Commit | Line | Data |
---|---|---|
96964f3e JB |
1 | #include <stdlib.h> |
2 | #include <stdio.h> | |
3 | #include <string.h> | |
4b580abd | 4 | #include <stdbool.h> |
6a19b8fe | 5 | #include <wchar.h> |
e9a51b68 | 6 | |
6a19b8fe JB |
7 | #define TOKEN_MAX 500 |
8 | ||
9 | struct token_s { | |
10 | const char* type; | |
11 | wint_t value[50]; | |
12 | }; | |
13 | ||
7f9d92f6 | 14 | struct token_s token[TOKEN_MAX] = {{NULL, {0}}}; |
dcfcd9ab JB |
15 | |
16 | FILE *source = NULL, *target = NULL; | |
6a19b8fe JB |
17 | wint_t c; |
18 | unsigned int tokenFound = 0; | |
4b580abd JB |
19 | enum TokenType { |
20 | MOTCLE, | |
21 | SECTION, | |
22 | SSECTION, | |
23 | NPARA, | |
24 | MOT, | |
25 | FIN | |
26 | } tokenType; | |
27 | const char* tokenTypestr[] = { "MOTCLE", "SECTION", "SSECTION", "NPARA", "MOT", "FIN" }; | |
28 | ||
91e46777 | 29 | /* It looks silly to check for each characters but for debugging, it's just the way to go */ |
4b580abd | 30 | bool istAlpha() { |
6a19b8fe JB |
31 | if (c == L'a' || c == L'b' || c == L'c' || c == L'd' || c == L'e' || c == L'f' || c == L'g' || \ |
32 | c == L'h' || c == L'i' || c == L'j' || c == L'k' || c == L'l' || c == L'm' || c == L'n' || \ | |
33 | c == L'o' || c == L'p' || c == L'q' || c == L'r' || c == L's' || c == L't' || c == L'u' || \ | |
34 | c == L'v' || c == L'w' || c == L'x' || c == L'y' || c == L'z' || \ | |
35 | c == L'A' || c == L'B' || c == L'C' || c == L'D' || c == L'E' || c == L'F' || c == L'G' || \ | |
36 | c == L'H' || c == L'I' || c == L'J' || c == L'K' || c == L'L' || c == L'M' || c == L'N' || \ | |
37 | c == L'O' || c == L'P' || c == L'Q' || c == L'R' || c == L'S' || c == L'T' || c == L'U' || \ | |
38 | c == L'V' || c == L'W' || c == L'X' || c == L'Y' || c == L'Z' || \ | |
39 | c == L'.' || c == L'?' || c == L'!' || c == L',' || c == L';' || c == L':' || c == L'-' || \ | |
40 | c == L'\''|| c == L'#' || \ | |
41 | c == L'0' || c == L'1' || c == L'2' || c == L'3' || c == L'4' || c == L'5' || c == L'6' || \ | |
42 | c == L'7' || c == L'8' || c == L'9' || \ | |
fa60d3b4 | 43 | // FIXME: Accentued characters (aka multibytes characters) support is still buggy |
6a19b8fe JB |
44 | c == L'à' || c == L'â' || c == L'ç' || c == L'è' || c == L'é' || c == L'î' || c == L'ô' || \ |
45 | c == L'ù' || c == L'û' || \ | |
46 | c == L'À' || c == L'Â' || c == L'Ç' || c == L'È' || c == L'É' || c == L'Î' || c == L'Ô' || \ | |
47 | c == L'Ù' || c == L'Û') { | |
4b580abd JB |
48 | return true; |
49 | } | |
50 | return false; | |
51 | } | |
52 | ||
53 | bool isSeparator() { | |
6a19b8fe | 54 | if (c == L'\t' || c == L' ' || c == L'\n') { |
4b580abd JB |
55 | return true; |
56 | } | |
57 | return false; | |
58 | } | |
59 | ||
60 | int scanner() { | |
62426b89 JB |
61 | unsigned int i = 0; |
62 | wchar_t m[6]; | |
4b580abd | 63 | |
4b580abd | 64 | init: |
6a19b8fe JB |
65 | if (c == L' ' || c == L'\t') { |
66 | c = fgetwc(source); | |
4b580abd JB |
67 | goto init; |
68 | } | |
6a19b8fe JB |
69 | if (c == L'\n') { |
70 | c = fgetwc(source); | |
fde9417f JB |
71 | goto initLV1; |
72 | } | |
6a19b8fe JB |
73 | if (c == L'>') { |
74 | c = fgetwc(source); | |
4b580abd JB |
75 | goto MC1; |
76 | } | |
6a19b8fe JB |
77 | if (c == L'=') { |
78 | c = fgetwc(source); | |
4b580abd JB |
79 | goto S1SS1; |
80 | } | |
81 | if (istAlpha()) { | |
6a19b8fe | 82 | token[tokenFound].value[i] = c; |
4b580abd | 83 | i++; |
62426b89 | 84 | c = fgetwc(source); |
4b580abd JB |
85 | goto M1; |
86 | } | |
6a19b8fe | 87 | if (c == WEOF) { |
4b580abd | 88 | goto FIN; |
4b580abd | 89 | } |
62426b89 | 90 | goto error; |
4b580abd JB |
91 | |
92 | MC1: | |
62426b89 JB |
93 | if (c == L'A' && !wcscmp(fgetws(m, 6, source), L"uteur")) { |
94 | wcscpy((wchar_t*)token[tokenFound].value, L">Auteur"); | |
6a19b8fe | 95 | c = fgetwc(source); |
62426b89 | 96 | goto MC2; |
4b580abd | 97 | } |
62426b89 JB |
98 | if (c == L'T' && !wcscmp(fgetws(m, 5, source), L"itre")) { |
99 | wcscpy((wchar_t*)token[tokenFound].value, L">Titre"); | |
6a19b8fe | 100 | c = fgetwc(source); |
4b580abd JB |
101 | goto MC2; |
102 | } | |
62426b89 | 103 | goto error; |
4b580abd JB |
104 | |
105 | S1SS1: | |
6a19b8fe JB |
106 | if (c == L'=') { |
107 | c = fgetwc(source); | |
4b580abd JB |
108 | goto SS2; |
109 | } | |
6a19b8fe | 110 | if (isSeparator() || c == WEOF) { |
d3eb30ef JB |
111 | goto SECTION; |
112 | } | |
62426b89 | 113 | goto error; |
d3eb30ef | 114 | |
4b580abd | 115 | SS2: |
6a19b8fe | 116 | if (isSeparator() || c == WEOF) { |
d3eb30ef | 117 | goto SSECTION; |
4b580abd | 118 | } |
62426b89 | 119 | goto error; |
4b580abd JB |
120 | |
121 | SECTION: | |
122 | tokenType = SECTION; | |
b5cd8f86 | 123 | return EXIT_SUCCESS; |
4b580abd | 124 | |
d3eb30ef JB |
125 | SSECTION: |
126 | tokenType = SSECTION; | |
b5cd8f86 | 127 | return EXIT_SUCCESS; |
d3eb30ef | 128 | |
4b580abd JB |
129 | M1: |
130 | if (istAlpha()) { | |
6a19b8fe | 131 | token[tokenFound].value[i] = c; |
4b580abd | 132 | i++; |
62426b89 | 133 | c = fgetwc(source); |
4b580abd JB |
134 | goto M1; |
135 | } | |
6a19b8fe | 136 | if (isSeparator() || c == WEOF) { |
4b580abd JB |
137 | goto MOT; |
138 | } | |
62426b89 | 139 | goto error; |
4b580abd JB |
140 | |
141 | initLV1: | |
6a19b8fe JB |
142 | if (c == L' ' || c == L'\t') { |
143 | c = fgetwc(source); | |
4b580abd JB |
144 | goto initLV1; |
145 | } | |
6a19b8fe JB |
146 | if (c == L'\n') { |
147 | c = fgetwc(source); | |
fde9417f JB |
148 | goto initLV1LV2; |
149 | } | |
4b580abd | 150 | if (istAlpha()) { |
6a19b8fe | 151 | token[tokenFound].value[i] = c; |
4b580abd | 152 | i++; |
62426b89 | 153 | c = fgetwc(source); |
4b580abd JB |
154 | goto M1; |
155 | } | |
6a19b8fe JB |
156 | if (c == L'=') { |
157 | c = fgetwc(source); | |
4b580abd JB |
158 | goto S1SS1; |
159 | } | |
6a19b8fe JB |
160 | if (c == L'>') { |
161 | c = fgetwc(source); | |
4b580abd JB |
162 | goto MC1; |
163 | } | |
6a19b8fe | 164 | if (c == WEOF) { |
4b580abd JB |
165 | goto FIN; |
166 | } | |
62426b89 | 167 | goto error; |
4b580abd JB |
168 | |
169 | initLV1LV2: | |
170 | if (isSeparator()) { | |
6a19b8fe | 171 | c = fgetwc(source); |
4b580abd JB |
172 | goto initLV1LV2; |
173 | } | |
fde9417f JB |
174 | if (istAlpha()) { |
175 | goto NPARA; | |
176 | } | |
6a19b8fe JB |
177 | if (c == L'>') { |
178 | c = fgetwc(source); | |
4b580abd JB |
179 | goto MC1; |
180 | } | |
6a19b8fe JB |
181 | if (c == L'=') { |
182 | c = fgetwc(source); | |
4b580abd JB |
183 | goto S1SS1; |
184 | } | |
6a19b8fe | 185 | if (c == WEOF) { |
d3eb30ef JB |
186 | goto FIN; |
187 | } | |
62426b89 | 188 | goto error; |
4b580abd JB |
189 | |
190 | NPARA: | |
191 | tokenType = NPARA; | |
b5cd8f86 | 192 | return EXIT_SUCCESS; |
4b580abd JB |
193 | |
194 | MOT: | |
195 | tokenType = MOT; | |
b5cd8f86 | 196 | return EXIT_SUCCESS; |
4b580abd JB |
197 | |
198 | MC2: | |
6a19b8fe | 199 | if (isSeparator() || c == WEOF) { |
4b580abd JB |
200 | goto MOTCLE; |
201 | } | |
62426b89 | 202 | goto error; |
4b580abd JB |
203 | |
204 | MOTCLE: | |
205 | tokenType = MOTCLE; | |
b5cd8f86 | 206 | return EXIT_SUCCESS; |
4b580abd JB |
207 | |
208 | FIN: | |
209 | tokenType = FIN; | |
b5cd8f86 | 210 | return EXIT_SUCCESS; |
4b580abd JB |
211 | |
212 | error: | |
213 | tokenType = FIN; | |
b5cd8f86 | 214 | return EXIT_FAILURE; |
4b580abd | 215 | } |
96964f3e | 216 | |
6a19b8fe | 217 | int main() { |
96964f3e | 218 | // Ouvre le fichier test.txt en lecture seulement (le fichier doit exister) : |
62426b89 | 219 | source = fopen("test.txt", "r+"); |
96964f3e JB |
220 | // Cree et ouvre un fichier target.html en lecture/ecriture |
221 | // avec suppression du contenu au prealable : | |
222 | target = fopen("target.html", "w+"); | |
223 | ||
224 | if (source == NULL) { | |
225 | printf("Impossible d'ouvrir le fichier source\n"); | |
226 | return -1; | |
227 | } | |
228 | ||
229 | if (target == NULL) { | |
230 | printf("Impossible d'ouvrir le fichier target\n"); | |
231 | return -1; | |
232 | } | |
233 | ||
62426b89 | 234 | c = fgetwc(source); // lecture du premier caractere |
4b580abd | 235 | do { |
4b580abd | 236 | int scanrt = scanner(); |
b5cd8f86 | 237 | if (scanrt == EXIT_FAILURE) { |
6a19b8fe | 238 | wprintf(L"Scanner error with token value: %ls\n", token[tokenFound].value); |
4b580abd JB |
239 | exit(EXIT_FAILURE); |
240 | } | |
3db137c9 | 241 | if (tokenType == MOT || tokenType == MOTCLE) { |
62426b89 | 242 | wprintf(L"%20s: %ls\n", tokenTypestr[tokenType], token[tokenFound].value); |
4b580abd | 243 | } else { |
62426b89 | 244 | wprintf(L"%20s\n", tokenTypestr[tokenType]); |
4b580abd | 245 | } |
6a19b8fe | 246 | token[tokenFound].type = tokenTypestr[tokenType]; |
28280a4c | 247 | tokenFound++; |
62426b89 | 248 | } while (tokenType != FIN); // tant que la fin du fichier n'est pas atteinte |
96964f3e JB |
249 | |
250 | if (source != NULL) fclose(source); // fermeture du fichier source | |
251 | if (target != NULL) fclose(target); // fermeture du fichier target | |
252 | ||
b5cd8f86 | 253 | return EXIT_SUCCESS; |
96964f3e | 254 | } |