diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 5f96ee7e3377d01c9eb188520bed08d099f23469..1f9a3072d826769442f390d1cc718db9d640c604 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -276,6 +276,7 @@ get-collation-results:
- 'curl --output jobs.json "https://gitlab.gwdg.de/api/v4/projects/$COLLATEX_REPO_ID/pipelines/$PIPELINE_ID/jobs?private_token=$GITLAB_TOKEN"'
- RELEVANT_JOB_ID=$(jq ".[]|select(.stage==\"summarize\")" < jobs.json | jq .id)
- 'curl --output collation-results.zip "https://gitlab.gwdg.de/api/v4/projects/$COLLATEX_REPO_ID/jobs/$RELEVANT_JOB_ID/artifacts?private_token=$GITLAB_TOKEN"'
+ - rm -rf backend/exist-app/data/collation-results
- unzip -u -d backend/exist-app/data/collation-results collation-results.zip
- cd backend
- git add exist-app/data/collation-results/* && git commit -m "update collation results" && git push
diff --git a/CHANGELOG.md b/CHANGELOG.md
index e96cdfd859a2322c12bbae395d5a322732e004ff..840935996c76753768c652236c3aa7001fbe85bf 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,12 @@ All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+## [5.11.0] - 2021-04-27
+
+### Added
+
+- the variants have been added to the AnnotationAPI.
+
## [5.10.0] - 2021-04-22
### Changed
diff --git a/exist-app/build.properties b/exist-app/build.properties
index 64537ba59d850e01f546d92943a647038d0ed827..62f71104d5e62adf7f1eb95879b7d2fec943d32c 100644
--- a/exist-app/build.properties
+++ b/exist-app/build.properties
@@ -1,5 +1,5 @@
project.name=https://ahikar-test.sub.uni-goettingen.de/
-project.version=5.10.0
+project.version=5.11.0
project.title=Ahiqar
project.abbrev=ahikar-test
project.processorversion=5.2.0
diff --git a/exist-app/modules/AnnotationAPI/annotations-variants.xqm b/exist-app/modules/AnnotationAPI/annotations-variants.xqm
new file mode 100644
index 0000000000000000000000000000000000000000..edafce0410f0c943b026dc64ad26ae9c0e666b18
--- /dev/null
+++ b/exist-app/modules/AnnotationAPI/annotations-variants.xqm
@@ -0,0 +1,296 @@
+xquery version "3.1";
+
+(:~
+ : This module is responsible for retrieving the variants of a given text on a
+ : given page. For this, it leverages the tokenization of the text that takes
+ : place during commons:get-page-fragment.
+ :
+ : The output of the main function, vars:get-variants, is a sequence of maps of
+ : the following format:
+ :
+ : map {
+ : "body"
+ : map {
+ : "x-content-type" : "Variant",
+ : "value" : [ map {
+ : "entry" : "${text-of-respective-witness}",
+ : "witness" : "${ms-id}"
+ : }],
+ : "format" : "text/plain",
+ : "type" : "TextualBody"
+ : },
+ : "target" : map {
+ : "format" : "text/xml",
+ : "language" : "syc",
+ : "id" : "http://ahikar.sub.uni-goettingen.de/ns/annotations/${teixml-uri}/${token-id}"
+ : },
+ : "type" : "Annotation",
+ : "id" : "http://ahikar.sub.uni-goettingen.de/ns/annotations/${teixml-uri}/annotation-variants-${token-id}"
+ : }
+ :
+ :)
+
+module namespace vars="http://ahikar.sub.uni-goettingen.de/ns/annotations/variants";
+
+declare namespace tei="http://www.tei-c.org/ns/1.0";
+
+import module namespace commons="http://ahikar.sub.uni-goettingen.de/ns/commons" at "../commons.xqm";
+import module namespace functx = "http://www.functx.com";
+
+declare variable $vars:ns := "http://ahikar.sub.uni-goettingen.de/ns/annotations";
+
+declare function vars:get-variants($teixml-uri as xs:string,
+ $page as xs:string)
+as map(*)* {
+ let $variants := vars:get-variants-on-page-as-maps($teixml-uri, $page)
+
+ for $map in $variants
+ let $id := map:get($map, "current") => map:get("id")
+ return
+ map {
+ "id": $vars:ns || "/" || $teixml-uri || "/annotation-variants-" || $id,
+ "type": "Annotation",
+ "body": vars:get-body-object($map),
+ "target": vars:get-target-information($map, $teixml-uri, $id)
+ }
+};
+
+declare function vars:get-body-object($map as map(*))
+as map() {
+ map {
+ "type": "TextualBody",
+ "value": vars:make-annotation-value($map),
+ "format": "text/plain",
+ "x-content-type": "Variant"
+ }
+};
+
+declare function vars:make-annotation-value($map as map(*))
+as map(*)+ {
+ for $variant in map:get($map, "variants") return
+ let $entry := map:get($variant, "entry")
+ let $witness := map:get($variant, "witness")
+ return
+ map {
+ "entry": if ($entry instance of xs:string) then $entry else "omisit",
+ "witness": map:get($commons:idno-to-sigils-map, $witness)
+ }
+};
+
+declare function vars:get-target-information($map as map(*),
+ $teixml-uri as xs:string,
+ $id as xs:string)
+as map(*) {
+ map {
+ "id": $vars:ns || "/" || $teixml-uri || "/"|| $id,
+ "format": "text/xml",
+ "language": vars:get-target-language($teixml-uri)
+ }
+};
+
+declare function vars:get-target-language($teixml-uri as xs:string)
+as xs:string {
+ let $doc := commons:open-tei-xml($teixml-uri)
+ let $language := $doc//tei:text[@xml:lang][matches(descendant::text(), "[\w]")]/@xml:lang
+ return
+ $language
+};
+
+(:~
+ : Returns a map holding the current token as well as all its resp. variants per
+ : page. The exact value of the current token depends on the manuscript we are
+ : currently looking at.
+ :
+ : @param $teixml-uri The current TEI document's base URI, e.g. "12345"
+ : @param $page The page as given in tei:pb/@n
+ : @return A map with information about the current token and its variants
+ :)
+declare function vars:get-variants-on-page-as-maps($teixml-uri as xs:string,
+ $page as xs:string)
+as map()* {
+ let $ms-id := vars:get-ms-id-from-idno($teixml-uri)
+ let $relevant-files-for-ms-id := vars:get-relevant-files($ms-id)
+ (: the MS identifier position is the same in all relevant files, so we simply
+ choose the first for looking it up :)
+ let $ms-id-position := vars:determine-id-position($ms-id, $relevant-files-for-ms-id[1])
+ let $tokens := vars:get-token-ids-on-page($teixml-uri, $page)
+ let $files-relevant-for-page := vars:get-files-relevant-for-page($relevant-files-for-ms-id, $ms-id-position, $tokens)
+
+ for $file in $files-relevant-for-page return
+ let $table := map:get($file, "table")
+ let $no-of-sequence := array:size($table)
+ let $indices-relevant-for-page := vars:get-indices-relevant-for-page($table, $no-of-sequence, $ms-id-position, $tokens)
+ let $non-ms-id-positions := vars:get-non-ms-id-positions-in-array($file, $ms-id-position)
+
+ for $iii in $indices-relevant-for-page return
+ vars:make-map-for-token($file, $table, $iii, $ms-id-position, $non-ms-id-positions)
+};
+
+
+(:~
+ : Each token (= relevant word) is encoded in a tei:w which has an @xml:id
+ : attribute.
+ :
+ : @param $teixml-uri The base URI of the document, e.g. "12345"
+ : @param $page The current page as provided in tei:pb/@n
+ :)
+declare function vars:get-token-ids-on-page($teixml-uri as xs:string,
+ $page as xs:string)
+as xs:string+ {
+ let $page-chunks := commons:get-transcription-and-transliteration-per-page($teixml-uri, $page)
+ return
+ $page-chunks//tei:w/@xml:id
+};
+
+declare function vars:get-ms-id-from-idno($teixml-uri as xs:string)
+as xs:string {
+ let $TEI := commons:open-tei-xml($teixml-uri)//tei:TEI
+ return
+ commons:make-id-from-idno($TEI)
+};
+
+(:~
+ : In the CollateX results, we first have an aray of witnesses before the result
+ : of the collation is given in the JSON file. The position of a MS identifier
+ : in this array is reflected in the collation result; The witness on position 2
+ : will always have its tokens in the second position of the result.
+ :
+ : @param $ms-id The MS identifier as given in the array of witnesses
+ : @param $json The collation result as JSON
+ : @return The position of the current MS identifier in the witnesses array
+ :)
+declare function vars:determine-id-position($ms-id as xs:string,
+ $json as map(*))
+as xs:integer {
+ let $witnesses := map:get($json, "witnesses")
+ return
+ index-of($witnesses?*, $ms-id)
+};
+
+declare function vars:get-relevant-files($ms-id as xs:string)
+as item()+ {
+ let $collation-collection := collection("/db/apps/ahikar/data/collation-results")
+ let $relevant-base-uris :=
+ for $doc in $collation-collection return
+ if(matches(base-uri($doc), replace($ms-id, "/", ""))
+ and matches(base-uri($doc), "json")) then
+ base-uri($doc)
+ else
+ ()
+ for $uri in $relevant-base-uris return
+ util:binary-doc($uri)
+ => util:base64-decode()
+ => parse-json()
+};
+
+
+declare function vars:get-files-relevant-for-page($relevant-files-for-ms-id as map()+,
+ $ms-id-position as xs:integer,
+ $tokens as xs:string+)
+as map()+ {
+ let $first-token := $tokens[1]
+ let $last-token := $tokens[last()]
+ for $file in $relevant-files-for-ms-id return
+ let $table := map:get($file, "table")
+ let $no-of-sequences := array:size($table)
+ for $iii in 1 to $no-of-sequences return
+ let $sequence-entry := $table?($iii)
+ let $ms-id-entry := $sequence-entry?($ms-id-position)
+ let $this-manuscripts-ids :=
+ if (array:size($ms-id-entry) gt 0) then
+ $ms-id-entry?(1)
+ => map:get("id")
+ else
+ ()
+ return
+ if ($this-manuscripts-ids = ($first-token, $last-token)) then
+ $file
+ else
+ ()
+};
+
+declare function vars:get-witness($file as map(),
+ $witness-position as xs:integer)
+as xs:string {
+ map:get($file, "witnesses")
+ => array:get($witness-position)
+};
+
+declare function vars:get-non-ms-id-positions-in-array($file as map(),
+ $witness-position as xs:integer)
+as xs:integer+ {
+ let $no-of-witnesses := map:get($file, "witnesses") => array:size()
+ let $no-of-witnesses-as-sequence :=
+ for $iii in 1 to $no-of-witnesses return
+ $iii
+ return
+ functx:value-except($no-of-witnesses-as-sequence, $witness-position)
+};
+
+declare function vars:get-witness-entry($table as array(*),
+ $entry-no as xs:integer,
+ $witness-position as xs:integer)
+as array(*) {
+ let $sequence-entry := $table?($entry-no)
+ return
+ $sequence-entry?($witness-position)
+};
+
+(:~
+ : Returns a sequence of index positions that are relevant for the current witness.
+ : If a witness doesn't have an entry in the collation result or if the entry is
+ : not on the current page, no index position is returned.
+ :
+ : @param $table The CollateX results
+ : @param $no-of-sequences The number of result sequences in $table
+ : @param $ms-id-position An integer determining which position within the result entry is relevant for the current witness
+ : @param $tokens A list of tokens relevant for the current page
+ :)
+declare function vars:get-indices-relevant-for-page($table as array(*),
+ $no-of-sequences as xs:integer,
+ $ms-id-position as xs:integer,
+ $tokens as xs:string+)
+as xs:integer+ {
+ for $iii in 1 to $no-of-sequences return
+ let $sequence-entry := $table?($iii)
+ let $ms-id-entry := $sequence-entry?($ms-id-position)
+ return
+ if (array:size($ms-id-entry) gt 0
+ and array:get($ms-id-entry, 1) => map:get("id") = $tokens) then
+ $iii
+ else
+ ()
+};
+
+(:~
+ : Creates a map in for every token on a page in which all variants are listed.
+ :
+ : @param $table The CollateX results
+ : @param $no-of-sequences The number of result sequences in $table
+ : @param $ms-id-position An integer determining which position within the result entry is relevant for the current witness
+ : @param $tokens A list of tokens relevant for the current page
+ :)
+declare function vars:make-map-for-token($file as map(),
+ $table as array(*),
+ $entry-pos as xs:integer,
+ $ms-id-position as xs:integer,
+ $non-ms-id-positions as xs:integer*)
+as map(*) {
+ let $sequence-entry := $table?($entry-pos)
+ let $ms-id-entry := $sequence-entry?($ms-id-position)
+ let $tokens-in-sequence := array:size($ms-id-entry)
+ for $token-no in $tokens-in-sequence return
+ map {
+ "current": $sequence-entry?($ms-id-position) => array:get($token-no),
+ "variants":
+ for $jjj in $non-ms-id-positions return
+ map {
+ "witness": vars:get-witness($file, $jjj),
+ "entry":
+ if ($sequence-entry?($jjj) => array:size() ge $token-no) then
+ $sequence-entry?($jjj) => array:get($token-no) => map:get("t")
+ else
+ []
+ }
+ }
+};
diff --git a/exist-app/modules/AnnotationAPI/annotations.xqm b/exist-app/modules/AnnotationAPI/annotations.xqm
index 26e5877f6c63292743daf7a29f48ada2ad92b32d..45912a95bec7eb99e0eca42c3d71cede6e7cd7d5 100644
--- a/exist-app/modules/AnnotationAPI/annotations.xqm
+++ b/exist-app/modules/AnnotationAPI/annotations.xqm
@@ -18,6 +18,7 @@ declare namespace tgmd="http://textgrid.info/namespaces/metadata/core/2010";
import module namespace commons="http://ahikar.sub.uni-goettingen.de/ns/commons" at "../commons.xqm";
import module namespace functx = "http://www.functx.com";
+import module namespace vars="http://ahikar.sub.uni-goettingen.de/ns/annotations/variants" at "annotations-variants.xqm";
declare variable $anno:ns := "http://ahikar.sub.uni-goettingen.de/ns/annotations";
@@ -263,8 +264,8 @@ as map() {
anno:find-in-map($anno:uris, $document) => anno:get-all-xml-uris-for-submap()
let $annotations :=
for $xml in $xmls return
- for $page in anno:get-pages-in-TEI($xml)return
- anno:get-annotations($xml, $page)
+ for $page in anno:get-pages-in-TEI($xml) return
+ anno:make-complete-annotations($xml, $page)
return
map {
@@ -374,7 +375,7 @@ as map() {
"next": $nextPageURL,
"prev": $prevPageURL,
"startIndex": anno:determine-start-index-for-page($document, $page),
- "items": anno:get-annotations($xml, $page)
+ "items": anno:make-complete-annotations($xml, $page)
}
}
};
@@ -392,15 +393,7 @@ as map() {
declare function anno:get-annotations($teixml-uri as xs:string,
$page as xs:string)
as map()* {
- let $xml-doc := commons:open-tei-xml($teixml-uri)
- let $langs := $xml-doc//tei:text[@xml:lang[. = ("syc", "ara", "karshuni")]]/@xml:lang/string()
- let $pageChunks :=
- if ($langs = "karshuni") then
- (anno:get-page-fragment($teixml-uri, $page, "transcription"),
- anno:get-page-fragment($teixml-uri, $page, "transliteration"))
- else
- anno:get-page-fragment($teixml-uri, $page, "transcription")
-
+ let $pageChunks := commons:get-transcription-and-transliteration-per-page($teixml-uri, $page)
let $annotation-elements :=
for $chunk in $pageChunks return
@@ -418,23 +411,6 @@ as map()* {
}
};
-
-(:~
- : Returns a single page from a TEI resource, i.e. all content from the given $page
- : up to the next page break.
- :
- : @param $documentURI The resource's URI. Attention: This refers to the TEI file itself!
- : @param $page The page to be returned as tei:pb/@n/string()
- :)
-declare function anno:get-page-fragment($documentURI as xs:string,
- $page as xs:string,
- $text-type as xs:string)
-as element(tei:TEI)? {
- let $nodeURI := commons:get-document($documentURI, "data")/base-uri()
- return
- commons:get-page-fragment($nodeURI, $page, $text-type)
-};
-
(:~
: Returns the Body Object for an annotation.
:
@@ -918,3 +894,22 @@ as xs:integer {
return
sum($noOfAnnotationsPerElement)
};
+
+(:~
+ : Gets a complete list of all relevant annotations per page.
+ :
+ : @param $teixml-uri The current document's URI, e.g. "12345"
+ : @param $page The current page
+ : @return A sequence of maps containing the annotations
+ :)
+declare function anno:make-complete-annotations($teixml-uri as xs:string,
+ $page as xs:string)
+as map(*)* {
+ anno:get-annotations($teixml-uri, $page),
+ (: the sample data has to be excluded here since it doesn't have any variants
+ and no CollateX output. :)
+ if (starts-with(map:get($anno:lang-aggs, "syriac"), "sample")) then
+ ()
+ else
+ vars:get-variants($teixml-uri, $page)
+};
\ No newline at end of file
diff --git a/exist-app/modules/commons.xqm b/exist-app/modules/commons.xqm
index 6f9b3aeb9f40c59eb5d814b6c58bc2f228287ce6..11acf5ca8a55d45b30a6209558e5c3b9bcdf76e1 100644
--- a/exist-app/modules/commons.xqm
+++ b/exist-app/modules/commons.xqm
@@ -22,6 +22,49 @@ declare variable $commons:appHome := "/db/apps/ahikar";
declare variable $commons:ns := "http://ahikar.sub.uni-goettingen.de/ns/commons";
+declare variable $commons:idno-to-sigils-map :=
+ map {
+ "Borg_ar_201": "Borg. ar. 201",
+ "Add_2020": "C",
+ "Sachau_162": "S",
+ "syr_611": "K",
+ "syr_612": "I",
+ "syr_434": "B",
+ "Add_7200": "L",
+ "Brit_Mus_Add_7209": "Brit. Add. 7209",
+ "Brit_Libr_Or_9321": "Brit. Or. 9321",
+ "Cambrigde_Add_3497": "Cam. Add. 3497",
+ "Camb_Add_2886": "Cam. Add. 2886",
+ "Cod_Arab_236": "Cod. Arab. 236",
+ "Paris_Arabe_3637": "Paris. ar. 3637",
+ "Sachau_290_Sachau_339": "Sach. 339",
+ "DFM_00614": "DFM 614",
+ "GCAA_00486": "GCAA 486",
+ "Mingana_syr_133": "Ming. syr. 133",
+ "Mingana_ar_christ_93_84": "Ming. ar. 93",
+ "Mingana_Syr_258": "Ming. syr. 258",
+ "433": "M",
+ "Ms_orient_A_2652": "Gotha 2652",
+ "430": "D",
+ "Or_1292b": "Leiden Or. 1292",
+ "Paris_Arabe_3656": "Paris. ar. 3656",
+ "syr_422": "N",
+ "Sado_no_9": "P",
+ "Salhani": "Salhani",
+ "Sbath_25": "Sbath 25",
+ "Ar_7/229": "A",
+ "Manuscrit_4122": "T",
+ "Or_2313": "O",
+ "Syr_80": "H",
+ "162": "J",
+ "Sachau_336": "U",
+ "Vat_ar_74_Scandar_40": "Vat. ar. 74",
+ "Vat_ar_2054": "Vat. ar. 2054",
+ "Vat_sir_159": "Vat. syr. 159",
+ "Vat_sir_199": "Vat. syr. 199",
+ "Vat_sir_424": "Vat. syr. 424"
+ };
+
declare variable $commons:responseHeader200 :=
<rest:response>
<http:response xmlns:http="http://expath.org/ns/http-client" status="200">
@@ -75,6 +118,35 @@ as xs:string* {
()
};
+declare function commons:get-transcription-and-transliteration-per-page($teixml-uri as xs:string,
+ $page as xs:string)
+as element(tei:TEI)+ {
+ let $xml-doc := commons:open-tei-xml($teixml-uri)
+ let $langs := $xml-doc//tei:text[@xml:lang[. = ("syc", "ara", "karshuni")]]/@xml:lang/string()
+ return
+ if ($langs = "karshuni") then
+ (commons:get-page-fragment-from-uri($teixml-uri, $page, "transcription"),
+ commons:get-page-fragment-from-uri($teixml-uri, $page, "transliteration"))
+ else
+ commons:get-page-fragment-from-uri($teixml-uri, $page, "transcription")
+};
+
+(:~
+ : Returns a single page from a TEI resource, i.e. all content from the given $page
+ : up to the next page break.
+ :
+ : @param $documentURI The resource's URI
+ : @param $page The page to be returned as tei:pb/@n/string()
+ :)
+declare function commons:get-page-fragment-from-uri($documentURI as xs:string,
+ $page as xs:string,
+ $text-type as xs:string)
+as element(tei:TEI)? {
+ let $nodeURI := commons:get-document($documentURI, "data")/base-uri()
+ return
+ commons:get-page-fragment($nodeURI, $page, $text-type)
+};
+
(:~
: Returns a given page from a requested TEI document and from the requested text type.
: In some cases the requested text type isn't available or doesn't have any text, so that
@@ -91,8 +163,9 @@ declare function commons:get-page-fragment($tei-xml-base-uri as xs:string,
as element() {
if (local:has-text-content($tei-xml-base-uri, $page, $text-type)) then
let $node := doc($tei-xml-base-uri)/tei:TEI
+ => tokenize:main()
=> commons:add-IDs()
- => tokenize:main(),
+ ,
$start-node := $node//tei:text[@type = $text-type]//tei:pb[@n = $page],
$end-node := commons:get-end-node($start-node),
$wrap-in-first-common-ancestor-only := false(),
@@ -235,7 +308,7 @@ declare %private function local:create-textgrid-session-id() {
declare function commons:compress-to-zip($collection-uri as xs:string)
as xs:string* {
- if (commons:does-zip-need-update()) then
+(: if (commons:does-zip-need-update()) then:)
let $valid-uris :=
for $doc in collection($collection-uri) return
if (contains(base-uri($doc), "sample")) then
@@ -245,11 +318,11 @@ as xs:string* {
let $zip := compression:zip($valid-uris, false())
return
(
- commons:make-last-zip-created(),
+(: commons:make-last-zip-created(),:)
xmldb:store-as-binary("/db/data", "ahikar-json.zip", $zip)
)
- else
- ()
+(: else:)
+(: ():)
};
declare function commons:does-zip-need-update()
diff --git a/exist-app/modules/tei2json.xqm b/exist-app/modules/tei2json.xqm
index a656d7445f433d0b6deddeefe37d4284a904aaf4..3757d0443d72232d5eb02af3cc826736871eae75 100644
--- a/exist-app/modules/tei2json.xqm
+++ b/exist-app/modules/tei2json.xqm
@@ -106,18 +106,23 @@ declare variable $tei2json:lines-of-transmission :=
declare function tei2json:main()
as xs:string+ {
- tei2json:create-json-collection-if-not-available(),
+ tei2json:remove-old-jsons(),
+ tei2json:create-json-collection(),
tei2json:tokenize-teis()
=> tei2json:make-jsons-per-section-and-transmission-line()
};
+declare function tei2json:remove-old-jsons()
+as item()? {
+ if (xmldb:collection-available($commons:tg-collection || "/json")) then
+ xmldb:remove($commons:tg-collection || "/json")
+ else
+ ()
+};
-declare function tei2json:create-json-collection-if-not-available()
+declare function tei2json:create-json-collection()
as xs:string {
- if (xmldb:collection-available($commons:json)) then
- $commons:json
- else
- xmldb:create-collection($commons:tg-collection, "json")
+ xmldb:create-collection($commons:tg-collection, "json")
};
diff --git a/exist-app/modules/testtrigger.xqm b/exist-app/modules/testtrigger.xqm
index 1de99e1d3f12de1c3bb1d2d81127d32210ad8630..876318aed23c589f32faf31b92714b60b20183bc 100644
--- a/exist-app/modules/testtrigger.xqm
+++ b/exist-app/modules/testtrigger.xqm
@@ -15,6 +15,7 @@ import module namespace rest="http://exquery.org/ns/restxq";
import module namespace test="http://exist-db.org/xquery/xqsuite" at "resource:org/exist/xquery/lib/xqsuite/xqsuite.xql";
import module namespace at="http://ahikar.sub.uni-goettingen.de/ns/annotations/tests" at "../tests/annotation-tests.xqm";
+import module namespace av="http://ahikar.sub.uni-goettingen.de/ns/annotations/variants/tests" at "../tests/annotation-variants-tests.xqm";
import module namespace art="http://ahikar.sub.uni-goettingen.de/ns/annotations/rest/tests" at "../tests/annotation-rest-tests.xqm";
import module namespace ct="http://ahikar.sub.uni-goettingen.de/ns/commons-tests" at "../tests/commons-tests.xqm";
import module namespace tct="http://ahikar.sub.uni-goettingen.de/ns/tapi/collection/tests" at "../tests/tapi-collection-tests.xqm";
@@ -79,6 +80,7 @@ as element()+ {
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/tei2html-tests")),
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/tei2html-textprocessing-tests")),
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/annotations/tests")),
+ test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/annotations/variants/tests")),
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/tapi/txt/normalization/tests")),
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/annotations/rest/tests")),
test:suite(util:list-functions("http://ahikar.sub.uni-goettingen.de/ns/tapi/images/tests")),
@@ -126,6 +128,7 @@ as xs:string? {
case "http://ahikar.sub.uni-goettingen.de/ns/tei2html-textprocessing-tests" return "TEI2HTML text processing"
case "http://ahikar.sub.uni-goettingen.de/ns/tapi/txt/normalization/tests" return "TXT normalization"
case "http://ahikar.sub.uni-goettingen.de/ns/annotations/tests" return "AnnotationAPI"
+ case "http://ahikar.sub.uni-goettingen.de/ns/annotations/variants/tests" return "Annotation variants"
case "http://ahikar.sub.uni-goettingen.de/ns/annotations/rest/tests" return "AnnotationAPI REST"
case "http://ahikar.sub.uni-goettingen.de/ns/tapi/images/tests" return "Image Sections"
case "http://ahikar.sub.uni-goettingen.de/ns/tokenize/tests" return "Tokenize"
diff --git a/exist-app/tests/annotation-tests.xqm b/exist-app/tests/annotation-tests.xqm
index 9c19518907d3ba994098557bb4b29fa5a726e99e..acd87dbcd3946fcc0b11cc98c5823ec689060500 100644
--- a/exist-app/tests/annotation-tests.xqm
+++ b/exist-app/tests/annotation-tests.xqm
@@ -137,15 +137,6 @@ as map() {
anno:make-annotationCollection-for-manifest($collection, $document, $page, $server)
};
-declare
- %test:args("sample_teixml", "82a", "transcription") %test:assertXPath("$result//* = 'حقًا'")
- %test:args("sample_teixml", "82a", "transliteration") %test:assertXPath("$result//* = 'الحاسوب'")
-function at:get-page-fragment($documentURI as xs:string,
- $page as xs:string,
- $text-type as xs:string)
-as element(tei:TEI) {
- anno:get-page-fragment($documentURI, $page, $text-type)
-};
declare
%test:args("sample_main_edition") %test:assertEquals("476")
@@ -292,7 +283,7 @@ as map() {
declare
%test:args("sample_teixml", "84a")
- %test:assertXPath("$result = 'http://ahikar.sub.uni-goettingen.de/ns/annotations/sample_teixml/annotation-N4.4.2.4.4.354.2'")
+ %test:assertXPath("$result = 'http://ahikar.sub.uni-goettingen.de/ns/annotations/sample_teixml/annotation-N1.2.1.2.3.100.3'")
function at:get-annotations-detailed-id($teixml-uri as xs:string,
$page as xs:string)
as xs:string {
diff --git a/exist-app/tests/annotation-variants-tests.xqm b/exist-app/tests/annotation-variants-tests.xqm
new file mode 100644
index 0000000000000000000000000000000000000000..500dc930ab5c6fdead99522f9ef90b0ff9231280
--- /dev/null
+++ b/exist-app/tests/annotation-variants-tests.xqm
@@ -0,0 +1,165 @@
+xquery version "3.1";
+
+module namespace t="http://ahikar.sub.uni-goettingen.de/ns/annotations/variants/tests";
+
+import module namespace test="http://exist-db.org/xquery/xqsuite" at "resource:org/exist/xquery/lib/xqsuite/xqsuite.xql";
+import module namespace vars="http://ahikar.sub.uni-goettingen.de/ns/annotations/variants" at "../modules/AnnotationAPI/annotations-variants.xqm";
+
+declare variable $t:sample-map :=
+ map {
+ "variants": (map {
+ "entry": "ܐܘ",
+ "witness": "430"
+ },map {
+ "entry": "ܕܐܚܛܛ",
+ "witness": "syr_422"
+ },map {
+ "entry": [],
+ "witness": "syr_611"
+ },map {
+ "entry": [],
+ "witness": "syr_612"
+ }),
+ "current": map {
+ "t": "ܐܡܪ",
+ "id": "syr_434_N4.4.2.6.4.8.1.3_1"
+ }
+ };
+
+declare variable $t:sample-file :=
+ util:binary-doc("/db/apps/ahikar/data/collation-results/ara-karshuni_Sbath_25_Vat_sir_424_Vat_sir_199_parables_result.json")
+ => util:base64-decode()
+ => parse-json();
+
+declare
+ %test:args("sample_teixml", "82a") %test:assertXPath("count($result) = 349")
+function t:get-token-ids-on-page($teixml-uri as xs:string,
+ $page as xs:string)
+as xs:string+ {
+ vars:get-token-ids-on-page($teixml-uri, $page)
+};
+
+
+declare
+ %test:args("sample_teixml") %test:assertEquals("Add_2020")
+function t:get-ms-id-from-idno($teixml-uri as xs:string)
+as xs:string {
+ vars:get-ms-id-from-idno($teixml-uri)
+};
+
+declare
+ %test:args("Sachau_336") %test:assertXPath("count($result) = 5")
+function t:get-relevant-files($ms-id as xs:string)
+as item()+ {
+ vars:get-relevant-files($ms-id)
+};
+
+declare
+ %test:args("Ar_7/229") %test:assertEquals("4")
+function t:determine-id-position($ms-id as xs:string)
+as xs:integer {
+ let $json := vars:get-relevant-files($ms-id)
+ return
+ vars:determine-id-position($ms-id, $json[1])
+};
+
+declare
+ %test:assertEquals("Variant")
+function t:get-body-object() {
+ vars:get-body-object($t:sample-map)
+ => map:get("x-content-type")
+};
+
+declare
+ %test:assertXPath("count($result) = 2")
+function t:get-files-relevant-for-page() {
+ let $relevant-files-for-ms-id := vars:get-relevant-files("Sbath_25")
+ let $ms-id-position := 1
+ let $tokens := ("Sbath_25_N4.4.2.4.4.2552.1_1", "Sbath_25_N4.4.2.4.4.500.1_1")
+ return
+ vars:get-files-relevant-for-page($relevant-files-for-ms-id, $ms-id-position, $tokens)
+};
+
+declare
+ %test:assertEquals("1")
+function t:get-indices-relevant-for-page() {
+ let $table :=
+ $t:sample-file
+ => map:get("table")
+ let $no-of-sequences := array:size($table)
+ let $ms-id-position := 1
+ let $tokens := ("Sbath_25_N4.4.2.4.4.2276.1_1", "Sbath_25_N4.4.2.4.4.2276.2.1_1", "Sbath_25_N4.4.2.4.4.2276.3_1", "Sbath_25_N4.4.2.4.4.2276.3_2", "Sbath_25_N4.4.2.4.4.2276.3_3")
+ return
+ vars:get-indices-relevant-for-page($table, $no-of-sequences, $ms-id-position, $tokens)
+};
+
+declare
+ %test:assertXPath("count($result) = 2 and $result = (2, 3)")
+function t:get-non-ms-id-positions-in-array() {
+ vars:get-non-ms-id-positions-in-array($t:sample-file, 1)
+};
+
+declare
+ %test:assertXPath("$result = ('ܕܐܚܛܛ', 'N')")
+function t:get-target-information() {
+ vars:make-annotation-value($t:sample-map)[2]
+ => map:get("entry"),
+ vars:make-annotation-value($t:sample-map)[2]
+ => map:get("witness")
+};
+
+declare
+ %test:assertEquals("ara")
+function t:get-target-language() {
+ vars:get-target-language("kant_sample_teixml")
+};
+
+declare
+ %test:assertEquals("Sbath_25")
+function t:get-witness() {
+ vars:get-witness($t:sample-file, 1)
+};
+
+declare
+ %test:assertEquals("5")
+function t:get-witness-entry() {
+ let $table :=
+ $t:sample-file
+ => map:get("table")
+ let $entry-no := "1"
+ let $witness-position := "1"
+ return
+ vars:get-witness-entry($table, $entry-no, $witness-position)
+ => array:size()
+};
+
+declare
+ %test:assertEquals("D")
+function t:make-annotation-value() {
+ vars:make-annotation-value($t:sample-map)[1]
+ => map:get("witness")
+};
+
+declare
+ %test:assertXPath("$result = ('Sbath_25_N4.4.2.4.4.2276.3_3', 'Vat_sir_424', 'Vat_sir_199')")
+function t:make-map-for-token() {
+ let $table := map:get($t:sample-file, "table")
+ let $entry-pos := 1
+ let $ms-id-position := 1
+ let $non-ms-id-positions := (2, 3)
+ let $current :=
+ vars:make-map-for-token($t:sample-file, $table, $entry-pos, $ms-id-position, $non-ms-id-positions)
+ => map:get("current")
+ => map:get("id")
+ let $variants :=
+ vars:make-map-for-token($t:sample-file, $table, $entry-pos, $ms-id-position, $non-ms-id-positions)
+ => map:get("variants")
+ return
+ (
+ $current,
+ $variants[1] => map:get("witness"),
+ $variants[2] => map:get("witness")
+
+ )
+
+};
\ No newline at end of file
diff --git a/exist-app/tests/commons-tests.xqm b/exist-app/tests/commons-tests.xqm
index 6cc032d36bfd2be582c90be7b50588b0731bf386..e614fe1529ae868d0df72f6c8bf946e2d2c67a74 100644
--- a/exist-app/tests/commons-tests.xqm
+++ b/exist-app/tests/commons-tests.xqm
@@ -87,6 +87,25 @@ as document-node() {
commons:get-metadata-file($uri)
};
+declare
+ %test:args("sample_teixml", "82a", "transcription") %test:assertXPath("$result//* = 'حقًا'")
+ %test:args("sample_teixml", "82a", "transliteration") %test:assertXPath("$result//* = 'الحاسوب'")
+function ct:get-page-fragment($documentURI as xs:string,
+ $page as xs:string,
+ $text-type as xs:string)
+as element(tei:TEI) {
+ commons:get-page-fragment-from-uri($documentURI, $page, $text-type)
+};
+
+declare
+ %test:args("sample_teixml", "82a") %test:assertXPath("count($result) = 2")
+ %test:args("sample_syriac_teixml", "86r") %test:assertXPath("count($result) = 1")
+function ct:get-transcription-and-transliteration-per-page($documentURI as xs:string,
+ $page as xs:string)
+as element(tei:TEI)+ {
+ commons:get-transcription-and-transliteration-per-page($documentURI, $page)
+};
+
declare
%test:assertXPath("$result = 'Add_2020'")
%test:assertXPath("$result = 'Sachau_290_Sachau_339'")
diff --git a/exist-app/tests/tei2json-tests.xqm b/exist-app/tests/tei2json-tests.xqm
index 9498cbf5e8fbe65a662a2722c2055407eedf05a3..07d259cec41c4df8fa4c3ee6ec44b35f095ca667 100644
--- a/exist-app/tests/tei2json-tests.xqm
+++ b/exist-app/tests/tei2json-tests.xqm
@@ -22,8 +22,8 @@ function t:main() {
declare
%test:assertTrue
-function t:create-json-collection-if-not-available() {
- let $create-collection := tei2json:create-json-collection-if-not-available()
+function t:create-json-collection() {
+ let $create-collection := tei2json:create-json-collection()
return
xmldb:collection-available("/db/data/textgrid/json")
};
diff --git a/exist-app/tests/tokenize-tests.xqm b/exist-app/tests/tokenize-tests.xqm
index 05588d407f91488c11052f2ab21766c5194e1f7d..1f7c438b3f4faaacf49c57a894ecf6ee59994b8b 100644
--- a/exist-app/tests/tokenize-tests.xqm
+++ b/exist-app/tests/tokenize-tests.xqm
@@ -20,7 +20,6 @@ function t:main() {
deep-equal(tokenize:main($src-TEI), $target-TEI)
};
-
declare
%test:assertXPath("$result/local-name() = 'ab' and count($result//text()) = 2 ")
function t:add-ids-single-node()
@@ -103,42 +102,6 @@ as element(tei:TEI) {
</TEI>
};
-declare function local:get-tei-header-1()
-as element(tei:TEI) {
- <TEI xmlns="http://www.tei-c.org/ns/1.0">
- <teiHeader>
- <fileDesc>
- <sourceDesc>
- <msDesc>
- <msIdentifier>
- <institution>University of Cambridge - Cambridge University Library</institution>
- <idno>Sachau 290 (=Sachau 339)</idno>
- </msIdentifier>
- </msDesc>
- </sourceDesc>
- </fileDesc>
- </teiHeader>
- </TEI>
-};
-
-declare function local:get-tei-header-2()
-as element(tei:TEI) {
- <TEI xmlns="http://www.tei-c.org/ns/1.0">
- <teiHeader>
- <fileDesc>
- <sourceDesc>
- <msDesc>
- <msIdentifier>
- <institution>University of Cambridge - Cambridge University Library</institution>
- <idno>Mingana ar. christ. 93[84]</idno>
- </msIdentifier>
- </msDesc>
- </sourceDesc>
- </fileDesc>
- </teiHeader>
- </TEI>
-};
-
declare function local:get-sample-result() {
<TEI xmlns="http://www.tei-c.org/ns/1.0">
<teiHeader>