From 584609647e51d0c120d6d12473696fe15e638ee4 Mon Sep 17 00:00:00 2001
From: Robert Bossy <Robert.Bossy@inra.fr>
Date: Fri, 19 Feb 2021 02:42:55 +0100
Subject: [PATCH] try to fix everything (defunct method)
---
bacdive2alvisnlp.xslt | 7 ++
dsmz-match.plan | 286 +++++++++++++++++++++++++++++++++++++-----
2 files changed, 265 insertions(+), 28 deletions(-)
diff --git a/bacdive2alvisnlp.xslt b/bacdive2alvisnlp.xslt
index bdb35ea..98926a5 100644
--- a/bacdive2alvisnlp.xslt
+++ b/bacdive2alvisnlp.xslt
@@ -15,7 +15,14 @@
</xsl:template>
<xsl:template match="taxonomy_name/strains/list-item">
+ <a:section name="subspecies_epithet" xpath-contents="subspecies_epithet"/>
<a:section name="species" xpath-contents="species"/>
+ <a:section name="genus" xpath-contents="genus"/>
+ <a:section name="family" xpath-contents="family"/>
+ <a:section name="ordo" xpath-contents="ordo"/>
+ <a:section name="class" xpath-contents="class"/>
+ <a:section name="phylum" xpath-contents="phylum"/>
+ <a:section name="domain" xpath-contents="domain"/>
<a:section name="full_scientific_name" xpath-contents="normalize-space(full_scientific_name)"/>
<a:section name="designation" xpath-contents="designation"/>
</xsl:template>
diff --git a/dsmz-match.plan b/dsmz-match.plan
index 9379fae..69f520f 100644
--- a/dsmz-match.plan
+++ b/dsmz-match.plan
@@ -31,21 +31,153 @@
<createSections/>
</species-and-number>
- <match class="FileMapper">
+ <mark-strains class="Action">
<target>documents.(sections:catalog-number | sections:species-and-number)</target>
+ <action>set:feat:strain("yes")</action>
+ <setFeatures/>
+ </mark-strains>
+
+ <mark-candidates class="Action">
+ <target>documents.(sections[@strain] | sections:species | sections:genus | sections:family | sections:ordo | sections:class | sections:phylum | sections:domain)</target>
+ <action>set:feat:candidate("yes")</action>
+ <setFeatures/>
+ </mark-candidates>
+
+ <match class="FileMapper">
+ <target>documents.sections[@candidate]</target>
<form>contents</form>
<targetFeatures>,taxid,canonical-name,path,pos,rank</targetFeatures>
</match>
+ <dispatch>
+ <equivalent class="Action">
+ <target>documents[(not @dispatch) and sections[@strain and @rank == "no rank"]]</target>
+ <action>
+ set:feat:dispatch("equivalent")
+ | set:feat:taxid(sections[@strain and @rank == "no rank"]{0}.@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </equivalent>
+
+ <separate-subspecies-type-material class="Action">
+ <target>documents[(not @dispatch) and sections[@strain and @rank == "subspecies"]]</target>
+ <action>
+ set:feat:dispatch("new")
+ | set:feat:taxid(sections[@strain and @rank == "subspecies"]{0}.@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </separate-subspecies-type-material>
+
+ <separate-species-type-material class="Action">
+ <target>documents[(not @dispatch) and sections[@strain and @rank == "species"]]</target>
+ <action>
+ set:feat:dispatch("new")
+ | set:feat:taxid(sections[@strain and @rank == "species"]{0}.@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </separate-species-type-material>
+
+ <new-strain-for-subspecies class="Action">
+ <target>documents[(not @dispatch) and sections:species[@rank == "subspecies"]]</target>
+ <action>
+ set:feat:dispatch("new")
+ | set:feat:taxid(sections:species[@rank == "subspecies"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-subspecies>
+
+ <new-strain-for-species class="Action">
+ <target>documents[(not @dispatch) and sections:species[@rank == "species"]]</target>
+ <action>
+ set:feat:dispatch("new")
+ | set:feat:taxid(sections:species[@rank == "species"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-species>
+
+ <new-strain-for-genus class="Action">
+ <target>documents[(not @dispatch) and sections:genus[@rank == "genus"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:genus[@rank == "genus"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-genus>
+
+ <new-strain-for-family class="Action">
+ <target>documents[(not @dispatch) and sections:family[@rank == "family"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:family[@rank == "family"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-family>
+
+ <new-strain-for-ordo class="Action">
+ <target>documents[(not @dispatch) and sections:ordo[@rank == "order"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:ordo[@rank == "order"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-ordo>
+
+ <new-strain-for-class class="Action">
+ <target>documents[(not @dispatch) and sections:class[@rank == "class"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:class[@rank == "class"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-class>
+
+ <new-strain-for-phylum class="Action">
+ <target>documents[(not @dispatch) and sections:phylum[@rank == "phylum"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:phylum[@rank == "phylum"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-phylum>
+
+ <new-strain-for-domain class="Action">
+ <target>documents[(not @dispatch) and sections:domain[@rank == "superkingdom"]]</target>
+ <action>
+ set:feat:dispatch("new-species")
+ | set:feat:taxid(sections:domain[@rank == "superkingdom"].@taxid)
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </new-strain-for-domain>
+
+ <no-match class="Action">
+ <target>documents[not @dispatch]</target>
+ <action>
+ set:feat:dispatch("no-match")
+ | set:feat:taxid("")
+ | set:feat:rule(module:id)
+ </action>
+ <setFeatures/>
+ </no-match>
+ </dispatch>
+
<export>
- <name-match class="TabularExport">
+ <full-report class="TabularExport">
<outDir>.</outDir>
- <corpusFile>all.txt</corpusFile>
- <lines>documents.sections[@taxid]</lines>
+ <corpusFile>full-report.txt</corpusFile>
+ <lines>documents.sections[@candidate]</lines>
<columns>
document.@id,
- document.sections:species.contents,
- document.sections:full_scientific_name.contents,
@name,
contents,
@taxid,
@@ -53,47 +185,145 @@
@rank
</columns>
<headers>
- "ID",
- "SPECIES NAME",
- "FULL NAME",
+ "BACDIVE ID",
"FIELD",
"NAME",
"NCBI TAXID",
"NCBI CANONICAL",
"NCBI RANK"
</headers>
- </name-match>
+ </full-report>
+
+ <dispatch-report class="TabularExport">
+ <outDir>.</outDir>
+ <corpusFile>dispatch-report.txt</corpusFile>
+ <lines>documents</lines>
+ <columns>
+ @id,
+ @dispatch,
+ @taxid,
+ @rule
+ </columns>
+ <headers>
+ "BACDIVE ID",
+ "DISPATCH",
+ "NCBI TAXID",
+ "RULE"
+ </headers>
+ </dispatch-report>
<equivalent-strains class="TabularExport">
<outDir>.</outDir>
<corpusFile>equivalent-strains.txt</corpusFile>
- <lines>documents[sections[@rank == "no rank"]].sections:catalog-number[contents ?= " "]</lines> <!-- exclude ym20-087 kondo51 etc -->
+ <lines>documents</lines>
<columns separator=";">
- str:replace(str:lower(contents), " ", ":");
- document.sections[@rank == "no rank"]{0}.@taxid
+ (if @dispatch == "equivalent" then @taxid ^ "\t" else "") ^
+ str:join:'\t'(sections:catalog-number[contents ?= " "], str:replace(str:lower(contents), " ", ":")) <!-- exclude ym20-087 kondo51 etc -->
</columns>
</equivalent-strains>
<new-strains>
- <identifier class="Action">
- <target>documents[sections[@rank == "species" or @rank == "subspecies"]]</target>
- <action>set:feat:new-taxid(str:replace(str:lower(sections:catalog-number[contents ^= "DSM"].contents), " ", ":"))</action>
+ <canonical>
+ <first-dsm class="Action">
+ <target>documents[@dispatch == "new" or @dispatch == "new-species"].sections:catalog-number[contents ^= "DSM"]{0}</target>
+ <action>set:feat:canonical("yes")</action>
+ <setFeatures/>
+ </first-dsm>
+
+ <first-any class="Action">
+ <target>documents[(@dispatch == "new" or @dispatch == "new-species") and not sections:catalog-number[@canonical]].sections:catalog-number{0}</target>
+ <action>set:feat:canonical("yes")</action>
+ <setFeatures/>
+ </first-any>
+
+ <new-taxid class="Action">
+ <target>documents[@dispatch == "new" or @dispatch == "new-species"].sections:catalog-number[@canonical]</target>
+ <action>document.set:feat:new-taxid(str:replace(str:lower(target.contents), " ", ":"))</action>
+ <setFeatures/>
+ </new-taxid>
+
+ <scientific-name class="Action">
+ <target>documents[@dispatch == "new" or @dispatch == "new-species"].sections:catalog-number[@canonical]</target>
+ <action>document.sections:species-and-number[contents ?= target.contents]{0}.set:feat:scientific-name("yes")</action>
+ <setFeatures/>
+ </scientific-name>
+ </canonical>
+
+ <new-species-id class="Action">
+ <target>documents[@dispatch == "new-species"]</target>
+ <action>set:feat:new-species-id("prov:" ^ str:replace(sections:species[contents != "unclassified"].contents, " ", "-"))</action>
<setFeatures/>
- </identifier>
+ </new-species-id>
+
+ <export-nodes class="TabularExport">
+ <outDir>.</outDir>
+ <corpusFile>dsmz-nodes.dmp</corpusFile>
+ <lines>documents[@dispatch == "new" or @dispatch == "new-species"]</lines>
+ <separator trim="false"> | </separator>
+ <columns>
+ @new-taxid,
+ if @dispatch == "new" then @taxid else @new-species-id,
+ "no rank",
+ "",
+ "0",
+ "1",
+ "11",
+ "1",
+ "0",
+ "1",
+ "1",
+ "0",
+ ""
+ </columns>
+ </export-nodes>
+
+ <export-species-nodes class="TabularExport">
+ <outDir>.</outDir>
+ <corpusFile>dsmz-species-nodes.dmp</corpusFile>
+ <lines>sort:nsval(documents[@dispatch == "new-species"], @new-species-id ^ "___" ^ @taxid)</lines>
+ <separator trim="false"> | </separator>
+ <columns>
+ @new-species-id,
+ @taxid,
+ if sections:subspecies_epithet.contents == "" then "species" else "subspecies",
+ "",
+ "0",
+ "1",
+ "11",
+ "1",
+ "0",
+ "1",
+ "1",
+ "0",
+ ""
+ </columns>
+ </export-species-nodes>
+
+ <export-names class="TabularExport">
+ <outDir>.</outDir>
+ <corpusFile>dsmz-names.dmp</corpusFile>
+ <lines>documents[@dispatch == "new" or @dispatch == "new-species"].sections[@strain]</lines>
+ <separator trim="false"> | </separator>
+ <columns>
+ document.@new-taxid,
+ contents,
+ "",
+ if @scientific-name then "scientific name" else "equivalent catalog"
+ </columns>
+ </export-names>
- <export class="TabularExport">
+ <export-species-names class="TabularExport">
<outDir>.</outDir>
- <corpusFile>taxa+id_dsmz.txt</corpusFile>
- <lines>documents[@new-taxid].(sections:catalog-number | sections:species-and-number)</lines>
- <columns separator=";">
- contents;
- document.@new-taxid;
- document.sections:species-and-number[contents ?= "DSM"].contents;
- document.sections[@taxid]{0}.@path ^ "/" ^ document.@new-taxid;
- document.sections[@taxid]{0}.@pos;
- "no rank"
+ <corpusFile>dsmz-species-names.dmp</corpusFile>
+ <lines>documents[@dispatch == "new-species"].sections:species</lines>
+ <separator trim="false"> | </separator>
+ <columns>
+ document.@new-species-id,
+ contents,
+ "",
+ "scientific name"
</columns>
- </export>
+ </export-species-names>
</new-strains>
</export>
</alvisnlp-plan>
--
GitLab