accents_de.txt 2.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169
  1. # À => A
  2. "\u00C0" => "A"
  3. # Á => A
  4. "\u00C1" => "A"
  5. # Â => A
  6. "\u00C2" => "A"
  7. # Ã => A
  8. "\u00C3" => "A"
  9. # Ä => Ae
  10. # Not needed if German2 Porter stemmer is used. This way word lists like
  11. # stopwords are easier to maintain.
  12. # Å => A
  13. "\u00C5" => "A"
  14. # Ą => A
  15. "\u0104" => "A"
  16. # Æ => AE
  17. "\u00C6" => "AE"
  18. # Ç => C
  19. "\u00C7" => "C"
  20. # Ć => C
  21. "\U0106" => "C"
  22. # È => E
  23. "\u00C8" => "E"
  24. # É => E
  25. "\u00C9" => "E"
  26. # Ê => E
  27. "\u00CA" => "E"
  28. # Ë => E
  29. "\u00CB" => "E"
  30. # Ę => E
  31. "\u0118" => "E"
  32. # Ì => I
  33. "\u00CC" => "I"
  34. # Í => I
  35. "\u00CD" => "I"
  36. # Î => I
  37. "\u00CE" => "I"
  38. # Ï => I
  39. "\u00CF" => "I"
  40. # IJ => IJ
  41. "\u0132" => "IJ"
  42. # Ð => D
  43. "\u00D0" => "D"
  44. # Ł => L
  45. "\u0141" => "L"
  46. # Ñ => N
  47. "\u00D1" => "N"
  48. # Ń => N
  49. "\u0143" => "N"
  50. # Ò => O
  51. "\u00D2" => "O"
  52. # Ó => O
  53. "\u00D3" => "O"
  54. # Ô => O
  55. "\u00D4" => "O"
  56. # Õ => O
  57. "\u00D5" => "O"
  58. # Ö => Oe
  59. # Not needed if German2 Porter stemmer is used. This way word lists like
  60. # stopwords are easier to maintain.
  61. # Ø => O
  62. "\u00D8" => "O"
  63. # Œ => OE
  64. "\u0152" => "OE"
  65. # Þ
  66. "\u00DE" => "TH"
  67. # Ù => U
  68. "\u00D9" => "U"
  69. # Ú => U
  70. "\u00DA" => "U"
  71. # Û => U
  72. "\u00DB" => "U"
  73. # Ü => Ue
  74. # Not needed if German2 Porter stemmer is used. This way word lists like
  75. # stopwords are easier to maintain.
  76. # Ý => Y
  77. "\u00DD" => "Y"
  78. # Ÿ => Y
  79. "\u0178" => "Y"
  80. # à => a
  81. "\u00E0" => "a"
  82. # á => a
  83. "\u00E1" => "a"
  84. # â => a
  85. "\u00E2" => "a"
  86. # ã => a
  87. "\u00E3" => "a"
  88. # ä => ae
  89. # Not needed if German2 Porter stemmer is used. This way word lists like
  90. # stopwords are easier to maintain.
  91. # å => a
  92. "\u00E5" => "a"
  93. # æ => ae
  94. "\u00E6" => "ae"
  95. # ç => c
  96. "\u00E7" => "c"
  97. # è => e
  98. "\u00E8" => "e"
  99. # é => e
  100. "\u00E9" => "e"
  101. # ê => e
  102. "\u00EA" => "e"
  103. # ë => e
  104. "\u00EB" => "e"
  105. # ì => i
  106. "\u00EC" => "i"
  107. # í => i
  108. "\u00ED" => "i"
  109. # î => i
  110. "\u00EE" => "i"
  111. # ï => i
  112. "\u00EF" => "i"
  113. # ij => ij
  114. "\u0133" => "ij"
  115. # ð => d
  116. "\u00F0" => "d"
  117. # ñ => n
  118. "\u00F1" => "n"
  119. # ò => o
  120. "\u00F2" => "o"
  121. # ó => o
  122. "\u00F3" => "o"
  123. # ô => o
  124. "\u00F4" => "o"
  125. # õ => o
  126. "\u00F5" => "o"
  127. # ö => oe
  128. # Not needed if German2 Porter stemmer is used. This way word lists like
  129. # stopwords are easier to maintain.
  130. # ø => o
  131. "\u00F8" => "o"
  132. # œ => oe
  133. "\u0153" => "oe"
  134. # ß => ss
  135. # Not needed if German2 Porter stemmer is used. This way word lists like
  136. # stopwords are easier to maintain.
  137. # þ => th
  138. "\u00FE" => "th"
  139. # ù => u
  140. "\u00F9" => "u"
  141. # ú => u
  142. "\u00FA" => "u"
  143. # û => u
  144. "\u00FB" => "u"
  145. # ü => ue
  146. # Not needed if German2 Porter stemmer is used. This way word lists like
  147. # stopwords are easier to maintain.
  148. # ý => y
  149. "\u00FD" => "y"
  150. # ÿ => y
  151. "\u00FF" => "y"
  152. # Ź => Z
  153. "\u0179" => "Z"
  154. # Ż => Z
  155. "\u017b" => "Z"
  156. # ff => ff
  157. "\uFB00" => "ff"
  158. # fi => fi
  159. "\uFB01" => "fi"
  160. # fl => fl
  161. "\uFB02" => "fl"
  162. # ffi => ffi
  163. "\uFB03" => "ffi"
  164. # ffl => ffl
  165. "\uFB04" => "ffl"
  166. # ſt => st
  167. "\uFB05" => "st"
  168. # st => st
  169. "\uFB06" => "st"