aboutsummaryrefslogtreecommitdiff
path: root/ext/bg/data
diff options
context:
space:
mode:
authortoasted-nutbread <toasted-nutbread@users.noreply.github.com>2021-01-10 14:43:06 -0500
committerGitHub <noreply@github.com>2021-01-10 14:43:06 -0500
commitf6a38f40dc52c4517e41ddb381278ecf5efba056 (patch)
tree0b56e9224ee25c0b6cc2c18cf8ae8ab891427569 /ext/bg/data
parent083da93142ec6302021ee1c29428121b54fc9e68 (diff)
Customizable sentence parsing (#1217)
* Add new sentenceParsing options * Update TextScanner.setOptions * Assign terminator/quote maps * Pass sentence parsing info to extractSentence * Simplify setting * Add setting for enableTerminationCharacters * Create new settings for sentence termination characters
Diffstat (limited to 'ext/bg/data')
-rw-r--r--ext/bg/data/options-schema.json61
1 files changed, 60 insertions, 1 deletions
diff --git a/ext/bg/data/options-schema.json b/ext/bg/data/options-schema.json
index 9053ebb1..151a7fe3 100644
--- a/ext/bg/data/options-schema.json
+++ b/ext/bg/data/options-schema.json
@@ -846,13 +846,72 @@
"sentenceParsing": {
"type": "object",
"required": [
- "scanExtent"
+ "scanExtent",
+ "enableTerminationCharacters",
+ "terminationCharacters"
],
"properties": {
"scanExtent": {
"type": "integer",
"minimum": 0,
"default": 200
+ },
+ "enableTerminationCharacters": {
+ "type": "boolean",
+ "default": true
+ },
+ "terminationCharacters": {
+ "type": "array",
+ "items": {
+ "type": "object",
+ "required": [
+ "enabled",
+ "character1",
+ "character2",
+ "includeCharacterAtStart",
+ "includeCharacterAtEnd"
+ ],
+ "properties": {
+ "enabled": {
+ "type": "boolean",
+ "default": true
+ },
+ "character1": {
+ "type": "string",
+ "default": "\"",
+ "minLength": 1,
+ "maxLength": 1
+ },
+ "character2": {
+ "type": ["string", "null"],
+ "default": "\"",
+ "minLength": 1,
+ "maxLength": 1
+ },
+ "includeCharacterAtStart": {
+ "type": "boolean",
+ "default": false
+ },
+ "includeCharacterAtEnd": {
+ "type": "boolean",
+ "default": false
+ }
+ }
+ },
+ "default": [
+ {"enabled": true, "character1": "「", "character2": "」", "includeCharacterAtStart": false, "includeCharacterAtEnd": false},
+ {"enabled": true, "character1": "『", "character2": "』", "includeCharacterAtStart": false, "includeCharacterAtEnd": false},
+ {"enabled": true, "character1": "\"", "character2": "\"", "includeCharacterAtStart": false, "includeCharacterAtEnd": false},
+ {"enabled": true, "character1": "'", "character2": "'", "includeCharacterAtStart": false, "includeCharacterAtEnd": false},
+ {"enabled": true, "character1": ".", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "!", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "?", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": ".", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "。", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "!", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "?", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true},
+ {"enabled": true, "character1": "…", "character2": null, "includeCharacterAtStart": false, "includeCharacterAtEnd": true}
+ ]
}
}
}