@mailwoman/corpus 9.4.0 → 10.1.0

package/out/be/adapters/vlaanderen/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/be/adapters/vlaanderen/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkEG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAClE,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,cAAc,EAAE,MAAM,4BAA4B,CAAA;AAC3D,OAAO,EAAsB,oBAAoB,EAAE,MAAM,+BAA+B,CAAA;AACxF,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AAErC,OAAO,EAAE,uBAAuB,EAAE,MAAM,kBAAkB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAA;AAC1D,OAAO,EACN,cAAc,EACd,cAAc,EACd,UAAU,EACV,iBAAiB,EACjB,SAAS,EACT,oBAAoB,EACpB,gBAAgB,GAChB,MAAM,kBAAkB,CAAA;AACzB,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EAAE,WAAW,EAA8D,aAAa,EAAE,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,YAAY,CAAA;AAEjD;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iDAAiD,CAAA;AAE3F;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAsB,CAAC,IAAI,CAAC,CAAA;AAE7D;;;;;;;;;GASG;AACH,MAAM,uBAAuB,GAAG,0BAA0B,CAAA;AAE1D;;;;;;;;GAQG;AACH,MAAM,kBAAkB,GAAG,UAAU,CAAA;AAErC;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,cAAc,GAAG,GAAG,CAAA;AA8E1B;;;;;;GAMG;AACH,MAAM,OAAO,sBAAuB,SAAQ,KAAK;IAChD,YAAY,OAAe;QAC1B,KAAK,CAAC,GAAG,qBAAqB,aAAa,OAAO,EAAE,CAAC,CAAA;QAErD,IAAI,CAAC,IAAI,GAAG,wBAAwB,CAAA;IACrC,CAAC;CACD;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY;IAC/C,IAAI,GAAQ,CAAA;IAEZ,IAAI,CAAC;QACJ,GAAG,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,CAAA;IACpB,CAAC;IAAC,MAAM,CAAC;QACR,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,MAAM,QAAQ,GAAG,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEhF,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,IAAI,CAAA;IAEpC,OAAO,GAAG,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;AAC/C,CAAC;AAmCD;;;;;;;GAOG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa;IAC5C,MAAM,SAAS,GAAG,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAA;IAExC,IAAI,SAAS,KAAK,CAAC,CAAC,EAAE,CAAC;QACtB,MAAM,IAAI,sBAAsB,CAC/B,yBAAyB,aAAa,CAAC,KAAK,CAAC,6BAA6B;YACzE,8FAA8F;YAC9F,6EAA6E,CAC9E,CAAA;IACF,CAAC;IAED,OAAO,EAAE,MAAM,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,IAAI,EAAE,EAAE,aAAa,EAAE,KAAK,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,CAAA;AACtG,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,aAAa,CAAC,OAAsB;IACnD,OAAO,UAAU,CAAC,iBAAiB,CAAC,OAAO,CAAC,EAAE,uBAAuB,CAAC,CAAA;AACvE,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,kBAAkB,CAChC,IAAiB,EACjB,KAAwB,EACxB,QAAgB,EAChB,IAA+C,EAC/C,MAAoB;IAEpB,MAAM,KAAK,GAAG,IAAI,GAAG,EAAa,CAAA;IAElC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,CAAA;QAE3B,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,oBAAoB,CAAC,cAAc,CAAC,QAAQ,CAAC,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,EAAE,CAAC;YAC7G,0FAA0F;YAC1F,gFAAgF;YAChF,MAAM,EAAE,GAAG,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAA;YAEvC,IAAI,CAAC,EAAE,EAAE,CAAC;gBACT,MAAM,IAAI,sBAAsB,CAC/B,KAAK,QAAQ,eAAe,QAAQ,CAAC,QAAQ,EAAE,iDAAiD,CAChG,CAAA;YACF,CAAC;YAED,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC,CAAA;YAE3B,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACzB,KAAK,CAAC,GAAG,CAAC,EAAE,EAAE,KAAK,CAAC,CAAA;YACrB,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAA;AACb,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,kBAAkB,CAChC,IAAiB,EACjB,OAA0B,EAC1B,MAAoB;IAEpB,MAAM,OAAO,GAAG,MAAM,kBAAkB,CACvC,IAAI,EACJ,OAAO,CAAC,UAAU,CAAC,gBAAgB,EACnC,qBAAqB,EACrB,gBAAgB,EAChB,MAAM,CACN,CAAA;IAED,MAAM,SAAS,GAAG,MAAM,kBAAkB,CACzC,IAAI,EACJ,OAAO,CAAC,UAAU,CAAC,gBAAgB,EACnC,qBAAqB;IACrB,wCAAwC;IACxC,wFAAwF;IACxF,+DAA+D;IAC/D,8CAA8C;IAC9C,oBAAoB,EACpB,MAAM,CACN,CAAA;IAED,MAAM,cAAc,GAAG,MAAM,kBAAkB,CAC9C,IAAI,EACJ,OAAO,CAAC,UAAU,CAAC,aAAa,EAChC,kBAAkB,EAClB,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,cAAc,CAAC,OAAO,CAAC,KAAK,kBAAkB,CAAC,CAAC,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,EAC9F,MAAM,CACN,CAAA;IAED,KAAK,MAAM,CAAC,QAAQ,EAAE,KAAK,CAAC,IAAI;QAC/B,CAAC,qBAAqB,EAAE,OAAO,CAAC;QAChC,CAAC,qBAAqB,EAAE,SAAS,CAAC;QAClC,CAAC,kBAAkB,EAAE,cAAc,CAAC;KAC3B,EAAE,CAAC;QACZ,uFAAuF;QACvF,8CAA8C;QAC9C,IAAI,MAAM,EAAE,OAAO;YAAE,MAAK;QAE1B,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;YACjB,MAAM,IAAI,sBAAsB,CAC/B,iBAAiB,QAAQ,8DAA8D;gBACtF,yFAAyF;gBACzF,wCAAwC,CACzC,CAAA;QACF,CAAC;IACF,CAAC;IAED,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,cAAc,EAAE,CAAA;AAC9C,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,qBAAqB,CAAC,IAAiB;IAC5D,MAAM,YAAY,GAAG,IAAI,CAAC,cAAc,CAAC,CAAA;IACzC,MAAM,OAAO,GAAG,MAAM,iBAAiB,CAAoB,YAAY,CAAC,CAAA;IACxE,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,CAAA;IAErC,IACC,CAAC,UAAU,EAAE,gBAAgB,EAAE,MAAM;QACrC,CAAC,UAAU,CAAC,gBAAgB,EAAE,MAAM;QACpC,CAAC,UAAU,CAAC,aAAa,EAAE,MAAM,EAChC,CAAC;QACF,MAAM,IAAI,sBAAsB,CAC/B,GAAG,YAAY,CAAC,QAAQ,EAAE,qEAAqE;YAC9F,8FAA8F;YAC9F,iCAAiC,CAClC,CAAA;IACF,CAAC;IAED,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,MAAM,EAAE,CAAC;QAC1E,MAAM,IAAI,sBAAsB,CAAC,GAAG,YAAY,CAAC,QAAQ,EAAE,wBAAwB,CAAC,CAAA;IACrF,CAAC;IAED,OAAO,OAAO,CAAA;AACf,CAAC;AA0BD,MAAM,UAAU,uBAAuB;IACtC,OAAO;QACN,EAAE,EAAE,qBAAqB;QACzB,cAAc,EAAE,0BAA0B;QAC1C,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,0BAA0B;QACnD,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,0GAA0G;QAE3G,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,oBAAoB,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAClE,MAAM,IAAI,uBAAuB,CAAC,qBAAqB,EAAE,oBAAoB,EAAE,IAAI,CAAC,OAAO,CAAC,CAAA;YAC7F,CAAC;YAED,uFAAuF;YACvF,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;gBAAE,OAAM;YAEhC,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAC7C,MAAM,OAAO,GAAG,MAAM,qBAAqB,CAAC,IAAI,CAAC,CAAA;YACjD,MAAM,KAAK,GAAG,MAAM,kBAAkB,CAAC,IAAI,EAAE,OAAO,EAAE,IAAI,CAAC,MAAM,CAAC,CAAA;YAElE,IAAI,OAAO,GAAG,CAAC,CAAA;YACf,MAAM,OAAO,GAAa,EAAE,aAAa,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,aAAa,EAAE,CAAC,EAAE,UAAU,EAAE,CAAC,EAAE,CAAA;YAE5F,IAAI,CAAC;gBACJ,2EAA2E;gBAC3E,4CAA4C;gBAC5C,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;oBACzC,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAE5D,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBAChC,IAAI,IAAI,GAAG,CAAC,CAAA;oBAEZ,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,oBAAoB,CAAC,cAAc,CAAC,QAAQ,CAAC,EAAE,YAAY,EAAE;wBACxF,GAAG,EAAE,IAAI;wBACT,MAAM,EAAE,IAAI,CAAC,MAAM;qBACnB,CAAC,EAAE,CAAC;wBACJ,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;4BAAE,MAAK;wBAE/B,IAAI,EAAE,CAAA;wBAEN,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;4BAAE,MAAK;wBAE5D,MAAM,KAAK,GAAG,aAAa,CAAC,OAAO,CAAC,CAAA;wBAEpC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;4BACzB,OAAO,CAAC,aAAa,EAAE,CAAA;4BAEvB,SAAQ;wBACT,CAAC;wBAED,IAAI,MAA0B,CAAA;wBAC9B,IAAI,QAA4B,CAAA;wBAChC,IAAI,QAA4B,CAAA;wBAChC,IAAI,UAAU,GAAG,KAAK,CAAA;wBAEtB,KAAK,MAAM,IAAI,IAAI,cAAc,CAAC,OAAO,CAAC,EAAE,CAAC;4BAC5C,MAAM,EAAE,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAA;4BAEpC,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;gCACjB,UAAU,GAAG,IAAI,CAAA;gCAEjB,MAAK;4BACN,CAAC;4BAED,MAAM,WAAW,GAAG,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;4BAEzC,IAAI,WAAW,KAAK,SAAS,EAAE,CAAC;gCAC/B,MAAM,GAAG,WAAW,CAAA;gCAEpB,SAAQ;4BACT,CAAC;4BAED,MAAM,aAAa,GAAG,KAAK,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;4BAE7C,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;gCACjC,QAAQ,GAAG,aAAa,CAAA;gCAExB,SAAQ;4BACT,CAAC;4BAED,0EAA0E;4BAC1E,yDAAyD;4BACzD,QAAQ,GAAG,KAAK,CAAC,cAAc,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,QAAQ,CAAA;wBACpD,CAAC;wBAED,8DAA8D;wBAC9D,uCAAuC;wBACvC,kFAAkF;wBAClF,IAAI,UAAU,IAAI,MAAM,KAAK,SAAS,IAAI,QAAQ,KAAK,SAAS,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;4BAC5F,OAAO,CAAC,QAAQ,EAAE,CAAA;4BAElB,SAAQ;wBACT,CAAC;wBAED,IACC,MAAM,CAAC,QAAQ,CAAC,cAAc,CAAC;4BAC/B,QAAQ,CAAC,QAAQ,CAAC,cAAc,CAAC;4BACjC,QAAQ,CAAC,QAAQ,CAAC,cAAc,CAAC,EAChC,CAAC;4BACF,OAAO,CAAC,aAAa,EAAE,CAAA;4BAEvB,SAAQ;wBACT,CAAC;wBAED,MAAM,QAAQ,GAAG,gBAAgB,CAChC;4BACC,YAAY,EAAE,kBAAkB,CAAC,KAAK,EAAE,EAAE,CAAC;4BAC3C,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,MAAM;4BACtC,QAAQ;4BACR,QAAQ;yBACR,EACD,IAAI,EACJ,EAAE,UAAU,EAAE,IAAI,EAAE,CACpB,CAAA;wBAED,IAAI,CAAC,QAAQ,EAAE,CAAC;4BACf,OAAO,CAAC,UAAU,EAAE,CAAA;4BAEpB,SAAQ;wBACT,CAAC;wBAED,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,QAAQ,CAAA;wBAC7C,uFAAuF;wBACvF,MAAM,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAA;wBAE/C,MAAM;4BACL,GAAG;4BACH,UAAU,EAAE,OAAO;4BACnB,OAAO,EAAE,IAAI;4BACb,wEAAwE;4BACxE,gEAAgE;4BAChE,MAAM,EAAE,OAAO;4BACf,MAAM,EAAE,qBAAqB;4BAC7B,SAAS,EAAE,IAAI,CAAC,CAAC,CAAC,GAAG,qBAAqB,IAAI,IAAI,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC,qBAAqB,EAAE,OAAO,CAAC;4BACrG,cAAc,EAAE,EAAE;4BAClB,OAAO,EAAE,0BAA0B;yBACnC,CAAA;wBAED,OAAO,EAAE,CAAA;oBACV,CAAC;oBAED,wDAAwD;oBACxD,6DAA6D;oBAC7D,qDAAqD;oBACrD,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;oBAEzE,IAAI,CAAC,OAAO,IAAI,IAAI,KAAK,IAAI,CAAC,cAAc,EAAE,CAAC;wBAC9C,MAAM,IAAI,sBAAsB,CAC/B,GAAG,QAAQ,CAAC,QAAQ,EAAE,UAAU,IAAI,iDAAiD;4BACpF,kBAAkB,IAAI,CAAC,cAAc,mBAAmB,IAAI,CAAC,UAAU,EAAE,CAC1E,CAAA;oBACF,CAAC;gBACF,CAAC;YACF,CAAC;oBAAS,CAAC;gBACV,MAAM,OAAO,GAAG,OAAO,CAAC,aAAa,GAAG,OAAO,CAAC,QAAQ,GAAG,OAAO,CAAC,aAAa,GAAG,OAAO,CAAC,UAAU,CAAA;gBAErG,IAAI,OAAO,GAAG,CAAC,EAAE,CAAC;oBACjB,OAAO,CAAC,MAAM,CAAC,KAAK,CACnB,iBAAiB,OAAO,eAAe,OAAO,WAAW;wBACxD,IAAI,OAAO,CAAC,aAAa,4BAA4B,OAAO,CAAC,QAAQ,oBAAoB;wBACzF,8BAA8B,OAAO,CAAC,aAAa,0BAA0B;wBAC7E,GAAG,OAAO,CAAC,UAAU,oBAAoB,CAC1C,CAAA;gBACF,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,uBAAuB,EAAE,CAAA"}

package/out/be/adapters/wallonie/adapter.d.ts

@@ -0,0 +1,196 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `wallonie`: the INSPIRE Addresses (AD) theme the Service public de Wallonie publishes out of its
+ * ICAR address-point register.
+ *
+ * Coverage is the Walloon Region rather than Belgium. Belgium needs three readers, because each
+ * region publishes the theme through its own service: Flanders through a WFS (`vlaanderen`),
+ * Brussels through a bare HTTP download (`brussels`), and Wallonia through the ATOM service below.
+ * Every row this adapter emits is `BE`, and any other `opts.country` raises
+ * {@linkcode UnsupportedCountryError}.
+ *
+ * The schema-level reading is `#inspire/address`, which owns every part of an `ad:Address` the
+ * publishers spell alike. This module holds only what is Wallonia's own: the two-pass join its
+ * document order forces, and the number extension it writes on every address.
+ *
+ * ## The publication
+ *
+ * The ATOM service document is `https://geoservices.wallonie.be/inspire/atom/AD_Service.xml`, 9,935
+ * bytes, with one entry. That entry's dataset feed offers one archive for the whole region:
+ * `…/spwdatadownload/results/b62f8405-2ce3-449f-8a61-5fc2f38d8b73/AD.Addresses.gml.zip`,
+ * 106,665,832 bytes, sha256 `8eebae71d793632c22c9902e8420bc50e0e572c97de15b7d10051b29b59e208a`,
+ * holding one deflate member `AD.Addresses.gml` of 5,163,917,131 bytes. The archive is 48 times
+ * smaller than what it holds, so the member is streamed and never buffered.
+ *
+ * The host ignores `Range`: a `bytes=0-4194303` request answers HTTP 200 with the whole body and no
+ * `content-range`, so there is no prefix to sample and every measurement below is over all
+ * 5,163,917,131 bytes.
+ *
+ * Counts, read 2026-10-02 with {@linkcode streamMarkupElements} and reconciled against a second
+ * instrument that extracted `gml:featureMember` blocks from the same stream as text: 1,772,312
+ * `ad:Address`, 55,766 `ad:ThoroughfareName`, 1,963 `ad:AddressAreaName`, 598 `ad:PostalDescriptor`
+ * and 264 `ad:AdminUnitName`. The two instruments agree, and they reconcile: 1,772,312 addresses plus
+ * 58,591 component features is the 1,830,903 `gml:featureMember` elements the file holds, and the
+ * component total is the number of distinct component `gml:id` values. The address count is also the
+ * one the address-source register records for `be-property-building-2`.
+ *
+ * Every component `gml:id` is distinct. Over 58,591 component features, 0 ids repeat, so the index
+ * below states no collision rule and raises on a collision instead. A replaced entry would
+ * substitute one street for another on every address referencing it.
+ *
+ * ## Why the join takes two passes
+ *
+ * Wallonia writes all 1,772,312 addresses before the first component feature. Measured by resolving
+ * each reference against the components indexed so far, all 10,633,872 references are forward
+ * references: 0 resolve in a single pass. ČÚZK's one-pass reader with a deferred tail cannot be used
+ * here, because deferring every address means holding 1.77 million subtrees in memory.
+ *
+ * So the member is inflated twice. The first pass indexes the four component types — 58,591 entries,
+ * bounded and small — and the second streams the addresses and joins locally. Two inflations of
+ * 5,163,917,131 bytes cost less than one buffered document.
+ *
+ * Each address writes exactly 6 `ad:component` references, 10,633,872 over the file, and each one
+ * addresses a `gml:id` in the same document. Four are the region's own
+ * (`BE.WL.ICAR.ThoroughfareName`, `AdminUnitName`, `PostalDescriptor`, `AddressAreaName`) and two are
+ * the country and the region, `#BE.AdminUnitName.1000` (`Belgique`, `1stOrder`) and
+ * `#BE.AdminUnitName.3000` (`Région wallonne`, `2ndOrder`). The index keeps an administrative unit
+ * only at {@link MUNICIPALITY_LEVEL}, so those two resolve to no entry and are a reference the
+ * adapter declines rather than a defect — the same shape Flanders' `refnis1995_1000` reference has.
+ *
+ * ## The number extension is the box number, rendered as `unit`
+ *
+ * Every address writes two inline designators: `addressIdentifierGeneral` holds the house
+ * number and `addressNumberExtension` holds a four-character code. `ad:level` is
+ * `postalDeliveryPoint` on every address, so a row is a letterbox rather than a building, and the
+ * extension distinguishes letterboxes at one number: `93` appears with `B001` and `C003`, and `29`
+ * with `0014`, `B004`, `B007`, `C009` and `D015`. Other observed values are `0001`, `A001`, `008B`,
+ * `BP3` and `0RCH`.
+ *
+ * The element is written on all 1,772,312 addresses and holds a value on 347,859 of them, measured by
+ * running this adapter over the whole archive. The other 1,424,453, 80.4%, write
+ * `addressNumberExtension` with its `ad:designator` empty. That empty element is the publisher
+ * stating that the address has no extension rather than omitting the field.
+ *
+ * ICAR, the register this publication is drawn from, models a sub-address below the police number
+ * and states what identifies it: the SPW's ICAR presentation to the Club des utilisateurs du PICC
+ * (27 February 2015) lists the communes' validation of sub-addresses identified by their
+ * `numéro de boîte`, the box number, and its ICAR address form records each sub-address's `Numéro` apart from the `Index`
+ * field. A box number identifies one delivery point inside the numbered premise, which is the
+ * concept the `unit` tag carries. It is not a floor, a building, an entrance or a staircase, even
+ * where a code such as `0RCH` (rez-de-chaussée) names the box after the floor it sits on.
+ *
+ * The value is rendered as the publisher writes it, with no designator word: `Rue de la Station 29
+ * B004, 1420 Braine-l'Alleud`. Wallonia publishes no formatted line to copy a designator from, so a
+ * word such as `bte` would be a surface this source does not state. Two letterboxes at one number
+ * therefore render two lines, and deduplication keeps both.
+ *
+ * ## Voids
+ *
+ * Every void in this file states a reason. Over the whole member, 1,830,903 `ad:validTo` and
+ * 1,830,903 `ad:endLifespanVersion` elements, 57,997 `gn:pronunciation`, `gn:grammaticalGender` and
+ * `gn:grammaticalNumber` each, 2,823 `ad:validFrom` and 264 `ad:adminUnit` are `xsi:nil="true"` with
+ * a `nilReason`, and 0 nil elements omit the reason. No value this adapter reads was void in the file
+ * measured. Brussels is the opposite case, writing no reason on any void, and `isVoid` in
+ * `#inspire/address` keys on `xsi:nil` rather than on a reason for that reason. This adapter reads
+ * the same primitives, so a Walloon void that stopped stating a reason would still read as a void.
+ *
+ * ## Coordinates
+ *
+ * This adapter emits text and components only, because {@linkcode CanonicalRow} declares no position
+ * field. The file's axis order is recorded here for whoever does emit one. Wallonia declares
+ * `srsName="http://www.opengis.net/def/crs/EPSG/0/3035"` and complies with that authority's
+ * northing-then-easting order: `gml:pos` `3079047.229041581 3924386.42904076` reads as latitude
+ * 50.686064, longitude 4.381708, in Braine-l'Alleud. That is the municipality the same address's own
+ * `ad:AdminUnitName` names. Its `gml:boundedBy` lower corner, `2942208.441255148 3816649.462808032`,
+ * puts the smaller number first in the same way. Brussels declares the same CRS and writes
+ * easting-northing, so the two publishers disagree and a reader cannot take the declaration at its
+ * word. `brussels/adapter.ts` records that measurement.
+ *
+ * ## Terms
+ *
+ * The address-source register elects `Licence CC-BY 4.0, as the Service public de Wallonie states it`
+ * for `be-property-building-2`, under license id
+ * `unchecked-access-free-be-service-public-de-wallonie-spw`, and that decision records
+ * `spdx: "CC-BY-4.0"`. `electedLicenseLabel` answers an elected decision's `spdx` before its
+ * `electedTerms`, so {@linkcode WALLONIE_DEFAULT_LICENSE} is the SPDX id rather than the French
+ * sentence. The register's own words for what the publisher permits are use, modification into a
+ * derived work, and publication of both, on condition that the sources are cited, so a model card
+ * states `Source : Service public de Wallonie (SPW) - INSPIRE - Points d'adresses en Wallonie (BE)
+ * (2017-11-20) https://geodata.wallonie.be/id/b62f8405-2ce3-449f-8a61-5fc2f38d8b73`.
+ *
+ * That row declares no `adapterID`, so `createIneligibilityReader` refuses every row this adapter
+ * emits until a review sets `adapterID` to `wallonie` on it. The refusal is by construction and
+ * states itself in the build manifest.
+ */
+import { type AdapterOptions, type CorpusAdapter } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ * The name is the region as its own publisher writes it, as `vlaanderen`'s is.
+ */
+export declare const WALLONIE_ADAPTER_ID = "wallonie";
+/**
+ * The stable id of the publication behind every row.
+ *
+ * It belongs in `SourceRegister` beside `be-vlaanderen-inspire-ad`, and this constant
+ * stands in until that entry lands, because a built corpus stores the id on every row
+ * and the id must not change afterwards.
+ * Replace the constant with the `SourceRegister` member on integration.
+ */
+export declare const WALLONIE_SOURCE_REGISTER: "be-wallonie-inspire-ad";
+/**
+ * The one jurisdiction this adapter emits.
+ *
+ * The service publishes the Walloon Region, to which ISO 3166-1 assigns no
+ * code of its own, so the set holds one code and a caller naming another gets
+ * {@linkcode UnsupportedCountryError} rather than zero rows.
+ */
+export declare const WALLONIE_COUNTRIES: readonly string[];
+/**
+ * The license recorded on every row.
+ *
+ * The register's elected decision for `unchecked-access-free-be-service-public-de-wallonie-spw`
+ * records `spdx: "CC-BY-4.0"`, and `electedLicenseLabel` answers the SPDX id when one is recorded.
+ */
+export declare const WALLONIE_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+/**
+ * The components an address references, keyed by the `gml:id` its `#` fragment names.
+ */
+export interface WallonieComponentIndex {
+    /**
+     * `ad:ThoroughfareName` → the street name the publisher spells.
+     */
+    streets: ReadonlyMap<string, string>;
+    /**
+     * `ad:PostalDescriptor` → its `ad:postCode`.
+     */
+    postcodes: ReadonlyMap<string, string>;
+    /**
+     * `ad:AdminUnitName` at {@link MUNICIPALITY_LEVEL} → the municipality name.
+     *
+     * A unit at another level is left out, so the country and region references cannot answer for a locality.
+     */
+    municipalities: ReadonlyMap<string, string>;
+    /**
+     * `ad:AddressAreaName` → the name of the part of the municipality, such as `Hamme-Mille` in `Beauvechain`.
+     */
+    areas: ReadonlyMap<string, string>;
+}
+/**
+ * Indexes the four component types in one pass over the member.
+ *
+ * @throws {@linkcode InspireArchiveError} when a component states no `gml:id`,
+ * when two components share one, or when the pass ends with an empty index.
+ */
+export declare function readWallonieComponents(inputPath: AdapterOptions["inputPath"], signal?: AbortSignal): Promise<WallonieComponentIndex>;
+export declare function createWallonieAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const wallonieAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/be/adapters/wallonie/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/be/adapters/wallonie/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8HG;AAsBH,OAAO,EAEN,KAAK,cAAc,EAEnB,KAAK,aAAa,EAGlB,MAAM,QAAQ,CAAA;AAEf;;;;;GAKG;AACH,eAAO,MAAM,mBAAmB,aAAa,CAAA;AAE7C;;;;;;;GAOG;AACH,eAAO,MAAM,wBAAwB,0BAA0C,CAAA;AAE/E;;;;;;GAMG;AACH,eAAO,MAAM,kBAAkB,EAAE,SAAS,MAAM,EAAW,CAAA;AAE3D;;;;;GAKG;AACH,eAAO,MAAM,wBAAwB,cAAc,CAAA;AAmDnD;;;GAGG;AACH,MAAM,WAAW,sBAAsB;IACtC;;OAEG;IACH,OAAO,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACpC;;OAEG;IACH,SAAS,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACtC;;;;OAIG;IACH,cAAc,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAC3C;;OAEG;IACH,KAAK,EAAE,WAAW,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAClC;AAED;;;;;GAKG;AACH,wBAAsB,sBAAsB,CAC3C,SAAS,EAAE,cAAc,CAAC,WAAW,CAAC,EACtC,MAAM,CAAC,EAAE,WAAW,GAClB,OAAO,CAAC,sBAAsB,CAAC,CAgFjC;AAwBD,wBAAgB,qBAAqB,IAAI,aAAa,CAwJrD;AAED;;GAEG;AACH,eAAO,MAAM,eAAe,eAA0B,CAAA"}

package/out/be/adapters/wallonie/adapter.js

@@ -0,0 +1,437 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `wallonie`: the INSPIRE Addresses (AD) theme the Service public de Wallonie publishes out of its
+ * ICAR address-point register.
+ *
+ * Coverage is the Walloon Region rather than Belgium. Belgium needs three readers, because each
+ * region publishes the theme through its own service: Flanders through a WFS (`vlaanderen`),
+ * Brussels through a bare HTTP download (`brussels`), and Wallonia through the ATOM service below.
+ * Every row this adapter emits is `BE`, and any other `opts.country` raises
+ * {@linkcode UnsupportedCountryError}.
+ *
+ * The schema-level reading is `#inspire/address`, which owns every part of an `ad:Address` the
+ * publishers spell alike. This module holds only what is Wallonia's own: the two-pass join its
+ * document order forces, and the number extension it writes on every address.
+ *
+ * ## The publication
+ *
+ * The ATOM service document is `https://geoservices.wallonie.be/inspire/atom/AD_Service.xml`, 9,935
+ * bytes, with one entry. That entry's dataset feed offers one archive for the whole region:
+ * `…/spwdatadownload/results/b62f8405-2ce3-449f-8a61-5fc2f38d8b73/AD.Addresses.gml.zip`,
+ * 106,665,832 bytes, sha256 `8eebae71d793632c22c9902e8420bc50e0e572c97de15b7d10051b29b59e208a`,
+ * holding one deflate member `AD.Addresses.gml` of 5,163,917,131 bytes. The archive is 48 times
+ * smaller than what it holds, so the member is streamed and never buffered.
+ *
+ * The host ignores `Range`: a `bytes=0-4194303` request answers HTTP 200 with the whole body and no
+ * `content-range`, so there is no prefix to sample and every measurement below is over all
+ * 5,163,917,131 bytes.
+ *
+ * Counts, read 2026-10-02 with {@linkcode streamMarkupElements} and reconciled against a second
+ * instrument that extracted `gml:featureMember` blocks from the same stream as text: 1,772,312
+ * `ad:Address`, 55,766 `ad:ThoroughfareName`, 1,963 `ad:AddressAreaName`, 598 `ad:PostalDescriptor`
+ * and 264 `ad:AdminUnitName`. The two instruments agree, and they reconcile: 1,772,312 addresses plus
+ * 58,591 component features is the 1,830,903 `gml:featureMember` elements the file holds, and the
+ * component total is the number of distinct component `gml:id` values. The address count is also the
+ * one the address-source register records for `be-property-building-2`.
+ *
+ * Every component `gml:id` is distinct. Over 58,591 component features, 0 ids repeat, so the index
+ * below states no collision rule and raises on a collision instead. A replaced entry would
+ * substitute one street for another on every address referencing it.
+ *
+ * ## Why the join takes two passes
+ *
+ * Wallonia writes all 1,772,312 addresses before the first component feature. Measured by resolving
+ * each reference against the components indexed so far, all 10,633,872 references are forward
+ * references: 0 resolve in a single pass. ČÚZK's one-pass reader with a deferred tail cannot be used
+ * here, because deferring every address means holding 1.77 million subtrees in memory.
+ *
+ * So the member is inflated twice. The first pass indexes the four component types — 58,591 entries,
+ * bounded and small — and the second streams the addresses and joins locally. Two inflations of
+ * 5,163,917,131 bytes cost less than one buffered document.
+ *
+ * Each address writes exactly 6 `ad:component` references, 10,633,872 over the file, and each one
+ * addresses a `gml:id` in the same document. Four are the region's own
+ * (`BE.WL.ICAR.ThoroughfareName`, `AdminUnitName`, `PostalDescriptor`, `AddressAreaName`) and two are
+ * the country and the region, `#BE.AdminUnitName.1000` (`Belgique`, `1stOrder`) and
+ * `#BE.AdminUnitName.3000` (`Région wallonne`, `2ndOrder`). The index keeps an administrative unit
+ * only at {@link MUNICIPALITY_LEVEL}, so those two resolve to no entry and are a reference the
+ * adapter declines rather than a defect — the same shape Flanders' `refnis1995_1000` reference has.
+ *
+ * ## The number extension is the box number, rendered as `unit`
+ *
+ * Every address writes two inline designators: `addressIdentifierGeneral` holds the house
+ * number and `addressNumberExtension` holds a four-character code. `ad:level` is
+ * `postalDeliveryPoint` on every address, so a row is a letterbox rather than a building, and the
+ * extension distinguishes letterboxes at one number: `93` appears with `B001` and `C003`, and `29`
+ * with `0014`, `B004`, `B007`, `C009` and `D015`. Other observed values are `0001`, `A001`, `008B`,
+ * `BP3` and `0RCH`.
+ *
+ * The element is written on all 1,772,312 addresses and holds a value on 347,859 of them, measured by
+ * running this adapter over the whole archive. The other 1,424,453, 80.4%, write
+ * `addressNumberExtension` with its `ad:designator` empty. That empty element is the publisher
+ * stating that the address has no extension rather than omitting the field.
+ *
+ * ICAR, the register this publication is drawn from, models a sub-address below the police number
+ * and states what identifies it: the SPW's ICAR presentation to the Club des utilisateurs du PICC
+ * (27 February 2015) lists the communes' validation of sub-addresses identified by their
+ * `numéro de boîte`, the box number, and its ICAR address form records each sub-address's `Numéro` apart from the `Index`
+ * field. A box number identifies one delivery point inside the numbered premise, which is the
+ * concept the `unit` tag carries. It is not a floor, a building, an entrance or a staircase, even
+ * where a code such as `0RCH` (rez-de-chaussée) names the box after the floor it sits on.
+ *
+ * The value is rendered as the publisher writes it, with no designator word: `Rue de la Station 29
+ * B004, 1420 Braine-l'Alleud`. Wallonia publishes no formatted line to copy a designator from, so a
+ * word such as `bte` would be a surface this source does not state. Two letterboxes at one number
+ * therefore render two lines, and deduplication keeps both.
+ *
+ * ## Voids
+ *
+ * Every void in this file states a reason. Over the whole member, 1,830,903 `ad:validTo` and
+ * 1,830,903 `ad:endLifespanVersion` elements, 57,997 `gn:pronunciation`, `gn:grammaticalGender` and
+ * `gn:grammaticalNumber` each, 2,823 `ad:validFrom` and 264 `ad:adminUnit` are `xsi:nil="true"` with
+ * a `nilReason`, and 0 nil elements omit the reason. No value this adapter reads was void in the file
+ * measured. Brussels is the opposite case, writing no reason on any void, and `isVoid` in
+ * `#inspire/address` keys on `xsi:nil` rather than on a reason for that reason. This adapter reads
+ * the same primitives, so a Walloon void that stopped stating a reason would still read as a void.
+ *
+ * ## Coordinates
+ *
+ * This adapter emits text and components only, because {@linkcode CanonicalRow} declares no position
+ * field. The file's axis order is recorded here for whoever does emit one. Wallonia declares
+ * `srsName="http://www.opengis.net/def/crs/EPSG/0/3035"` and complies with that authority's
+ * northing-then-easting order: `gml:pos` `3079047.229041581 3924386.42904076` reads as latitude
+ * 50.686064, longitude 4.381708, in Braine-l'Alleud. That is the municipality the same address's own
+ * `ad:AdminUnitName` names. Its `gml:boundedBy` lower corner, `2942208.441255148 3816649.462808032`,
+ * puts the smaller number first in the same way. Brussels declares the same CRS and writes
+ * easting-northing, so the two publishers disagree and a reader cannot take the declaration at its
+ * word. `brussels/adapter.ts` records that measurement.
+ *
+ * ## Terms
+ *
+ * The address-source register elects `Licence CC-BY 4.0, as the Service public de Wallonie states it`
+ * for `be-property-building-2`, under license id
+ * `unchecked-access-free-be-service-public-de-wallonie-spw`, and that decision records
+ * `spdx: "CC-BY-4.0"`. `electedLicenseLabel` answers an elected decision's `spdx` before its
+ * `electedTerms`, so {@linkcode WALLONIE_DEFAULT_LICENSE} is the SPDX id rather than the French
+ * sentence. The register's own words for what the publisher permits are use, modification into a
+ * derived work, and publication of both, on condition that the sources are cited, so a model card
+ * states `Source : Service public de Wallonie (SPW) - INSPIRE - Points d'adresses en Wallonie (BE)
+ * (2017-11-20) https://geodata.wallonie.be/id/b62f8405-2ce3-449f-8a61-5fc2f38d8b73`.
+ *
+ * That row declares no `adapterID`, so `createIneligibilityReader` refuses every row this adapter
+ * emits until a review sets `adapterID` to `wallonie` on it. The refusal is by construction and
+ * states itself in the build manifest.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { streamMarkupElements } from "@mailwoman/core/html/elements";
+import { stringifyJSON } from "@mailwoman/core/json";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { stableSourceID } from "#adapters/source-id";
+import { adminUnitLevel, componentHrefs, componentJoinKey, designator, designatorsByType, placeName, postalDescriptorCode, thoroughfareName, voidDesignatorTypes, } from "#inspire/address";
+import { inspireGMLChunks } from "#inspire/archive";
+import { InspireArchiveError } from "#inspire/errors";
+import { SourceRegister } from "#registers";
+import { AddressRole, countDropped, SurfaceOrigin, } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ * The name is the region as its own publisher writes it, as `vlaanderen`'s is.
+ */
+export const WALLONIE_ADAPTER_ID = "wallonie";
+/**
+ * The stable id of the publication behind every row.
+ *
+ * It belongs in `SourceRegister` beside `be-vlaanderen-inspire-ad`, and this constant
+ * stands in until that entry lands, because a built corpus stores the id on every row
+ * and the id must not change afterwards.
+ * Replace the constant with the `SourceRegister` member on integration.
+ */
+export const WALLONIE_SOURCE_REGISTER = SourceRegister.WallonieInspireAddresses;
+/**
+ * The one jurisdiction this adapter emits.
+ *
+ * The service publishes the Walloon Region, to which ISO 3166-1 assigns no
+ * code of its own, so the set holds one code and a caller naming another gets
+ * {@linkcode UnsupportedCountryError} rather than zero rows.
+ */
+export const WALLONIE_COUNTRIES = ["BE"];
+/**
+ * The license recorded on every row.
+ *
+ * The register's elected decision for `unchecked-access-free-be-service-public-de-wallonie-spw`
+ * records `spdx: "CC-BY-4.0"`, and `electedLicenseLabel` answers the SPDX id when one is recorded.
+ */
+export const WALLONIE_DEFAULT_LICENSE = "CC-BY-4.0";
+/**
+ * The archive member holding the GML.
+ *
+ * A pattern rather than the literal `AD.Addresses.gml`, so a republication
+ * that renames the member still reads.
+ * The archive measured holds exactly one member.
+ */
+const GML_MEMBER = /\.gml$/iu;
+/**
+ * The designator type Wallonia writes its house number under.
+ *
+ * `ad:LocatorDesignator` repeats and is typed, so the number is selected by this term
+ * rather than by position.
+ * Czechia and Brussels write `buildingIdentifier` and the Netherlands `addressNumber` for the same part.
+ *
+ * `#inspire/address` therefore leaves the vocabulary to the caller.
+ */
+const HOUSE_NUMBER_DESIGNATOR = "addressIdentifierGeneral";
+/**
+ * The designator type carrying the box number beside the house number.
+ *
+ * Emitted as `unit`, for the reason this file's header states.
+ */
+const NUMBER_EXTENSION_DESIGNATOR = "addressNumberExtension";
+/**
+ * The `ad:level` of the administrative unit carrying the municipality name.
+ *
+ * An address references three administrative units: the municipality at `5thOrder`,
+ * the region at `2ndOrder` (`Région wallonne`) and the country at `1stOrder` (`Belgique`).
+ * A reader that took whichever arrived first would write `Belgique` into the `locality` of some rows.
+ */
+const MUNICIPALITY_LEVEL = "5thOrder";
+/**
+ * The component feature types the first pass indexes.
+ *
+ * All four are read in one pass, because the publisher writes them interleaved
+ * in the file's tail rather than grouped by type.
+ */
+const COMPONENT_TYPES = [
+    "ad:ThoroughfareName",
+    "ad:PostalDescriptor",
+    "ad:AdminUnitName",
+    "ad:AddressAreaName",
+];
+/**
+ * Indexes the four component types in one pass over the member.
+ *
+ * @throws {@linkcode InspireArchiveError} when a component states no `gml:id`,
+ * when two components share one, or when the pass ends with an empty index.
+ */
+export async function readWallonieComponents(inputPath, signal) {
+    const streets = new Map();
+    const postcodes = new Map();
+    const municipalities = new Map();
+    const areas = new Map();
+    const seen = new Set();
+    for await (const feature of streamMarkupElements(inspireGMLChunks(inputPath, GML_MEMBER), COMPONENT_TYPES, {
+        xml: true,
+        signal,
+    })) {
+        const id = feature.attributes["gml:id"];
+        if (!id) {
+            throw new InspireArchiveError(WALLONIE_ADAPTER_ID, `a ${feature.name} feature carries no gml:id, so no address can reference it`);
+        }
+        // Every one of 58,591 component ids measured is distinct.
+        // An overwritten entry would substitute one street for another on every address naming it.
+        if (seen.has(id)) {
+            throw new InspireArchiveError(WALLONIE_ADAPTER_ID, `two component features share the gml:id ${stringifyJSON(id)}. All 58,591 ids in the file ` +
+                `measured are distinct, so a reference can no longer be resolved to one feature.`);
+        }
+        seen.add(id);
+        if (feature.name === "ad:ThoroughfareName") {
+            const street = thoroughfareName(feature);
+            if (street !== undefined) {
+                streets.set(id, street);
+            }
+        }
+        else if (feature.name === "ad:PostalDescriptor") {
+            const postcode = postalDescriptorCode(feature);
+            if (postcode !== undefined) {
+                postcodes.set(id, postcode);
+            }
+        }
+        else if (feature.name === "ad:AddressAreaName") {
+            const area = placeName(feature);
+            if (area !== undefined) {
+                areas.set(id, area);
+            }
+        }
+        else if (adminUnitLevel(feature) === MUNICIPALITY_LEVEL) {
+            const municipality = placeName(feature);
+            if (municipality !== undefined) {
+                municipalities.set(id, municipality);
+            }
+        }
+    }
+    // An aborted pass stopped mid-member, so an empty index states that the caller cancelled
+    // rather than that the publication is incomplete.
+    if (signal?.aborted)
+        return { streets, postcodes, municipalities, areas };
+    for (const [typeName, index] of [
+        ["ad:ThoroughfareName", streets],
+        ["ad:PostalDescriptor", postcodes],
+        ["ad:AdminUnitName", municipalities],
+        ["ad:AddressAreaName", areas],
+    ]) {
+        if (!index.size) {
+            throw new InspireArchiveError(WALLONIE_ADAPTER_ID, `the archive holds no ${typeName} feature this adapter can join. Every address references all ` +
+                `four component types, so an empty index turns every row into a row with no street rather ` +
+                `than into fewer rows.`);
+        }
+    }
+    return { streets, postcodes, municipalities, areas };
+}
+/**
+ * The reasons the adapter refuses an address, as the keys it increments in `dropped`.
+ */
+const Refusal = {
+    /**
+     * The publisher stated no house number, having written no `addressIdentifierGeneral` designator.
+     */
+    AbsentHouseNumber: "row:house-number-absent",
+    /**
+     * The publisher marked the house number void, so the value could not be read.
+     */
+    VoidHouseNumber: "row:house-number-void",
+    /**
+     * A reference the component index cannot answer, or a component the address never referenced.
+     */
+    Unjoined: "row:component-unjoined",
+    /**
+     * An address the codex could not render into a line.
+     */
+    Unrendered: "row:render-failed",
+};
+export function createWallonieAdapter() {
+    return {
+        id: WALLONIE_ADAPTER_ID,
+        defaultLicense: WALLONIE_DEFAULT_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: WALLONIE_SOURCE_REGISTER,
+        surface: SurfaceOrigin.Rendered,
+        description: "INSPIRE Addresses (Service public de Wallonie): house-number-level addresses for the Walloon Region.",
+        async *rows(opts) {
+            if (opts.country && !WALLONIE_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(WALLONIE_ADAPTER_ID, WALLONIE_COUNTRIES, opts.country);
+            }
+            // Checked before the archive is opened, so a caller that cancelled first pays no inflation.
+            if (opts.signal?.aborted)
+                return;
+            const index = await readWallonieComponents(opts.inputPath, opts.signal);
+            if (opts.signal?.aborted)
+                return;
+            let emitted = 0;
+            let extensions = 0;
+            // A caller that passes no map still gets the stderr summary below, so the adapter keeps its own.
+            const tally = { dropped: opts.dropped ?? new Map() };
+            try {
+                for await (const address of streamMarkupElements(inspireGMLChunks(opts.inputPath, GML_MEMBER), "ad:Address", {
+                    xml: true,
+                    signal: opts.signal,
+                })) {
+                    if (opts.signal?.aborted)
+                        break;
+                    if (opts.limit !== undefined && emitted >= opts.limit)
+                        break;
+                    // `designatorsByType` leaves out a designator the publisher marked void, so a void number
+                    // and an absent one both read as undefined and `voidDesignatorTypes` separates them below.
+                    const byType = designatorsByType(address);
+                    const house = designator(byType, HOUSE_NUMBER_DESIGNATOR);
+                    if (house === undefined) {
+                        // A void number is a value this run could not read and an absent one is
+                        // a number the publisher states the address has none of.
+                        // Neither is emitted, and the two are counted apart so a run reports which it met.
+                        if (voidDesignatorTypes(address).has(HOUSE_NUMBER_DESIGNATOR)) {
+                            countDropped(tally, Refusal.VoidHouseNumber);
+                        }
+                        else {
+                            countDropped(tally, Refusal.AbsentHouseNumber);
+                        }
+                        continue;
+                    }
+                    const box = designator(byType, NUMBER_EXTENSION_DESIGNATOR);
+                    if (box !== undefined) {
+                        extensions++;
+                    }
+                    let street;
+                    let postcode;
+                    let locality;
+                    let area;
+                    for (const href of componentHrefs(address)) {
+                        const key = componentJoinKey(href);
+                        if (key === undefined)
+                            continue;
+                        street = index.streets.get(key) ?? street;
+                        postcode = index.postcodes.get(key) ?? postcode;
+                        // The country and region references are indexed out by level,
+                        // so they reach here with no municipality and are not defects.
+                        locality = index.municipalities.get(key) ?? locality;
+                        area = index.areas.get(key) ?? area;
+                    }
+                    // A component the index cannot answer is a street, postcode or locality this run could not read.
+                    // A row emitted without it would record an absence the publisher never stated.
+                    if (street === undefined || postcode === undefined || locality === undefined) {
+                        countDropped(tally, Refusal.Unjoined);
+                        continue;
+                    }
+                    const components = {
+                        house_number: house,
+                        street,
+                        postcode,
+                        locality,
+                    };
+                    if (box !== undefined) {
+                        components.unit = box;
+                    }
+                    // The `partie de commune` is a dependent locality only where it names
+                    // something other than the municipality.
+                    // `Braine-l'Alleud` has an area of the same name, and `Beauvechain` has `Hamme-Mille`.
+                    if (area !== undefined && area !== locality) {
+                        components.dependent_locality = area;
+                    }
+                    const rendered = formatAddressRow(components, "BE", { singleLine: true });
+                    if (!rendered) {
+                        countDropped(tally, Refusal.Unrendered);
+                        continue;
+                    }
+                    const { raw, components: aligned } = rendered;
+                    // `gml:id` is `BE.WL.ICAR.Address.<localId>`, unique in the publication and written
+                    // in the file rather than derived, so two runs over one archive agree on it.
+                    const seed = address.attributes["gml:id"]?.trim();
+                    yield {
+                        raw,
+                        components: aligned,
+                        country: "BE",
+                        // Every component name is read in French: `gn:language` is `fre` on the first
+                        // `ad:name` of each feature, and the country and region features add `ger`
+                        // and `dut` spellings after it that `placeName` does not reach.
+                        locale: "fr-BE",
+                        source: WALLONIE_ADAPTER_ID,
+                        source_id: seed ? `${WALLONIE_ADAPTER_ID}-${seed}` : stableSourceID(WALLONIE_ADAPTER_ID, aligned),
+                        corpus_version: "",
+                        license: WALLONIE_DEFAULT_LICENSE,
+                    };
+                    emitted++;
+                }
+            }
+            finally {
+                const count = (reason) => tally.dropped.get(reason) ?? 0;
+                const dropped = Object.values(Refusal).reduce((sum, reason) => sum + count(reason), 0);
+                if (dropped > 0) {
+                    process.stderr.write(`  wallonie: ${emitted} rows kept, ${dropped} refused ` +
+                        `(${count(Refusal.Unjoined)} hold a reference the archive cannot answer, ` +
+                        `${count(Refusal.VoidHouseNumber)} state a void house number, ` +
+                        `${count(Refusal.AbsentHouseNumber)} state no house number, ` +
+                        `${count(Refusal.Unrendered)} did not render)\n`);
+                }
+                if (extensions > 0) {
+                    process.stderr.write(`  wallonie: ${extensions} of ${emitted} rows kept carry a box number as unit\n`);
+                }
+            }
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const wallonieAdapter = createWallonieAdapter();
+//# sourceMappingURL=adapter.js.map