bool SpellCheckSentence()
{
WORDP D,E;
fixedSpell = false;
bool lowercase = false;
int language = ENGLISH;
char* lang = GetUserVariable((char*)"$cs_language");
if (lang && !stricmp(lang,(char*)"spanish")) language = SPANISH;
// check for all uppercase
for (int i = FindOOBEnd(1) + 1; i <= wordCount; ++i) // skip start of sentence
{
char* word = wordStarts[i];
size_t len = strlen(word);
for (int j = 0; j < (int)len; ++j)
{
if (IsLowerCase(word[j]))
{
lowercase = true;
i = j = 1000;
}
}
}
if (!lowercase && wordCount > 2) // must have several words in uppercase
{
for (int i = FindOOBEnd(1); i <= wordCount; ++i)
{
char* word = wordStarts[i];
MakeLowerCase(word);
}
}
int startWord = FindOOBEnd(1);
for (int i = startWord; i <= wordCount; ++i)
{
char* word = wordStarts[i];
if (!word || !word[1] || *word == '"' ) continue; // illegal or single char or quoted thingy
size_t len = strlen(word);
// dont spell check uppercase not at start or joined word
if (IsUpperCase(word[0]) && (i != startWord || strchr(word,'_')) && tokenControl & NO_PROPER_SPELLCHECK) continue;
// dont spell check email or other things with @ or . in them
if (strchr(word,'@') || strchr(word,'.') || strchr(word,'$')) continue;
// dont spell check names of json objects or arrays
if (!strnicmp(word,"ja-",3) || !strnicmp(word,"jo-",3)) continue;
char* known = ProbableKnownWord(word);
if (known && !strcmp(known,word)) continue; // we know it
if (known && strcmp(known,word))
{
char* tokens[2];
if (!IsUpperCase(*known)) // revised the word to lower case (avoid to upper case like "fields" to "Fields"
{
WORDP D = FindWord(known,0,LOWERCASE_LOOKUP);
if (D)
{
tokens[1] = D->word;
ReplaceWords(i,1,1,tokens);
fixedSpell = true;
continue;
}
}
else // is uppercase a concept member? then revise upwards
{
WORDP D = FindWord(known,0,UPPERCASE_LOOKUP);
if (IsConceptMember(D))
{
tokens[1] = D->word;
ReplaceWords(i,1,1,tokens);
fixedSpell = true;
continue;
}
}
}
char* p = word -1;
unsigned char c;
char* hyphen = 0;
while ((c = *++p) != 0)
{
++len;
if (c == '-') hyphen = p; // note is hyphenated - use trailing
}
if (len == 0 || GetTemperatureLetter(word)) continue; // bad ignore utf word or llegal length - also no composite words
if (c && c != '@' && c != '.') // illegal word character
{
if (IsDigit(word[0]) || len == 1){;} // probable numeric?
// accidental junk on end of word we do know immedately?
else if (i > 1 && !IsAlphaUTF8OrDigit(wordStarts[i][len-1]) )
{
WORDP entry,canonical;
char word[MAX_WORD_SIZE];
strcpy(word,wordStarts[i]);
word[len-1] = 0;
uint64 sysflags = 0;
uint64 cansysflags = 0;
WORDP revise;
GetPosData(i,word,revise,entry,canonical,sysflags,cansysflags,true,true); // dont create a non-existent word
if (entry && entry->properties & PART_OF_SPEECH)
{
wordStarts[i] = reuseAllocation(wordStarts[i],entry->word);
fixedSpell = true;
continue; // not a legal word character, leave it alone
}
}
}
// see if we know the other case
if (!(tokenControl & (ONLY_LOWERCASE|STRICT_CASING)) || (i == startSentence && !(tokenControl & ONLY_LOWERCASE)))
{
WORDP E = FindWord(word,0,SECONDARY_CASE_ALLOWED);
bool useAlternateCase = false;
if (E && E->systemFlags & PATTERN_WORD) useAlternateCase = true;
if (E && E->properties & (PART_OF_SPEECH|FOREIGN_WORD))
{
// if the word we find is UPPER case, and this might be a lower case noun plural, don't change case.
size_t len = strlen(word);
if (word[len-1] == 's' )
{
WORDP F = FindWord(word,len-1);
if (!F || !(F->properties & (PART_OF_SPEECH|FOREIGN_WORD))) useAlternateCase = true;
else continue;
}
else useAlternateCase = true;
}
else if (E) // does it have a member concept fact
{
if (IsConceptMember(E))
{
useAlternateCase = true;
break;
}
}
if (useAlternateCase)
{
char* tokens[2];
tokens[1] = E->word;
ReplaceWords(i,1,1,tokens);
fixedSpell = true;
continue;
}
}
// merge with next token?
char join[MAX_WORD_SIZE * 3];
if (i != wordCount && *wordStarts[i+1] != '"' )
{
// direct merge as a single word
strcpy(join,word);
strcat(join,wordStarts[i+1]);
WORDP D = FindWord(join,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
strcpy(join,word);
if (!D || !(D->properties & PART_OF_SPEECH) ) // merge these two, except "going to" or wordnet composites of normal words // merge as a compound word
{
strcat(join,(char*)"_");
strcat(join,wordStarts[i+1]);
D = FindWord(join,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
}
if (D && D->properties & PART_OF_SPEECH && !(D->properties & AUX_VERB)) // merge these two, except "going to" or wordnet composites of normal words
{
WORDP P1 = FindWord(word,0,LOWERCASE_LOOKUP);
WORDP P2 = FindWord(wordStarts[i+1],0,LOWERCASE_LOOKUP);
if (!P1 || !P2 || !(P1->properties & PART_OF_SPEECH) || !(P2->properties & PART_OF_SPEECH))
{
char* tokens[2];
tokens[1] = D->word;
ReplaceWords(i,2,1,tokens);
fixedSpell = true;
continue;
}
}
}
// break apart slashed pair like eat/feed
char* slash = strchr(word,'/');
if (slash && slash != word && slash[1]) // break apart word/word
{
if ((wordCount + 2 ) >= REAL_SENTENCE_LIMIT) continue; // no room
*slash = 0;
D = StoreWord(word);
*slash = '/';
E = StoreWord(slash+1);
char* tokens[4];
tokens[1] = D->word;
tokens[2] = "/";
tokens[3] = E->word;
ReplaceWords(i,1,3,tokens);
fixedSpell = true;
--i;
continue;
}
// see if hypenated word should be separate or joined (ignore obvious adjective suffix)
if (hyphen && !stricmp(hyphen,(char*)"-like"))
{
StoreWord(word,ADJECTIVE_NORMAL|ADJECTIVE); // accept it as a word
continue;
}
else if (hyphen && (hyphen-word) > 1)
{
char test[MAX_WORD_SIZE];
char first[MAX_WORD_SIZE];
// test for split
*hyphen = 0;
strcpy(test,hyphen+1);
strcpy(first,word);
*hyphen = '-';
WORDP E = FindWord(test,0,LOWERCASE_LOOKUP);
WORDP D = FindWord(first,0,LOWERCASE_LOOKUP);
if (*first == 0)
{
wordStarts[i] = AllocateString(wordStarts[i] + 1); // -pieces want to lose the leading hypen (2-pieces)
fixedSpell = true;
}
else if (D && E) // 1st word gets replaced, we added another word after
{
if ((wordCount + 1 ) >= REAL_SENTENCE_LIMIT) continue; // no room
char* tokens[3];
tokens[1] = D->word;
tokens[2] = E->word;
ReplaceWords(i,1,2,tokens);
fixedSpell = true;
--i;
}
else if (!stricmp(test,(char*)"old") || !stricmp(test,(char*)"olds")) // break apart 5-year-old
{
if ((wordCount + 1 ) >= REAL_SENTENCE_LIMIT) continue; // no room
D = StoreWord(first);
E = StoreWord(test);
char* tokens[3];
tokens[1] = D->word;
tokens[2] = E->word;
ReplaceWords(i,1,2,tokens);
fixedSpell = true;
--i;
}
else // remove hyphen entirely?
{
strcpy(test,first);
strcat(test,hyphen+1);
D = FindWord(test,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
if (D)
{
wordStarts[i] = D->word;
fixedSpell = true;
--i;
}
}
continue; // ignore hypenated errors that we couldnt solve, because no one mistypes a hypen
}
// leave uppercase in first position if not adjusted yet... but check for lower case spell error
if (IsUpperCase(word[0]) && tokenControl & NO_PROPER_SPELLCHECK)
{
char lower[MAX_WORD_SIZE];
MakeLowerCopy(lower,word);
WORDP D = FindWord(lower,0,LOWERCASE_LOOKUP);
if (!D && i == startWord)
{
char* okword = SpellFix(lower,i,PART_OF_SPEECH,language);
if (okword)
{
char* tokens[2];
WORDP E = StoreWord(okword);
tokens[1] = E->word;
ReplaceWords(i,1,1,tokens);
fixedSpell = true;
}
}
continue;
}
if (*word != '\'' && (!FindCanonical(word, i,true) || IsUpperCase(word[0]))) // dont check quoted or findable words unless they are capitalized
{
word = SpellCheck(i,language);
// dont spell check proper names to improper, if word before or after is lower case originally
if (word && i != 1 && originalCapState[i] && !IsUpperCase(*word))
{
if (!originalCapState[i-1]) return false;
else if (i != wordCount && !originalCapState[i+1]) return false;
}
if (word && !*word) // performed substitution on prior word, restart this one
{
fixedSpell = true;
--i;
continue;
}
if (word)
{
char* tokens[2];
tokens[1] = word;
ReplaceWords(i,1,1,tokens);
fixedSpell = true;
continue;
}
}
}
return fixedSpell;
}
bool SpellCheckSentence()
{
WORDP D,E;
fixedSpell = false;
bool lowercase = false;
// check for all uppercase (capslock)
for (int i = FindOOBEnd(1); i <= wordCount; ++i) // skip start of sentence
{
char* word = wordStarts[i];
if (!word[1]) continue; // autoconversion of letters to lower case should be ignored (eg A)
if (!stricmp(word, "the")) continue;
size_t len = strlen(word);
for (int j = 0; j < (int)len; ++j)
{
if (IsLowerCase(word[j]))
{
lowercase = true;
i = j = len+1000; // len might be BIG (oob data) so make sure beyond it)
}
}
}
if (!lowercase && wordCount > 2) // must have multiple words all in uppercase
{
for (int i = FindOOBEnd(1); i <= wordCount; ++i)
{
char* word = wordStarts[i];
char myword[MAX_WORD_SIZE];
MakeLowerCopy(myword,word);
if (strcmp(word, myword))
{
char* tokens[2];
tokens[1] = myword;
ReplaceWords("caplocWord", i, 1, 1, tokens);
originalCapState[i] = false;
}
}
}
int startWord = FindOOBEnd(1);
for (int i = startWord; i <= wordCount; ++i)
{
char* word = wordStarts[i];
char* tokens[2];
// change any \ to /
char newword[MAX_WORD_SIZE];
bool altered = false;
if (strlen(word) < MAX_WORD_SIZE)
{
strcpy(newword, word);
char* at = newword;
while ((at = strchr(at,'\\')))
{
*at = '/';
altered = true;
}
if (altered) word = wordStarts[i] = StoreWord(newword, AS_IS)->word;
}
if (*word == '\'' && !word[1] && i != startWord && IsDigit(*wordStarts[i - 1]) && !stricmp(language, "english")) // fails if not digit bug
{
tokens[1] = (char*)"foot";
ReplaceWords("' as feet", i, 1, 1, tokens);
fixedSpell = true;
continue;
}
if (*word == '"' && !word[1] && i != startWord && IsDigit(*wordStarts[i - 1]) && !stricmp(language, "english")) // fails if not digit bug
{
tokens[1] = (char*)"inch";
ReplaceWords("' as feet", i, 1, 1, tokens);
fixedSpell = true;
continue;
}
if (!word || !word[1] || *word == '"' ) continue; // illegal or single char or quoted thingy
size_t len = strlen(word);
// dont spell check uppercase not at start or joined word
if (IsUpperCase(word[0]) && (i != startWord || strchr(word,'_')) && tokenControl & NO_PROPER_SPELLCHECK) continue;
// dont spell check email or other things with @ or . in them
if (strchr(word,'@') || strchr(word, '&') || strchr(word,'.') || strchr(word,'$')) continue;
// dont spell check names of json objects or arrays
if (!strnicmp(word,"ja-",3) || !strnicmp(word,"jo-",3)) continue;
// dont spell check web addresses
if (!strnicmp(word,"http",4) || !strnicmp(word,"www",3)) continue;
// nor fractions
if (IsFraction(word)) continue; // fraction?
// joined number words like 100dollars
char* at = word - 1;
while (IsDigit(*++at) || *at == numberPeriod);
if (IsDigit(*word) && strlen(at) > 3 && ProbableKnownWord(at))
{
char first[MAX_WORD_SIZE];
strncpy(first, word, (at - word));
first[at - word] = 0;
char* tokens[3];
tokens[1] = first;
tokens[2] = at;
ReplaceWords("joined number word", i, 1, 2, tokens);
continue;
}
// nor model numbers
if (IsModelNumber(word))
{
WORDP X = FindWord(word, 0, UPPERCASE_LOOKUP);
if (IsConceptMember(X) && !strcmp(word,X->word))
{
char* tokens[2];
tokens[1] = X->word;
ReplaceWords("KnownUpperModelNumber", i, 1, 1, tokens);
fixedSpell = true;
}
continue;
}
char* number;
if (GetCurrency((unsigned char*)word, number)) continue; // currency
if (!stricmp(word, (char*)"am") && i != startWord &&
(IsDigit(*wordStarts[i-1]) || IsNumber(wordStarts[i-1]) ==REAL_NUMBER) && !stricmp(language,"english")) // fails if not digit bug
{
char* tokens[2];
tokens[1] = (char*)"a.m.";
ReplaceWords("am as time", i, 1, 1, tokens);
fixedSpell = true;
continue;
}
char* known = ProbableKnownWord(word);
if (known && !strcmp(known,word)) continue; // we know it
if (known && strcmp(known,word))
{
WORDP D = FindWord(known);
char* tokens[2];
if ((!D || !(D->internalBits & UPPERCASE_HASH)) && !IsUpperCase(*known)) // revised the word to lower case (avoid to upper case like "fields" to "Fields"
{
WORDP X = FindWord(known,0,LOWERCASE_LOOKUP);
if (X)
{
tokens[1] = X->word;
ReplaceWords("KnownWord",i,1,1,tokens);
fixedSpell = true;
continue;
}
}
else // is uppercase a concept member? then revise upwards
{
WORDP X = FindWord(known,0,UPPERCASE_LOOKUP);
if (IsConceptMember(X) || stricmp(language,"english")) // all german nouns are uppercase
{
tokens[1] = X->word;
ReplaceWords("KnownUpper",i,1,1,tokens);
fixedSpell = true;
continue;
}
}
}
char* p = word -1;
unsigned char c;
char* hyphen = 0;
while ((c = *++p) != 0)
{
++len;
if (c == '-') hyphen = p; // note is hyphenated - use trailing
}
if (len == 0 || GetTemperatureLetter(word)) continue; // bad ignore utf word or llegal length - also no composite words
if (c && c != '@' && c != '.') // illegal word character
{
if (IsDigit(word[0]) || len == 1){;} // probable numeric?
// accidental junk on end of word we do know immedately?
else if (i > 1 && !IsAlphaUTF8OrDigit(wordStarts[i][len-1]) )
{
WORDP entry,canonical;
char word[MAX_WORD_SIZE];
strcpy(word,wordStarts[i]);
word[len-1] = 0;
uint64 sysflags = 0;
uint64 cansysflags = 0;
WORDP revise;
GetPosData(i,word,revise,entry,canonical,sysflags,cansysflags,true,true); // dont create a non-existent word
if (entry && entry->properties & PART_OF_SPEECH)
{
wordStarts[i] = entry->word;
fixedSpell = true;
continue; // not a legal word character, leave it alone
}
}
}
// see if we know the other case
if (!(tokenControl & (ONLY_LOWERCASE|STRICT_CASING)) || (i == startSentence && !(tokenControl & ONLY_LOWERCASE)))
{
WORDP E = FindWord(word,0,SECONDARY_CASE_ALLOWED);
bool useAlternateCase = false;
if (E && E->systemFlags & PATTERN_WORD) useAlternateCase = true;
if (E && E->properties & (PART_OF_SPEECH|FOREIGN_WORD))
{
// if the word we find is UPPER case, and this might be a lower case noun plural, don't change case.
size_t len = strlen(word);
if (word[len-1] == 's' )
{
WORDP F = FindWord(word,len-1);
if (!F || !(F->properties & (PART_OF_SPEECH|FOREIGN_WORD))) useAlternateCase = true;
else continue;
}
else useAlternateCase = true;
}
else if (E) // does it have a member concept fact
{
if (IsConceptMember(E))
{
useAlternateCase = true;
break;
}
}
if (useAlternateCase)
{
char* tokens[2];
tokens[1] = E->word;
ReplaceWords("Alternatecase",i,1,1,tokens);
fixedSpell = true;
continue;
}
}
// merge with next token?
char join[MAX_WORD_SIZE * 3];
if (i != wordCount && *wordStarts[i+1] != '"' )
{
// direct merge as a single word
strcpy(join,word);
strcat(join,wordStarts[i+1]);
WORDP D = FindWord(join,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
strcpy(join,word);
// if (!D || !(D->properties & PART_OF_SPEECH) ) // merge these two, except "going to" or wordnet composites of normal words // merge as a compound word
// {
// strcat(join,(char*)"_");
// strcat(join,wordStarts[i+1]);
// D = FindWord(join,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
// } DONT CREATE _ words, let sequence handle it
if (D && D->properties & PART_OF_SPEECH && !(D->properties & AUX_VERB)) // merge these two, except "going to" or wordnet composites of normal words
{
WORDP P1 = FindWord(word,0,LOWERCASE_LOOKUP);
WORDP P2 = FindWord(wordStarts[i+1],0,LOWERCASE_LOOKUP);
if (!P1 || !P2 || !(P1->properties & PART_OF_SPEECH) || !(P2->properties & PART_OF_SPEECH))
{
char* tokens[2];
tokens[1] = D->word;
ReplaceWords("merge",i,2,1,tokens);
fixedSpell = true;
continue;
}
}
}
// break apart slashed pair like eat/feed
char* slash = strchr(word,'/');
if (slash && !slash[1] && len < MAX_WORD_SIZE) // remove trailing slash
{
strcpy(newword, word);
newword[slash - word] = 0;
word = wordStarts[i] = StoreWord(newword, AS_IS)->word;
}
if (slash && slash != word && slash[1]) // break apart word/word
{
if ((wordCount + 2 ) >= REAL_SENTENCE_LIMIT) continue; // no room
*slash = 0;
D = StoreWord(word);
*slash = '/';
E = StoreWord(slash+1);
char* tokens[4];
tokens[1] = D->word;
tokens[2] = "/";
tokens[3] = E->word;
ReplaceWords("Split",i,1,3,tokens);
fixedSpell = true;
--i;
continue;
}
// see if hypenated word should be separate or joined (ignore obvious adjective suffix)
if (hyphen && !stricmp(hyphen,(char*)"-like"))
{
StoreWord(word,ADJECTIVE_NORMAL|ADJECTIVE); // accept it as a word
continue;
}
else if (hyphen && (hyphen-word) > 1 && !IsPlaceNumber(word)) // dont break up fifty-second
{
char test[MAX_WORD_SIZE];
char first[MAX_WORD_SIZE];
// test for split
*hyphen = 0;
strcpy(test,hyphen+1);
strcpy(first,word);
*hyphen = '-';
WORDP E = FindWord(test,0,LOWERCASE_LOOKUP);
WORDP D = FindWord(first,0,LOWERCASE_LOOKUP);
if (*first == 0)
{
wordStarts[i] = AllocateHeap(wordStarts[i] + 1); // -pieces want to lose the leading hypen (2-pieces)
fixedSpell = true;
}
else if (D && E) // 1st word gets replaced, we added another word after
{
if ((wordCount + 1 ) >= REAL_SENTENCE_LIMIT) continue; // no room
char* tokens[3];
tokens[1] = D->word;
tokens[2] = E->word;
ReplaceWords("Pair",i,1,2,tokens);
fixedSpell = true;
--i;
}
else if (!stricmp(test,(char*)"old") || !stricmp(test,(char*)"olds")) // break apart 5-year-old
{
if ((wordCount + 1 ) >= REAL_SENTENCE_LIMIT) continue; // no room
D = StoreWord(first);
E = StoreWord(test);
char* tokens[3];
tokens[1] = D->word;
tokens[2] = E->word;
ReplaceWords("Break old",i,1,2,tokens);
fixedSpell = true;
--i;
}
else // remove hyphen entirely?
{
strcpy(test,first);
strcat(test,hyphen+1);
D = FindWord(test,0,(tokenControl & ONLY_LOWERCASE) ? PRIMARY_CASE_ALLOWED : STANDARD_LOOKUP);
if (D)
{
wordStarts[i] = D->word;
fixedSpell = true;
--i;
}
}
continue; // ignore hypenated errors that we couldnt solve, because no one mistypes a hypen
}
// see if number in front of unit split like 10mg
if (IsDigit(*word))
{
char* at = word;
while (*++at && IsDigit(*at)) {;}
WORDP E = FindWord(at);
if (E && strlen(at) > 2 && *at != 'm') // number in front of known word ( but must be longer than 2 char, 5th) but allow mg
{
char token1[MAX_WORD_SIZE];
int len = at - word;
strncpy(token1,word,len);
token1[len] = 0;
D = StoreWord(token1);
char* tokens[4];
tokens[1] = D->word;
tokens[2] = E->word;
ReplaceWords("Split",i,1,2,tokens);
fixedSpell = true;
continue;
}
}
// leave uppercase in first position if not adjusted yet... but check for lower case spell error
if (IsUpperCase(word[0]) && tokenControl & NO_PROPER_SPELLCHECK)
{
char lower[MAX_WORD_SIZE];
MakeLowerCopy(lower,word);
WORDP D = FindWord(lower,0,LOWERCASE_LOOKUP);
if (!D && i == startWord)
{
char* okword = SpellFix(lower,i,PART_OF_SPEECH);
if (okword)
{
char* tokens[2];
WORDP E = StoreWord(okword);
tokens[1] = E->word;
ReplaceWords("Spell",i,1,1,tokens);
fixedSpell = true;
}
}
continue;
}
// see if smooshed word pair
size_t len1 = strlen(word);
int j;
if (!IsDigit(*word))
{
for (j = 1; j <= len1 - 1; ++j)
{
WORDP X1 = FindWord(word, j); // any case
WORDP X2 = FindWord(word + j, len1 - i); // any case
if (X1 && X2 && (X1->word[1] || X1->word[0] == 'i' || X1->word[0] == 'I' || X1->word[0] == 'a'))
{
char* tokens[3];
tokens[1] = X1->word;
tokens[2] = X2->word;
ReplaceWords("Split", i, 1, 2, tokens);
fixedSpell = true;
break;
}
}
if (j != len1) continue;
}
if (*word != '\'' && (!FindCanonical(word, i,true) || IsUpperCase(word[0]))) // dont check quoted or findable words unless they are capitalized
{
word = SpellCheck(i);
// dont spell check proper names to improper, if word before or after is lower case originally
if (word && i != 1 && originalCapState[i] && !IsUpperCase(*word))
{
if (!originalCapState[i-1]) continue;
else if (i != wordCount && !originalCapState[i+1]) continue;
}
if (word && !*word) // performed substitution on prior word, restart this one
{
fixedSpell = true;
--i;
continue;
}
if (word)
{
char* tokens[2];
tokens[1] = word;
ReplaceWords("Spell",i,1,1,tokens);
fixedSpell = true;
continue;
}
}
}
return fixedSpell;
}