--- psiconv/trunk/program/psiconv/psiconv.c 2000/10/21 00:49:13 53 +++ psiconv/trunk/program/psiconv/psiconv.c 2004/01/09 22:20:03 185 @@ -17,6 +17,22 @@ Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ +/* + 2002/Apr. Keita KAWABE + * Support for narrow build Asian Psions added. + + Now psiconv understands -u or --UTF8 option. + This option is passed to output_function as the fourth parameter + (thus the typedef of output_function was changed). + + If a called generator (gen_*.c) can handle UTF8, then it handles + the character conversion etc. accordingly. + + * psiconv seemed to want to take the argument of the options + case-insensitively (i.e. LaTeX -> LATEX), but that was not working + under my environment. Fixed it. +*/ + /* Driver program */ #include "config.h" @@ -33,30 +49,30 @@ #endif #ifdef IMAGEMAGICK -#include +#include "magick-aux.h" #endif -#include "data.h" -#include "parse.h" -#include "gen.h" +#include +#include +#include +#include #include "psiconv.h" +#include "gen.h" static void print_help(void); static void print_version(void); static void strtoupper(char *str); -int psiconv_verbosity = PSICONV_VERB_WARN; - void print_help(void) { - psiconv_fileformat ff; + fileformat ff; int i,j; puts("Syntax: psiconv [OPTIONS..] [FILE]"); - puts("Convert the psion Word file FILE to a HTML file"); + puts("Convert the psion file FILE to other formats"); puts("If FILE is not specified, use stdin"); puts(" -d, --debug Show debug information on stderr"); - puts(" -e, --exact Quit on any parse errors or warnings"); + puts(" -e, --encoding Output encoding (default: UTF8)"); puts(" -h, --help Display this help and exit"); puts(" -o, --outputfile Output to file instead of stdout"); puts(" -s, --silent Do not even show warnings on stderr"); @@ -64,6 +80,12 @@ puts(" -V, --version Display the program version and exit"); puts(" -v, --verbose Show progress indicators on stderr"); puts(""); + puts("The following encodings are currently supported:"); + puts(" UTF8 Variable length Unicode encoding"); + puts(" UCS2 Fixed 16-bit length Unicode encoding"); + puts(" Psion The encoding your Psion uses (as in psiconv.conf)"); + puts(" ASCII 7-bit ASCII (other symbols are substituted by '?')"); + puts(""); puts("The following output types are known:"); for (i = 0; i < psiconv_list_length(fileformat_list); i ++) { ff = psiconv_list_get(fileformat_list,i); @@ -72,6 +94,9 @@ putchar(' '); puts(ff->description); } + puts(""); + puts("When using UTF8 with LaTeX type, the resulting LaTeX source should be converted"); + puts(" to a suitable encoding for your LaTeX installation before being typeset"); } void print_version(void) @@ -97,10 +122,10 @@ {"silent",no_argument,NULL,'s'}, {"outputfile",required_argument,NULL,'o'}, {"type",required_argument,NULL,'T'}, - {"exact",no_argument,NULL,'e'}, + {"encoding",no_argument,NULL,'e'}, {0,0,0,0} }; - const char* short_options = "hVvsdo:eT:"; + const char* short_options = "hVvsdo:e:T:"; int option_index; FILE * f; struct stat fbuf; @@ -108,19 +133,30 @@ const char *inputfilename = ""; const char *outputfilename = ""; char *type = strdup("HTML3"); - int exact=0; + encoding encoding_type=ENCODING_UTF8; + psiconv_list outputlist; int c,i,res; psiconv_buffer buf; psiconv_file file; - psiconv_fileformat ff; + fileformat ff = NULL; + + psiconv_config config = psiconv_config_default(); + psiconv_config_read(NULL,&config); + + if (!(fileformat_list = psiconv_list_new(sizeof(struct fileformat_s)))) { + fputs("Out of memory error",stderr); + exit(1); + } - fileformat_list = psiconv_list_new(sizeof(struct psiconv_fileformat)); init_txt(); +#if 0 init_html(); init_html4(); + init_latex(); init_rtf(); init_image(); +#endif while(1) { c = getopt_long(argc,argv,short_options, long_options, &option_index); @@ -129,14 +165,27 @@ switch(c) { case 'h': print_help(); exit(0); case 'V': print_version(); exit(0); - case 'v': if (psiconv_verbosity < PSICONV_VERB_PROGRESS) - psiconv_verbosity = PSICONV_VERB_PROGRESS; + case 'v': if (config->verbosity < PSICONV_VERB_PROGRESS) + config->verbosity = PSICONV_VERB_PROGRESS; break; - case 'd': psiconv_verbosity = PSICONV_VERB_DEBUG; break; - case 's': psiconv_verbosity = PSICONV_VERB_SILENT; break; - case 'e': exact = 1; break; + case 'd': config->verbosity = PSICONV_VERB_DEBUG; break; + case 's': config->verbosity = PSICONV_VERB_ERROR; break; case 'o': outputfilename = strdup(optarg); break; case 'T': type = strdup(optarg); break; + case 'e': if(!strcmp(optarg,"UTF8")) + encoding_type = ENCODING_UTF8; + else if (!strcmp(optarg,"UCS2")) + encoding_type = ENCODING_UCS2; + else if (!strcmp(optarg,"ASCII")) + encoding_type = ENCODING_ASCII; + else if (!strcmp(optarg,"Psion")) + encoding_type = ENCODING_PSION; + else { + fputs("Unknown encoding type " + "(try '-h' for more information\n",stderr); + exit(1); + } + break; case '?': case ':': fputs("Try `-h' for more information\n",stderr); exit(1); default: fprintf(stderr,"Internal error: getopt_long returned character " @@ -149,7 +198,11 @@ "Try `-h' for more information\n",stderr); exit(1); } else if (optind == argc-1) - inputfilename = strdup(argv[optind]); + if (!(inputfilename = strdup(argv[optind]))) { + fputs("Out of memory error",stderr); + exit(1); + } + /* Open inputfile for reading */ @@ -166,9 +219,14 @@ } else f = stdin; - buf = psiconv_list_new(sizeof(psiconv_u8)); - while (! feof(f)) - psiconv_list_fread(buf,1024,f); + if (!(buf = psiconv_buffer_new())) { + fputs("Out of memory error",stderr); + exit(1); + } + if (psiconv_buffer_fread_all(buf,f)) { + fprintf(stderr,"Failure reading file"); + exit(1); + } if (strlen(inputfilename) != 0) if (fclose(f)) { @@ -176,24 +234,16 @@ exit(1); } - if (psiconv_parse(buf,&file)) - if(exact) - exit(1); - - /* Set correct output file */ - if (strlen(outputfilename) == 0) - outputfilename = "/dev/stdout"; + if (psiconv_parse(config,buf,&file) || (file->type == psiconv_unknown_file)) + { + fprintf(stderr,"Parse error\n"); + exit(1); + } strtoupper(type); for (i = 0; i < psiconv_list_length(fileformat_list); i ++) { ff = psiconv_list_get(fileformat_list,i); - if (! strcmp(type,ff->name)) { - res = ff->output(outputfilename,file,type); - if (res) { - fprintf(stderr, - "Output format `%s' not permitted for this file type\n",type); - exit(1); - } + if (! strcasecmp(type,ff->name)) { break; } } @@ -203,7 +253,37 @@ exit(1); } + if (!(outputlist = psiconv_list_new(sizeof(psiconv_u8)))) { + fputs("Out of memory error\n",stderr); + exit(1); + } + + res = ff->output(config,outputlist,file,type,encoding_type); + if (res) { + fprintf(stderr, + "Output format `%s' not permitted for this file type\n",type); + exit(1); + } + psiconv_free_file(file); + if (strlen(outputfilename) != 0) { + f = fopen(outputfilename,"w"); + if (! f) { + perror(inputfilename); + exit(1); + } + } else + f = stdout; + + psiconv_list_fwrite_all(outputlist,f); + + if (fclose(f)) { + perror(inputfilename); + exit(1); + } + + psiconv_list_free(outputlist); + exit(0); }