. To wit, neither altIdent nor equiv are permitted, there can be at most one gloss and must be one and only one desc.

 adding altIdent to content model because it was removed from model.identSynonyms adding altIdent to content model because it was removed from model.identSynonyms same as vanilla except impose order and max of 1 each for members of glossLike and descLike in a customization ODD, the mode= attribute of ＜elementSpec＞ should be specified the module= attribute of ＜elementSpec＞ must be specified anytime the mode= is not 'add' Current ODD processors will not correctly handle more than one ＜elementSpec＞ with the same @ident Removing ＜TEI＞ from your schema guarantees it is not TEI conformant Removing ＜teiHeader＞ from your schema guarantees it is not TEI conformant Removing ＜fileDesc＞ from your schema guarantees it is not TEI conformant Removing ＜titleStmt＞ from your schema guarantees it is not TEI conformant Removing ＜title＞ from your schema guarantees it is not TEI conformant Removing ＜publicationStmt＞ from your schema guarantees it is not TEI conformant Removing ＜sourceDesc＞ from your schema guarantees it is not TEI conformant When adding a new element (whether replacing an old one or not), a content model must be specified; but this ＜elementSpec＞ does not have a ＜content＞ child. When used to delete an element from your schema, the ＜elementSpec＞ should be empty When used to add an element, ＜elementSpec＞ (or its ancestor ＜schemaSpec＞) should have an @ns attribute. The only difference here is that TEI does not permit <egXML> to be a descendant of <egXML>; but since no one knows why that restriction exists, we've removed it.

For our purposes, constraints must be expressed in ISO Schematron. (TEI permits others, including non-XML expressions, although there is no processing of anything other than ISO Schematron.)

 in examples we'd like to be able to show use of <xi:include>; of course, we can't put in an <xi:include> element, or it will be processed! Thus, here we use a bogus namespace xml of course <xi:include> refers to <xi:fallback>, so we need to declare that, too, just in case. Extreme conference proceedings markup Docbook XHTML Catalogue of Attributes Catalogue of Attribute Classes Catalogue of Elements Catalogue of Macros Catalogue of Model Classes Catalogue of Deprecations Table of Contents general entity name RELAX NG identifier DTD fragment identifier namespace parameter entity name Digital Object Identifier International Standard Book Number any form of web address The @xml:id "" on ＜＞ duplicates an @xml:id found earlier in the document Constrains the source attribute of various tagset documentation elements to those values recommended by TEI The @source attribute of ＜＞ is not in the recommended format, which is either "tei:current" or "tei:x.y.z", where x.y.z is a version number.