<?xml version="1.0"?>
<!--
-@(#)💄📝 Les·M·L xslt/lesml.xslt 2026-03-31T01:28:11Z
+@(#)💄📝 Les·M·L xslt/lesml.xslt 2026-03-31T01:29:32Z
SPDX-FileCopyrightText: 2024, 2025, 2026 Lady <https://www.ladys.computer/about/#lady>
SPDX-License-Identifier: MPL-2.0
-->
This forms the “canonical” definition of the Les·M·L syntax;
no features will be added to the language which are not feasible to
be implemented here.
+
+§ Implementation
+
+❦ On `LesML-´ processing instructions
+
+Processing instructions which begin with `LesML-´ are used internally
+ for bookkeeping as, unlike attributes, the Les·M·L format provides
+ no means by which authors could generate them.
+All of these instructions should be removed by the end of the
+ transformation, but understanding them is important to understanding
+ how this code operates.
+The processing instructions are as follows :—
+
+• `<?LesML-Footnote?>´ identifies list items which are also footnotes.
+
+• `<?LesML-Level?>´ gives the level of the containing block or
+ paragraph.
+
+❦ Doctype, entity, and namespace definitions
+
+The `&block-types;´ and `¶graph-types;´ entities provide X·Path
+ tests for determining if the current node is a block or paragraph
+ element.
+The `&level-pi;´ entity provides an X·Path selector for
+ `<?LesML-Level?>´ processing instructions.
</!-->
<!DOCTYPE transform [
version="1.0"
>
<书社:id>urn:fdc:ladys.computer:20240512:LesML:parser.xslt</书社:id>
+
+<!-->
+❦ Template parameters
+
+The `LESML_SECTION_BREAK_CHARS´ parameter is provided to allow users to
+ configure which characters are treated as section break characters.
+Changing this value isn¦t recommended, but the option is provided for
+ now.
+</!-->
+
<param name="LESML_SECTION_BREAK_CHARS" select="'*-.=_~ ·․‥…⁂⋯─━┄┅┈┉╌╍═╴╶╸╺☙❧ ・＊－．＝＿～'"/>
+
+<!-->
+❦ Functions
+
+The `LesML:split´ function returns a nodeset of H·T·M·L `<span>´
+ elements, with each one giving the next item whivh results from
+ splitting the provided source text on the provided separator.
+This is just a convenience wrapper for the `LesML:do-split´ named
+ template.
+</!-->
+
<exslfunc:function name="LesML:split">
<param name="source" select="string()"/>
<param name="separator" select="'
'"/>
</variable>
<exslfunc:result select="exsl:node-set($result-fragment)/node()"/>
</exslfunc:function>
+
+<!-->
+❦ Named templates
+
+In contrast to matching templates, named templates generally operate on
+ a single string or a flat list of lines, which are provided as
+ parameters.
+Some of these are small utility templates, whereas others provide the
+ bulk of the block‐level processing.
+
+✠ `LesML:do-split´
+
+The `LesML:do-split´ template provides the internal implementation for
+ the `LesML:split´ function.
+</!-->
+
<template name="LesML:do-split">
<param name="source"/>
<param name="separator" select="'
'"/>
</otherwise>
</choose>
</template>
+
+<!-->
+✠ `LesML:comment-out´
+
+The `LesML:comment-out´ template simply produces an X·M·L comment
+ containing the provided text.
+This is only nontrivial because it has to escape any `{U+2D}{U+2D}´
+ sequences, which are not permitted in X·M·L.
+
+The escaping simply inserts `U+034F COMBINING GRAPHEME JOINER´, which
+ is (per Unicode) intended to be used when disambiguating digraphs
+ from two independent characters in sequence.
+</!-->
+
<template name="LesML:comment-out">
<param name="source"/>
<comment>
</for-each>
</comment>
</template>
+
+<!-->
+✠ `LesML:unescape´
+
+The `LesML:unescape´ template takes the provided source text and
+ processes the Unicode character escapes within it.
+The result of this template is raw, unescaped X·M·L, because the only
+ way to implement Unicode character escapes is to convert them to
+ X·M·L entities, which do not have a representation in the X·S·L·T
+ data model.
+Consequently, this template must be among the last ones called and its
+ output cannot be fed back into X·S·L·T safely.
+</!-->
+
<template name="LesML:unescape">
<param name="source"/>
<choose>
</otherwise>
</choose>
</template>
+
+<!-->
+✠ `LesML:expand-sigils´
+
+The `LesML:expand-sigils´ template converts a list of block sigils into
+ the resulting H·T·M·L structure.
+It requires a number of parameters :—
+
+• The sigils themselves, as a flat list of nodes.
+
+• The level of the resulting (outermost) block.
+
+• The type of paragraph to be created within the (innermost) block.
+
+• For preformatted code paragraphs, the associated syntax name.
+
+• The lines of content within the paragraph, as a flat list of nodes.
+
+For as long as the sigils are nonempty, the template is simply called
+ again with the level incremented and the outer sigil removed, with
+ the result wrapped appropriately.
+When no sigils remain, an appropriate paragraph element is created,
+ processing as necessary a leading `¶´ only if it appears on the first
+ line.
+
+Paragraphs and comments are “wrapped” in a `<div>´ to enable them to
+ participate in level nesting, and this `<div>´ is assigned their
+ `@id´ and `@lang´.
+They will later be “unwrapped”, and reclaim these attributes, if they
+ are not footnotes or comments and do not have any block children.
+If a block is nested within the wrapping `<div>´ of a paragraph, the
+ wrapping `<div>´ is preserved as the wrapper for both the paragraph
+ and its nested contents.
+</!-->
+
<template name="LesML:expand-sigils">
<param name="sigils" select="/.."/>
<param name="level" select="0"/>
</otherwise>
</choose>
</template>
+
+<!-->
+✠ `LesML:chunk´
+
+The `LesML:chunk´ template converts the provided list of lines into the
+ list of block elements that they represent.
+This comprises the following steps :—
+
+• Discovering the “last” lines of each block, which are nonempty lines
+ that are not followed by a nonempty line.
+
+• Collecting all of the lines in the block; i·e all lines since the
+ previous “last” line that are not empty.
+
+• Determining the level of the block and its sigils.
+
+• Passing the lines of the block, with any leading sigils removed, to
+ `LesML:expand-sigils´ to generate the actual block element.
+This also requires processing any paragraph‐type indicators and
+ stripping them if necessary to get the type of the underlying
+ paragraph.
+</!-->
+
<template name="LesML:chunk">
<param name="lines" select="/.."/>
<variable name="last-lines" select="$lines[normalize-space()!='' and (normalize-space(following-sibling::*[1])='' or position()=last())]"/>
</call-template>
</for-each>
</template>
+
+<!-->
+✠ `LesML:paragraphize´
+
+The `LesML:paragraphize´ template first turns the provided lines into
+ blocks via `LesML:chunk´, arranges those blocks by applying templates
+ in the `LesML:blockify´ mode, then processes the inline contents of
+ those blocks by applying templates in the `LesML:inlinify´ mode.
+This does much of the processing for the body of a Les·M·L document,
+ altho some things, like footnote handling, can¦t be fully finalized
+ until later.
+</!-->
+
<template name="LesML:paragraphize">
<param name="lines" select="/.."/>
<variable name="chunked-fragment">
</variable>
<apply-templates select="exsl:node-set($chunked-fragment)" mode="LesML:blockify"/>
</template>
+
+<!-->
+✠ `LesML:parse´
+
+The `LesML:parse´ template is the main entrypoint for Les·M·L handling,
+ taking a list of input lines and returning a series of resulting
+ Les·M·L documents.
+This template operates by outputting the result of the first document
+ in the provided lines, and then calling itself recursively with any
+ remaining documents until the lines are exhausted.
+</!-->
+
<template name="LesML:parse">
<param name="lines" select="/.."/>
<param name="parent-params" select="/.."/>
</call-template>
</if>
</template>
+
+<!-->
+❦ Templates in the default mode
+
+This file provides a single template in the default mode: one which
+ matches H·T·M·L `<script>´ elements with a `@type="text/lesml"´ and
+ passes the contents thru to `LesML:parse´.
+</!-->
+
<template match="html:script[@type='text/lesml']">
<variable name="source">
<for-each select=".//text()">
</call-template>
</element>
</template>
+
+<!-->
+❦ Templates in the `LesML:blockify´ mode
+
+There is only one template in the `LesML:blockify´ mode and it can only
+ process an entire document at a time.
+It successively applies templates in the `LesML:nesting-blocks´ and
+ `LesML:inlinify´ modes to convert the relatively flat structure
+ produced by chunking into approximately the final tree.
+</!-->
+
<template match="/" mode="LesML:blockify">
<variable name="nested-fragment">
<apply-templates mode="LesML:nesting-blocks"/>
<with-param name="footnote-ids" select="$nested-document/*/processing-instruction()[local-name()='LesML-Footnote']"/>
</apply-templates>
</template>
+
+<!-->
+❦ Templates in the `LesML:nesting-blocks´ mode
+
+For any given block element, this template moves any following elements
+ with a higher level inside and then reprocesses the resulting
+ children.
+Elements which are preceded by elements of a lower level are assumed to
+ be nested and are not rendered.
+(At the time that these templates run, block elements can only have
+ block siblings.)
+</!-->
+
<template match="*[&level-pi;]" mode="LesML:nesting-blocks">
<variable name="current-level" select="number(&level-pi;)"/>
<if test="not(preceding-sibling::*[&level-pi; and $current-level>&level-pi;])">
<apply-templates select="node()" mode="LesML:nesting-blocks"/>
</copy>
</template>
+
+<!-->
+❦ Templates in the `LesML:inlinify´ and related modes
+
+These modes break up the lines of a paragraph into constituent inline
+ nodes.
+</!-->
+
<template match="*[&level-pi;]" mode="LesML:inlinify">
<param name="footnote-ids" select="/.."/>
<copy>
Lady’s Gitweb / LesML / commitdiff