… | |
… | |
36 | could be added) |
36 | could be added) |
37 | |
37 | |
38 | * Also does formatting commands (italics -> \emph, bold -> \textbf, |
38 | * Also does formatting commands (italics -> \emph, bold -> \textbf, |
39 | underline -> \underline), and character translation |
39 | underline -> \underline), and character translation |
40 | |
40 | |
|
|
41 | */ |
|
|
42 | |
|
|
43 | /* |
|
|
44 | 2002/Apr. Keita KAWABE |
|
|
45 | Support for narrow build Asian Psions added (experimental). |
|
|
46 | |
|
|
47 | If the encoding_type is PSICONV_ENCODING_UTF8, use utf8_table for |
|
|
48 | character conversion. Otherwise use char_table. |
|
|
49 | |
|
|
50 | When using UTF8 for LaTeX type, note that the resulting latex source |
|
|
51 | is not at all usable unless the user him/herself converts it to |
|
|
52 | appropriate encoding. This is because the encoding in localized TeXs |
|
|
53 | differ from system to system. |
|
|
54 | For example, if the user wants to generate a Japanese LaTeX source |
|
|
55 | from a UTF8 Psion Word, he/she first has to convert the Word document by |
|
|
56 | |
|
|
57 | psiconv -TLaTeX -u word | iconv --from-code=utf8 --to-code=EUC-JP > test.tex |
|
|
58 | |
|
|
59 | or something similar. After that, just proceed as normal. |
|
|
60 | |
|
|
61 | If you want to use Japanese-localized documentclass "jarticle" rather |
|
|
62 | than "article", please uncomment all the commented-out lines that are |
|
|
63 | explicitly stated as "Uncomment if you'd like to use jarticle" in this |
|
|
64 | source. |
41 | */ |
65 | */ |
42 | |
66 | |
43 | #include "config.h" |
67 | #include "config.h" |
44 | #include <stdio.h> |
68 | #include <stdio.h> |
45 | #include <string.h> |
69 | #include <string.h> |
46 | #include <stdlib.h> |
70 | #include <stdlib.h> |
47 | #include "psiconv/data.h" |
71 | #include "psiconv/data.h" |
48 | #include "psiconv/list.h" |
72 | #include "psiconv/list.h" |
49 | #include "gen.h" |
73 | #include "gen.h" |
50 | #include "psiconv.h" |
74 | #include "psiconv.h" |
|
|
75 | |
|
|
76 | #ifdef DMALLOC |
|
|
77 | #include "dmalloc.h" |
|
|
78 | #endif |
51 | |
79 | |
52 | /* This is incomplete at the moment. |
80 | /* This is incomplete at the moment. |
53 | |
81 | |
54 | Most of the translation/faking of chars is borrowed from the |
82 | Most of the translation/faking of chars is borrowed from the |
55 | HTML::Latex.pm perl module. */ |
83 | HTML::Latex.pm perl module. */ |
… | |
… | |
109 | "\\`{\\i}","\\'{\\i}","\\^{\\i}","\\\"{\\i}", |
137 | "\\`{\\i}","\\'{\\i}","\\^{\\i}","\\\"{\\i}", |
110 | /* 0xf0 */ "\\v{o}","\\~n","\\`o","\\'o", "\\^o","\\~o","\\\"o","$\\div$", |
138 | /* 0xf0 */ "\\v{o}","\\~n","\\`o","\\'o", "\\^o","\\~o","\\\"o","$\\div$", |
111 | /* 0xf8 */ "{\\o}","\\`u","\\'u","\\^u", "\\\"u","\\'y","","\\\"y" |
139 | /* 0xf8 */ "{\\o}","\\`u","\\'u","\\^u", "\\\"u","\\'y","","\\\"y" |
112 | }; |
140 | }; |
113 | |
141 | |
|
|
142 | static const char *utf_table[0x100] = |
|
|
143 | { |
|
|
144 | /* 0x00 */ "" ,"" ,"" ,"" ,"" ,"" ,"\n\n","\\\\" |
|
|
145 | , |
|
|
146 | /* 0x08 */ "\n\n" ," " ,"" ,"" ,"" ,"" ,"" ,"" |
|
|
147 | , |
|
|
148 | /* 0x10 */ " " ,"" ,"" ,"" ,"" ,"" ,"" ,"" |
|
|
149 | , |
|
|
150 | /* 0x18 */ "" ,"" ,"" ,"" ,"" ,"" ,"" ,"" |
|
|
151 | , |
|
|
152 | /* 0x20 */ " " ,"!" ,"\"" ,"\\#" ,"\\$" ,"\\%" ,"\\&","'" |
|
|
153 | , |
|
|
154 | /* 0x28 */ "(" ,")" ,"*" ,"+" ,"," ,"-" ,"." ,"/" |
|
|
155 | , |
|
|
156 | /* 0x30 */ "0" ,"1" ,"2" ,"3" ,"4" ,"5" ,"6" ,"7" |
|
|
157 | , |
|
|
158 | /* 0x38 */ "8" ,"9" ,":" ,";" ,"$<$" ,"=" ,"$>$" ,"?" , |
|
|
159 | /* 0x40 */ "@" ,"A" ,"B" ,"C" ,"D" ,"E" ,"F" ,"G" |
|
|
160 | , |
|
|
161 | /* 0x48 */ "H" ,"I" ,"J" ,"K" ,"L" ,"M" ,"N" ,"O" |
|
|
162 | , |
|
|
163 | /* 0x50 */ "P" ,"Q" ,"R" ,"S" ,"T" ,"U" ,"V" ,"W" |
|
|
164 | , |
|
|
165 | /* 0x58 */ "X" ,"Y" ,"Z" ,"[" ,"$\\backslash$" ,"]" ,"\\^{}" |
|
|
166 | ,"\\_" , |
|
|
167 | /* 0x60 */ "`" ,"a" ,"b" ,"c" ,"d" ,"e" ,"f" ,"g" |
|
|
168 | , |
|
|
169 | /* 0x68 */ "h" ,"i" ,"j" ,"k" ,"l" ,"m" ,"n" ,"o" |
|
|
170 | , |
|
|
171 | /* 0x70 */ "p" ,"q" ,"r" ,"s" ,"t" ,"u" ,"v" ,"w" |
|
|
172 | , |
|
|
173 | /* 0x78 */ "x" ,"y" ,"z" ,"\\{" ,"$|$" ,"\\}" ,"~" |
|
|
174 | ,"" , |
|
|
175 | /* 0x80 */ "\x80", "\x81", "\x82", "\x83", "\x84", "\x85", "\x86", "\x87", |
|
|
176 | /* 0x88 */ "\x88", "\x89", "\x8a", "\x8b", "\x8c", "\x8d", "\x8e", "\x8f", |
|
|
177 | /* 0x90 */ "\x90", "\x91", "\x92", "\x93", "\x94", "\x95", "\x96", "\x97", |
|
|
178 | /* 0x98 */ "\x98", "\x99", "\x9a", "\x9b", "\x9c", "\x9d", "\x9e", "\x9f", |
|
|
179 | /* 0xa0 */ "\xa0", "\xa1", "\xa2", "\xa3", "\xa4", "\xa5", "\xa6", "\xa7", |
|
|
180 | /* 0xa8 */ "\xa8", "\xa9", "\xaa", "\xab", "\xac", "\xad", "\xae", "\xaf", |
|
|
181 | /* 0xb0 */ "\xb0", "\xb1", "\xb2", "\xb3", "\xb4", "\xb5", "\xb6", "\xb7", |
|
|
182 | /* 0xb8 */ "\xb8", "\xb9", "\xba", "\xbb", "\xbc", "\xbd", "\xbe", "\xbf", |
|
|
183 | /* 0xc0 */ "\xc0", "\xc1", "\xc2", "\xc3", "\xc4", "\xc5", "\xc6", "\xc7", |
|
|
184 | /* 0xc8 */ "\xc8", "\xc9", "\xca", "\xcb", "\xcc", "\xcd", "\xce", "\xcf", |
|
|
185 | /* 0xd0 */ "\xd0", "\xd1", "\xd2", "\xd3", "\xd4", "\xd5", "\xd6", "\xd7", |
|
|
186 | /* 0xd8 */ "\xd8", "\xd9", "\xda", "\xdb", "\xdc", "\xdd", "\xde", "\xdf", |
|
|
187 | /* 0xe0 */ "\xe0", "\xe1", "\xe2", "\xe3", "\xe4", "\xe5", "\xe6", "\xe7", |
|
|
188 | /* 0xe8 */ "\xe8", "\xe9", "\xea", "\xeb", "\xec", "\xed", "\xee", "\xef", |
|
|
189 | /* 0xf0 */ "\xf0", "\xf1", "\xf2", "\xf3", "\xf4", "\xf5", "\xf6", "\xf7", |
|
|
190 | /* 0xf8 */ "\xf8", "\xf9", "\xfa", "\xfb", "\xfc", "\xfd", "\xfe", "\xff", |
|
|
191 | }; |
|
|
192 | |
|
|
193 | /* a flag to indicate the use of UTF8 */ |
|
|
194 | static psiconv_encoding encoding = PSICONV_ENCODING_CP1252; |
|
|
195 | |
114 | static psiconv_character_layout gen_base_char(const psiconv_font font, |
196 | static psiconv_character_layout gen_base_char(const psiconv_font font, |
115 | const psiconv_color color, |
197 | const psiconv_color color, |
116 | const psiconv_color |
198 | const psiconv_color |
117 | back_color); |
199 | back_color); |
118 | static void diff_char(FILE *of, const psiconv_character_layout old, |
200 | static void diff_char(FILE *of, const psiconv_character_layout old, |
… | |
… | |
218 | psiconv_in_line_layout inl; |
300 | psiconv_in_line_layout inl; |
219 | int flags = 0; |
301 | int flags = 0; |
220 | psiconv_word_style sty; |
302 | psiconv_word_style sty; |
221 | char *env = NULL; |
303 | char *env = NULL; |
222 | |
304 | |
|
|
305 | char ** table; |
|
|
306 | if (encoding == PSICONV_ENCODING_UTF8){ |
|
|
307 | table = (char**)utf_table; |
|
|
308 | }else{ |
|
|
309 | table = (char**)char_table; |
|
|
310 | } |
|
|
311 | |
223 | if (para->base_paragraph->bullet->on) { |
312 | if (para->base_paragraph->bullet->on) { |
224 | if (! bullet_switch_on) { |
313 | if (! bullet_switch_on) { |
225 | fputs("\\begin{itemize}\n\n", of); |
314 | fputs("\\begin{itemize}\n\n", of); |
226 | bullet_switch_on = psiconv_bool_true; |
315 | bullet_switch_on = psiconv_bool_true; |
227 | } |
316 | } |
… | |
… | |
262 | for (i = 0; i < psiconv_list_length(para->in_lines); i++) { |
351 | for (i = 0; i < psiconv_list_length(para->in_lines); i++) { |
263 | inl = psiconv_list_get(para->in_lines,i); |
352 | inl = psiconv_list_get(para->in_lines,i); |
264 | diff_char(of,cur_char,inl->layout,&flags); |
353 | diff_char(of,cur_char,inl->layout,&flags); |
265 | cur_char = inl->layout; |
354 | cur_char = inl->layout; |
266 | for (j = loc; j < inl->length + loc; j ++) { |
355 | for (j = loc; j < inl->length + loc; j ++) { |
267 | fputs(char_table[(unsigned char) (para->text[j])],of); |
356 | fputs(table[(unsigned char) (para->text[j])],of); |
268 | } |
357 | } |
269 | loc = j; |
358 | loc = j; |
270 | } |
359 | } |
271 | |
360 | |
272 | if (loc < strlen(para->text)) { |
361 | if (loc < strlen(para->text)) { |
273 | diff_char(of,cur_char,para->base_character,&flags); |
362 | diff_char(of,cur_char,para->base_character,&flags); |
274 | cur_char = para->base_character; |
363 | cur_char = para->base_character; |
275 | for (j = loc; j < strlen(para->text); j ++) { |
364 | for (j = loc; j < strlen(para->text); j ++) { |
276 | fputs(char_table[(unsigned char) (para->text[j])],of); |
365 | fputs(table[(unsigned char) (para->text[j])],of); |
277 | } |
366 | } |
278 | } |
367 | } |
279 | |
368 | |
280 | flags |= 2; |
369 | flags |= 2; |
281 | diff_char(of,cur_char,base_char,&flags); |
370 | diff_char(of,cur_char,base_char,&flags); |
… | |
… | |
288 | |
377 | |
289 | fputs("\n\n", of); |
378 | fputs("\n\n", of); |
290 | } |
379 | } |
291 | |
380 | |
292 | int psiconv_gen_latex(const char * filename,const psiconv_file file, |
381 | int psiconv_gen_latex(const char * filename,const psiconv_file file, |
293 | const char *dest) |
382 | const char *dest, const psiconv_encoding encoding_type) |
294 | { |
383 | { |
295 | FILE *of = fopen(filename,"w"); |
384 | FILE *of = fopen(filename,"w"); |
296 | if (! of) |
385 | if (! of) |
297 | return -1; |
386 | return -1; |
|
|
387 | |
|
|
388 | encoding = encoding_type; |
298 | |
389 | |
299 | if (file->type == psiconv_word_file) { |
390 | if (file->type == psiconv_word_file) { |
300 | psiconv_gen_latex_word(of,(psiconv_word_f) file->file); |
391 | psiconv_gen_latex_word(of,(psiconv_word_f) file->file); |
301 | } else if (file->type == psiconv_texted_file) { |
392 | } else if (file->type == psiconv_texted_file) { |
302 | psiconv_gen_latex_texted(of,(psiconv_texted_f) file->file); |
393 | psiconv_gen_latex_texted(of,(psiconv_texted_f) file->file); |
… | |
… | |
315 | int i; |
406 | int i; |
316 | |
407 | |
317 | /* We have nothing better */ |
408 | /* We have nothing better */ |
318 | base_char = psiconv_basic_character_layout(); |
409 | base_char = psiconv_basic_character_layout(); |
319 | |
410 | |
|
|
411 | /* Uncomment if you'd like to use jarticle... |
|
|
412 | if (encoding == PSICONV_ENCODING_UTF8){ |
|
|
413 | fputs("\\documentclass{jarticle}\n\n\\begin{document}\n\n", of); |
|
|
414 | }else{ |
|
|
415 | */ |
320 | fputs("\\documentclass{article}\n\n\\begin{document}\n\n", of); |
416 | fputs("\\documentclass{article}\n\n\\begin{document}\n\n", of); |
|
|
417 | /* Uncomment if you'd like to use jarticle... |
|
|
418 | } |
|
|
419 | */ |
321 | for (i = 0; i < psiconv_list_length(tf->texted_sec->paragraphs); i++) { |
420 | for (i = 0; i < psiconv_list_length(tf->texted_sec->paragraphs); i++) { |
322 | para = psiconv_list_get(tf->texted_sec->paragraphs,i); |
421 | para = psiconv_list_get(tf->texted_sec->paragraphs,i); |
323 | gen_para(of,para,base_char, NULL); |
422 | gen_para(of,para,base_char, NULL); |
324 | } |
423 | } |
325 | fputs("\\end{document}",of); |
424 | fputs("\\end{document}",of); |
… | |
… | |
347 | black,white); |
446 | black,white); |
348 | |
447 | |
349 | psiconv_free_color(black); |
448 | psiconv_free_color(black); |
350 | psiconv_free_color(white); |
449 | psiconv_free_color(white); |
351 | |
450 | |
|
|
451 | /* Uncomment if you'd like to use jarticle |
|
|
452 | if (encoding == PSICONV_ENCODING_UTF8){ |
|
|
453 | fputs("\\documentclass{jarticle}\n\n\\begin{document}\n\n", of); |
|
|
454 | }else{ |
|
|
455 | */ |
352 | fputs("\\documentclass{article}\n\n\\begin{document}\n\n", of); |
456 | fputs("\\documentclass{article}\n\n\\begin{document}\n\n", of); |
|
|
457 | /* Uncomment if you'd like to use jarticle |
|
|
458 | } |
|
|
459 | */ |
353 | |
460 | |
354 | for (i = 0; i < psiconv_list_length(wf->paragraphs); i++) { |
461 | for (i = 0; i < psiconv_list_length(wf->paragraphs); i++) { |
355 | para = psiconv_list_get(wf->paragraphs,i); |
462 | para = psiconv_list_get(wf->paragraphs,i); |
356 | gen_para(of,para,base_char,wf); |
463 | gen_para(of,para,base_char,wf); |
357 | } |
464 | } |