Rev 10 | Rev 8113 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed
Rev 10 | Rev 3808 | ||
---|---|---|---|
Line 43... | Line 43... | ||
43 | } |
43 | } |
44 | 44 | ||
45 | /* searches a list. Returns index if found, -1 if nomatch. |
45 | /* searches a list. Returns index if found, -1 if nomatch. |
46 | * This routine is faster than naive one by one comparisons, |
46 | * This routine is faster than naive one by one comparisons, |
47 | * and is especially suited for large lists. */ |
47 | * and is especially suited for large lists. */ |
48 | int suffix_list(void *list, int items, size_t item_size, const char *str) |
48 | int suffix_list(void *list, int items, size_t item_size, const unsigned char *str) |
49 | { |
49 | { |
50 | int i1,i2,j,k,t,v; |
50 | int i1,i2,j,k,t,v; |
51 | char c,d; |
51 | unsigned char c,d; |
52 | 52 | ||
53 | if(items<=0) return -1; |
53 | if(items<=0) return -1; |
54 | k=sufcomp(0,str); |
54 | k=sufcomp(0,str); |
55 | if(k==0) return 0; if(k>0) return -1; |
55 | if(k==0) return 0; if(k>0) return -1; |
56 | j=items-1; k=sufcomp(j,str); |
56 | j=items-1; k=sufcomp(j,str); |
Line 68... | Line 68... | ||
68 | if(t<sufminlen) return -1; if(t>=suf[j].olen) return j; |
68 | if(t<sufminlen) return -1; if(t>=suf[j].olen) return j; |
69 | for(j--,c=str[sufwordlen-1],d=str[sufwordlen-t]; |
69 | for(j--,c=str[sufwordlen-1],d=str[sufwordlen-t]; |
70 | j>=0 && suf[j].original[0]==c && suf[j].olen>t |
70 | j>=0 && suf[j].original[0]==c && suf[j].olen>t |
71 | && suf[j].original[t-1]==d;j--); |
71 | && suf[j].original[t-1]==d;j--); |
72 | if(j>=0 && suf[j].original[0]==c && |
72 | if(j>=0 && suf[j].original[0]==c && |
73 | strncmp( |
73 | strncmp((char*)suf[j].original,(char*)suf[v].original,suf[j].olen)==0) |
74 | return j; |
74 | return j; |
75 | else goto backcheck; |
75 | else goto backcheck; |
76 | } |
76 | } |
77 | 77 | ||
78 | /* Prepare dictionary. */ |
78 | /* Prepare dictionary. */ |
Line 97... | Line 97... | ||
97 | *pp++=0; |
97 | *pp++=0; |
98 | strip_trailing_spaces(p1); strip_trailing_spaces(pp); |
98 | strip_trailing_spaces(p1); strip_trailing_spaces(pp); |
99 | p1=find_word_start(p1); pp=find_word_start(pp); |
99 | p1=find_word_start(p1); pp=find_word_start(pp); |
100 | if(*p1==0) continue; |
100 | if(*p1==0) continue; |
101 | if(i>0) { |
101 | if(i>0) { |
102 | k=strcmp(suf[i-1].original,p1); |
102 | k=strcmp((char*)suf[i-1].original,p1); |
103 | if(k>0) { |
103 | if(k>0) { |
104 | pp=strrchr(sdicname,'/'); if(pp==NULL) pp=sdicname; else pp++; |
104 | pp=strrchr(sdicname,'/'); if(pp==NULL) pp=sdicname; else pp++; |
105 | error("unsorted_dictionary %s: %s > %s.\n", |
105 | error("unsorted_dictionary %s: %s > %s.\n", |
106 | pp,suf[i-1].original,p1); |
106 | pp,suf[i-1].original,p1); |
107 | } |
107 | } |
108 | if(k==0) { |
108 | if(k==0) { |
109 | pp=strrchr(sdicname,'/'); if(pp==NULL) pp=sdicname; else pp++; |
109 | pp=strrchr(sdicname,'/'); if(pp==NULL) pp=sdicname; else pp++; |
110 | error("duplication_in_dictionary %s: %s.\n",pp,p1); |
110 | error("duplication_in_dictionary %s: %s.\n",pp,p1); |
111 | } |
111 | } |
112 | } |
112 | } |
113 | suf[i].original=p1; suf[i].olen=l=strlen(p1); |
113 | suf[i].original=(unsigned char*)p1; suf[i].olen=l=strlen(p1); |
114 | if(l<sufminlen) sufminlen=l; |
114 | if(l<sufminlen) sufminlen=l; |
115 | suf[i].replace=pp; i++; |
115 | suf[i].replace=(unsigned char*)pp; i++; |
116 | } |
116 | } |
117 | suffixcnt=i; |
117 | suffixcnt=i; |
118 | } |
118 | } |
119 | 119 | ||
120 | /* Suffix translation. */ |
120 | /* Suffix translation. */ |
Line 128... | Line 128... | ||
128 | p1=p2) { |
128 | p1=p2) { |
129 | if(!isalpha(*p1)) {p2=p1+1; continue;} |
129 | if(!isalpha(*p1)) {p2=p1+1; continue;} |
130 | for(p2=p1;isalpha(*p2);p2++); |
130 | for(p2=p1;isalpha(*p2);p2++); |
131 | if(*p2!=0 && strchr(" ,.?!'\"\n`:;()[]{}<>",*p2)==NULL) continue; |
131 | if(*p2!=0 && strchr(" ,.?!'\"\n`:;()[]{}<>",*p2)==NULL) continue; |
132 | sufwordlen=p2-p1; |
132 | sufwordlen=p2-p1; |
133 | t=suffix_list(suf,suffixcnt,sizeof(suf[0]),p1); |
133 | t=suffix_list(suf,suffixcnt,sizeof(suf[0]),(unsigned char*)p1); |
134 | if(t<0) continue; |
134 | if(t<0) continue; |
135 | string_modify(p,p2-suf[t].olen,p2,suf[t].replace); |
135 | string_modify(p,p2-suf[t].olen,p2,(char*)suf[t].replace); |
136 | p2=p2-suf[t].olen+strlen(suf[t].replace); |
136 | p2=p2-suf[t].olen+strlen((char*)suf[t].replace); |
137 | } |
137 | } |
138 | p[MAX_LINELEN]=0; |
138 | p[MAX_LINELEN]=0; |
139 | } |
139 | } |
140 | 140 | ||
141 | void suffix(char *p, char *sdicname) |
141 | void suffix(char *p, char *sdicname) |