diff options
Diffstat (limited to 'tests/data/example_french.json')
| -rw-r--r-- | tests/data/example_french.json | 23 |
1 files changed, 22 insertions, 1 deletions
diff --git a/tests/data/example_french.json b/tests/data/example_french.json index f722cc9b..1e77a8a4 100644 --- a/tests/data/example_french.json +++ b/tests/data/example_french.json @@ -1 +1,22 @@ -{"sentences": [{"index": 0, "tokens": [{"index": 1, "word": "Cette", "originalText": "Cette", "characterOffsetBegin": 0, "characterOffsetEnd": 5, "pos": "DET", "before": "", "after": " "}, {"index": 2, "word": "enqu\u00eate", "originalText": "enqu\u00eate", "characterOffsetBegin": 6, "characterOffsetEnd": 13, "pos": "NOUN", "before": " ", "after": " "}, {"index": 3, "word": "pr\u00e9liminaire", "originalText": "pr\u00e9liminaire", "characterOffsetBegin": 14, "characterOffsetEnd": 26, "pos": "ADJ", "before": " ", "after": " "}, {"index": 4, "word": "fait", "originalText": "fait", "characterOffsetBegin": 27, "characterOffsetEnd": 31, "pos": "VERB", "before": " ", "after": " "}, {"index": 5, "word": "suite", "originalText": "suite", "characterOffsetBegin": 32, "characterOffsetEnd": 37, "pos": "NOUN", "before": " ", "after": " "}, {"index": 6, "word": "aux", "originalText": "aux", "characterOffsetBegin": 38, "characterOffsetEnd": 41, "pos": "ADJ", "before": " ", "after": " "}, {"index": 7, "word": "r\u00e9v\u00e9lations", "originalText": "r\u00e9v\u00e9lations", "characterOffsetBegin": 42, "characterOffsetEnd": 53, "pos": "NOUN", "before": " ", "after": " "}, {"index": 8, "word": "de", "originalText": "de", "characterOffsetBegin": 54, "characterOffsetEnd": 56, "pos": "ADP", "before": " ", "after": " "}, {"index": 9, "word": "l'hebdomadaire", "originalText": "l\u2019hebdomadaire", "characterOffsetBegin": 57, "characterOffsetEnd": 71, "pos": "PROPN", "before": " ", "after": " "}, {"index": 10, "word": "quelques", "originalText": "quelques", "characterOffsetBegin": 72, "characterOffsetEnd": 80, "pos": "DET", "before": " ", "after": " "}, {"index": 11, "word": "jours", "originalText": "jours", "characterOffsetBegin": 81, "characterOffsetEnd": 86, "pos": "NOUN", "before": " ", "after": " "}, {"index": 12, "word": "plus", "originalText": "plus", "characterOffsetBegin": 87, "characterOffsetEnd": 91, "pos": "ADV", "before": " ", "after": " "}, {"index": 13, "word": "t\u00f4t", "originalText": "t\u00f4t", "characterOffsetBegin": 92, "characterOffsetEnd": 95, "pos": "ADV", "before": " ", "after": ""}, {"index": 14, "word": ".", "originalText": ".", "characterOffsetBegin": 95, "characterOffsetEnd": 96, "pos": "PUNCT", "before": "", "after": ""}]}]} +{"sentences": + [{"index": 0, + "tokens": [ + {"index": 1, "word": "Cette", "originalText": "Cette", "characterOffsetBegin": 0, "characterOffsetEnd": 5, "pos": "DET", "before": "", "after": " "}, + {"index": 2, "word": "enquête", "originalText": "enquête", "characterOffsetBegin": 6, "characterOffsetEnd": 13, "pos": "NOUN", "before": " ", "after": " "}, + {"index": 3, "word": "préliminaire", "originalText": "préliminaire", "characterOffsetBegin": 14, "characterOffsetEnd": 26, "pos": "ADJ", "before": " ", "after": " "}, + {"index": 4, "word": "fait", "originalText": "fait", "characterOffsetBegin": 27, "characterOffsetEnd": 31, "pos": "VERB", "before": " ", "after": " "}, + {"index": 5, "word": "suite", "originalText": "suite", "characterOffsetBegin": 32, "characterOffsetEnd": 37, "pos": "NOUN", "before": " ", "after": " "}, + {"index": 6, "word": "à", "originalText": "à", "characterOffsetBegin": 38, "characterOffsetEnd": 41, "pos": "ADP", "before": " ", "after": " "}, + {"index": 7, "word": "les", "originalText": "les", "characterOffsetBegin": 38, "characterOffsetEnd": 41, "pos": "DET", "before": " ", "after": " "}, + {"index": 8, "word": "révélations", "originalText": "révélations", "characterOffsetBegin": 42, "characterOffsetEnd": 53, "pos": "NOUN", "before": " ", "after": " "}, + {"index": 9, "word": "de", "originalText": "de", "characterOffsetBegin": 54, "characterOffsetEnd": 56, "pos": "ADP", "before": " ", "after": " "}, + {"index": 10, "word": "l’", "originalText": "l’", "characterOffsetBegin": 57, "characterOffsetEnd": 59, "pos": "NOUN", "before": " ", "after": ""}, + {"index": 11, "word": "hebdomadaire", "originalText": "hebdomadaire", "characterOffsetBegin": 59, "characterOffsetEnd": 71, "pos": "ADJ", "before": "", "after": " "}, + {"index": 12, "word": "quelques", "originalText": "quelques", "characterOffsetBegin": 72, "characterOffsetEnd": 80, "pos": "DET", "before": " ", "after": " "}, + {"index": 13, "word": "jours", "originalText": "jours", "characterOffsetBegin": 81, "characterOffsetEnd": 86, "pos": "NOUN", "before": " ", "after": " "}, + {"index": 14, "word": "plus", "originalText": "plus", "characterOffsetBegin": 87, "characterOffsetEnd": 91, "pos": "ADV", "before": " ", "after": " "}, + {"index": 15, "word": "tôt", "originalText": "tôt", "characterOffsetBegin": 92, "characterOffsetEnd": 95, "pos": "ADV", "before": " ", "after": ""}, + {"index": 16, "word": ".", "originalText": ".", "characterOffsetBegin": 95, "characterOffsetEnd": 96, "pos": "PUNCT", "before": "", "after": ""} + ]} + ] +} |