| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 1 | package de.ids_mannheim.korap.resource; |
| 2 | |
| Michael Hanl | badd79c | 2015-06-19 07:41:03 +0200 | [diff] [blame] | 3 | import de.ids_mannheim.korap.config.KustvaktConfiguration; |
| Michael Hanl | 00b64e0 | 2016-05-24 20:24:27 +0200 | [diff] [blame] | 4 | import de.ids_mannheim.korap.config.Attributes; |
| Michael Hanl | e4fc775 | 2016-02-17 11:41:18 +0100 | [diff] [blame] | 5 | import de.ids_mannheim.korap.user.Userdata; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 6 | |
| 7 | /** |
| 8 | * @author hanl |
| 9 | * @date 14/10/2014 |
| 10 | */ |
| 11 | public class LayerMapper { |
| 12 | |
| Michael Hanl | e4fc775 | 2016-02-17 11:41:18 +0100 | [diff] [blame] | 13 | private Userdata settings; |
| Michael Hanl | badd79c | 2015-06-19 07:41:03 +0200 | [diff] [blame] | 14 | private KustvaktConfiguration config; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 15 | |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 16 | |
| 17 | public LayerMapper (KustvaktConfiguration config, Userdata settings) { |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 18 | this.settings = settings; |
| Michael Hanl | 1e18cb4 | 2015-08-06 20:57:35 +0200 | [diff] [blame] | 19 | this.config = config; |
| Michael Hanl | badd79c | 2015-06-19 07:41:03 +0200 | [diff] [blame] | 20 | } |
| 21 | |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 22 | |
| 23 | public LayerMapper (KustvaktConfiguration config) { |
| Michael Hanl | 1e18cb4 | 2015-08-06 20:57:35 +0200 | [diff] [blame] | 24 | this.config = config; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 25 | } |
| 26 | |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 27 | |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 28 | /** |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 29 | * find foundry entry in settings specific settings. Includes a |
| 30 | * call to #translateLayer to get the |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 31 | * correct mapping for the layer denomination! |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 32 | * |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 33 | * @param layer |
| 34 | * @return |
| 35 | */ |
| 36 | |
| 37 | //todo: make mapping configurable! |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 38 | public String findFoundry (String layer) { |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 39 | if (settings != null) { |
| 40 | switch (translateLayer(layer.toLowerCase().trim())) { |
| 41 | case "d": |
| Michael Hanl | 25aac54 | 2016-02-01 18:16:44 +0100 | [diff] [blame] | 42 | return (String) settings |
| 43 | .get(Attributes.DEFAULT_REL_FOUNDRY); |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 44 | case "c": |
| Michael Hanl | 25aac54 | 2016-02-01 18:16:44 +0100 | [diff] [blame] | 45 | return (String) settings |
| 46 | .get(Attributes.DEFAULT_CONST_FOUNDRY); |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 47 | case "pos": |
| Michael Hanl | 25aac54 | 2016-02-01 18:16:44 +0100 | [diff] [blame] | 48 | return (String) settings |
| 49 | .get(Attributes.DEFAULT_POS_FOUNDRY); |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 50 | case "lemma": |
| Michael Hanl | 25aac54 | 2016-02-01 18:16:44 +0100 | [diff] [blame] | 51 | return (String) settings |
| 52 | .get(Attributes.DEFAULT_LEMMA_FOUNDRY); |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 53 | case "surface": |
| 54 | return "opennlp"; |
| 55 | default: |
| 56 | // if the layer is not in this specific listing, assume a default layer |
| 57 | // like orth or other tokenization layers |
| Michael Hanl | c0ed00f | 2016-06-23 14:33:10 +0200 | [diff] [blame] | 58 | return null; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 59 | } |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 60 | } |
| 61 | else { |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 62 | switch (translateLayer(layer.toLowerCase().trim())) { |
| 63 | case "d": |
| 64 | return config.getDefault_dep(); |
| 65 | case "c": |
| 66 | return config.getDefault_const(); |
| 67 | case "pos": |
| 68 | return config.getDefault_pos(); |
| 69 | case "lemma": |
| 70 | return config.getDefault_lemma(); |
| 71 | case "surface": |
| Michael Hanl | f21773f | 2015-10-16 23:02:31 +0200 | [diff] [blame] | 72 | return config.getDefault_token(); |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 73 | // refers to "structure" and is used for paragraphs or sentence boundaries |
| Michael Hanl | 505f48c | 2015-06-26 19:04:27 +0200 | [diff] [blame] | 74 | case "s": |
| 75 | return "base"; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 76 | default: |
| 77 | // if the layer is not in this specific listing, assume a default layer |
| 78 | // like orth or other tokenization layers |
| Michael Hanl | c0ed00f | 2016-06-23 14:33:10 +0200 | [diff] [blame] | 79 | return null; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 80 | } |
| 81 | } |
| 82 | } |
| 83 | |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 84 | |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 85 | // relevance: map to access control id references. p is usually mapped to pos, l to lemma, etc. |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 86 | public String translateLayer (String layer) { |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 87 | switch (layer.toLowerCase().trim()) { |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 88 | // case "pos": |
| 89 | // return "p"; |
| 90 | // case "lemma": |
| 91 | // return "l"; |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 92 | case "m": |
| 93 | return "msd"; |
| Michael Hanl | 8abaf9e | 2016-05-23 16:46:35 +0200 | [diff] [blame] | 94 | //todo the orth layer does not need a foundry entry |
| Michael Hanl | 704211e | 2015-06-19 07:26:04 +0200 | [diff] [blame] | 95 | case "orth": |
| 96 | return "surface"; |
| 97 | case "t": |
| 98 | return "surface"; |
| 99 | case "const": |
| 100 | return "c"; |
| 101 | case "p": |
| 102 | return "pos"; |
| 103 | case "l": |
| 104 | return "lemma"; |
| 105 | default: |
| 106 | return layer; |
| 107 | } |
| 108 | } |
| 109 | |
| 110 | } |