1 | #include <stdlib.h> |
---|
2 | #include <ctype.h> |
---|
3 | #include "../lib/iotools.h" |
---|
4 | #define _CMDLINE_FILE "../cor/cmdline.h" |
---|
5 | #include "../common/common.h" |
---|
6 | #include "common_cor.h" |
---|
7 | #include "corr.hh" |
---|
8 | #include "cmdline.h" |
---|
9 | #include <locale.h> |
---|
10 | |
---|
11 | |
---|
12 | int main(int argc, char** argv) { |
---|
13 | |
---|
14 | // setlocale(LC_CTYPE,""); |
---|
15 | // setlocale(LC_COLLATE,""); |
---|
16 | |
---|
17 | gengetopt_args_info args; |
---|
18 | |
---|
19 | if(cmdline_parser(argc, argv, &args) != 0) |
---|
20 | exit(1); |
---|
21 | |
---|
22 | process_config_files(&args,argv[0]); |
---|
23 | process_common_options(&args,argv[0]); |
---|
24 | process_cor_options(&args); |
---|
25 | |
---|
26 | Corr cor; |
---|
27 | |
---|
28 | cor.load(dictionary); |
---|
29 | cor.t=args.distance_arg; |
---|
30 | |
---|
31 | char line[MAX_LINE+1]; |
---|
32 | long line_count = 0; |
---|
33 | |
---|
34 | Segment seg; |
---|
35 | Words tab; |
---|
36 | char form1[MAX_LINE]; |
---|
37 | char* form; |
---|
38 | int formcasing; |
---|
39 | char corfield[MAX_LINE]=""; |
---|
40 | |
---|
41 | while (fgets(line, MAX_LINE, inputf)) |
---|
42 | { |
---|
43 | // strcpy(outline,line); |
---|
44 | ++line_count; |
---|
45 | |
---|
46 | // if(!seg.parse(line)) |
---|
47 | // { |
---|
48 | // fprintf(stderr,"Input error in line %d.\n",line_count); |
---|
49 | // exit(1); |
---|
50 | // } |
---|
51 | |
---|
52 | char outline[128]; |
---|
53 | //printf("Starting cor... searching for %d fields\n", args.input_field_given); |
---|
54 | //for (int i=0; i<args.input_field_given; ++i) { |
---|
55 | // printf("\t%d. %s\n", i, args.input_field_arg[i]); |
---|
56 | //} |
---|
57 | |
---|
58 | if (!process_seg(line, args)) |
---|
59 | fputs(line, outputf); |
---|
60 | else |
---|
61 | { |
---|
62 | char form[MAX_FORM]; |
---|
63 | |
---|
64 | tab.clear(); |
---|
65 | getfield(line,input_field_prefix,form); |
---|
66 | if (form==NULL) continue; |
---|
67 | |
---|
68 | formcasing=3; |
---|
69 | cor.correct(form, tab); |
---|
70 | |
---|
71 | if( tab.count() == 0 ) |
---|
72 | { |
---|
73 | formcasing=casing(form); |
---|
74 | if( formcasing == 1 || formcasing == 2) |
---|
75 | tolowers(form, form1), cor.correct(form1, tab); |
---|
76 | } |
---|
77 | |
---|
78 | if ( tab.count() == 0) |
---|
79 | fputs(line, failedf); |
---|
80 | else |
---|
81 | { |
---|
82 | if(args.replace_flag) |
---|
83 | { |
---|
84 | char corfield[128]; |
---|
85 | strcpy(corfield, input_field_prefix); |
---|
86 | strcat(corfield, form); |
---|
87 | seg.aux[seg.auxn]=corfield; |
---|
88 | ++seg.auxn; |
---|
89 | for(int i=0; i<tab.count(); ++i) |
---|
90 | { |
---|
91 | seg.form=tab[i].form(); |
---|
92 | restorecasing(seg.form,seg.form,formcasing); |
---|
93 | seg.print(outline); |
---|
94 | fputs(outline, outputf); |
---|
95 | } |
---|
96 | --seg.auxn; |
---|
97 | } |
---|
98 | else |
---|
99 | { |
---|
100 | if(one_line) |
---|
101 | { |
---|
102 | char* p=corfield; |
---|
103 | for(int i=0; i<tab.count(); ++i) |
---|
104 | { |
---|
105 | restorecasing(tab[i].form(),tab[i].form(),formcasing); |
---|
106 | p += sprintf(p," %s%s",output_field_prefix,tab[i].form()); |
---|
107 | } |
---|
108 | sprintf(p,"\n"); |
---|
109 | |
---|
110 | strcpy(outline,line); |
---|
111 | outline[strlen(outline)-1]='\0'; |
---|
112 | strcat(outline,corfield); |
---|
113 | fputs(outline, outputf); |
---|
114 | } |
---|
115 | else if(one_field) |
---|
116 | { |
---|
117 | char* p=corfield; |
---|
118 | p += sprintf(p," %s",output_field_prefix); |
---|
119 | for(int i=0; i<tab.count(); ++i) |
---|
120 | { |
---|
121 | restorecasing(tab[i].form(),tab[i].form(),formcasing); |
---|
122 | p += sprintf(p,(i==0)?"%s":";%s",tab[i].form()); |
---|
123 | } |
---|
124 | |
---|
125 | sprintf(p,"\n"); |
---|
126 | |
---|
127 | strcpy(outline,line); |
---|
128 | outline[strlen(outline)-1]='\0'; |
---|
129 | strcat(outline,corfield); |
---|
130 | fputs(outline, outputf); |
---|
131 | } |
---|
132 | else |
---|
133 | { |
---|
134 | for(int i=0; i<tab.count(); ++i) |
---|
135 | { |
---|
136 | restorecasing(tab[i].form(),tab[i].form(),formcasing); |
---|
137 | sprintf(corfield," %s%s\n",output_field_prefix,tab[i].form()); |
---|
138 | strcpy(outline,line); |
---|
139 | outline[strlen(outline)-1]='\0'; |
---|
140 | strcat(outline,corfield); |
---|
141 | fputs(outline, outputf); |
---|
142 | } |
---|
143 | } |
---|
144 | } |
---|
145 | } |
---|
146 | } |
---|
147 | |
---|
148 | if(args.interactive_flag) |
---|
149 | { |
---|
150 | fflush(outputf); |
---|
151 | fflush(failedf); |
---|
152 | } |
---|
153 | } |
---|
154 | cmdline_parser_free(&args); |
---|
155 | } |
---|