indymedia
/
epicyon
mirror of https://gitlab.com/bashrc2/epicyon
7
4
Fork 1
Code Issues 14 Releases Wiki Activity

Handle double words

merge-requests/30/head
Bob Mottram 2022-07-05 20:45:18 +01:00
parent 85780153dc
commit 606fa14da1
1 changed files with 19 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
content.py
View File

@ -1043,8 +1043,10 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
ending = True ending = True
if ending: if ending:
prev_wrd = ''
for wrd in words: for wrd in words:
if wrd.endswith(whistle): wrd2 = (prev_wrd + ' ' + wrd).strip()
if wrd.endswith(whistle) or wrd2.endswith(whistle):
if not result.get(whistle): if not result.get(whistle):
result[whistle] = { result[whistle] = {
"count": 1, "count": 1,
@ -1052,6 +1054,7 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
} }
else: else:
result[whistle]['count'] += 1 result[whistle]['count'] += 1
prev_wrd = wrd
continue continue
if whistle.lower().endswith('-x'): if whistle.lower().endswith('-x'):
@ -1064,8 +1067,10 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
starting = True starting = True
if starting: if starting:
prev_wrd = ''
for wrd in words: for wrd in words:
if wrd.startswith(whistle): wrd2 = (prev_wrd + ' ' + wrd).strip()
if wrd.startswith(whistle) or wrd2.startswith(whistle):
if not result.get(whistle): if not result.get(whistle):
result[whistle] = { result[whistle] = {
"count": 1, "count": 1,
@ -1073,14 +1078,19 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
} }
else: else:
result[whistle]['count'] += 1 result[whistle]['count'] += 1
prev_wrd = wrd
continue continue
if '*' in whistle: if '*' in whistle:
whistle_start = whistle.split('*', 1)[0] whistle_start = whistle.split('*', 1)[0]
whistle_end = whistle.split('*', 1)[1] whistle_end = whistle.split('*', 1)[1]
prev_wrd = ''
for wrd in words: for wrd in words:
if wrd.startswith(whistle_start) and \ wrd2 = (prev_wrd + ' ' + wrd).strip()
wrd.endswith(whistle_end): if ((wrd.startswith(whistle_start) and
wrd.endswith(whistle_end)) or
(wrd2.startswith(whistle_start) and
wrd2.endswith(whistle_end))):
if not result.get(whistle): if not result.get(whistle):
result[whistle] = { result[whistle] = {
"count": 1, "count": 1,
@ -1088,10 +1098,13 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
} }
else: else:
result[whistle]['count'] += 1 result[whistle]['count'] += 1
prev_wrd = wrd
continue continue
prev_wrd = ''
for wrd in words: for wrd in words:
if wrd == whistle: wrd2 = (prev_wrd + ' ' + wrd).strip()
if whistle in (wrd, wrd2):
if not result.get(whistle): if not result.get(whistle):
result[whistle] = { result[whistle] = {
"count": 1, "count": 1,
@ -1099,6 +1112,7 @@ def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
} }
else: else:
result[whistle]['count'] += 1 result[whistle]['count'] += 1
prev_wrd = wrd
return result return result