// NEWSARC: News de-archiver // Copyright 1998 Sean Barrett // // Takes a file containing multiple articles written // out with [t]rn in non-mailbox format, and extracts // them into HTML files suitable for display on the web. // Also removes many headers #include #include #include #include #define BUFSIZE 4096 char buffer[BUFSIZE]; // more globals... this would have been a lot cleaner // if ALL the files were globals FILE *index_file; char subject[BUFSIZE]; char from[BUFSIZE]; void getline(FILE *f) { buffer[0] = 0; fgets(buffer, BUFSIZE, f); buffer[BUFSIZE-1] = 0; } int article_first_line(void) { char buf2[BUFSIZE]; int x; if (sscanf(buffer, "Article %d of %s", &x, buf2) == 2) return 1; return 0; } int is_empty(void) { char *s = buffer; while (*s && isspace(*s)) ++s; if (*s) return 0; return 1; } char *trim_headers[] = { "Xref: ", "Path: ", "Reply-to: ", "Followup-to: ", "Distribution: ", "NNTP-Posting-Host: ", "Mime-Version: ", "Content-Type: ", "Content-Transfer-Encoding: ", }; void copy_line(char *str, FILE *g) { char *s; for (s=str; *s; ++s) { switch(*s) { case '<': fprintf(g, "<"); break; case '>': fprintf(g, ">"); break; case '&': fprintf(g, "&"); break; default: fputc(*s, g); break; } } } void process_line(char *str, FILE *g, int *in_header) { if (*in_header) { int i; if (is_empty()) { *in_header = 0; fprintf(g, "\n"); return; } if (article_first_line()) return; for (i=0; i < sizeof(trim_headers)/sizeof(trim_headers[0]); ++i) { if (strncmp(trim_headers[i], str, strlen(trim_headers[i])) == 0) return; } if (strncmp(str, "Subject: ", 9)==0) { strcpy(subject, str+9); if (subject[strlen(subject)-1] == '\n') subject[strlen(subject)-1] = 0; } else if (strncmp(str, "From: ", 6) == 0) { strcpy(from, str+6); if (from[strlen(from)-1] == '\n') from[strlen(from)-1] = 0; } } copy_line(str, g); } char *ARTICLE_HEADER = "" "" "" "" "

"
     "\n";

char *ARTICLE_TAIL =
     "

" "" "" "\n"; // leading spaces in the index file mean // you can run it out to 'sort' and it will // leave the head at the top and the tail at the bottom char *INDEX_HEAD = " " "

\n"; char *INDEX_TAIL = ""; void copy_article(FILE *f, char *output_dir, int n) { char filename[256]; FILE *g; int in_header; // skip whitespace while (is_empty() && !feof(f)) { getline(f); } if (feof(f)) return; sprintf(filename, "%s/news%04d.html", output_dir, n); g = fopen(filename, "w"); if (!g) { fprintf(stderr, "Failed to open output file '%s'\n", filename); exit(1); } fputs(ARTICLE_HEADER, g); in_header = 1; strcpy(subject, ""); strcpy(from, ""); do { process_line(buffer, g, &in_header); getline(f); } while (!feof(f) && !article_first_line()); fputs(ARTICLE_TAIL, g); fclose(g); fprintf(index_file, "

"); process_line(from, index_file, &in_header); // tricky little hack fprintf(index_file, ": ", n); process_line(subject, index_file, &in_header); // tricky little hack fprintf(index_file, "\n"); } int main(int argc, char **argv) { FILE *f; int n; if (argc < 2 || argc > 3) { fprintf(stderr, "Usage: newsarc archive-file [output-directory]\n"); return 1; } f = fopen(argv[1], "r"); if (!f) { fprintf(stderr, "Couldn't open input file '%s'\n"); return 2; } { char index_name[256]; sprintf(index_name, "%s/index.html", argc == 3 ? argv[2] : "."); index_file = fopen(index_name, "w"); if (!index_file) { fprintf(stderr, "Failed to open output file '%s'\n", index_name); return 3; } fputs(INDEX_HEAD, index_file); } getline(f); for (n=1; !feof(f); ++n) { copy_article(f, argc == 3 ? argv[2] : ".", n); } fputs(INDEX_TAIL, index_file); fclose(index_file); fclose(f); return 0; }