| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 1 | <xsl:stylesheet version="3.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" |
| 2 | xmlns:saxon="http://saxon.sf.net/" |
| Marc Kupietz | a553865 | 2024-04-21 15:49:30 +0200 | [diff] [blame] | 3 | xmlns:xs="http://www.w3.org/2001/XMLSchema" |
| 4 | xmlns:TextClassifier="java:de.ids_mannheim.TextClassifier" |
| 5 | exclude-result-prefixes="saxon xs TextClassifier"> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 6 | |
| 7 | <xsl:output method="xml" indent="yes" saxon:line-length="1000" |
| 8 | doctype-public="-//IDS//DTD IDS-I5 1.0//EN" |
| 9 | doctype-system="http://corpora.ids-mannheim.de/I5/DTD/i5.dtd" |
| 10 | /> |
| 11 | |
| Marc Kupietz | a553865 | 2024-04-21 15:49:30 +0200 | [diff] [blame] | 12 | <xsl:variable name="domainClassifier" select="TextClassifier:new('models/dereko_domains_s.classifier')"/> |
| 13 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 14 | <xsl:mode streamable="yes" on-no-match="shallow-copy"/> |
| 15 | <xsl:mode streamable="no" on-no-match="shallow-copy" name="no-Streaming"/> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 16 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 17 | <xsl:template match="idsText"> |
| 18 | <xsl:variable name="idsText" as="node()"> |
| 19 | <xsl:copy-of select="."/> |
| 20 | </xsl:variable> |
| 21 | <xsl:apply-templates select="$idsText" mode="no-Streaming"/> |
| 22 | </xsl:template> |
| 23 | |
| 24 | <xsl:template match="textClass" mode="no-Streaming"> |
| Marc Kupietz | a553865 | 2024-04-21 15:49:30 +0200 | [diff] [blame] | 25 | <textClass> |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 26 | <xsl:text>
</xsl:text> |
| 27 | <xsl:value-of disable-output-escaping="yes" |
| 28 | select="TextClassifier:insertCatRefs($domainClassifier, 'topic', ../../../text, 0.0001)"/> |
| 29 | <xsl:apply-templates mode="no-Streaming"/> |
| Marc Kupietz | a553865 | 2024-04-21 15:49:30 +0200 | [diff] [blame] | 30 | </textClass> |
| 31 | </xsl:template> |
| 32 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 33 | <xsl:template match="p[not(normalize-space())]" priority="1.0" mode="no-Streaming"/> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 34 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 35 | <xsl:template match="div[not(normalize-space())]" priority="1.0" mode="no-Streaming"/> |
| Marc Kupietz | 164a283 | 2024-04-14 21:00:48 +0200 | [diff] [blame] | 36 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 37 | <xsl:template match="p[descendant::div|descendant::p and not(ancestor::item)]" priority="0.9" mode="no-Streaming"> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 38 | <div type="section"> |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 39 | <xsl:apply-templates mode="no-Streaming"/> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 40 | </div> |
| 41 | </xsl:template> |
| 42 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 43 | <xsl:template match="(ref|emph|hi|text())[parent::div]" priority="0.9" mode="no-Streaming"> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 44 | <p> |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 45 | <xsl:apply-templates mode="no-Streaming"/> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 46 | </p> |
| 47 | </xsl:template> |
| 48 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 49 | <xsl:template match="head[parent::p]" mode="no-Streaming"> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 50 | <hi rend="bold"> |
| 51 | <xsl:value-of select="."/> |
| 52 | </hi> |
| 53 | </xsl:template> |
| 54 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 55 | <xsl:template match="hi[parent::div]" mode="no-Streaming"> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 56 | <p> |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 57 | <xsl:apply-templates mode="no-Streaming"/> |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 58 | </p> |
| 59 | </xsl:template> |
| 60 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 61 | <xsl:template match="div[ancestor::item]" mode="no-Streaming"> |
| Marc Kupietz | a6d175b | 2025-02-20 16:58:23 +0100 | [diff] [blame] | 62 | <p> |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 63 | <xsl:apply-templates mode="no-Streaming"/> |
| Marc Kupietz | a6d175b | 2025-02-20 16:58:23 +0100 | [diff] [blame] | 64 | </p> |
| 65 | </xsl:template> |
| 66 | |
| Marc Kupietz | 764658b | 2025-03-28 16:20:55 +0100 | [diff] [blame^] | 67 | <xsl:template match="p[normalize-space(.) = ' ']" mode="no-Streaming"/> |
| Marc Kupietz | 8d29363 | 2024-04-15 06:54:26 +0200 | [diff] [blame] | 68 | |
| Marc Kupietz | 28f48e1 | 2024-04-14 16:17:05 +0200 | [diff] [blame] | 69 | </xsl:stylesheet> |