diff --git a/CHANGELOG.md b/CHANGELOG.md
index 5c3eb56..a8020d3 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -4,11 +4,13 @@
 
 ### Added
 
+- Krill `externalLink` is now derived for sources where the full-text link is not explicitly encoded as a `ref[@type=page_url]`/`link` element, mirroring the predecessor tool (KorAP-XML-Krill): DGD/AGD/FOLK transcripts (title `_DF_<n>` suffix stripped, title "DGD"), Süddeutsche Zeitung (`U<yy>` sigles → szarchiv link from the biblNote `ID:`, title "Süddeutsche Zeitung"), and Genios newspapers (corpus sigle → `orikuerzel` via a built-in mapping + biblNote `ID:` → genios.de link, title "GENIOS"). Explicitly encoded links still take precedence. Link titles are carried via a new `externalLinkTitle` helper field.
 - Sentence segmentation fallback: when a corpus has no `s` spans in `structure.xml` (common for TEI conversions of custom corpora), sentence boundaries now fall back to other segment-like TEI elements, in order of preference: `posting` (chat/CMC), `l` (verse line), `seg` (segment), `u` (utterance). This makes integrated taggers/parsers and KorAP sentence-based queries work on such corpora.
 - New `dck_sample.zip` test resource: two-text excerpt from the CC BY licensed Dortmunder Chat-Korpus with custom `cmc` tokenization/annotations (in one text the `s` spans are removed to exercise the sentence fallback)
 
 ### Fixed
 
+- Krill `externalLink` is now picked up from older Wikipedia corpora (wpd17/wdd17) where the article URL is only present in the `reference[type=complete]` text rather than encoded as a `ref`/`link` element ([#47](https://github.com/KorAP/korapxmltool/issues/47)). The URL is extracted with title "Wikipedia".
 - `tokenSource` in Krill JSON output no longer resolves to a stand-off annotation foundry ([#48](https://github.com/KorAP/korapxmltool/issues/48)): token source identification and token-map updates are now restricted to base archive processing, so stand-off annotation foundries can no longer overwrite the base tokenization spans or steal the `tokenSource`.
 - Krill output now drops texts that contain no tokens instead of emitting empty, unindexable documents ([#46](https://github.com/KorAP/korapxmltool/issues/46)). Such texts (e.g. articles with empty `data.xml`/`base/tokens.xml`) are skipped with a per-text warning and a summary count; texts with at least one token are kept. New `m21_empty_sample.zip` regression fixture (one empty text, one single-token text).
 - Corpora with custom tokenization and annotations inside the base ZIP (e.g. `cmc/morpho.xml` from TEI conversions, with no `base/tokens.xml`) are now handled correctly: the foundry is derived from the annotation folder name instead of the ZIP file name, so Krill output indexes the annotations (e.g. `cmc/p`, `cmc/l`) instead of silently dropping them, the token stream is no longer empty (`tokenSource` is set to e.g. `cmc#morpho`), and CoNLL-U output reports `# foundry = cmc` instead of `# foundry = base`
diff --git a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolver.kt b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolver.kt
new file mode 100644
index 0000000..e89baac
--- /dev/null
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolver.kt
@@ -0,0 +1,77 @@
+package de.ids_mannheim.korapxmltools
+
+/**
+ * Derives a text's external full-text link for Krill output in cases where the link is not
+ * explicitly encoded as a `ref[@type=page_url]`/`link` element, mirroring the behaviour of the
+ * predecessor tool (KorAP-XML-Krill, KorAP::XML::Meta::I5):
+ *
+ *  - Wikipedia: the URL is only present in the `reference[type=complete]` text.
+ *  - DGD/AGD/FOLK: a DGD access link is built from the corpus sigle and the transcript title.
+ *  - Süddeutsche Zeitung (sigle `U<yy>`): an szarchiv link is built from the biblNote `ID:`.
+ *  - Genios newspapers: a genios.de link is built from the corpus sigle (via [GeniosKuerzelMap])
+ *    and the biblNote `ID:`.
+ *
+ * All rules are pure functions of the header values so they can be unit-tested without ZIP
+ * fixtures. [resolve] applies them in priority order; the first match wins.
+ */
+object ExternalLinkResolver {
+    data class ResolvedLink(val url: String, val title: String)
+
+    // Wikipedia references end in "... URL:<url>: Wikipedia, <year>" (http or https).
+    private val WIKIPEDIA_REFERENCE = Regex("""URL:(https?:.+?):\s+Wikipedia,\s+\d+\s*$""")
+    // biblNote carries the source id as "ID: <id> ..." (id may itself start with a letter, e.g. SZ).
+    private val BIBL_NOTE_ID = Regex("""ID:\s*(\S+)""")
+    // Local newspaper corpus sigles always look like [A-Z]{1,3}[0-9]{2}; the two digits are the
+    // year volume. Genios detection requires this shape so arbitrary sigles can't trigger a link.
+    private val NEWSPAPER_SIGLE = Regex("""^[A-Z]{1,3}\d{2}$""")
+    private val SUEDDEUTSCHE_SIGLE = Regex("""^U\d{2}$""")
+    private val DGD_SIGLE = Regex("""^(?:[AD]GD|FOLK)$""")
+    private val DGD_TRANSCRIPT_SUFFIX = Regex("""_DF_\d+$""", RegexOption.IGNORE_CASE)
+
+    /** Extract the `ID:` value from a biblNote text, or null if absent. */
+    fun biblNoteId(text: String?): String? =
+        text?.let { BIBL_NOTE_ID.find(it)?.groupValues?.get(1) }
+
+    /** The Genios botkuerzel for a corpus sigle: its leading letters, lowercased (e.g. W24 -> "w"). */
+    private fun botkuerzel(corpusSigle: String?): String? =
+        corpusSigle?.takeWhile { it.isLetter() }?.lowercase()?.takeIf { it.isNotEmpty() }
+
+    fun wikipedia(reference: String?): ResolvedLink? =
+        reference?.let { WIKIPEDIA_REFERENCE.find(it) }
+            ?.let { ResolvedLink(it.groupValues[1], "Wikipedia") }
+
+    fun dgd(corpusSigle: String?, title: String?): ResolvedLink? {
+        if (corpusSigle == null || !DGD_SIGLE.matches(corpusSigle)) return null
+        val transcript = title?.trim()?.takeIf { it.isNotEmpty() }
+            ?.replace(DGD_TRANSCRIPT_SUFFIX, "") ?: return null
+        return ResolvedLink(
+            "https://dgd.ids-mannheim.de/DGD2Web/ExternalAccessServlet?command=displayData&id=$transcript",
+            "DGD"
+        )
+    }
+
+    fun sueddeutsche(corpusSigle: String?, biblNoteId: String?): ResolvedLink? {
+        if (corpusSigle == null || !SUEDDEUTSCHE_SIGLE.matches(corpusSigle) || biblNoteId == null) return null
+        return ResolvedLink(
+            "https://archiv.szarchiv.de/Portal/restricted/Start.act?articleId=$biblNoteId",
+            "Süddeutsche Zeitung"
+        )
+    }
+
+    fun genios(corpusSigle: String?, biblNoteId: String?): ResolvedLink? {
+        if (corpusSigle == null || biblNoteId == null || !NEWSPAPER_SIGLE.matches(corpusSigle)) return null
+        val orikuerzel = botkuerzel(corpusSigle)?.let { GeniosKuerzelMap.byBotkuerzel[it] } ?: return null
+        return ResolvedLink("https://www.genios.de/document/${orikuerzel}__$biblNoteId", "GENIOS")
+    }
+
+    /**
+     * Try every derivation rule in priority order and return the first match. Wikipedia is
+     * content-detected and wins over the sigle-based rules; Süddeutsche is checked before Genios
+     * so `U<yy>` sigles resolve to szarchiv rather than a colliding Genios botkuerzel.
+     */
+    fun resolve(corpusSigle: String?, reference: String?, biblNoteId: String?, title: String?): ResolvedLink? =
+        wikipedia(reference)
+            ?: dgd(corpusSigle, title)
+            ?: sueddeutsche(corpusSigle, biblNoteId)
+            ?: genios(corpusSigle, biblNoteId)
+}
diff --git a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/GeniosKuerzelMap.kt b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/GeniosKuerzelMap.kt
new file mode 100644
index 0000000..714cace
--- /dev/null
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/GeniosKuerzelMap.kt
@@ -0,0 +1,320 @@
+// GENERATED FILE - DO NOT EDIT BY HAND.
+// Genios newspaper short-code (botkuerzel) -> Genios document prefix (orikuerzel) mapping,
+// generated from the IDS Genios metadata.xml. Used to build genios.de full-text links for
+// DeReKo corpora that originate from Genios (see ExternalLinkResolver.genios).
+// To regenerate, re-run the extraction over an updated metadata.xml.
+package de.ids_mannheim.korapxmltools
+
+object GeniosKuerzelMap {
+    // Keyed by the lowercased alphabetic prefix of a corpus sigle (the botkuerzel).
+    val byBotkuerzel: Map<String, String> = mapOf(
+        "a" to "STG",
+        "aan" to "AAN",
+        "aaz" to "AAZ",
+        "abm" to "ALBB",
+        "abo" to "AARB",
+        "aez" to "AEZT",
+        "afz" to "AFZ",
+        "agf" to "AGEF",
+        "agz" to "ED",
+        "aho" to "AUON",
+        "ait" to "AUTI",
+        "alz" to "ALLZ",
+        "art" to "GART",
+        "aue" to "AUEL",
+        "auf" to "AF",
+        "auh" to "AUTH",
+        "aui" to "AUIN",
+        "aup" to "AUPR",
+        "aut" to "AUTO",
+        "azm" to "MUAZ",
+        "b" to "BEZE",
+        "baz" to "BAZ",
+        "bdw" to "BIWI",
+        "bdz" to "BADZ",
+        "bee" to "BEEF",
+        "bei" to "BE",
+        "beo" to "BEOB",
+        "bez" to "BERN",
+        "bil" to "ABIL",
+        "bkr" to "BKR",
+        "bku" to "BKU",
+        "bla" to "BLAU",
+        "bli" to "BLI",
+        "bmo" to "BGM",
+        "bna" to "BNA",
+        "bot" to "BVH",
+        "boz" to "BOEZ",
+        "brg" to "BRIG",
+        "brm" to "BRIM",
+        "bru" to "BR",
+        "brw" to "BRIW",
+        "bsz" to "BSTZ",
+        "bue" to "BUER",
+        "bun" to "BUND",
+        "bup" to "BUPU",
+        "bwa" to "BWAI",
+        "bze" to "BZ",
+        "bzg" to "BEZG",
+        "cap" to "CAPI",
+        "cav" to "CAV",
+        "chk" to "CHEK",
+        "chm" to "CHMM",
+        "cho" to "CHMO",
+        "cht" to "CHET",
+        "cic" to "CICE",
+        "cid" to "CID",
+        "cio" to "CIOD",
+        "cit" to "CITL",
+        "cou" to "COU",
+        "ct" to "CT",
+        "ctb" to "COBU",
+        "dae" to "DAE",
+        "dak" to "DIAK",
+        "daz" to "DAZ",
+        "dbt" to "MCDB",
+        "dec" to "DECH",
+        "dgl" to "DGL",
+        "dib" to "DIBA",
+        "dki" to "DIKI",
+        "dnn" to "DNN",
+        "dnv" to "DNV",
+        "dog" to "DOGS",
+        "dol" to "DOL",
+        "dpr" to "PRIG",
+        "dsz" to "DSTZ",
+        "dtz" to "DTZ",
+        "dvz" to "DVZ",
+        "dww" to "DWW",
+        "e" to "TAG",
+        "edf" to "EDF",
+        "eft" to "ETB",
+        "eid" to "EID",
+        "ein" to "EVIN",
+        "elf" to "ELTF",
+        "eli" to "EIND",
+        "elo" to "ELEO",
+        "elt" to "ELTN",
+        "ema" to "EMA",
+        "emt" to "EMT",
+        "ene" to "ENER",
+        "epp" to "EPP",
+        "epr" to "EPRA",
+        "et" to "ETEC",
+        "ett" to "EUTD",
+        "eut" to "EUT",
+        "euw" to "EUZW",
+        "eze" to "EZE",
+        "fis" to "FRIM",
+        "fk" to "FK",
+        "flt" to "FALT",
+        "fmt" to "FORM",
+        "fnp" to "FNP",
+        "foc" to "FOCU",
+        "fom" to "FOCM",
+        "fpc" to "FEPR",
+        "fra" to "FRA",
+        "frt" to "FRT",
+        "fuw" to "FLOW",
+        "gal" to "GALA",
+        "gaz" to "GAZ",
+        "geo" to "GEO",
+        "ges" to "GEOS",
+        "ggt" to "GG",
+        "giz" to "GIAN",
+        "gng" to "GONG",
+        "gob" to "DAGO",
+        "gsp" to "GESP",
+        "gta" to "GTB",
+        "gtb" to "GETA",
+        "gwp" to "GWP",
+        "haa" to "HA",
+        "hab" to "HB",
+        "hat" to "HATA",
+        "hau" to "HAEU",
+        "hdz" to "HZ",
+        "hfz" to "HOFZ",
+        "hgz" to "AHGZ",
+        "hhz" to "HOZE",
+        "hkr" to "HK",
+        "hna" to "HNA",
+        "hrz" to "HOER",
+        "hst" to "HST",
+        "htb" to "HOT",
+        "hzs" to "HERZ",
+        "hzw" to "HZWI",
+        "hzz" to "HOZG",
+        "i" to "TITA",
+        "iee" to "IEE",
+        "imw" to "IMWI",
+        "imz" to "IMMO",
+        "itb" to "ITB",
+        "itt" to "ITT",
+        "ix" to "IX",
+        "jue" to "JUEZ",
+        "k" to "KLEI",
+        "kaz" to "KRAN",
+        "ke" to "KE",
+        "kem" to "KEM",
+        "kfz" to "KFZB",
+        "khh" to "KH",
+        "kir" to "KIRZ",
+        "kn" to "KN",
+        "kru" to "KR",
+        "ksa" to "KSTA",
+        "ktz" to "KTZ",
+        "kur" to "KUR",
+        "kxp" to "EXPR",
+        "l" to "BMP",
+        "lab" to "LB",
+        "lah" to "LIAH",
+        "lan" to "LAAN",
+        "laz" to "LAZE",
+        "lhz" to "LAZ",
+        "lit" to "LIN",
+        "lmd" to "LEMO",
+        "lmz" to "LMZ",
+        "ln" to "LBN",
+        "log" to "HLOG",
+        "lru" to "LR",
+        "ltb" to "LUXT",
+        "lvz" to "LVZ",
+        "mag" to "MAER",
+        "mav" to "MAV",
+        "maz" to "MAZ",
+        "mdr" to "MDR",
+        "mep" to "MEP",
+        "met" to "MUTE",
+        "mib" to "MIB",
+        "mid" to "MID",
+        "mm" to "MM",
+        "mme" to "MUME",
+        "mpo" to "MPW",
+        "msp" to "MASP",
+        "mt" to "MT",
+        "mtk" to "MATK",
+        "mut" to "MUAR",
+        "muv" to "MUVB",
+        "mwo" to "MAC",
+        "mze" to "MZ",
+        "n" to "SN",
+        "nas" to "NECH",
+        "nbk" to "NBK",
+        "ndo" to "NIDO",
+        "neo" to "NEON",
+        "neu" to "NEUL",
+        "new" to "ANEW",
+        "ngz" to "NGVZ",
+        "nku" to "NKU",
+        "nlz" to "NLZ",
+        "nnn" to "NNN",
+        "nnp" to "NNP",
+        "now" to "NOW",
+        "noz" to "NOZ",
+        "npr" to "NPHA",
+        "nun" to "NN",
+        "nuz" to "NZ",
+        "nvb" to "NVB",
+        "nvt" to "NVT",
+        "nwe" to "NEUW",
+        "nws" to "AGZ",
+        "nwt" to "NEWE",
+        "nwz" to "NWZ",
+        "nzf" to "FOLI",
+        "nzs" to "NZZS",
+        "nzz" to "NZZ",
+        "o" to "KRON",
+        "oaz" to "OAZ",
+        "ohz" to "OHEZ",
+        "osz" to "OSZ",
+        "otz" to "OTZ",
+        "ovz" to "OVZ",
+        "p" to "PRE",
+        "paz" to "PAZ",
+        "pco" to "PCW",
+        "pha" to "PHAP",
+        "pmm" to "GJPM",
+        "pnn" to "PNN",
+        "pnp" to "PNP",
+        "prd" to "PRTR",
+        "prf" to "PROF",
+        "qe" to "QE",
+        "r" to "FR",
+        "rbs" to "RBS",
+        "rga" to "GEA",
+        "rhz" to "RZTG",
+        "rlh" to "RLH",
+        "rln" to "RLNR",
+        "rn" to "RN",
+        "rpo" to "RP",
+        "rsw" to "NKR",
+        "rue" to "RELO",
+        "rvz" to "RVZ",
+        "saz" to "SAZE",
+        "sbl" to "SBLI",
+        "sbz" to "SAAR",
+        "sch" to "SHF",
+        "sct" to "SWT",
+        "scv" to "SVER",
+        "scw" to "SOWO",
+        "scz" to "SWAZ",
+        "sku" to "SK",
+        "smp" to "SOMP",
+        "soa" to "STP",
+        "spz" to "DSZ",
+        "sta" to "STA",
+        "stb" to "STAG",
+        "ste" to "STER",
+        "stg" to "STGL",
+        "stn" to "STN",
+        "stz" to "STZ",
+        "svz" to "SVZ",
+        "swp" to "SWP",
+        "sze" to "SZO",
+        "t" to "TAZ",
+        "tai" to "TAI",
+        "tas" to "TAS",
+        "tbz" to "TAB",
+        "ter" to "TERE",
+        "tet" to "TETE",
+        "tew" to "TWA",
+        "tha" to "TA",
+        "thb" to "THBE",
+        "tlz" to "TLZ",
+        "tnz" to "TZH",
+        "toz" to "TZ",
+        "tre" to "TRE",
+        "tru" to "TRUC",
+        "tsg" to "TSPG",
+        "tsp" to "TSP",
+        "tue" to "TUE",
+        "tvd" to "TVDT",
+        "tvf" to "TV",
+        "uan" to "USAN",
+        "v" to "VN",
+        "vbw" to "VBW",
+        "vku" to "VKU",
+        "vra" to "VRAG",
+        "vru" to "VR",
+        "vsw" to "VW",
+        "vzs" to "VZS",
+        "w" to "WELT",
+        "was" to "WAMS",
+        "wbl" to "WB",
+        "weo" to "WEON",
+        "wez" to "WEST",
+        "wfb" to "WFB",
+        "wiz" to "WZ",
+        "wku" to "WK",
+        "wor" to "WORZ",
+        "wtb" to "WITA",
+        "wwo" to "WEWO",
+        "wwt" to "WWT",
+        "x" to "OOEN",
+        "z" to "ZEIT",
+        "zca" to "ZTCS",
+        "zcw" to "CUW",
+        "zge" to "ZTGS",
+        "zwi" to "ZTWI",
+    )
+}
diff --git a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXmlTool.kt b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXmlTool.kt
index b96ffbd..f339440 100644
--- a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXmlTool.kt
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/KorapXmlTool.kt
@@ -5664,6 +5664,25 @@
                 }
         }
 
+        // Derive the external link for cases where it is not explicitly encoded (Wikipedia URL in
+        // the reference text, DGD/FOLK transcripts, Süddeutsche and Genios newspapers). See
+        // ExternalLinkResolver. Explicitly encoded links (above) always take precedence.
+        if (!metadata.containsKey("externalLink")) {
+            val corpusSigle = headerRoot.firstText("textSigle")?.substringBefore("/")
+            val biblNoteId = headerRoot.childElements("biblNote")
+                .mapNotNull { ExternalLinkResolver.biblNoteId(it.textContent) }
+                .firstOrNull()
+            ExternalLinkResolver.resolve(
+                corpusSigle = corpusSigle,
+                reference = metadata["reference"] as? String,
+                biblNoteId = biblNoteId,
+                title = metadata["title"] as? String
+            )?.let { resolved ->
+                metadata["externalLink"] = resolved.url
+                metadata["externalLinkTitle"] = resolved.title
+            }
+        }
+
         val biblNoteElement = analytic.firstElement("biblNote") { it.getAttribute("n") == "url" }
             ?: monogr.firstElement("biblNote") { it.getAttribute("n") == "url" }
         biblNoteElement?.let {
diff --git a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/formatters/KrillJsonGenerator.kt b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/formatters/KrillJsonGenerator.kt
index cf22cbe..905a829 100644
--- a/app/src/main/kotlin/de/ids_mannheim/korapxmltools/formatters/KrillJsonGenerator.kt
+++ b/app/src/main/kotlin/de/ids_mannheim/korapxmltools/formatters/KrillJsonGenerator.kt
@@ -231,8 +231,10 @@
                 }
                 "externalLink" -> {
                     val url = value.toString()
-                    // Extract title from corpus/publisher metadata if available
-                    val title = resolvedHeaderMetadata["publisher"]?.toString() ?: "Link"
+                    // Prefer an explicit link title (e.g. "Wikipedia", "GENIOS", "Süddeutsche
+                    // Zeitung", "DGD" set by ExternalLinkResolver), then publisher, then "Link".
+                    val title = resolvedHeaderMetadata["externalLinkTitle"]?.toString()
+                        ?: resolvedHeaderMetadata["publisher"]?.toString() ?: "Link"
                     val encodedUrl = url.replace(":", "%3A").replace("/", "%2F")
                     "type:attachement" to jsonString("data:application/x.korap-link;title=$title,$encodedUrl")
                 }
diff --git a/app/src/test/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolverTest.kt b/app/src/test/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolverTest.kt
new file mode 100644
index 0000000..ab3380b
--- /dev/null
+++ b/app/src/test/kotlin/de/ids_mannheim/korapxmltools/ExternalLinkResolverTest.kt
@@ -0,0 +1,111 @@
+package de.ids_mannheim.korapxmltools
+
+import kotlin.test.Test
+import kotlin.test.assertEquals
+import kotlin.test.assertNull
+
+/**
+ * Unit tests for the external-link derivation rules (issue #47). These are pure functions of the
+ * header values, so they are tested directly without ZIP fixtures.
+ */
+class ExternalLinkResolverTest {
+
+    @Test
+    fun extractsBiblNoteId() {
+        assertEquals("216199263", ExternalLinkResolver.biblNoteId("ID: 216199263 file: Originaldaten/2024/x@ Categories: ..."))
+        assertEquals("A124349227", ExternalLinkResolver.biblNoteId("ID: A124349227"))
+        assertNull(ExternalLinkResolver.biblNoteId("no id here"))
+        assertNull(ExternalLinkResolver.biblNoteId(null))
+    }
+
+    @Test
+    fun derivesWikipediaLinkFromReferenceText() {
+        val reference = "WDD17/B06.45592 Diskussion:Berlinische Grammatik, In: Wikipedia - " +
+            "URL:http://de.wikipedia.org/wiki/Diskussion:Berlinische_Grammatik: Wikipedia, 2017"
+        val link = ExternalLinkResolver.wikipedia(reference)
+        assertEquals("http://de.wikipedia.org/wiki/Diskussion:Berlinische_Grammatik", link?.url)
+        assertEquals("Wikipedia", link?.title)
+    }
+
+    @Test
+    fun ignoresNonWikipediaReference() {
+        assertNull(ExternalLinkResolver.wikipedia("U24/JUN.00442 Süddeutsche Zeitung, 06.06.2024, S. 14"))
+        assertNull(ExternalLinkResolver.wikipedia(null))
+    }
+
+    @Test
+    fun derivesGeniosLinkFromSigleAndId() {
+        // W24 -> botkuerzel "w" -> orikuerzel WELT
+        val link = ExternalLinkResolver.genios("W24", "216199263")
+        assertEquals("https://www.genios.de/document/WELT__216199263", link?.url)
+        assertEquals("GENIOS", link?.title)
+    }
+
+    @Test
+    fun returnsNoGeniosLinkForUnknownSigle() {
+        assertNull(ExternalLinkResolver.genios("ZZ99", "123"))
+        assertNull(ExternalLinkResolver.genios("W24", null))
+    }
+
+    @Test
+    fun geniosRequiresWellFormedNewspaperSigle() {
+        // "w" maps to WELT, but only a [A-Z]{1,3}[0-9]{2} sigle may trigger a Genios link.
+        assertNull(ExternalLinkResolver.genios("W", "216199263"))             // no year volume
+        assertNull(ExternalLinkResolver.genios("W2024", "216199263"))         // three digits
+        assertNull(ExternalLinkResolver.genios("w24", "216199263"))           // lowercase
+        assertNull(ExternalLinkResolver.genios("WELT24", "216199263"))        // four letters
+        assertNull(ExternalLinkResolver.genios("W24/SEP.00359", "216199263")) // full text sigle, not corpus prefix
+        // The well-formed sigle still works.
+        assertEquals("https://www.genios.de/document/WELT__216199263", ExternalLinkResolver.genios("W24", "216199263")?.url)
+    }
+
+    @Test
+    fun derivesSueddeutscheLinkFromSigleAndId() {
+        val link = ExternalLinkResolver.sueddeutsche("U24", "A124349227")
+        assertEquals("https://archiv.szarchiv.de/Portal/restricted/Start.act?articleId=A124349227", link?.url)
+        assertEquals("Süddeutsche Zeitung", link?.title)
+    }
+
+    @Test
+    fun sueddeutscheRequiresYearVolumeSigle() {
+        // The two digits stand for the year volume, so a bare "U" must not match.
+        assertNull(ExternalLinkResolver.sueddeutsche("U", "A1"))
+        assertNull(ExternalLinkResolver.sueddeutsche("UXX", "A1"))
+    }
+
+    @Test
+    fun derivesDgdLinkAndStripsTranscriptSuffix() {
+        val link = ExternalLinkResolver.dgd("FOLK", "FOLK_E_00010_SE_01_DF_01")
+        assertEquals(
+            "https://dgd.ids-mannheim.de/DGD2Web/ExternalAccessServlet?command=displayData&id=FOLK_E_00010_SE_01",
+            link?.url
+        )
+        assertEquals("DGD", link?.title)
+        assertNull(ExternalLinkResolver.dgd("W24", "whatever"))
+    }
+
+    @Test
+    fun resolvePrefersWikipediaOverIdBasedRules() {
+        // A text that is both a Wikipedia reference and carries an ID + Genios sigle resolves to Wikipedia.
+        val reference = "X URL:http://de.wikipedia.org/wiki/Foo: Wikipedia, 2017"
+        val link = ExternalLinkResolver.resolve(
+            corpusSigle = "W24", reference = reference, biblNoteId = "216199263", title = "Foo"
+        )
+        assertEquals("http://de.wikipedia.org/wiki/Foo", link?.url)
+        assertEquals("Wikipedia", link?.title)
+    }
+
+    @Test
+    fun resolvePrefersSueddeutscheOverGenios() {
+        // U-sigle texts go to szarchiv even though "u" could collide with a Genios botkuerzel.
+        val link = ExternalLinkResolver.resolve(
+            corpusSigle = "U24", reference = null, biblNoteId = "A124349227", title = null
+        )
+        assertEquals("Süddeutsche Zeitung", link?.title)
+    }
+
+    @Test
+    fun resolveReturnsNullWhenNothingMatches() {
+        assertNull(ExternalLinkResolver.resolve("ZZZ", "plain reference", null, "title"))
+    }
+}
diff --git a/app/src/test/kotlin/de/ids_mannheim/korapxmltools/KrillJsonGeneratorTest.kt b/app/src/test/kotlin/de/ids_mannheim/korapxmltools/KrillJsonGeneratorTest.kt
index 1316c14..9cbc88b 100644
--- a/app/src/test/kotlin/de/ids_mannheim/korapxmltools/KrillJsonGeneratorTest.kt
+++ b/app/src/test/kotlin/de/ids_mannheim/korapxmltools/KrillJsonGeneratorTest.kt
@@ -1315,4 +1315,62 @@
             "log should warn about the dropped empty text"
         )
     }
+
+    /**
+     * Regression test for https://github.com/KorAP/korapxmltool/issues/47
+     *
+     * In older Wikipedia corpora (wdd17/wpd17) the article URL is only present in the
+     * reference[type=complete] text, not encoded as a ref/link element. It must still be picked up
+     * as the Krill externalLink, with title "Wikipedia".
+     */
+    @Test
+    fun krillExtractsWikipediaExternalLinkFromReference() {
+        val baseZip = loadResource("wdd17sample.zip").path
+        val tar = ensureKrillTar("wdd17_externallink", "wdd17sample.krill.tar") { outputDir ->
+            arrayOf("-t", "krill", "-q", "-D", outputDir.path, baseZip)
+        }
+
+        val json = readKrillJson(tar).getValue("WDD17-B06-45592.json")
+        val externalLink = krillFieldValue(json, "externalLink")
+        assertNotNull(externalLink, "Wikipedia text should have an externalLink field")
+        assertTrue(externalLink.contains("title=Wikipedia"), "link title should be Wikipedia: $externalLink")
+        assertTrue(
+            externalLink.contains("de.wikipedia.org%2Fwiki%2FDiskussion%3ABerlinische_Grammatik"),
+            "link should contain the encoded Wikipedia URL: $externalLink"
+        )
+    }
+
+    /**
+     * Wiring test for the Genios newspaper case (issue #47): a corpus sigle whose botkuerzel is
+     * known (W24 -> WELT) plus a biblNote "ID:" yields a genios.de full-text link with title GENIOS.
+     */
+    @Test
+    fun krillDerivesGeniosExternalLinkForNewspaper() {
+        val tool = KorapXmlTool()
+        val metadata = collectKrillMetadata(
+            tool,
+            "W24_SEP.00359",
+            headerElement(
+                """
+                <idsHeader>
+                  <fileDesc>
+                    <titleStmt><textSigle>W24/SEP.00359</textSigle></titleStmt>
+                    <sourceDesc>
+                      <biblStruct>
+                        <analytic>
+                          <h.title type="main">Wir brauchen ein Smartphone-Verbot in Schulen</h.title>
+                          <biblNote n="1">ID: 216199263 file: Originaldaten/2024/WELT.xml.zip@ Categories: Ressort: Forum</biblNote>
+                        </analytic>
+                        <monogr><h.title type="main">Die Welt</h.title></monogr>
+                      </biblStruct>
+                      <reference type="complete" assemblage="regular">W24/SEP.00359 Die Welt, 12.09.2024, S. 7.</reference>
+                    </sourceDesc>
+                  </fileDesc>
+                </idsHeader>
+                """
+            )
+        )
+        assertEquals("https://www.genios.de/document/WELT__216199263", metadata["externalLink"])
+        assertEquals("GENIOS", metadata["externalLinkTitle"])
+    }
 }