Merge pull request #28 from dracor-org/20-speakers

Improve transformation of `who` attributes
dracor-org · Sep 2, 2024 · 9d0ab6b · 9d0ab6b
2 parents 9183657 + 7cef7ae
commit 9d0ab6b
Show file tree

Hide file tree

Showing 162 changed files with 950 additions and 1,479 deletions.
diff --git a/tc2dracor.xq b/tc2dracor.xq
@@ -14,7 +14,7 @@ declare variable $id-map := doc('ids.xml');
 (: XML document mapping original author info to normalized author names :)
 declare variable $author-map := doc('authors.xml');
 
-declare variable $who-tokenize-pattern := '/|,';
+declare variable $who-tokenize-pattern := '\s*(/|,|_)\s*';
 
 declare variable $comedy-genres := (
   "comédie ballet",
@@ -170,18 +170,10 @@ declare function local:attribute-to-comment($node as attribute()+) {
 };
 
 declare function local:translate($string as xs:string) as xs:string {
-  let $work :=
-    translate(lower-case($string), "*[]’' áàâéèêíìîóòôúùû", '------aaaeeeiiiooouuu')
-    => replace('\.', '')
-    => replace('^\-', '')
-    => replace('^(\d)', 'num') (: FIXME: this can effectively createe the same ID for different characters :)
-    => replace('^[|]$', '')
-    => replace('&#xfffd;', '')
-  return
-    (: quality assurance :)
-    if($work => matches('^\s*?$'))
-    then 'empty-string'
-    else $work
+  translate(lower-case($string), "_*[]’' áàâéèêíìîóòôúùû", '-------aaaeeeiiiooouuu')
+  => replace('\.', '')
+  => replace('^\-', '')
+  => replace('^(\d+(?:e|nde?|eme?|ere?)?)-(.+)$', '$2-$1')
 };
 
 declare function local:fix-type ($value) as xs:string {
@@ -228,29 +220,6 @@ declare function local:transform($nodes) {
   return
     typeswitch ( $node )
 
-      (:  👇 the incredible typo hack  :)
-      case element(SPEAKER) return
-        element {QName('', 'speaker')} {
-        $node/node()} => local:transform()
-      case element(P) return
-        element {QName('', 'p')} {
-        $node/node()} => local:transform()
-      case element(acheverImprimer) return
-        element {QName('', 'acheveImprime')} {
-        $node/node()} => local:transform()
-      case element(achevedImprime) return
-        element {QName('', 'acheveImprime')} {
-        $node/node()} => local:transform()
-      case element(acheveImprimer) return
-        element {QName('', 'acheveImprime')} {
-        $node/node()} => local:transform()
-      case element(appobation) return
-        element {QName('', 'approbation')} {
-        $node/node()} => local:transform()
-      case element(pinter) return
-        element {QName('', 'printer')} {
-        $node/node()} => local:transform()
-
       case text() return
         (: replace occurrences of 'quart_d_heure' :)
         (: see https://github.com/dracor-org/theatre-classique/commit/0e1f871dea95f4343895dad7e648de750c6dcf91 :)
@@ -284,24 +253,13 @@ declare function local:transform($nodes) {
         }
 
       case element(sp) return
-        let $exceptionsWho := (
-          $node/@who,
-          $node/@stwho,
-          $node/@givewho,
-          $node/@towardwho,
-          $node/@embarrassedwho,
-          $node/@breakwho,
-          $node/@ho,
-          $node/@w4ho
-        )
-        return
         if( not(exists($node/* except $node/*:speaker)) )
         then comment { 'ERROR: ', serialize($node)}
         else
         element {QName('http://www.tei-c.org/ns/1.0', 'sp')} {
-          $node/@* except ($node/@stage, $exceptionsWho, $node/@type),
+          $node/@* except ($node/@stage, $node/@who, $node/@type),
           attribute who {
-            let $easy := tokenize(string-join($exceptionsWho, ' '), $who-tokenize-pattern)
+            let $easy := tokenize(string-join($node/@who, ' '), $who-tokenize-pattern)
                               ! ('#' || local:translate(.))
             return
               if(string($easy[1]) != '')
@@ -717,8 +675,13 @@ declare function local:transform($nodes) {
 };
 
 declare function local:make-particDesc ($doc as element()) as node()* {
-  let $whos := ($doc//*:text//*:sp/tokenize(./@who || ./@ho || ./@w4ho, $who-tokenize-pattern) => distinct-values())
-  let $whos := if(string($whos[1]) != '') then $whos else (($doc//*:speaker/string(.)) => distinct-values())
+  let $whos := (for $sp in $doc//*:text//*:sp
+    return if ($sp/@who != '') then
+      tokenize(normalize-space($sp/@who), $who-tokenize-pattern)
+    else if ($sp/*:speaker[matches(., '\p{L}')]) then
+      replace($sp/*:speaker, "([-\p{L}' ]+).*", '$1')
+    else ()) => distinct-values()
+
   return if (count($whos)) then
     element {QName('http://www.tei-c.org/ns/1.0', 'particDesc')} {
       element {QName('http://www.tei-c.org/ns/1.0', 'listPerson')} {

diff --git a/tei/allainval-hiver.xml b/tei/allainval-hiver.xml
@@ -72,8 +72,8 @@
                     <person xml:id="la-volupte" sex="FEMALE">
                         <persName>La Volupté</persName>
                     </person>
-                    <person xml:id="jeux_ris_graces" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Jeux_ris_graces</persName>
+                    <person xml:id="jeux-ris" sex="UNKNOWN">
+                        <persName>Les Jeux et les Ris de la Suite de l'Hiver</persName>
                     </person>
                 </listPerson>
             </particDesc>
@@ -138,7 +138,7 @@
 	<castItem>
                     <role corresp="#bacchus">BACCHUS</role>.</castItem> 
 	<castItem>
-                    <role corresp="#jeux">Les Jeux et les Ris de la Suite de l'Hiver</role>.</castItem> 
+                    <role corresp="#jeux-ris">Les Jeux et les Ris de la Suite de l'Hiver</role>.</castItem> 
 	</castList>
 <!--TODO: usage of set in correct place but with unknown attributes <set location="Paris" country="France" periode="(indéterminé)" gps="48.856614, 2.352222">La scène est à Paris.</set>-->
 </front>
@@ -2144,7 +2144,7 @@
                     <head>DIVERTISSEMENT.</head>
 <stage type="entrance">Le Bal amène les Jeux, les Ris et les Grâces.</stage>
 <stage type="musique">MARCHE.</stage>
-<sp who="#jeux_ris_graces">
+<sp who="#jeux-ris">
                         <speaker>[JEUX, RIS et GRACES].</speaker>
                         <stage type="music">Air.</stage>
                         <l n="857">Venez plaisirs charmants et doux ;</l>
@@ -2176,7 +2176,7 @@
 </div>
 <div type="scene" n="15">
                     <head>VAUDEVILLE.</head>
-<sp who="#jeux_ris_graces">
+<sp who="#jeux-ris">
                         <speaker>[JEUX, RIS et GRACES].</speaker>
                         <l n="879">Quand un jeune amant vif et tendre,</l>
                         <l n="880">A trouvé l'art de nous surprendre,</l>

diff --git a/tei/andrieux-le-reve-du-mari.xml b/tei/andrieux-le-reve-du-mari.xml
@@ -53,7 +53,7 @@
                     <person xml:id="emilie" sex="FEMALE">
                         <persName>Emilie</persName>
                     </person>
-                    <person xml:id="gillot" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
+                    <person xml:id="gillot" sex="MALE">
                         <persName>Gillot</persName>
                     </person>
                     <person xml:id="la-baronne" sex="FEMALE">
@@ -143,7 +143,7 @@
 			<castItem>
                     <role corresp="#la-baronne">LA BARONNE</role>, soeur de Mathilde. Mlle DEMERSON.</castItem>
 			<castItem>
-                    <role corresp="#empty-string">GILLOT</role>, concierge. M. ARMAND-DAILLY.</castItem>
+                    <role corresp="#gillot">GILLOT</role>, concierge. M. ARMAND-DAILLY.</castItem>
 		</castList>
 	<!--TODO: usage of set in correct place but with unknown attributes <set location="Paris" country="France" periode="XVIIIème" gps="46.580224, 0.340375">La scène est en Poitou, dans un château, à la campagne.</set>-->
 </front>

diff --git a/tei/anonyme-sauvages-civilises.xml b/tei/anonyme-sauvages-civilises.xml
@@ -99,8 +99,8 @@
                     <person xml:id="louis" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Louis</persName>
                     </person>
-                    <person xml:id="guerriers-et-illinois" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Guerriers et Illinois</persName>
+                    <person xml:id="guerriers" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
+                        <persName>Guerriers</persName>
                     </person>
                     <person xml:id="choeur-general" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Choeur General</persName>
@@ -1123,7 +1123,7 @@ statue d'Ariski, et préparent le bûcher.</stage>
                         <l n="432">Rappelleront ta simplicité</l>
                         <l n="433">Des vertus du premier âge.</l>
                     </sp>
-<sp who="#guerriers-et-illinois">
+<sp who="#guerriers #illinois">
                         <speaker>GUERRIERS et ILLINOIS.</speaker>
                         <l n="434">Nous renonçons, </l>
                         <l n="435">Aux sanglantes conquêtes ;</l>

diff --git a/tei/barre-radet-candide.xml b/tei/barre-radet-candide.xml
@@ -73,9 +73,6 @@
                     <person xml:id="tous" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Tous</persName>
                     </person>
-                    <person xml:id="madame--de-candide" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Madame de Candide</persName>
-                    </person>
                     <person xml:id="osmin" sex="MALE">
                         <persName>Osmin</persName>
                     </person>

diff --git a/tei/bary-du-beau-sein.xml b/tei/bary-du-beau-sein.xml
@@ -45,12 +45,12 @@
         <profileDesc>
             <particDesc>
                 <listPerson>
-                    <person xml:id="sinope" sex="FEMALE">
-                        <persName>Sinope</persName>
-                    </person>
                     <person xml:id="tyrisias" sex="MALE">
                         <persName>Tyrisias</persName>
                     </person>
+                    <person xml:id="sinope" sex="FEMALE">
+                        <persName>Sinope</persName>
+                    </person>
                 </listPerson>
             </particDesc>
             <textClass>

diff --git a/tei/biancollelli-foire-galante.xml b/tei/biancollelli-foire-galante.xml
@@ -61,9 +61,6 @@
                     <person xml:id="une-des-graces" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Une des Graces</persName>
                     </person>
-                    <person xml:id="arlequin-et-pierrot" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Arlequin et Pierrot</persName>
-                    </person>
                     <person xml:id="le-docteur" sex="MALE">
                         <persName>Le Docteur</persName>
                     </person>
@@ -73,9 +70,6 @@
                     <person xml:id="marinette" sex="FEMALE">
                         <persName>Marinette</persName>
                     </person>
-                    <person xml:id="le-choeur-et-le-docteur" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Le Choeur et le Docteur</persName>
-                    </person>
                     <person xml:id="les-chanteuses" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Les Chanteuses</persName>
                     </person>
@@ -317,7 +311,7 @@
                         <l n="88">Et Tircis sans le cocuage,</l>
                         <l n="89">Serait-il ce qu'il est ?</l>
                     </sp>
-<sp who="#arlequin-et-pierrot">
+<sp who="#arlequin #pierrot">
                         <speaker>ARLEQUIN et PIERROT, sur le même ton, ensemble.</speaker>
                         <l n="90">Devenons par le cocuage</l>
                         <l n="91">Aussi riche qu'il est.</l>
@@ -574,7 +568,7 @@
                         <l n="226">Cette râpe de fer blanc,</l>
                         <l n="227">Cette andouille de Tabac.</l>
                     </sp>
-<sp who="#le-choeur-et-le-docteur">
+<sp who="#le-choeur #le-docteur">
                         <speaker>LE CHOEUR et LE DOCTEUR, à l'imitation de Aimez, aimez, belle bergère, ensemble</speaker>
                         <l n="228">Râpez, râpez, belle bergère,</l>
                         <l n="229">Si vous voulez charmer !</l>

diff --git a/tei/boissy-babillard.xml b/tei/boissy-babillard.xml
@@ -66,24 +66,18 @@
                     <person xml:id="la-fleur" sex="MALE">
                         <persName>La Fleur</persName>
                     </person>
-                    <person xml:id="doris_melite" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Doris_melite</persName>
-                    </person>
                     <person xml:id="doris" sex="FEMALE">
                         <persName>Doris</persName>
                     </person>
+                    <person xml:id="melite" sex="FEMALE">
+                        <persName>Mélite</persName>
+                    </person>
                     <person xml:id="ismene" sex="FEMALE">
                         <persName>Ismène</persName>
                     </person>
                     <person xml:id="cephise" sex="FEMALE">
                         <persName>Céphise</persName>
                     </person>
-                    <person xml:id="melite" sex="FEMALE">
-                        <persName>Mélite</persName>
-                    </person>
-                    <person xml:id="leandre_melite_doris" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Leandre_melite_doris</persName>
-                    </person>
                 </listPerson>
             </particDesc>
             <textClass>
@@ -779,7 +773,7 @@
 </div>
 <div type="scene" n="9">
                     <head>SCÈNE IX. Léandre, Céphise, Ismène, Hortense, Daphné, Drois, Mélite.</head>
-<sp who="#doris_melite">
+<sp who="#doris #melite">
                         <speaker>DORIS et MÉLITE, entrant les premières. </speaker>
                         <l n="326">Nous nous rendons, madame, et ne disputons plus. </l>
                     </sp>
@@ -1234,7 +1228,7 @@
                         <speaker>DAPHNÉ, à Léandre.</speaker>
                         <l n="409" part="F">Tenez bon. </l>
                     </sp>
-<sp who="#leandre_melite_doris">
+<sp who="#leandre #melite #doris">
                         <speaker>LÉANDRE, MÉLITE, DORIS.</speaker>
                         <l n="410" part="I">Madame... </l>
                     </sp>

diff --git a/tei/boursault-fete-de-la-seine.xml b/tei/boursault-fete-de-la-seine.xml
@@ -59,8 +59,11 @@
                     <person xml:id="deux-nereides" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Deux Nereides</persName>
                     </person>
-                    <person xml:id="une-jeune-fontaine-et-un-ruisseau" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Une Jeune Fontaine et un Ruisseau</persName>
+                    <person xml:id="une-jeune-fontaine" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
+                        <persName>Une Jeune Fontaine</persName>
+                    </person>
+                    <person xml:id="un-ruisseau" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
+                        <persName>Un Ruisseau</persName>
                     </person>
                     <person xml:id="deux-jeunes-ruisseaux" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Deux Jeunes Ruisseaux</persName>
@@ -80,9 +83,6 @@
                     <person xml:id="ensemble" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Ensemble</persName>
                     </person>
-                    <person xml:id="le-gange-et-le-nil" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
-                        <persName>Le Gange et le Nil</persName>
-                    </person>
                     <person xml:id="tous-ensemble" sex="UNKNOWN"><!--WARNING: no castItem found for reference in @who-->
                         <persName>Tous Ensemble</persName>
                     </person>
@@ -252,7 +252,7 @@
                         <l n="69">Trouvent la fin de leurs peines,</l>
                         <l n="70">Sans jamais changer leurs cours.</l>
                     </sp>
-<sp who="#une-jeune-fontaine-et-un-ruisseau">
+<sp who="#une-jeune-fontaine #un-ruisseau">
                         <speaker>UNE JEUNE FONTAINE ET UN RUISSEAU.</speaker>
                         <l n="71">Le temps d'aimer est un temps admirable ;</l>
                         <l n="72">Mais il ne dure pas assez :</l>
@@ -379,7 +379,7 @@
                         <l n="160">Consolez-vous, vos souffrances</l>
                         <l n="161">Ne dureront pas longtemps.</l>
                     </sp>
-<sp who="#le-gange-et-le-nil">
+<sp who="#le-gange #le-nil">
                         <speaker>LE GANGE et LE NIL.</speaker>
                         <l n="162">Puisse le Ciel qui l'a fait naître</l>
                         <l n="163">Pour affranchir du joug tant de peuples divers,</l>