successfully switched to ann_morpho for data.xml
Change-Id: Idb0612349cee0e4382b1e439ecc5fda1ced79578
diff --git a/nkjp2korap.xsl b/nkjp2korap.xsl
index 2f04231..37248ee 100644
--- a/nkjp2korap.xsl
+++ b/nkjp2korap.xsl
@@ -372,7 +372,7 @@
<xsl:variable name="my_dir" as="xs:string" select="replace(base-uri(),'/text\.xml','')"/>
<xsl:variable name="my_textID" as="xs:string" select="tokenize($my_dir,'/')[last()]"/>
<xsl:variable name="ann_morphosyntax.uri" select="$my_dir || '/ann_morphosyntax.xml'" as="xs:string"/>
- <xsl:variable name="ann_segmentation.uri" select="$my_dir || '/ann_segmentation.xml'" as="xs:string"/>
+ <!--<xsl:variable name="ann_segmentation.uri" select="$my_dir || '/ann_segmentation.xml'" as="xs:string"/>-->
<xsl:variable name="ann_named.uri" select="$my_dir || '/ann_named.xml'" as="xs:string"/>
<xsl:variable name="ann_groups.uri" select="$my_dir || '/ann_groups.xml'" as="xs:string"/>
<xsl:variable name="ann_words.uri" select="$my_dir || '/ann_words.xml'" as="xs:string"/>
@@ -386,8 +386,8 @@
<xsl:with-param name="text.doc" as="document-node()" select="."/>
<xsl:with-param name="ann_morphosyntax.doc" as="document-node()"
select="doc($ann_morphosyntax.uri)"/>
- <xsl:with-param name="ann_segmentation.xml" as="document-node()"
- select="doc($ann_segmentation.uri)"/>
+<!-- <xsl:with-param name="ann_segmentation.xml" as="document-node()"
+ select="doc($ann_segmentation.uri)"/>-->
<xsl:with-param name="my_textID" select="$my_textID" as="xs:string"/>
<!-- the following parameters may happen to be null -->
<xsl:with-param name="ann_named.doc" as="document-node()*"
@@ -406,7 +406,7 @@
<xsl:template name="process_single_sample">
<xsl:param name="text.doc" as="document-node()"/>
<xsl:param name="ann_morphosyntax.doc" as="document-node()"/>
- <xsl:param name="ann_segmentation.xml" as="document-node()"/>
+<!-- <xsl:param name="ann_segmentation.xml" as="document-node()"/>-->
<xsl:param name="my_textID" as="xs:string" select="'0-BAD_textID'"/>
<!-- empty textID should never happen, but if it does, it will be signalled at the top of the output -->
<xsl:param name="ann_named.doc" as="document-node()*"/>
@@ -420,29 +420,34 @@
<!-- this is what occurs in the text and data layers as @docid -->
<xsl:call-template name="create_data">
- <xsl:with-param name="ann_segmentation.xml" select="$ann_segmentation.xml" as="document-node()"/>
+ <xsl:with-param name="ann_morphosyntax.doc" select="$ann_morphosyntax.doc" as="document-node()"/>
<xsl:with-param name="compoundID" select="$compoundID" as="xs:string"/>
<xsl:with-param name="target" select="$targetBaseDir || '/data.xml'" as="xs:string"/>
</xsl:call-template>
- <xsl:call-template name="create_struct">
+ <!-- <xsl:call-template name="create_struct">
<xsl:with-param name="compoundID" select="$compoundID" as="xs:string"/>
<xsl:with-param name="ann_segmentation.xml" select="$ann_segmentation.xml"
as="document-node()"/>
<xsl:with-param name="target" select="$targetBaseDir || '/struct/structure.xml'" as="xs:string"
/>
- </xsl:call-template>
+ </xsl:call-template>-->
- <xsl:call-template name="create_morpho">
+<!-- <xsl:call-template name="create_morpho">
<xsl:with-param name="compoundID" select="$compoundID" as="xs:string"/>
<xsl:with-param name="ann_segmentation.xml" select="$ann_segmentation.xml"
as="document-node()"/>
<xsl:with-param name="ann_morphosyntax.doc" select="$ann_morphosyntax.doc"
as="document-node()"/>
<xsl:with-param name="target" select="$targetBaseDir || '/nkjp/morpho.xml'" as="xs:string"/>
- </xsl:call-template>
+ </xsl:call-template>-->
- <xsl:call-template name="create_text_header">
+<!--
+
+ THESE ARE SAFE
+
+
+ <xsl:call-template name="create_text_header">
<xsl:with-param name="text.doc" select="$text.doc" as="document-node()"/>
<xsl:with-param name="compoundID" select="$compoundID" as="xs:string"/>
<xsl:with-param name="target" select="$targetBaseDir || '/header.xml'" as="xs:string"/>
@@ -453,8 +458,8 @@
<xsl:for-each
select="$ann_named.doc//tei:seg/tei:ptr">
<xsl:variable name="trg" as="xs:string" select="fn:substring-before(@target, '#')"/>
- <!-- caution: as of 01-June-2022, some of the pointers are malformed (missing '#' when referencing locally).
- so we need to act around it but also sustainably - in case that error gets corrected -->
+ <!-\- caution: as of 01-June-2022, some of the pointers are malformed (missing '#' when referencing locally).
+ so we need to act around it but also sustainably - in case that error gets corrected -\->
<xsl:if test="fn:string-length($trg) and $trg eq 'ann_morphosyntax.xml'">
<xsl:sequence>
@@ -476,12 +481,10 @@
<xsl:with-param name="rev_lookup" select="$rev_lookup" as="map(*)"/>
</xsl:call-template>
</xsl:if>
-
+-->
<xsl:if test="$ann_words.doc and $ann_groups.doc">
<xsl:call-template name="create_groups">
<xsl:with-param name="compoundID" select="$compoundID" as="xs:string"/>
- <xsl:with-param name="ann_segmentation.xml" select="$ann_segmentation.xml"
- as="document-node()"/>
<xsl:with-param name="ann_morphosyntax.doc" select="$ann_morphosyntax.doc"
as="document-node()"/>
<xsl:with-param name="ann_words.doc" select="$ann_words.doc"
@@ -497,7 +500,7 @@
<!-- ************************** data.xml ******************* -->
<xsl:template name="create_data">
- <xsl:param name="ann_segmentation.xml" as="document-node()"/>
+ <xsl:param name="ann_morphosyntax.doc" as="document-node()"/>
<xsl:param name="compoundID" as="xs:string"/>
<xsl:param name="target" as="xs:string"/>
<!-- create the data.xml file -->
@@ -513,12 +516,13 @@
<xsl:element name="text" namespace="{$KorAP_namespace}">
<xsl:variable name="content" as="xs:string+">
- <xsl:for-each select="$ann_segmentation.xml/tei:teiCorpus/tei:TEI/tei:text/tei:body/tei:p/tei:s//tei:seg[count(@nkjp:rejected) eq 0]">
+ <!--<xsl:for-each select="$ann_morphosyntax.doc/tei:teiCorpus/tei:TEI/tei:text/tei:body/tei:p/tei:s//tei:seg[count(@nkjp:rejected) eq 0]">-->
+ <xsl:for-each select="$ann_morphosyntax.doc/tei:teiCorpus/tei:TEI/tei:text/tei:body/tei:p/tei:s/tei:seg[tei:fs/tei:f[@name eq 'disamb']]">
<xsl:sequence select="
if (f:is_preceded_by_ws(.,false())) then
' '
else
- '', ./tei:w"/>
+ '', ./tei:fs/tei:f[@name eq 'orth']/tei:string"/>
</xsl:for-each>
</xsl:variable>
<xsl:value-of select="string-join($content)"/>
@@ -756,7 +760,6 @@
<xsl:template name="create_groups">
<xsl:param name="compoundID" as="xs:string"/>
- <xsl:param name="ann_segmentation.xml" as="document-node()"/>
<xsl:param name="ann_morphosyntax.doc" as="document-node()"/>
<xsl:param name="ann_words.doc" as="document-node()"/>
<xsl:param name="ann_groups.doc" as="document-node()"/>