#86 #93 #106 #154 add bgnpcgn mappings for deu, fao, isl, sme

maps/bgnpcgn-deu-Latn-Latn-2000.yaml

@@ -0,0 +1,55 @@
+---
+authority_id: bgnpcgn
+id: 2000
+language: iso-639-2:deu
+source_script: Latn
+destination_script: Latn
+name: BGN/PCGN German 2000 Roman-Script Spelling Convention Agreement
+url: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/693794/ROMAN-SCRIPT_SPELLING_CONVENTIONS.pdf
+creation_date: 2000
+
+notes:
+  - |
+    The special letter β, known as eszett, [Unicode: 03B2] is a standard letter of the German alphabet and occurs
+    only in word-medial and word-final positions. It is a lowercase letter only, and when a word and/or a name is
+    written entirely in uppercase letters, it is always rendered as SS. As a result of the orthographic reform of
+    German, implemented in August 1998, the β is now rendered ss if it follows a short vowel, but it is still used
+    if it follows a long vowel or a diphthong. In those instances where β cannot be reproduced, the digraph ss
+    may be substituted for it. For alphabetization and sorting purposes, β should be treated as ss.
+  - |
+    In those instances when the vowel letters ä [Unicode: 00E4], ö [Unicode: 00F6], and ü [Unicode: 00FC]
+    cannot be reproduced, the alternate spellings ae, oe, and ue may be substituted.
+
+tests:
+  - source: Dein weiβes Fleisch erregt mich so
+    expected: Dein weisses Fleisch erregt mich so
+  - source: GROβSTÄDTE
+    expected: GROSSSTAEDTE
+  - source: Göttingen
+    expected: Goettingen
+  - source: Gütersloh
+    expected: Guetersloh
+  - source: Mährisch-Ostrau
+    expected: Maehrisch-Ostrau
+
+map:
+  rules:
+    - pattern: "(?<=[[:upper:]])\u03B2(?=[[:upper:]])?"
+      result: "SS"
+    - pattern: "(?<=[[:upper:]])\u00C4(?=[[:upper:]])?"
+      result: "AE"
+    - pattern: "(?<=[[:upper:]])\u00D6(?=[[:upper:]])?"
+      result: "OE"
+    - pattern: "(?<=[[:upper:]])\u00DC(?=[[:upper:]])?"
+      result: "UE"
+
+  characters:
+    "\u00C4": Ae # Ä
+    "\u00D6": Oe # Ö
+    "\u00DC": Ue # Ü
+
+    "\u00E4": ae # ä
+    "\u00F6": oe # ö
+    "\u00FC": ue # ü
+
+    "\u03B2": ss # β

maps/bgnpcgn-fao-Latn-Latn-1968.yaml

@@ -0,0 +1,18 @@
+---
+authority_id: bgnpcgn
+id: 1968
+language: iso-639-2:fao
+source_script: Latn
+destination_script: Latn
+name: BGN/PCGN Faroese 1968 Roman-Script Spelling Convention Agreement
+url: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/693794/ROMAN-SCRIPT_SPELLING_CONVENTIONS.pdf
+creation_date: 1968
+
+tests:
+  - source: Fyrirgefðu
+    expected: Fyrirgefdhu
+  - source: Þakka
+    expected: Þakka
+
+map:
+  inherit: bgnpcgn-fao-Latn-Latn-1964

maps/bgnpcgn-isl-Latn-Latn-1968.yaml

@@ -0,0 +1,29 @@
+---
+authority_id: bgnpcgn
+id: 1968
+language: iso-639-2:isl
+source_script: Latn
+destination_script: Latn
+name: BGN/PCGN Icelandic 1968 Roman-Script Spelling Convention Agreement
+url: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/693794/ROMAN-SCRIPT_SPELLING_CONVENTIONS.pdf
+creation_date: 1968
+
+notes:
+  - |
+    The special letter Ð ð, known as edh, [Unicode: 00D0, 00F0] and the special letter Þ þ, known as thorn,
+    [Unicode: 00DE, 00FE] should be reproduced in those forms whenever encountered. In those instances
+    when they cannot be reproduced, however, the digraph Dh dh may be substituted for Ð ð and the digraph
+    Th th may be substituted for Þ þ.
+
+tests:
+  - source: Fyrirgefðu
+    expected: Fyrirgefdhu
+  - source: þu ert velkominn
+    expected: thu ert velkominn
+  - source: GOÐAN DAGINN
+    expected: GODHAN DAGINN
+  - source: Þakka
+    expected: Thakka
+
+map:
+  inherited: bgnpcgn-isl-Latn-Latn-1964

maps/bgnpcgn-sme-Latn-Latn-1984.yaml

@@ -0,0 +1,50 @@
+---
+authority_id: bgnpcgn
+id: 1984
+language: iso-639-2:sme
+source_script: Latn
+destination_script: Latn
+name: BGN/PCGN Northern Sami (North Lappish) 1984 Roman-Script Spelling Convention Agreement
+url: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/693794/ROMAN-SCRIPT_SPELLING_CONVENTIONS.pdf
+creation_date: 1984
+
+notes:
+  - |
+    The special letter Ŋ ŋ, known as eng, [Unicode: 014A, 014B] should be reproduced in that form whenever
+    encountered. In those instances when it cannot be reproduced, however, the letter Ń ń [Unicode: 0143, 0144]
+    may be substituted for it.
+  - |
+    In a further note additional to the 1984 agreement, other special letters should be retained as found:
+      Á á [Unicode: 00C1, 00E1],
+      Č č [Unicode: 010C, 010D],
+      Đ đ [Unicode: 0110, 0111],
+      Š š [Unicode: 0160, 0161],
+      Ŧ ŧ [Unicode: 0166, 0167],
+      Ž ž [Unicode: 017D, 017E].
+
+tests: # https://web.archive.org/web/20120918094122/http://www.uta.fi/~km56049/same/svocab.html
+  - source: adjágas
+    expected: adjágas
+  - source: agálaš
+    expected: agálaš
+  - source: ÁEL
+    expected: ÁEL
+  - source: hčagastinárpu
+    expected: hčagastinárpu
+  - source: algŋa
+    expected: algńa
+  - source: Šveica
+    expected: Šveica
+  - source: MAŊŊIL
+    expected: MAŃŃIL
+  - source: giđa
+    expected: giđa
+  - source: ruoŧŧelaš
+    expected: ruoŧŧelaš
+  - source: skálžu
+    expected: skálžu
+
+map:
+  characters:
+    "\u014A": "\u0143" # Ŋ => Ń
+    "\u014B": "\u0144" # ŋ => ń