Advertisement
Guest User

Untitled

a guest
Jan 21st, 2012
2,123
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 8.28 KB | None | 0 0
  1. <scraperfunctions>
  2. <GetAMGDiscography dest="5">
  3. <RegExp input="$$2$$3" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
  4. <RegExp input="$$1" output="&lt;album&gt;&lt;year&gt;\1&lt;/year&gt;&lt;title&gt;\2&lt;/title&gt;&lt;label&gt;\3&lt;/label&gt;&lt;/album&gt;" dest="2">
  5. <expression repeat="yes" clear="yes" noclean="1,3,4">&lt;td class=&quot;sorted cell&quot;&gt;([^&lt;]*)&lt;/td&gt;.*?&lt;td class=&quot;cell&quot;&gt;&lt;a href=&quot;[^&gt;]*&gt;([^&lt;]*)&lt;.*?&lt;td class=&quot;cell&quot;&gt;([^&lt;]*)&lt;</expression>
  6. </RegExp>
  7. <RegExp input="$$2" output="\1&amp;amp;\2" dest="3">
  8. <expression repeat="yes" noclean="1,2">(.*?)&amp;(.+)</expression>
  9. </RegExp>
  10. <RegExp input="$$3" output="" dest="2">
  11. <expression>(.+)</expression>
  12. </RegExp>
  13. <expression noclean="1" />
  14. </RegExp>
  15. </GetAMGDiscography>
  16. <GetAMGBiography dest="5">
  17. <RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
  18. <RegExp input="$$1" output="&lt;biography&gt;\1&lt;/biography&gt;" dest="2">
  19. <expression clear="yes">Biography&lt;/h2&gt;.*?&lt;/p&gt;(.*?)&lt;/p&gt;</expression>
  20. </RegExp>
  21. <expression noclean="1"/>
  22. </RegExp>
  23. </GetAMGBiography>
  24. <GetAMGReview dest="5">
  25. <RegExp input="$$2" output="&lt;details&gt;\1&lt;/details&gt;" dest="5">
  26. <RegExp input="$$1" output="&lt;review&gt;\1&lt;/review&gt;" dest="2">
  27. <expression>&lt;h2 class=&quot;title&quot;&gt;Review&lt;/h2&gt;.*?&lt;/p&gt;(.*?)&lt;/p&gt;</expression>
  28. </RegExp>
  29. <expression noclean="1" />
  30. </RegExp>
  31. </GetAMGReview>
  32. <ParseAMGArtist dest="3">
  33. <RegExp input="$$5" output="&lt;details&gt;\1&lt;/details&gt;" dest="3">
  34. <RegExp input="$$1" output="&lt;name&gt;\1&lt;/name&gt;" dest="5">
  35. <expression noclean="1">&lt;meta name=&quot;title&quot; content=&quot;([^&quot;]*)</expression>
  36. </RegExp>
  37. <RegExp input="$$1" output="&lt;born&gt;\1&lt;/born&gt;" dest="5+">
  38. <expression>&lt;h3&gt;Born[^&gt;]*&gt;[^&gt;]*&gt;(.*?)&lt;/p&gt;</expression>
  39. </RegExp>
  40. <RegExp input="$$1" output="&lt;born&gt;\1&lt;/born&gt;" dest="5+">
  41. <expression>&lt;h3&gt;Birth[^&gt;]*&gt;[^&gt;]*&gt;(.*?)&lt;/p&gt;</expression>
  42. </RegExp>
  43. <RegExp input="$$1" output="&lt;died&gt;\1&lt;/died&gt;" dest="5+">
  44. <expression>&lt;h3&gt;Died[^&gt;]*&gt;[^&gt;]*&gt;(.*?)&lt;/p&gt;</expression>
  45. </RegExp>
  46. <RegExp input="$$1" output="&lt;formed&gt;\1&lt;/formed&gt;" dest="5+">
  47. <expression>&lt;h3&gt;Formed[^&gt;]*&gt;[^&gt;]*&gt;(.*?)&lt;/p&gt;</expression>
  48. </RegExp>
  49. <RegExp input="$$1" output="&lt;disbanded&gt;\1&lt;/disbanded&gt;" dest="5+">
  50. <expression>&lt;h3&gt;Disbanded[^&gt;]*&gt;[^&gt;]*&gt;(.*?)&lt;/p&gt;</expression>
  51. </RegExp>
  52. <RegExp input="$$1" output="\1" dest="6">
  53. <expression noclean="1">&lt;h3&gt;Years Active&lt;/h3&gt;(.*?)&lt;/p&gt;</expression>
  54. </RegExp>
  55. <RegExp input="$$6" output="\1's/" dest="7+">
  56. <expression repeat="yes">&lt;span class=&quot;active&quot;&gt;([0-9]+)&lt;/span&gt;</expression>
  57. </RegExp>
  58. <RegExp input="$$7" output="&lt;yearsactive&gt;\1&lt;/yearsactive&gt;" dest="5+">
  59. <expression noclean="1">(.+)/</expression>
  60. </RegExp>
  61. <RegExp input="$$1" output="\1" dest="6">
  62. <expression noclean="1">&lt;h3&gt;Genres&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  63. </RegExp>
  64. <RegExp input="$$6" output="&lt;genre&gt;\2&lt;/genre&gt;" dest="5+">
  65. <expression repeat="yes" noclean="1">&lt;a href=&quot;/explore/genre/[^&gt;]*&gt;(&lt;strong&gt;)?([^&lt;]*)</expression>
  66. </RegExp>
  67. <RegExp input="$$1" output="\1" dest="7">
  68. <expression noclean="1">&lt;h3&gt;Styles&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  69. </RegExp>
  70. <RegExp input="$$7" output="&lt;style&gt;\1&lt;/style&gt;" dest="5+">
  71. <expression repeat="yes" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  72. </RegExp>
  73. <RegExp input="$$1" output="\1" dest="8">
  74. <expression noclean="1">&lt;h3&gt;Moods&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  75. </RegExp>
  76. <RegExp input="$$8" output="&lt;mood&gt;\1&lt;/mood&gt;" dest="5+">
  77. <expression repeat="yes" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  78. </RegExp>
  79. <RegExp input="$$1" output="\1" dest="9">
  80. <expression noclean="1">&lt;h3&gt;Instruments&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  81. </RegExp>
  82. <RegExp input="$$9" output="&lt;instruments&gt;\1&lt;/instruments&gt;" dest="5+">
  83. <expression repeat="yes" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  84. </RegExp>
  85. <RegExp conditional="allmusicartistthumbs" input="$$1" output="&lt;thumb&gt;\1&lt;/thumb&gt;" dest="5+">
  86. <expression noclean="1">&lt;img id=&quot;artist_image&quot; src=&quot;([^&quot;]*)&quot;</expression>
  87. </RegExp>
  88. <expression noclean="1" />
  89. </RegExp>
  90. </ParseAMGArtist>
  91. <ParseAMGAlbum dest="3">
  92. <RegExp input="$$5" output="&lt;details&gt;\1&lt;/details&gt;" dest="3">
  93. <RegExp input="$$1" output="&lt;title&gt;\1&lt;/title&gt;" dest="5">
  94. <expression trim="1" noclean="1">&lt;meta name=&quot;title&quot; content=&quot;([^-]*)-[^&lt;]*&quot; /&gt;</expression>
  95. </RegExp>
  96. <RegExp input="$$1" output="&lt;artist&gt;\1&lt;/artist&gt;" dest="5+">
  97. <expression noclean="1">&lt;meta name=&quot;title&quot; content=&quot;[^-]*- ([^&lt;]*)&quot; /&gt;</expression>
  98. </RegExp>
  99. <RegExp input="$$1" output="&lt;rating&gt;\1&lt;/rating&gt;" dest="5+">
  100. <expression noclean="1">&lt;h3&gt;Rating&lt;/h3&gt;[^_]*_r([0-9^]*)</expression>
  101. </RegExp>
  102. <RegExp input="$$1" output="&lt;label&gt;\1&lt;/label&gt;" dest="5+">
  103. <expression noclean="1">&lt;h3&gt;Label&lt;/h3&gt;[^&gt;]*&gt;([^&lt;]*)</expression>
  104. </RegExp>
  105. <RegExp input="$$1" output="&lt;type&gt;\1&lt;/type&gt;" dest="5+">
  106. <expression noclean="1">&lt;h3&gt;Type&lt;/h3&gt;[^&gt;]*&gt;([^&lt;]*)</expression>
  107. </RegExp>
  108. <RegExp input="$$1" output="&lt;releasedate&gt;\1&lt;/releasedate&gt;" dest="5+">
  109. <expression noclean="1">&lt;h3&gt;Release Date&lt;/h3&gt;[^&gt;]*&gt;([^&lt;]*)</expression>
  110. </RegExp>
  111. <RegExp input="$$1" output="&lt;year&gt;\1&lt;/year&gt;" dest="5+">
  112. <expression noclean="1">&lt;span&gt;Release Date.*?([0-9]+)&lt;/</expression>
  113. </RegExp>
  114. <RegExp input="$$1" output="\1" dest="6">
  115. <expression noclean="1">&lt;h3&gt;Genre[s]?&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  116. </RegExp>
  117. <RegExp input="$$6" output="&lt;genre&gt;\1&lt;/genre&gt;" dest="5+">
  118. <expression repeat="yes" trim="1" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  119. </RegExp>
  120. <RegExp input="$$1" output="\1" dest="7">
  121. <expression noclean="1">&lt;h3&gt;Style[s]?&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  122. </RegExp>
  123. <RegExp input="$$7" output="&lt;style&gt;\1&lt;/style&gt;" dest="5+">
  124. <expression repeat="yes" trim="1" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  125. </RegExp>
  126. <RegExp input="$$1" output="\1" dest="8">
  127. <expression trim="1" noclean="1">&lt;h3&gt;Mood[s]?&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  128. </RegExp>
  129. <RegExp input="$$8" output="&lt;mood&gt;\1&lt;/mood&gt;" dest="5+">
  130. <expression repeat="yes" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  131. </RegExp>
  132. <RegExp input="$$1" output="\1" dest="9">
  133. <expression noclean="1">&lt;h3&gt;Theme[s]?&lt;/h3&gt;(.*?)&lt;/div&gt;</expression>
  134. </RegExp>
  135. <RegExp input="$$9" output="&lt;theme&gt;\1&lt;/theme&gt;" dest="5+">
  136. <expression repeat="yes" noclean="1">&lt;li&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/li&gt;</expression>
  137. </RegExp>
  138. <RegExp input="$$1" output="&lt;track&gt;&lt;position&gt;\1&lt;/position&gt;&lt;title&gt;\2&lt;/title&gt;&lt;duration&gt;\3&lt;/duration&gt;&lt;/track&gt;" dest="5+">
  139. <expression repeat="yes" noclean="1,2,3">&quot;cell&quot;&gt;([0-9]+)&lt;.*?&lt;td class=&quot;cell&quot;&gt;[^&gt;]*&gt;([^&lt;]*)&lt;/a&gt;&lt;/td&gt;.*?&lt;td class=&quot;cell&quot;&gt;([0-9]*:[0-9]*)&lt;/td&gt;</expression>
  140. </RegExp>
  141. <RegExp conditional="allmusicalbumthumbs" input="$$1" output="&lt;thumb&gt;http://image.allmusic.com/\1&lt;/thumb&gt;" dest="5+">
  142. <expression noclean="1">http://image.allmusic.com/([^&quot;]*)&quot;</expression>
  143. </RegExp>
  144. <expression noclean="1" />
  145. </RegExp>
  146. </ParseAMGAlbum>
  147. </scraperfunctions>
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement