xref: /plan9-contrib/sys/src/cmd/postscript/text2post/text2post.c (revision 9a747e4fd48b9f4522c70c07e8f882a15030f964)
1 #include <u.h>
2 #include <libc.h>
3 #include <ctype.h>
4 #include <bio.h>
5 #include <comments.h>
6 #include <path.h>
7 
8 #define UNKNOWNCHAR	"/sys/lib/postscript/prologues/pjw.char.ps"
9 
10 char	*optnames = "a:c:f:l:m:n:o:p:s:t:x:y:P:";
11 
12 Biobuf *bstdin, *bstdout, *bstderr;
13 Biobufhdr *Bstdin, *Bstdout, *Bstderr;
14 int char_no = 0;		/* character to be done on a line */
15 int line_no = 0;		/* line number on a page */
16 int page_no = 0;		/* page number in a document */
17 int in_string;		/* Boolean, to know whether or not we are inside a Postscript string */
18 int spaces = 0;
19 int tabs = 0;
20 int pages_printed;
21 double aspectratio = 1.0;
22 int copies = 1;
23 double magnification = 1.0;
24 int landscape = 0;
25 int formsperpage = 1;
26 int linesperpage = 66;
27 int pointsize = 10;
28 double xoffset = .25;
29 double yoffset = .25;
30 char *passthrough = 0;
31 static int pplistmaxsize=0;
32 
33 unsigned char *pplist=0;	/* bitmap list for storing pages to print */
34 
35 struct strtab {
36 	int size;
37 	char *str;
38 	int used;
39 };
40 
41 struct strtab charcode[256] = {
42 	{4, "\\000"}, {4, "\\001"}, {4, "\\002"}, {4, "\\003"},
43 	{4, "\\004"}, {4, "\\005"}, {4, "\\006"}, {4, "\\007"},
44 	{4, "\\010"}, {4, "\\011"}, {4, "\\012"}, {4, "\\013"},
45 	{4, "\\014"}, {4, "\\015"}, {4, "\\016"}, {4, "\\017"},
46 	{4, "\\020"}, {4, "\\021"}, {4, "\\022"}, {4, "\\023"},
47 	{4, "\\024"}, {4, "\\025"}, {4, "\\026"}, {4, "\\027"},
48 	{4, "\\030"}, {4, "\\031"}, {4, "\\032"}, {4, "\\033"},
49 	{4, "\\034"}, {4, "\\035"}, {4, "\\036"}, {4, "\\037"},
50 	{1, " "}, {1, "!"}, {1, "\""}, {1, "#"},
51 	{1, "$"}, {1, "%"}, {1, "&"}, {1, "'"},
52 	{2, "\\("}, {2, "\\)"}, {1, "*"}, {1, "+"},
53 	{1, ","}, {1, "-"}, {1, "."}, {1, "/"},
54 	{1, "0"}, {1, "1"}, {1, "2"}, {1, "3"},
55 	{1, "4"}, {1, "5"}, {1, "6"}, {1, "7"},
56 	{1, "8"}, {1, "9"}, {1, ":"}, {1, ";"},
57 	{1, "<"}, {1, "="}, {1, ">"}, {1, "?"},
58 	{1, "@"}, {1, "A"}, {1, "B"}, {1, "C"},
59 	{1, "D"}, {1, "E"}, {1, "F"}, {1, "G"},
60 	{1, "H"}, {1, "I"}, {1, "J"}, {1, "K"},
61 	{1, "L"}, {1, "M"}, {1, "N"}, {1, "O"},
62 	{1, "P"}, {1, "Q"}, {1, "R"}, {1, "S"},
63 	{1, "T"}, {1, "U"}, {1, "V"}, {1, "W"},
64 	{1, "X"}, {1, "Y"}, {1, "Z"}, {1, "["},
65 	{2, "\\\\"}, {1, "]"}, {1, "^"}, {1, "_"},
66 	{1, "`"}, {1, "a"}, {1, "b"}, {1, "c"},
67 	{1, "d"}, {1, "e"}, {1, "f"}, {1, "g"},
68 	{1, "h"}, {1, "i"}, {1, "j"}, {1, "k"},
69 	{1, "l"}, {1, "m"}, {1, "n"}, {1, "o"},
70 	{1, "p"}, {1, "q"}, {1, "r"}, {1, "s"},
71 	{1, "t"}, {1, "u"}, {1, "v"}, {1, "w"},
72 	{1, "x"}, {1, "y"}, {1, "z"}, {1, "{"},
73 	{1, "|"}, {1, "}"}, {1, "~"}, {4, "\\177"},
74 	{4, "\\200"}, {4, "\\201"}, {4, "\\202"}, {4, "\\203"},
75 	{4, "\\204"}, {4, "\\205"}, {4, "\\206"}, {4, "\\207"},
76 	{4, "\\210"}, {4, "\\211"}, {4, "\\212"}, {4, "\\213"},
77 	{4, "\\214"}, {4, "\\215"}, {4, "\\216"}, {4, "\\217"},
78 	{4, "\\220"}, {4, "\\221"}, {4, "\\222"}, {4, "\\223"},
79 	{4, "\\224"}, {4, "\\225"}, {4, "\\226"}, {4, "\\227"},
80 	{4, "\\230"}, {4, "\\231"}, {4, "\\232"}, {4, "\\233"},
81 	{4, "\\234"}, {4, "\\235"}, {4, "\\236"}, {4, "\\237"},
82 	{4, "\\240"}, {4, "\\241"}, {4, "\\242"}, {4, "\\243"},
83 	{4, "\\244"}, {4, "\\245"}, {4, "\\246"}, {4, "\\247"},
84 	{4, "\\250"}, {4, "\\251"}, {4, "\\252"}, {4, "\\253"},
85 	{4, "\\254"}, {4, "\\255"}, {4, "\\256"}, {4, "\\257"},
86 	{4, "\\260"}, {4, "\\261"}, {4, "\\262"}, {4, "\\263"},
87 	{4, "\\264"}, {4, "\\265"}, {4, "\\266"}, {4, "\\267"},
88 	{4, "\\270"}, {4, "\\271"}, {4, "\\272"}, {4, "\\273"},
89 	{4, "\\274"}, {4, "\\275"}, {4, "\\276"}, {4, "\\277"},
90 	{4, "\\300"}, {4, "\\301"}, {4, "\\302"}, {4, "\\303"},
91 	{4, "\\304"}, {4, "\\305"}, {4, "\\306"}, {4, "\\307"},
92 	{4, "\\310"}, {4, "\\311"}, {4, "\\312"}, {4, "\\313"},
93 	{4, "\\314"}, {4, "\\315"}, {4, "\\316"}, {4, "\\317"},
94 	{4, "\\320"}, {4, "\\321"}, {4, "\\322"}, {4, "\\323"},
95 	{4, "\\324"}, {4, "\\325"}, {4, "\\326"}, {4, "\\327"},
96 	{4, "\\330"}, {4, "\\331"}, {4, "\\332"}, {4, "\\333"},
97 	{4, "\\334"}, {4, "\\335"}, {4, "\\336"}, {4, "\\337"},
98 	{4, "\\340"}, {4, "\\341"}, {4, "\\342"}, {4, "\\343"},
99 	{4, "\\344"}, {4, "\\345"}, {4, "\\346"}, {4, "\\347"},
100 	{4, "\\350"}, {4, "\\351"}, {4, "\\352"}, {4, "\\353"},
101 	{4, "\\354"}, {4, "\\355"}, {4, "\\356"}, {4, "\\357"},
102 	{4, "\\360"}, {4, "\\361"}, {4, "\\362"}, {4, "\\363"},
103 	{4, "\\364"}, {4, "\\365"}, {4, "\\366"}, {4, "\\367"},
104 	{4, "\\370"}, {4, "\\371"}, {4, "\\372"}, {4, "\\373"},
105 	{4, "\\374"}, {4, "\\375"}, {4, "\\376"}, {4, "\\377"}
106 };
107 
108 #define FONTABSIZE 0x27
109 
110 struct strtab fontname[FONTABSIZE] = {
111 	{19, "LucidaSansUnicode00", 0},
112 	{19, "LucidaSansUnicode01", 0},
113 	{19, "LucidaSansUnicode02", 0},
114 	{19, "LucidaSansUnicode03", 0},
115 	{19, "LucidaSansUnicode04", 0},
116 	{19, "LucidaSansUnicode05", 0},
117 	{0, "", 0},
118 	{0, "", 0},
119 	{0, "", 0},
120 	{0, "", 0},
121 	{0, "", 0},
122 	{0, "", 0},
123 	{0, "", 0},
124 	{0, "", 0},
125 	{0, "", 0},
126 	{0, "", 0},
127 	{0, "", 0},
128 	{0, "", 0},
129 	{0, "", 0},
130 	{0, "", 0},
131 	{0, "", 0},
132 	{0, "", 0},
133 	{0, "", 0},
134 	{0, "", 0},
135 	{0, "", 0},
136 	{0, "", 0},
137 	{0, "", 0},
138 	{0, "", 0},
139 	{0, "", 0},
140 	{0, "", 0},
141 	{0, "", 0},
142 	{0, "", 0},
143 	{19, "LucidaSansUnicode20", 0},
144 	{19, "LucidaSansUnicode21", 0},
145 	{19, "LucidaSansUnicode22", 0},
146 	{0, "", 0},
147 	{19, "LucidaSansUnicode24", 0},
148 	{19, "LucidaSansUnicode25", 0},
149 	{7, "Courier", 0}
150 };
151 
152 /* This was taken from postprint */
153 
154 int
155 cat(char *filename) {
156 	Biobuf *bfile;
157 	Biobufhdr *Bfile;
158 	int n;
159 	static char buf[Bsize];
160 
161 	bstdin = Bopen(filename, 0);
162 	if (bstdin == 0) {
163 		return(1);
164 	}
165 	Bstdin = &(bstdin->Biobufhdr);
166 	if ((bfile = Bopen(filename, OREAD)) == 0) {
167 		return(1);
168 	}
169 	Bfile = &(bfile->Biobufhdr);
170 	while ((n=Bread(Bfile, buf, Bsize)) > 0) {
171 		if (Bwrite(Bstdout, buf, n) != n) {
172 			return(1);
173 		}
174 	}
175 	if (n != 0) {
176 		return(1);
177 	}
178 	return(0);
179 }
180 
181 void
182 prologues(void) {
183 	char *ts;
184 	int tabstop;
185 
186 	Bprint(Bstdout, "%s", CONFORMING);
187 	Bprint(Bstdout, "%s %s\n", VERSION, PROGRAMVERSION);
188 	Bprint(Bstdout, "%s %s\n", DOCUMENTFONTS, ATEND);
189 	Bprint(Bstdout, "%s %s\n", PAGES, ATEND);
190 	Bprint(Bstdout, "%s", ENDCOMMENTS);
191 
192 	if (cat(POSTPRINT)) {
193 		Bprint(Bstderr, "can't read %s", POSTPRINT);
194 		exits("prologue");
195 	}
196 
197 	if (DOROUND)
198 		cat(ROUNDPAGE);
199 
200 	tabstop = 0;
201 	ts = getenv("tabstop");
202 	if(ts != nil)
203 		tabstop = strtol(ts, nil, 0);
204 	if(tabstop == 0)
205 		tabstop = 8;
206 	Bprint(Bstdout, "/f {findfont pointsize scalefont setfont} bind def\n");
207 	Bprint(Bstdout, "/tabwidth /Courier f (");
208 	while(tabstop--)
209 		Bputc(Bstdout, 'n');
210 	Bprint(Bstdout, ") stringwidth pop def\n");
211 	Bprint(Bstdout, "/tab {tabwidth 0 ne {currentpoint 3 1 roll exch tabwidth mul add tabwidth\n");
212 	Bprint(Bstdout, "\tdiv truncate tabwidth mul exch moveto} if} bind def\n");
213 	Bprint(Bstdout, "/spacewidth /%s f ( ) stringwidth pop def\n", fontname[0].str);
214 	Bprint(Bstdout, "/sp {spacewidth mul 0 rmoveto} bind def\n");
215 	Bprint(Bstdout, "%s", ENDPROLOG);
216 	Bprint(Bstdout, "%s", BEGINSETUP);
217 	Bprint(Bstdout, "mark\n");
218 
219 	if (formsperpage > 1) {
220 		Bprint(Bstdout, "%s %d\n", FORMSPERPAGE, formsperpage);
221 		Bprint(Bstdout, "/formsperpage %d def\n", formsperpage);
222 	}
223 	if (aspectratio != 1) Bprint(Bstdout, "/aspectratio %g def\n", aspectratio);
224 	if (copies != 1) Bprint(Bstdout, "/#copies %d store\n", copies);
225 	if (landscape) Bprint(Bstdout, "/landscape true def\n");
226 	if (magnification != 1) Bprint(Bstdout, "/magnification %s def\n", magnification);
227 	if (pointsize != 10) Bprint(Bstdout, "/pointsize %d def\n", pointsize);
228 	if (xoffset != .25) Bprint(Bstdout, "/xoffset %g def\n", xoffset);
229 	if (yoffset != .25) Bprint(Bstdout, "/yoffset %g def\n", yoffset);
230 	cat(ENCODINGDIR"/Latin1.enc");
231 	if (passthrough != 0) Bprint(Bstdout, "%s\n", passthrough);
232 	Bprint(Bstdout, "setup\n");
233 	if (formsperpage > 1) {
234 		cat(FORMFILE);
235 		Bprint(Bstdout, "%d setupforms \n", formsperpage);
236 	}
237 	if (cat(UNKNOWNCHAR))
238 		Bprint(Bstderr, "cannot open %s\n", UNKNOWNCHAR);
239 	Bprint(Bstdout, "%s", ENDSETUP);
240 }
241 
242 int
243 pageon(void) {
244 	if (pplist == 0 && page_no != 0) return(1);	/* no page list, print all pages */
245 	if (page_no/8 < pplistmaxsize && (pplist[page_no/8] & 1<<(page_no%8)))
246 		return(1);
247 	else
248 		return(0);
249 }
250 
251 void
252 startpage(void) {
253 	++char_no;
254 	++line_no;
255 	++page_no;
256 	if (pageon()) {
257 		++pages_printed;
258 		Bprint(Bstdout, "%s %d %d\n", PAGE, page_no, pages_printed);
259 		Bprint(Bstdout, "/saveobj save def\n");
260 		Bprint(Bstdout, "mark\n");
261 		Bprint(Bstdout, "%d pagesetup\n", pages_printed);
262 	}
263 }
264 
265 void
266 endpage(void) {
267 	line_no = 0;
268 	char_no = 0;
269 	if (pageon()) {
270 		Bprint(Bstdout, "cleartomark\n");
271 		Bprint(Bstdout, "showpage\n");
272 		Bprint(Bstdout, "saveobj restore\n");
273 		Bprint(Bstdout, "%s %d %d\n", ENDPAGE, page_no, pages_printed);
274 	}
275 }
276 
277 void
278 startstring(void) {
279 	if (!in_string) {
280 		if (pageon()) Bprint(Bstdout, "(");
281 		in_string = 1;
282 	}
283 }
284 
285 void
286 endstring(void) {
287 	if (in_string) {
288 		if (pageon()) Bprint(Bstdout, ") show ");
289 		in_string = 0;
290 	}
291 }
292 
293 void
294 prspace(void) {
295 	if (spaces) {
296 		endstring();
297 		if (pageon()) Bprint(Bstdout, "%d sp ", spaces);
298 		spaces = 0;
299 	}
300 }
301 
302 void
303 prtab(void) {
304 	if (tabs) {
305 		endstring();
306 		if (pageon()) Bprint(Bstdout, "%d tab ", tabs);
307 		tabs = 0;
308 	}
309 }
310 
311 void
312 txt2post(void) {
313 	int lastfont = -1;
314 	int lastchar = -1;
315 	int thisfont, thischar;
316 	long r;
317 
318 	in_string = 0;
319 	char_no = 0;
320 	line_no = 0;
321 	page_no = 0;
322 	spaces = 0;
323 	fontname[0].used++;
324 	while ((r=Bgetrune(Bstdin)) >= 0) {
325 		thischar = r & 0xff;
326 		thisfont = (r>>8) & 0xff;
327 
328 		if (line_no == 0 && char_no == 0)
329 			startpage();
330 
331 		if (line_no == 1 && char_no == 1) {
332 			if (pageon()) Bprint(Bstdout, " /%s f\n", fontname[thisfont].str);
333 			lastfont = thisfont;
334 		}
335 
336 		switch (r) {
337 		case ' ':
338 			prtab();
339 			if (lastfont > 0) {
340 				spaces++;
341 				continue;
342 			}
343 			break;
344 		case '\n':
345 		case '\f':
346 			startstring();
347 			if (pageon()) Bprint(Bstdout, ")l\n");
348 			char_no = 1;
349 			in_string = 0;
350 			spaces = 0;
351 			tabs = 0;
352 			if (++line_no > linesperpage || r == '\f') {
353 				endpage();
354 			}
355 			lastchar = -1;
356 			continue;
357 		case '\t':
358 			prspace();
359 			tabs++;
360 			char_no++;
361 			lastchar = -1;
362 			continue;
363 		case '\b':
364 			/* just toss out backspaces for now */
365 			if (lastchar != -1) {
366 				endstring();
367 				if (pageon()) Bprint(Bstdout, "(%s) stringwidth pop neg 0 rmoveto ", charcode[lastchar].str);
368 			}
369 			char_no++;
370 			lastchar = -1;
371 			continue;
372 		}
373 
374 		/* do something if font is out of table range */
375 		if (thisfont>=FONTABSIZE || fontname[thisfont].size == 0) {
376 			prspace();
377 			prtab();
378 			endstring();
379 			Bprint(Bstdout, "pw ");
380 			char_no++;
381 			lastchar = -1;
382 			continue;
383 		}
384 
385 		if (thisfont != lastfont) {
386 			endstring();
387 			if (pageon()) {
388 				Bprint(Bstdout, "/%s f\n", fontname[thisfont].str);
389 			}
390 			fontname[thisfont].used++;
391 		}
392 		prspace();
393 		prtab();
394 		startstring();
395 		if (pageon()) Bprint(Bstdout, "%s", charcode[thischar].str);
396 /*		if (pageon()) Bprint(Bstdout, "%2.2x", thischar);	/* try hex strings*/
397 		char_no++;
398 		lastchar = thischar;
399 		lastfont = thisfont;
400 	}
401 	if (line_no != 0 || char_no != 0) {
402 		if (char_no != 1) {
403 			Bprint(Bstderr, "premature EOF: newline appended\n");
404 			startstring();
405 			if (pageon()) Bprint(Bstdout, ")l\n");
406 		}
407 		endpage();
408 	}
409 }
410 
411 void
412 pagelist(char *list) {
413 	char c;
414 	int n, m;
415 	int state, start, end;
416 
417 	if (list == 0) return;
418 	state = 1;
419 	while ((c=*list) != '\0') {
420 		n = 0;
421 		while (isdigit(c)) {
422 			n = n * 10 + c - '0';
423 			c = *++list;
424 		}
425 		switch (state) {
426 		case 1:
427 			start = n;
428 		case 2:
429 			if (n/8+1 > pplistmaxsize) {
430 				pplistmaxsize = n/8+1;
431 				if ((pplist = realloc(pplist, n/8+1)) == 0) {
432 					Bprint(Bstderr, "cannot allocate memory for page list\n");
433 					exits("malloc");
434 				}
435 			}
436 			for (m=start; m<=n; m++)
437 				pplist[m/8] |= 1<<(m%8);
438 			break;
439 		}
440 		switch (c) {
441 		case '-':
442 			state = 2;
443 			list++;
444 			break;
445 		case ',':
446 			state = 1;
447 			list++;
448 			break;
449 		case '\0':
450 			break;
451 		}
452 	}
453 }
454 
455 void
456 finish(void) {
457 	int i;
458 
459 	Bprint(Bstdout, "%s", TRAILER);
460 	Bprint(Bstdout, "done\n");
461 	Bprint(Bstdout, "%s", DOCUMENTFONTS);
462 
463 	for (i=0; i<FONTABSIZE; i++)
464 		if (fontname[i].used)
465 			Bprint(Bstdout, " %s", fontname[i].str);
466 	Bprint(Bstdout, "\n");
467 
468 	Bprint(Bstdout, "%s %d\n", PAGES, pages_printed);
469 
470 }
471 
472 main(int argc, char *argv[]) {
473 	int i;
474 	char *t;
475 	Biobuf *input;
476 
477 	if ((bstderr = (Biobuf *)malloc(sizeof(Biobuf))) < (Biobuf *)0)
478 		exits("malloc");
479 	if (Binit(bstderr, 2, OWRITE) == Beof)
480 		exits("Binit");
481 	Bstderr = &(bstderr->Biobufhdr);
482 
483 	if ((bstdout = (Biobuf *)malloc(sizeof(Biobuf))) < (Biobuf *)0)
484 		exits("malloc");
485 	if (Binit(bstdout, 1, OWRITE) == Beof)
486 		exits("Binit");
487 	Bstdout = &(bstdout->Biobufhdr);
488 
489 	ARGBEGIN{
490 		case 'a':			/* aspect ratio */
491 			aspectratio = atof(ARGF());
492 			break;
493 		case 'c':			/* copies */
494 			copies = atoi(ARGF());
495 			break;
496 		case 'f':			/* primary font, for now */
497 			t = ARGF();
498 			fontname[0].str = malloc(strlen(t)+1);
499 			strcpy(fontname[0].str, t);
500 			break;
501 		case 'l':			/* lines per page */
502 			linesperpage = atoi(ARGF());
503 			break;
504 		case 'm':			/* magnification */
505 			magnification = atof(ARGF());
506 			break;
507 		case 'n':			/* forms per page */
508 			formsperpage = atoi(ARGF());
509 			break;
510 		case 'o':			/* output page list */
511 			pagelist(ARGF());
512 			break;
513 		case 'p':			/* landscape or portrait mode */
514 			if ( ARGF()[0] == 'l' )
515 				landscape = 1;
516 			else
517 				landscape = 0;
518 			break;
519 		case 's':			/* point size */
520 			pointsize = atoi(ARGF());
521 			break;
522 		case 'x':			/* shift things horizontally */
523 			xoffset = atof(ARGF());
524 			break;
525 
526 		case 'y':			/* and vertically on the page */
527 			yoffset = atof(ARGF());
528 			break;
529 		case 'P':			/* PostScript pass through */
530 			t = ARGF();
531 			i = strlen(t) + 1;
532 			passthrough = malloc(i);
533 			if (passthrough == 0) {
534 				Bprint(Bstderr, "cannot allocate memory for argument string\n");
535 				exits("malloc");
536 			}
537 			strncpy(passthrough, t, i);
538 			break;
539 		default:			/* don't know what to do for ch */
540 			Bprint(Bstderr, "unknown option %C\n", ARGC());
541 			break;
542 	}ARGEND;
543 	prologues();
544 	if (argc <= 0) {
545 		if ((bstdin = (Biobuf *)malloc(sizeof(Biobuf))) < (Biobuf *)0)
546 			exits("malloc");
547 		if (Binit(bstdin, 0, OREAD) == Beof) {
548 			fprint(2, "cannot Binit stdin\n");
549 			exits("Binit");
550 		}
551 		Bstdin = &(bstdin->Biobufhdr);
552 		txt2post();
553 	}
554 	for (i=0; i<argc; i++) {
555 		bstdin = Bopen(argv[i], 0);
556 		if (bstdin == 0) {
557 			fprint(2, "cannot open file %s\n", argv[i]);
558 			continue;
559 		}
560 		Bstdin = &(bstdin->Biobufhdr);
561 		txt2post();
562 	}
563 	finish();
564 	exits("");
565 }
566