From 08036123933559b54b877de30572bf89ebf50b22 Mon Sep 17 00:00:00 2001
From: Alexander Dyuzhev <dyuzhev@gmail.com>
Date: Tue, 14 May 2024 19:48:34 +0300
Subject: [PATCH] SOURCE parsing added, #181

---
 README.adoc                            | 56 +++++++++----------
 pom.xml                                |  2 +-
 src/main/resources/stepmod2mn.adoc.xsl | 76 ++++++++++++++++++++++----
 3 files changed, 95 insertions(+), 39 deletions(-)

diff --git a/README.adoc b/README.adoc
index 5f63164..f90333c 100644
--- a/README.adoc
+++ b/README.adoc
@@ -17,14 +17,14 @@ You will need the `maven` build tool and `make`.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <XML-FileName>
+java -Xss5m -jar target/stepmod2mn-1.59.jar <XML-FileName>
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/resource_docs/draughting_elements/resource.xml 
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/resource_docs/draughting_elements/resource.xml 
 ----
 
 NOTE: resulted AsciiDoc found in data/resource_docs/draughting_elements/document.adoc
@@ -34,28 +34,28 @@ NOTE: resulted AsciiDoc found in data/resource_docs/draughting_elements/document
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <XML-FileName> [--output <AsciiDoc-FileName>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <XML-FileName> [--output <AsciiDoc-FileName>]
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/resource_docs/draughting_elements/resource.xml --output result_dir/draughting_elements/document.adoc
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/resource_docs/draughting_elements/resource.xml --output result_dir/draughting_elements/document.adoc
 ----
 
 === Convert all resource.xml and module.xml in the specified folder and sub-folders into the Metanorma AsciiDoc format:
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Input-folder>
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Input-folder>
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/
 ----
 
 NOTE: the result AsciiDocs will be saved in the folder with the source resource.xml and module.xml files.
@@ -65,14 +65,14 @@ NOTE: the result AsciiDocs will be saved in the folder with the source resource.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Input-folder> [--output <Output-folder>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Input-folder> [--output <Output-folder>]
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/ --output documents/
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/ --output documents/
 ----
 
 
@@ -80,42 +80,42 @@ java -Xss5m -jar target/stepmod2mn-1.58.jar data/ --output documents/
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Input-folder> [--exclude <documents or part numbers list>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Input-folder> [--exclude <documents or part numbers list>]
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/ --exclude "machining_features 104"
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/ --exclude "machining_features 104"
 ----
 
 === Convert only specified documents resource.xml and module.xml in the specified folder and sub-folders into the Metanorma AsciiDoc format:
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Input-folder> [--include-only <documents list>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Input-folder> [--include-only <documents list>]
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar data/ --include-only "fundamentals_of_product_description_and_support kinematics"
+java -Xss5m -jar target/stepmod2mn-1.59.jar data/ --include-only "fundamentals_of_product_description_and_support kinematics"
 ----
 
 === Generate Metanorma collection files (metanorma.yml and collection.yml) for the documents specified in the publication index xml file in the tags 'resource_docs' and 'modules':
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Publication-Index-XML-FileName>
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Publication-Index-XML-FileName>
 ----
 
 e.g.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar publication/part1000/CR_PMI_5/publication_index.xml
+java -Xss5m -jar target/stepmod2mn-1.59.jar publication/part1000/CR_PMI_5/publication_index.xml
 ----
 
 NOTE: the Metanorma collection files .yml will be saved in the 3 level up folder (usually, source repository's root folder).
@@ -125,14 +125,14 @@ and save them into the output folder:
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Publication-Index-XML-FileName> [--output <Output-folder>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Publication-Index-XML-FileName> [--output <Output-folder>]
 ----
 
 e.g.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar iso-10303-srl/iso-10303-stepmod-wg12/publication/part1000/CR_PMI_5/publication_index.xml --output iso-10303-srl/
+java -Xss5m -jar target/stepmod2mn-1.59.jar iso-10303-srl/iso-10303-stepmod-wg12/publication/part1000/CR_PMI_5/publication_index.xml --output iso-10303-srl/
 ----
 
 NOTE: the output folder should contains the folder with the generated .adoc (usually, 'documents'). You can specify this folder obviously by the parameter '--input-documents':
@@ -141,7 +141,7 @@ e.g.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar iso-10303-srl/iso-10303-stepmod-wg12/publication/part1000/CR_PMI_5/publication_index.xml --output iso-10303-srl/CR_PMI_5 --input-documents iso-10303-srl/documents
+java -Xss5m -jar target/stepmod2mn-1.59.jar iso-10303-srl/iso-10303-stepmod-wg12/publication/part1000/CR_PMI_5/publication_index.xml --output iso-10303-srl/CR_PMI_5 --input-documents iso-10303-srl/documents
 ----
 
 NOTE: the Metanorma collection files (.yml) will be saved in the folder 'iso-10303-srl/CR_PMI_5' and the references point to the documents in the folder iso-10303-srl/documents.
@@ -151,14 +151,14 @@ NOTE: the Metanorma collection files (.yml) will be saved in the folder 'iso-103
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar <Publication-Index-XML-FileName> [--type <resource_docs or modules>]
+java -Xss5m -jar target/stepmod2mn-1.59.jar <Publication-Index-XML-FileName> [--type <resource_docs or modules>]
 ----
 
 e.g.
 
 [source,sh]
 ----
-java -Xss5m -jar target/stepmod2mn-1.58.jar publication/part1000/CR_PMI_5/publication_index.xml --type resource_docs
+java -Xss5m -jar target/stepmod2mn-1.59.jar publication/part1000/CR_PMI_5/publication_index.xml --type resource_docs
 ----
 
 NOTE: the Metanorma collection file 'metanorma.yml' will be saved in the 3 level up folder (usually, source repository's root folder).
@@ -168,42 +168,42 @@ NOTE: the Metanorma collection file 'metanorma.yml' will be saved in the 3 level
 
 [source,sh]
 ----
-java -jar target/stepmod2mn-1.58.jar <start folder to process xml maps files> --svg
+java -jar target/stepmod2mn-1.59.jar <start folder to process xml maps files> --svg
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -jar target/stepmod2mn-1.58.jar data/resources --svg
+java -jar target/stepmod2mn-1.59.jar data/resources --svg
 ----
 
 === Generate SVG images for the XML in the specified folder and sub-folders, and save them in the output folder:
 
 [source,sh]
 ----
-java -jar target/stepmod2mn-1.58.jar <start folder to process xml maps files> --svg [--output <Output-folder>]
+java -jar target/stepmod2mn-1.59.jar <start folder to process xml maps files> --svg [--output <Output-folder>]
 ----
 
 e.g. 
 
 [source,sh]
 ----
-java -jar target/stepmod2mn-1.58.jar data/resources --svg --output schemas/
+java -jar target/stepmod2mn-1.59.jar data/resources --svg --output schemas/
 ----
 
 === Generate SVG image for Express Imagemap XML and Image:
 
 [source,sh]
 ----
-java -jar stepmod2mn-1.58.jar --xml <Express Imagemap XML file path> --image <Image file name> [--svg <resulted SVG map file or folder>]
+java -jar stepmod2mn-1.59.jar --xml <Express Imagemap XML file path> --image <Image file name> [--svg <resulted SVG map file or folder>]
 ----
 
 e.g.
 
 [source,sh]
 ----
-java -jar stepmod2mn-1.58.jar --xml data\resource_docs\fundamentals_of_product_description_and_support\schema_diagexpg1.xml --image schema_diagexpg1.gif --svg schema_diagexpg1.svg
+java -jar stepmod2mn-1.59.jar --xml data\resource_docs\fundamentals_of_product_description_and_support\schema_diagexpg1.xml --image schema_diagexpg1.gif --svg schema_diagexpg1.svg
 ----
 
 
@@ -223,7 +223,7 @@ Update version in `pom.xml`, e.g.:
 ----
 <groupId>org.metanorma</groupId>
 <artifactId>stepmod2mn</artifactId>
-<version>1.58</version>
+<version>1.59</version>
 <name>STEPmod XML to Metanorma AsciiDoc converter</name>
 ----
 
@@ -234,8 +234,8 @@ Tag the same version in Git:
 
 [source,xml]
 ----
-git tag v1.58
-git push origin v1.58
+git tag v1.59
+git push origin v1.59
 ----
 
 Then the corresponding GitHub release will be automatically created at:
diff --git a/pom.xml b/pom.xml
index c98e87f..5241baf 100644
--- a/pom.xml
+++ b/pom.xml
@@ -5,7 +5,7 @@
 	<modelVersion>4.0.0</modelVersion>
 	<groupId>org.metanorma</groupId>
 	<artifactId>stepmod2mn</artifactId>
-	<version>1.58</version>
+	<version>1.59</version>
 	<name>STEPmod XML to Metanorma AsciiDoc converter</name>
 	<packaging>jar</packaging>
 	<url>https://www.metanorma.org</url>
diff --git a/src/main/resources/stepmod2mn.adoc.xsl b/src/main/resources/stepmod2mn.adoc.xsl
index 79c216e..66d2a6c 100644
--- a/src/main/resources/stepmod2mn.adoc.xsl
+++ b/src/main/resources/stepmod2mn.adoc.xsl
@@ -409,8 +409,20 @@
 	<xsl:template name="insertParagraph">
 		<xsl:param name="text"/>
 		<xsl:param name="keep-with-next" select="'false'"/>
-		<!-- <xsl:value-of select="normalize-space($text)"/> -->
-		<xsl:apply-templates select="xalan:nodeset($text)" mode="linearize"/>
+		<xsl:variable name="paragraph">
+			<xsl:apply-templates select="xalan:nodeset($text)" mode="linearize"/>
+		</xsl:variable>
+		<xsl:choose>
+			<!-- for [SOURCE:...], see https://github.com/metanorma/stepmod2mn/issues/179#issuecomment-2083273335-->
+			<!-- <xsl:when test="starts-with($text,'[SOURCE:')">+++[+++<xsl:value-of select="substring($text,2)"/></xsl:when> -->
+			<xsl:when test="starts-with($paragraph,'[SOURCE:') and java:endsWith(java:java.lang.String.new($paragraph),']')">
+				<xsl:call-template name="parseSource">
+					<xsl:with-param name="text" select="$paragraph"/>
+				</xsl:call-template>
+			</xsl:when>
+			<xsl:when test="starts-with($paragraph,'[SOURCE:')">+++[+++<xsl:value-of select="substring($paragraph,2)"/></xsl:when>
+			<xsl:otherwise><xsl:copy-of select="$paragraph"/></xsl:otherwise>
+		</xsl:choose>
 		<xsl:if test="$keep-with-next = 'false'">
 			<xsl:text>&#xa;&#xa;</xsl:text>
 		</xsl:if>
@@ -1015,14 +1027,58 @@
 				</xsl:otherwise>
 			</xsl:choose>
 		</xsl:variable>
-		<xsl:variable name="text1">
-			<xsl:choose>
-				<!-- for [SOURCE:...], see https://github.com/metanorma/stepmod2mn/issues/179#issuecomment-2083273335-->
-				<xsl:when test="starts-with($text,'[SOURCE:')">+++[+++<xsl:value-of select="substring($text,2)"/></xsl:when>
-				<xsl:otherwise><xsl:value-of select="$text"/></xsl:otherwise>
-			</xsl:choose>
-		</xsl:variable>
-		<xsl:value-of select="java:org.metanorma.RegExEscaping.escapeFormattingCommands($text1)"/>
+		<xsl:value-of select="java:org.metanorma.RegExEscaping.escapeFormattingCommands($text)"/>
+	</xsl:template>
+	
+	<xsl:template name="parseSource">
+		<xsl:param name="text"/>
+		<!-- Example:
+		 [SOURCE: ISO/TS 10303-4000:2023, 3.1.1, Modified – Notes to entry removed]
+	 -->
+	 <xsl:variable name="text1" select="substring-after(substring($text,1,string-length($text) - 1), '[SOURCE:')"/>
+	 <xsl:variable name="standard_number">
+		<xsl:choose>
+			<xsl:when test="contains($text1,',')"><xsl:value-of select="normalize-space(substring-before($text1,','))"/></xsl:when>
+			<xsl:otherwise><xsl:value-of select="normalize-space($text1)"/></xsl:otherwise>
+		</xsl:choose>
+	 </xsl:variable>
+	 <xsl:variable name="standard_number_without_year">
+		<xsl:choose>
+			<xsl:when test="contains($standard_number, ':')"><xsl:value-of select="substring-before($standard_number, ':')"/></xsl:when>
+			<xsl:otherwise><xsl:value-of select="$standard_number"/></xsl:otherwise>
+		</xsl:choose>
+	 </xsl:variable>
+	 <!-- standard_number='<xsl:value-of select="$standard_number"/>'- -->
+	 <xsl:variable name="standard_number_after" select="substring-after($text1,',')"/> <!-- Example: 3.1.1, Modified – Notes to entry removed -->
+	 <xsl:variable name="section">
+		<xsl:choose>
+			<xsl:when test="contains($standard_number_after, ',')"><xsl:value-of select="normalize-space(substring-before($standard_number_after, ','))"/></xsl:when>
+			<xsl:otherwise><xsl:value-of select="$standard_number_after"/></xsl:otherwise>
+		</xsl:choose>
+	 </xsl:variable>
+	 <!-- section='<xsl:value-of select="$section"/>' -->
+	 <xsl:variable name="modified_" select="translate(normalize-space(substring-after($standard_number_after, ',')),'&#x2013;','-')"/>
+	 <xsl:variable name="modified__">
+		<xsl:choose>
+			<xsl:when test="contains($modified_, 'Modified -')"><xsl:value-of select="substring-after($modified_, 'Modified -')"/></xsl:when>
+			<xsl:when test="contains($modified_, 'modified -')"><xsl:value-of select="substring-after($modified_, 'modified -')"/></xsl:when>
+			<xsl:otherwise><xsl:value-of select="$modified_"/></xsl:otherwise>
+		</xsl:choose>
+	 </xsl:variable>
+	 <xsl:variable name="modified" select="normalize-space($modified__)"/>
+	 <!-- modified='<xsl:value-of select="$modified"/>' -->
+	 <xsl:if test="$standard_number != ''">
+		<xsl:text>[.source]</xsl:text>
+		<br/>
+		<xsl:text>&lt;&lt;</xsl:text>
+		<!-- determine bibitem id -->
+		<xsl:variable name="bibliography_xml_document" select="document(concat($path, '../../../data/basic/bibliography.xml'))"/>
+		<xsl:variable name="bibitem_id" select="$bibliography_xml_document/bibitem.list/bibitem[stdnumber = $standard_number or stdnumber = $standard_number_without_year][1]/@id"/>
+		<xsl:value-of select="concat('bibitem_',$bibitem_id)"/>
+		<xsl:if test="$section != ''">, section <xsl:value-of select="$section"/></xsl:if>
+		<xsl:text>&gt;&gt;</xsl:text>
+		<xsl:if test="$modified != ''">, <xsl:value-of select="$modified"/></xsl:if>
+	 </xsl:if>
 	</xsl:template>
 	
 	<xsl:template name="repeat">