whisper.cat stuff

2023-10-05 23:28:32 +11:00
parent 2a6e28637c
commit f127b4fea8
5224 changed files with 919361 additions and 0 deletions
--- a/weboasis/text/cleaner/js/modules/diacritics.js
+++ b/weboasis/text/cleaner/js/modules/diacritics.js
@ -0,0 +1 @@
+for(var diacriticsList=[{base:"A",chars:"ÁÀÂǍĂÃẢẠÄÅĀĄẤẦẪẨẬẮẰẴẲẶǺ"},{base:"C",chars:"ĆĈČĊÇ"},{base:"D",chars:"ĎĐÐ"},{base:"E",chars:"ÉÈÊĚĔẼẺĖËĒĘẾỀỄỂẸỆ"},{base:"G",chars:"ĞĜĠĢ"},{base:"H",chars:"ĤĦ"},{base:"I",chars:"ÍÌĬÎǏÏĨĮĪỈỊ"},{base:"J",chars:"Ĵ"},{base:"K",chars:"Ķ"},{base:"L",chars:"ĹĽĻŁĿ"},{base:"N",chars:"ŃŇÑŅ"},{base:"O",chars:"ÓÒŎÔỐỒỖỔǑÖŐÕØǾŌỎƠỚỜỠỞỢỌỘ"},{base:"P",chars:"ṔṖ"},{base:"R",chars:"ŔŘŖ"},{base:"S",chars:"ŚŜŠŞ"},{base:"T",chars:"ŤŢŦ"},{base:"U",chars:"ÚÙŬÛǓŮÜǗǛǙǕŰŨŲŪỦƯỨỪỮỬỰỤ"},{base:"W",chars:"ẂẀŴẄ"},{base:"Y",chars:"ÝỲŶŸỸỶỴ"},{base:"Z",chars:"ŹŽŻ"},{base:"a",chars:"áàâǎăãảạäåāąấầẫẩậắằẵẳặǻ"},{base:"C",chars:"ćĉčċç"},{base:"d",chars:"ďđ"},{base:"e",chars:"éèêěĕẽẻėëēęếềễểẹệ"},{base:"g",chars:"ğĝġģ"},{base:"h",chars:"ĥħ"},{base:"i",chars:"íìĭîǐïĩįīỉị"},{base:"j",chars:"ĵ"},{base:"k",chars:"ķ"},{base:"l",chars:"ĺľļłŀ"},{base:"n",chars:"ńňñņ"},{base:"o",chars:"óòŏôốồỗổǒöőõøǿōỏơớờỡởợọộ"},{base:"p",chars:"ṕṗ"},{base:"r",chars:"ŕřŗ"},{base:"s",chars:"śŝšş"},{base:"t",chars:"ťţŧ"},{base:"u",chars:"úùŭûǔůüǘǜǚǖűũųūủưứừữửựụ"},{base:"w",chars:"ẃẁŵẅ"},{base:"y",chars:"ýỳŷÿỹỷỵ"},{base:"z",chars:"źžż"}],diacriticsMap={},i=0;i<diacriticsList.length;i+=1)for(var chars=diacriticsList[i].chars,j=0;j<chars.length;j+=1)diacriticsMap[chars[j]]=diacriticsList[i].base;function removeDiacritics(a){return a.replace(/[^-~]/g,function(a){return diacriticsMap[a]||a})}
--- a/weboasis/text/cleaner/js/modules/shorthands.js
+++ b/weboasis/text/cleaner/js/modules/shorthands.js
@ -0,0 +1 @@
+SMS_map={"2day":"today","2f4u":"too fast for you","2morow":"tomorrow","2nite":"tonight",afaik:"as far as I know",afair:"as far as I remember",afk:"away from keyboard",aight:"alright",aka:"also known as","b/c":"because",bf:"boyfriend",btw:"by the way",dgmw:"don't get me wrong",faq:"frequently asked questions",ftw:"for the win",fwiw:"for what it's worth",fyi:"for your information",gf:"girlfriend",gr8:"great",gtg:"got to go",icymi:"in case you missed it",idc:"I don't care",idk:"I don't know",ily:"i love you",imho:"in my humble opinion",imo:"in my opinion",iow:"in other words",irl:"in real life",jfyi:"just for your information",l8r:"later",lol:"laughing out loud",mmw:"mark my words","n/a":"not applicable",nsfw:"not safe for work",nvm:"nevermind",omg:"oh my god",tbh:"to be honest",tnx:"thanks",ttyl:"talk to you later",tyt:"take your time",ur:"your",w8:"wait",wtf:"what the fuck",wth:"what the hell",xoxo:"hugs and kisses",ymmd:"you made my day",ymmv:"your mileage may vary"};var replaceSMS=function(o){return SMS_map[o.toLowerCase()]||o};
--- a/weboasis/text/cleaner/js/modules/unorm.js
+++ b/weboasis/text/cleaner/js/modules/unorm.js
				`@ -0,0 +1 @@`
				for(var diacriticsList=[{base:"A",chars:"ÁÀÂǍĂÃẢẠÄÅĀĄẤẦẪẨẬẮẰẴẲẶǺ"},{base:"C",chars:"ĆĈČĊÇ"},{base:"D",chars:"ĎĐÐ"},{base:"E",chars:"ÉÈÊĚĔẼẺĖËĒĘẾỀỄỂẸỆ"},{base:"G",chars:"ĞĜĠĢ"},{base:"H",chars:"ĤĦ"},{base:"I",chars:"ÍÌĬÎǏÏĨĮĪỈỊ"},{base:"J",chars:"Ĵ"},{base:"K",chars:"Ķ"},{base:"L",chars:"ĹĽĻŁĿ"},{base:"N",chars:"ŃŇÑŅ"},{base:"O",chars:"ÓÒŎÔỐỒỖỔǑÖŐÕØǾŌỎƠỚỜỠỞỢỌỘ"},{base:"P",chars:"ṔṖ"},{base:"R",chars:"ŔŘŖ"},{base:"S",chars:"ŚŜŠŞ"},{base:"T",chars:"ŤŢŦ"},{base:"U",chars:"ÚÙŬÛǓŮÜǗǛǙǕŰŨŲŪỦƯỨỪỮỬỰỤ"},{base:"W",chars:"ẂẀŴẄ"},{base:"Y",chars:"ÝỲŶŸỸỶỴ"},{base:"Z",chars:"ŹŽŻ"},{base:"a",chars:"áàâǎăãảạäåāąấầẫẩậắằẵẳặǻ"},{base:"C",chars:"ćĉčċç"},{base:"d",chars:"ďđ"},{base:"e",chars:"éèêěĕẽẻėëēęếềễểẹệ"},{base:"g",chars:"ğĝġģ"},{base:"h",chars:"ĥħ"},{base:"i",chars:"íìĭîǐïĩįīỉị"},{base:"j",chars:"ĵ"},{base:"k",chars:"ķ"},{base:"l",chars:"ĺľļłŀ"},{base:"n",chars:"ńňñņ"},{base:"o",chars:"óòŏôốồỗổǒöőõøǿōỏơớờỡởợọộ"},{base:"p",chars:"ṕṗ"},{base:"r",chars:"ŕřŗ"},{base:"s",chars:"śŝšş"},{base:"t",chars:"ťţŧ"},{base:"u",chars:"úùŭûǔůüǘǜǚǖűũųūủưứừữửựụ"},{base:"w",chars:"ẃẁŵẅ"},{base:"y",chars:"ýỳŷÿỹỷỵ"},{base:"z",chars:"źžż"}],diacriticsMap={},i=0;i<diacriticsList.length;i+=1)for(var chars=diacriticsList[i].chars,j=0;j<chars.length;j+=1)diacriticsMap[chars[j]]=diacriticsList[i].base;function removeDiacritics(a){return a.replace(/[^-~]/g,function(a){return diacriticsMap[a]\|\|a})}
				`@ -0,0 +1 @@`
				SMS_map={"2day":"today","2f4u":"too fast for you","2morow":"tomorrow","2nite":"tonight",afaik:"as far as I know",afair:"as far as I remember",afk:"away from keyboard",aight:"alright",aka:"also known as","b/c":"because",bf:"boyfriend",btw:"by the way",dgmw:"don't get me wrong",faq:"frequently asked questions",ftw:"for the win",fwiw:"for what it's worth",fyi:"for your information",gf:"girlfriend",gr8:"great",gtg:"got to go",icymi:"in case you missed it",idc:"I don't care",idk:"I don't know",ily:"i love you",imho:"in my humble opinion",imo:"in my opinion",iow:"in other words",irl:"in real life",jfyi:"just for your information",l8r:"later",lol:"laughing out loud",mmw:"mark my words","n/a":"not applicable",nsfw:"not safe for work",nvm:"nevermind",omg:"oh my god",tbh:"to be honest",tnx:"thanks",ttyl:"talk to you later",tyt:"take your time",ur:"your",w8:"wait",wtf:"what the fuck",wth:"what the hell",xoxo:"hugs and kisses",ymmd:"you made my day",ymmv:"your mileage may vary"};var replaceSMS=function(o){return SMS_map[o.toLowerCase()]\|\|o};