On 5/5/16 12:45 AM, Alexander Law wrote:
Done (previous patch cleaned). This patch optimizes XSL transformations contained in docbook-xsl (1.78.1).
I have looked through this patch, and it's awesome. I have tweaked it a bit more along the lines you guys have started, and now the build time is pretty much the same as with DSSSL. Attached is my final patch, which I plan to commit as soon as the new branch opens.
(I only have Alexander Lakhin as credit right now. Please let me know if anyone else contributed.)
-- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
From bc7fc7693e3a054ab6e46c3dc931c1d2535f2b40 Mon Sep 17 00:00:00 2001 From: Peter Eisentraut <pete...@gmx.net> Date: Fri, 3 Jun 2016 14:47:38 -0400 Subject: [PATCH] doc: Speed up XSLT builds The upstream XSLT stylesheets use some very general XPath expressions in some places that end up being very slow. We can optimize them with knowledge about the DocBook document structure and our particular use thereof. For example, when counting preceding chapters to get a number for the current chapter, we only need to count preceding sibling nodes (more or less) instead of searching through the entire node tree for chapter elements. This change attacks the slowest pieces as identified by xsltproc --profile. This makes the HTML build roughly 10 times faster, resulting in the new total build time being about the same as the old DSSSL-based build. Some of the non-HTML build targets (especially FO) will also benefit a bit, but they have not been specifically analyzed. With this, also remove the pg.fast parameter, which was previously a hack to get the build to a manageable speed. Alexander Lakhin <a.lak...@postgrespro.ru>, with some additional tweaking by me --- doc/src/sgml/stylesheet-common.xsl | 8 +- doc/src/sgml/stylesheet-speedup-common.xsl | 94 +++++++++ doc/src/sgml/stylesheet-speedup-xhtml.xsl | 293 +++++++++++++++++++++++++++++ doc/src/sgml/stylesheet.xsl | 1 + 4 files changed, 391 insertions(+), 5 deletions(-) create mode 100644 doc/src/sgml/stylesheet-speedup-common.xsl create mode 100644 doc/src/sgml/stylesheet-speedup-xhtml.xsl diff --git a/doc/src/sgml/stylesheet-common.xsl b/doc/src/sgml/stylesheet-common.xsl index de36376..e384113 100644 --- a/doc/src/sgml/stylesheet-common.xsl +++ b/doc/src/sgml/stylesheet-common.xsl @@ -7,11 +7,10 @@ all output formats (HTML, HTML Help, XSL-FO, etc.). --> +<xsl:include href="stylesheet-speedup-common.xsl" /> <!-- Parameters --> -<xsl:param name="pg.fast" select="'0'"/> - <!-- <xsl:param name="draft.mode"> <xsl:choose> @@ -31,9 +30,8 @@ <xsl:param name="callout.graphics" select="'0'"></xsl:param> <xsl:param name="toc.section.depth">2</xsl:param> <xsl:param name="linenumbering.extension" select="'0'"></xsl:param> -<xsl:param name="generate.index" select="1 - $pg.fast"></xsl:param> -<xsl:param name="section.autolabel" select="1 - $pg.fast"></xsl:param> -<xsl:param name="section.label.includes.component.label" select="1 - $pg.fast"></xsl:param> +<xsl:param name="section.autolabel" select="1"></xsl:param> +<xsl:param name="section.label.includes.component.label" select="1"></xsl:param> <xsl:param name="refentry.xref.manvolnum" select="0"/> <xsl:param name="formal.procedures" select="0"></xsl:param> <xsl:param name="punct.honorific" select="''"></xsl:param> diff --git a/doc/src/sgml/stylesheet-speedup-common.xsl b/doc/src/sgml/stylesheet-speedup-common.xsl new file mode 100644 index 0000000..007fdf6 --- /dev/null +++ b/doc/src/sgml/stylesheet-speedup-common.xsl @@ -0,0 +1,94 @@ +<?xml version="1.0" encoding="utf-8"?> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"; + version='1.0'> + +<!-- Performance-optimized versions of some upstream templates from common/ + directory --> + +<!-- from common/labels.xsl --> + +<xsl:template match="chapter" mode="label.markup"> + <xsl:choose> + <xsl:when test="@label"> + <xsl:value-of select="@label"/> + </xsl:when> + <xsl:when test="string($chapter.autolabel) != 0"> + <xsl:if test="$component.label.includes.part.label != 0 and + ancestor::part"> + <xsl:variable name="part.label"> + <xsl:apply-templates select="ancestor::part" + mode="label.markup"/> + </xsl:variable> + <xsl:if test="$part.label != ''"> + <xsl:value-of select="$part.label"/> + <xsl:apply-templates select="ancestor::part" + mode="intralabel.punctuation"> + <xsl:with-param name="object" select="."/> + </xsl:apply-templates> + </xsl:if> + </xsl:if> + <xsl:variable name="format"> + <xsl:call-template name="autolabel.format"> + <xsl:with-param name="format" select="$chapter.autolabel"/> + </xsl:call-template> + </xsl:variable> + <xsl:choose> + <xsl:when test="$label.from.part != 0 and ancestor::part"> + <xsl:number from="part" count="chapter" format="{$format}" level="any"/> + </xsl:when> + <xsl:otherwise> + <!-- Optimization for pgsql-docs: When counting to get label for + this chapter, preceding chapters can only be our siblings or + children of a preceding part, so only count those instead of + scanning the entire node tree. --> + <!-- <xsl:number from="book" count="chapter" format="{$format}" level="any"/> --> + <xsl:number value="count(../preceding-sibling::part/chapter) + count(preceding-sibling::chapter) + 1" format="{$format}"/> + </xsl:otherwise> + </xsl:choose> + </xsl:when> + </xsl:choose> +</xsl:template> + +<xsl:template match="appendix" mode="label.markup"> + <xsl:choose> + <xsl:when test="@label"> + <xsl:value-of select="@label"/> + </xsl:when> + <xsl:when test="string($appendix.autolabel) != 0"> + <xsl:if test="$component.label.includes.part.label != 0 and + ancestor::part"> + <xsl:variable name="part.label"> + <xsl:apply-templates select="ancestor::part" + mode="label.markup"/> + </xsl:variable> + <xsl:if test="$part.label != ''"> + <xsl:value-of select="$part.label"/> + <xsl:apply-templates select="ancestor::part" + mode="intralabel.punctuation"> + <xsl:with-param name="object" select="."/> + </xsl:apply-templates> + </xsl:if> + </xsl:if> + <xsl:variable name="format"> + <xsl:call-template name="autolabel.format"> + <xsl:with-param name="format" select="$appendix.autolabel"/> + </xsl:call-template> + </xsl:variable> + <xsl:choose> + <xsl:when test="$label.from.part != 0 and ancestor::part"> + <xsl:number from="part" count="appendix" format="{$format}" level="any"/> + </xsl:when> + <xsl:otherwise> + <!-- Optimization for pgsql-docs: When counting to get label for + this appendix, preceding appendixes can only be our siblings or + children of a preceding part, so only count those instead of + scanning the entire node tree. --> + <!-- <xsl:number from="book|article" count="appendix" format="{$format}" level="any"/> --> + <xsl:number value="count(../preceding-sibling::part/appendix) + count(preceding-sibling::appendix) + 1" format="{$format}"/> + </xsl:otherwise> + </xsl:choose> + </xsl:when> + </xsl:choose> +</xsl:template> + +</xsl:stylesheet> diff --git a/doc/src/sgml/stylesheet-speedup-xhtml.xsl b/doc/src/sgml/stylesheet-speedup-xhtml.xsl new file mode 100644 index 0000000..7c391a0 --- /dev/null +++ b/doc/src/sgml/stylesheet-speedup-xhtml.xsl @@ -0,0 +1,293 @@ +<?xml version="1.0" encoding="utf-8"?> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"; + xmlns="http://www.w3.org/1999/xhtml"; + version='1.0'> + +<!-- Performance-optimized versions of some upstream templates from xhtml/ + directory --> + +<!-- from xhtml/autoidx.xsl --> + +<xsl:template match="indexterm" mode="reference"> + <xsl:param name="scope" select="."/> + <xsl:param name="role" select="''"/> + <xsl:param name="type" select="''"/> + <xsl:param name="position"/> + <xsl:param name="separator" select="''"/> + + <xsl:variable name="term.separator"> + <xsl:call-template name="index.separator"> + <xsl:with-param name="key" select="'index.term.separator'"/> + </xsl:call-template> + </xsl:variable> + + <xsl:variable name="number.separator"> + <xsl:call-template name="index.separator"> + <xsl:with-param name="key" select="'index.number.separator'"/> + </xsl:call-template> + </xsl:variable> + + <xsl:variable name="range.separator"> + <xsl:call-template name="index.separator"> + <xsl:with-param name="key" select="'index.range.separator'"/> + </xsl:call-template> + </xsl:variable> + + <xsl:choose> + <xsl:when test="$separator != ''"> + <xsl:value-of select="$separator"/> + </xsl:when> + <xsl:when test="$position = 1"> + <xsl:value-of select="$term.separator"/> + </xsl:when> + <xsl:otherwise> + <xsl:value-of select="$number.separator"/> + </xsl:otherwise> + </xsl:choose> + + <xsl:choose> + <xsl:when test="@zone and string(@zone)"> + <xsl:call-template name="reference"> + <xsl:with-param name="zones" select="normalize-space(@zone)"/> + <xsl:with-param name="position" select="position()"/> + <xsl:with-param name="scope" select="$scope"/> + <xsl:with-param name="role" select="$role"/> + <xsl:with-param name="type" select="$type"/> + </xsl:call-template> + </xsl:when> + <xsl:otherwise> + <a> + <xsl:apply-templates select="." mode="class.attribute"/> + <xsl:variable name="title"> + <xsl:choose> + <xsl:when test="$index.prefer.titleabbrev != 0"> + <xsl:apply-templates select="(ancestor-or-self::set|ancestor-or-self::book|ancestor-or-self::part|ancestor-or-self::reference|ancestor-or-self::partintro|ancestor-or-self::chapter|ancestor-or-self::appendix|ancestor-or-self::preface|ancestor-or-self::article|ancestor-or-self::section|ancestor-or-self::sect1|ancestor-or-self::sect2|ancestor-or-self::sect3|ancestor-or-self::sect4|ancestor-or-self::sect5|ancestor-or-self::refentry|ancestor-or-self::refsect1|ancestor-or-self::refsect2|ancestor-or-self::refsect3|ancestor-or-self::simplesect|ancestor-or-self::bibliography|ancestor-or-self::glossary|ancestor-or-self::index|ancestor-or-self::webpage|ancestor-or-self::topic)[last()]" mode="titleabbrev.markup"/> + </xsl:when> + <xsl:otherwise> + <xsl:apply-templates select="(ancestor-or-self::set|ancestor-or-self::book|ancestor-or-self::part|ancestor-or-self::reference|ancestor-or-self::partintro|ancestor-or-self::chapter|ancestor-or-self::appendix|ancestor-or-self::preface|ancestor-or-self::article|ancestor-or-self::section|ancestor-or-self::sect1|ancestor-or-self::sect2|ancestor-or-self::sect3|ancestor-or-self::sect4|ancestor-or-self::sect5|ancestor-or-self::refentry|ancestor-or-self::refsect1|ancestor-or-self::refsect2|ancestor-or-self::refsect3|ancestor-or-self::simplesect|ancestor-or-self::bibliography|ancestor-or-self::glossary|ancestor-or-self::index|ancestor-or-self::webpage|ancestor-or-self::topic)[last()]" mode="title.markup"/> + </xsl:otherwise> + </xsl:choose> + </xsl:variable> + + <xsl:attribute name="href"> + <xsl:choose> + <xsl:when test="$index.links.to.section = 1"> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="(ancestor-or-self::set|ancestor-or-self::book|ancestor-or-self::part|ancestor-or-self::reference|ancestor-or-self::partintro|ancestor-or-self::chapter|ancestor-or-self::appendix|ancestor-or-self::preface|ancestor-or-self::article|ancestor-or-self::section|ancestor-or-self::sect1|ancestor-or-self::sect2|ancestor-or-self::sect3|ancestor-or-self::sect4|ancestor-or-self::sect5|ancestor-or-self::refentry|ancestor-or-self::refsect1|ancestor-or-self::refsect2|ancestor-or-self::refsect3|ancestor-or-self::simplesect|ancestor-or-self::bibliography|ancestor-or-self::glossary|ancestor-or-self::index|ancestor-or-self::webpage|ancestor-or-self::topic)[last()]"/> + <!-- Optimization for pgsql-docs: We only have an index as a + child of book, so look that up directly instead of + scanning the entire node tree. Also, don't look for + setindex. --> + <!-- <xsl:with-param name="context" select="(//index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))] | //setindex[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))])[1]"/> --> + <xsl:with-param name="context" select="(/book/index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))])[1]"/> + </xsl:call-template> + </xsl:when> + <xsl:otherwise> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="."/> + <xsl:with-param name="context" select="(//index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))] | //setindex[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))])[1]"/> + </xsl:call-template> + </xsl:otherwise> + </xsl:choose> + + </xsl:attribute> + + <xsl:value-of select="$title"/> <!-- text only --> + </a> + + <xsl:variable name="id" select="(@id|@xml:id)[1]"/> + <xsl:if test="key('endofrange', $id)[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))]"> + <xsl:apply-templates select="key('endofrange', $id)[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))][last()]" mode="reference"> + <xsl:with-param name="position" select="position()"/> + <xsl:with-param name="scope" select="$scope"/> + <xsl:with-param name="role" select="$role"/> + <xsl:with-param name="type" select="$type"/> + <xsl:with-param name="separator" select="$range.separator"/> + </xsl:apply-templates> + </xsl:if> + </xsl:otherwise> + </xsl:choose> +</xsl:template> + +<xsl:template name="reference"> + <xsl:param name="scope" select="."/> + <xsl:param name="role" select="''"/> + <xsl:param name="type" select="''"/> + <xsl:param name="zones"/> + + <xsl:choose> + <xsl:when test="contains($zones, ' ')"> + <xsl:variable name="zone" select="substring-before($zones, ' ')"/> + <xsl:variable name="target" select="key('sections', $zone)"/> + + <a> + <xsl:apply-templates select="." mode="class.attribute"/> + <xsl:call-template name="id.attribute"/> + <xsl:attribute name="href"> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="$target[1]"/> + <xsl:with-param name="context" select="//index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))][1]"/> + </xsl:call-template> + </xsl:attribute> + <xsl:apply-templates select="$target[1]" mode="index-title-content"/> + </a> + <xsl:text>, </xsl:text> + <xsl:call-template name="reference"> + <xsl:with-param name="zones" select="substring-after($zones, ' ')"/> + <xsl:with-param name="position" select="position()"/> + <xsl:with-param name="scope" select="$scope"/> + <xsl:with-param name="role" select="$role"/> + <xsl:with-param name="type" select="$type"/> + </xsl:call-template> + </xsl:when> + <xsl:otherwise> + <xsl:variable name="zone" select="$zones"/> + <xsl:variable name="target" select="key('sections', $zone)"/> + + <a> + <xsl:apply-templates select="." mode="class.attribute"/> + <xsl:call-template name="id.attribute"/> + <xsl:attribute name="href"> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="$target[1]"/> + <!-- Optimization for pgsql-docs: Only look for index under book + instead of searching the whole node tree. --> + <!-- <xsl:with-param name="context" select="//index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))][1]"/> --> + <xsl:with-param name="context" select="/book/index[count(ancestor::node()|$scope) = count(ancestor::node()) and ($role = @role or $type = @type or (string-length($role) = 0 and string-length($type) = 0))][1]"/> + </xsl:call-template> + </xsl:attribute> + <xsl:apply-templates select="$target[1]" mode="index-title-content"/> + </a> + </xsl:otherwise> + </xsl:choose> +</xsl:template> + + +<!-- from xhtml/chunk-common.xsl --> + +<xsl:template name="chunk-all-sections"> + <xsl:param name="content"> + <xsl:apply-imports/> + </xsl:param> + + <!-- Optimization for pgsql-docs: Since we set a fixed $chunk.section.depth, + we can do away with a bunch of complicated XPath searches for the + previous and next sections at various levels. --> + + <xsl:if test="$chunk.section.depth != 1"> + <xsl:message terminate="yes"> + <xsl:text>Error: If you change $chunk.section.depth, then you must update the performance-optimized chunk-all-sections-template.</xsl:text> + </xsl:message> + </xsl:if> + + <xsl:variable name="prev" + select="(preceding::book[1] + |preceding::preface[1] + |preceding::chapter[1] + |preceding::appendix[1] + |preceding::part[1] + |preceding::reference[1] + |preceding::refentry[1] + |preceding::colophon[1] + |preceding::article[1] + |preceding::topic[1] + |preceding::bibliography[parent::article or parent::book or parent::part][1] + |preceding::glossary[parent::article or parent::book or parent::part][1] + |preceding::index[$generate.index != 0] + [parent::article or parent::book or parent::part][1] + |preceding::setindex[$generate.index != 0][1] + |ancestor::set + |ancestor::book[1] + |ancestor::preface[1] + |ancestor::chapter[1] + |ancestor::appendix[1] + |ancestor::part[1] + |ancestor::reference[1] + |ancestor::article[1] + |ancestor::topic[1] + |preceding::sect1[1] + |ancestor::sect1[1])[last()]"/> + + <xsl:variable name="next" + select="(following::book[1] + |following::preface[1] + |following::chapter[1] + |following::appendix[1] + |following::part[1] + |following::reference[1] + |following::refentry[1] + |following::colophon[1] + |following::bibliography[parent::article or parent::book or parent::part][1] + |following::glossary[parent::article or parent::book or parent::part][1] + |following::index[$generate.index != 0] + [parent::article or parent::book][1] + |following::article[1] + |following::topic[1] + |following::setindex[$generate.index != 0][1] + |descendant::book[1] + |descendant::preface[1] + |descendant::chapter[1] + |descendant::appendix[1] + |descendant::article[1] + |descendant::topic[1] + |descendant::bibliography[parent::article or parent::book][1] + |descendant::glossary[parent::article or parent::book or parent::part][1] + |descendant::index[$generate.index != 0] + [parent::article or parent::book][1] + |descendant::colophon[1] + |descendant::setindex[$generate.index != 0][1] + |descendant::part[1] + |descendant::reference[1] + |descendant::refentry[1] + |following::sect1[1] + |descendant::sect1[1])[1]"/> + + <xsl:call-template name="process-chunk"> + <xsl:with-param name="prev" select="$prev"/> + <xsl:with-param name="next" select="$next"/> + <xsl:with-param name="content" select="$content"/> + </xsl:call-template> +</xsl:template> + +<xsl:template name="html.head"> + <xsl:param name="prev" select="/foo"/> + <xsl:param name="next" select="/foo"/> + + <!-- Optimization for pgsql-docs: Cut out a bunch of things we don't need + here, including an expensive //legalnotice search. --> + + <head> + <xsl:call-template name="system.head.content"/> + <xsl:call-template name="head.content"/> + + <xsl:if test="$prev"> + <link rel="prev"> + <xsl:attribute name="href"> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="$prev"/> + </xsl:call-template> + </xsl:attribute> + <xsl:attribute name="title"> + <xsl:apply-templates select="$prev" mode="object.title.markup.textonly"/> + </xsl:attribute> + </link> + </xsl:if> + + <xsl:if test="$next"> + <link rel="next"> + <xsl:attribute name="href"> + <xsl:call-template name="href.target"> + <xsl:with-param name="object" select="$next"/> + </xsl:call-template> + </xsl:attribute> + <xsl:attribute name="title"> + <xsl:apply-templates select="$next" mode="object.title.markup.textonly"/> + </xsl:attribute> + </link> + </xsl:if> + + <xsl:call-template name="user.head.content"/> + </head> +</xsl:template> + +</xsl:stylesheet> diff --git a/doc/src/sgml/stylesheet.xsl b/doc/src/sgml/stylesheet.xsl index 7967b36..631fcc4 100644 --- a/doc/src/sgml/stylesheet.xsl +++ b/doc/src/sgml/stylesheet.xsl @@ -6,6 +6,7 @@ <xsl:import href="http://docbook.sourceforge.net/release/xsl/current/xhtml/chunk.xsl"/> <xsl:include href="stylesheet-common.xsl" /> +<xsl:include href="stylesheet-speedup-xhtml.xsl" /> <!-- Parameters --> -- 2.8.3
-- Sent via pgsql-docs mailing list (pgsql-docs@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-docs
