Spaces:
Sleeping
Sleeping
Claude commited on
Commit ·
5d7e317
1
Parent(s): 4233f08
fix: gruffalo lowercase, remove dangerous so->Go word correction
Browse files
smoke_signal/scripts/punct_corrector.py
CHANGED
|
@@ -46,8 +46,8 @@ WORD_CORRECTIONS: dict[str, str] = {
|
|
| 46 |
"Tn": "In",
|
| 47 |
"Ts": "Is",
|
| 48 |
# G misread as s at word start
|
| 49 |
-
"sruffalo": "
|
| 50 |
-
"sruffalo's": "
|
| 51 |
"sood": "good",
|
| 52 |
"sreat": "great",
|
| 53 |
"sot": "got",
|
|
@@ -55,7 +55,6 @@ WORD_CORRECTIONS: dict[str, str] = {
|
|
| 55 |
"srew": "grew",
|
| 56 |
"save": "gave",
|
| 57 |
"soing": "going",
|
| 58 |
-
"so": "Go", # only catches capitalised — won't corrupt "so" mid sentence
|
| 59 |
}
|
| 60 |
|
| 61 |
MIN_PUNCT_PER_10_WORDS = 0.8
|
|
|
|
| 46 |
"Tn": "In",
|
| 47 |
"Ts": "Is",
|
| 48 |
# G misread as s at word start
|
| 49 |
+
"sruffalo": "gruffalo",
|
| 50 |
+
"sruffalo's": "gruffalo's",
|
| 51 |
"sood": "good",
|
| 52 |
"sreat": "great",
|
| 53 |
"sot": "got",
|
|
|
|
| 55 |
"srew": "grew",
|
| 56 |
"save": "gave",
|
| 57 |
"soing": "going",
|
|
|
|
| 58 |
}
|
| 59 |
|
| 60 |
MIN_PUNCT_PER_10_WORDS = 0.8
|