WebSVN – SvarDOS – Diff – /svarlang.lib/trunk/tlumacz.c


/*
 * Copyright (C) 2021-2023 Mateusz Viste
 *
 * usage: tlumacz en fr pl etc
 *
 * computes an out.lng file that contains all language ressources.
 *
 * DAT format:
 *
 * 4-bytes signature:
 * "SvL\x1b"
 *
 * Then "LANG BLOCKS" follow. Each LANG BLOCK is prefixed with 4 bytes:
 * II LL    - II is the LANG identifier ("EN", "PL", etc) and LL is the size
 *            of the block (65535 bytes max).
 *
 * Inside a LANG BLOCK is a set of strings:
 *
 * II LL S  where II is the string's 16-bit identifier, LL is its length
 *          (1-65535) and S is the actual string. All strings are ASCIIZ (ie.
 *          they end with a NULL terminator).
 *
 * The list of strings ends with a single 0-long string.
 */
 
 
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
 
#include "version.h"
 
 
struct bitmap {
  unsigned char bits[8192];
};
 
static void bitmap_set(struct bitmap *b, unsigned short id) {
  b->bits[id >> 3] |= 1 << (id & 7);
}
 
static int bitmap_get(const struct bitmap *b, unsigned short id) {
  return(b->bits[id >> 3] & (1 << (id & 7)));
}
 
static void bitmap_init(struct bitmap *b) {
  bzero(b, sizeof(struct bitmap));
}
 
 
 
/* read a single line from fd and fills it into dst, returns line length
 * ending CR/LF is trimmed, as well as any trailing spaces */
static unsigned short readl(char *dst, size_t dstsz, FILE *fd) {
  unsigned short l, lastnonspace = 0;
 
  if (fgets(dst, dstsz, fd) == NULL) return(0xffff); /* EOF */
  /* trim at first CR or LF and return len */
  for (l = 0; (dst[l] != 0) && (dst[l] != '\r') && (dst[l] != '\n'); l++) {
    if (dst[l] != ' ') lastnonspace = l;
  }
 
  if (lastnonspace < l) l = lastnonspace + 1; /* rtrim */
  dst[l] = 0;
 
  return(l);
}
 
 
/* parse a line in format "[?]1.50:somestring". fills id and returns a pointer to
 * the actual string part on success, or NULL on error */
static const char *parseline(unsigned short *id, const char *s) {
  int i;
  int dotpos = 0, colpos = 0, gotdigits = 0;
 
  /* strings prefixed by '?' are flagged as "dirty": ignore this flag here */
  if (*s == '?') s++;
 
  /* I must have a . and a : in the first 9 bytes */
  for (i = 0;; i++) {
    if (s[i] == '.') {
      if ((dotpos != 0) || (gotdigits == 0)) break;
      dotpos = i;
      gotdigits = 0;
    } else if (s[i] == ':') {
      if (gotdigits != 0) colpos = i;
      break;
    } else if ((s[i] < '0') || (s[i] > '9')) {
      break;
    }
    gotdigits++;
  }
  /* did I collect everything? */
  if ((dotpos == 0) || (colpos == 0)) return(NULL);
  if (s[colpos + 1] == 0) return(NULL);
 
  *id = atoi(s);
  *id <<= 8;
  *id |= atoi(s + dotpos + 1);
 
  /* printf("parseline(): %04X = '%s'\r\n", *id, s + colpos + 1); */
 
  return(s + colpos + 1);
}
 
 
/* converts escape sequences like "\n" or "\t" into actual bytes, returns
 * the new length of the string. */
static unsigned short unesc_string(char *linebuff) {
  unsigned short i;
  for (i = 0; linebuff[i] != 0; i++) {
    if (linebuff[i] != '\\') continue;
    strcpy(linebuff + i, linebuff + i + 1);
    if (linebuff[i] == 0) break;
    switch (linebuff[i]) {
      case 'n':
        linebuff[i] = '\n';
        break;
      case 'r':
        linebuff[i] = '\r';
        break;
      case 't':
        linebuff[i] = '\t';
        break;
    }
  }
  return(i);
}
 
 
/* opens a CATS-style file and compiles it into a ressources lang block
 * returns 0 on error, or the size of the generated data block otherwise */
static unsigned short gen_langstrings(unsigned char *buff, const char *langid, struct bitmap *b, const struct bitmap *refb, const unsigned char *refblock) {
  unsigned short len = 0, linelen;
  FILE *fd;
  char fname[] = "XX.TXT";
  static char linebuf[8192];
  const char *ptr;
  unsigned short id, linecount;
 
  bitmap_init(b);
 
  memcpy(fname + strlen(fname) - 6, langid, 2);
 
  fd = fopen(fname, "rb");
  if (fd == NULL) {
    printf("ERROR: FAILED TO OPEN '%s'\r\n", fname);
    return(0);
  }
 
  for (linecount = 1;; linecount++) {
 
    linelen = readl(linebuf, sizeof(linebuf), fd);
    if (linelen == 0xffff) break; /* EOF */
    if ((linelen == 0) || (linebuf[0] == '#')) continue;
 
    /* convert escaped chars to actual bytes (\n -> newline, etc) */
    linelen = unesc_string(linebuf);
 
    /* read id and get ptr to actual string ("1.15:string") */
    ptr = parseline(&id, linebuf);
    if (ptr == NULL) {
      printf("ERROR: line #%u of %s is malformed (linelen = %u):\r\n", linecount, fname, linelen);
      puts(linebuf);
      len = 0;
      break;
    }
 
    /* warn about dirty lines */
    if (linebuf[0] == '?') {
      printf("WARNING: %s[#%u] string id %u.%u is flagged as 'dirty'\r\n", fname, linecount, id >> 8, id & 0xff);
    }
 
    /* write string into block (II LL S) */
    memcpy(buff + len, &id, 2);
    len += 2;
    {
      unsigned short slen = strlen(ptr) + 1;
      memcpy(buff + len, &slen, 2);
      len += 2;
      memcpy(buff + len, ptr, slen);
      len += slen;
    }
 
    /* if reference bitmap provided: check that the id is valid */
    if ((refb != NULL) && (bitmap_get(refb, id) == 0)) {
      printf("WARNING: %s[#%u] has an invalid id (%u.%u not present in ref lang)\r\n", fname, linecount, id >> 8, id & 0xff);
    }
 
    /* make sure this id is not already present */
    if (bitmap_get(b, id) == 0) {
      /* set bit in bitmap to remember I have this string */
      bitmap_set(b, id);
    } else {
      printf("WARNING: %s[#%u] has a duplicated id (%u.%u)\r\n", fname, linecount, id >> 8, id & 0xff);
    }
  }
 
  fclose(fd);
 
  /* if refblock provided, pull missing strings from it */
  if (refblock != NULL) {
    for (;;) {
      unsigned short slen;
      id = ((unsigned short *)refblock)[0];
      slen = ((unsigned short *)refblock)[1];
      if ((id == 0) && (slen == 0)) break;
      if (bitmap_get(b, id) == 0) {
        printf("WARNING: %s is missing string %u.%u (pulled from ref lang)\r\n", fname, id >> 8, id & 0xff);
        /* copy missing string from refblock */
        memcpy(buff + len, refblock, slen + 4);
        len += slen + 4;
      }
      refblock += slen + 4;
    }
  }
 
  /* write the block terminator (0-long string) */
  buff[len++] = 0; /* id */
  buff[len++] = 0; /* id */
  buff[len++] = 0; /* len */
  buff[len++] = 0; /* len */
  buff[len++] = 0; /* empty string */
 
  return(len);
}
 
 
#define MEMBLOCKSZ 65000
 
int main(int argc, char **argv) {
  FILE *fd;
  int ecode = 0;
  char *buff, *refblock;
  unsigned short refblocksz = 0;
  static struct bitmap bufbitmap;
  static struct bitmap refbitmap;
  unsigned short i;
  unsigned short biggest_langsz = 0;
 
  if (argc < 2) {
    puts("tlumacz ver " SVARLANGVER " - this tool is part of the SvarLANG project.");
    puts("converts a set of CATS-style translations in files EN.TXT, PL.TXT, etc");
    puts("into a single resource file (OUT.LNG).");
    puts("");
    puts("usage: tlumacz en fr pl ...");
    return(1);
  }
 
  buff = malloc(MEMBLOCKSZ);
  refblock = malloc(MEMBLOCKSZ);
  if ((buff == NULL) || (refblock == NULL)) {
    puts("out of memory");
    return(1);
  }
 
  fd = fopen("out.lng", "wb");
  if (fd == NULL) {
    puts("ERR: failed to open or create SVARCOM.LNG");
    return(1);
  }
 
  /* write sig */
  fwrite("SvL\x1b", 1, 4, fd);
 
  /* write lang blocks */
  for (i = 1; i < argc; i++) {
    unsigned short sz;
    char id[3];
 
    if (strlen(argv[i]) != 2) {
      printf("INVALID LANG SPECIFIED: %s\r\n", argv[i]);
      ecode = 1;
      break;
    }
 
    id[0] = argv[i][0];
    id[1] = argv[i][1];
    id[2] = 0;
    if (id[0] >= 'a') id[0] -= 'a' - 'A';
    if (id[1] >= 'a') id[1] -= 'a' - 'A';
 
    sz = gen_langstrings(buff, id, &bufbitmap, (i != 1)?&refbitmap:NULL, (i != 1)?refblock:NULL);
    if (sz == 0) {
      printf("ERROR COMPUTING LANG '%s'\r\n", id);
      ecode = 1;
      break;
    } else {
      printf("computed %s lang block of %u bytes\r\n", id, sz);
      if (sz > biggest_langsz) biggest_langsz = sz;
    }
    /* write lang ID to file, followed by block size and then the actual block */
    if ((fwrite(id, 1, 2, fd) != 2) ||
        (fwrite(&sz, 1, 2, fd) != 2) ||
        (fwrite(buff, 1, sz, fd) != sz)) {
      printf("ERROR WRITING TO OUTPUT FILE\r\n");
      ecode = 1;
      break;
    }
    /* remember reference data for other languages */
    if (i == 1) {
      refblocksz = sz;
      memcpy(refblock, buff, MEMBLOCKSZ);
      memcpy(&refbitmap, &bufbitmap, sizeof(struct bitmap));
    }
  }
 
  fclose(fd);
 
  /* compute the deflang.c file containing a dump of the reference block */
  fd = fopen("DEFLANG.C", "wb");
  if (fd == NULL) {
    puts("ERROR: FAILED TO OPEN OR CREATE DEFLANG.C");
    ecode = 1;
  } else {
    unsigned short allocsz = biggest_langsz + (biggest_langsz / 20);
    printf("biggest lang block is %u bytes -> allocating a %u bytes buffe (5%% safety margin)r\n", biggest_langsz, allocsz);
    fprintf(fd, "/* THIS FILE HAS BEEN GENERATED BY TLUMACZ (PART OF THE SVARLANG LIBRARY) */\r\n");
    fprintf(fd, "const unsigned short svarlang_memsz = %uu;\r\n", allocsz);
    fprintf(fd, "char svarlang_mem[%u] = {\r\n", allocsz);
    for (i = 0; i < refblocksz; i++) {
      fprintf(fd, "%u", refblock[i]);
      if (i + 1 < refblocksz) fprintf(fd, ",");
      if ((i & 15) == 15) fprintf(fd, "\r\n");
    }
    fprintf(fd, "};\r\n");
    fclose(fd);
  }
 
  return(ecode);
}
 

Rev 1114	Rev 1247
1	`/*`	1	`/*`
2	`* Copyright (C) 2021-2022 Mateusz Viste`	2	`* Copyright (C) 2021-2023 Mateusz Viste`
3	`*`	3	`*`
4	`* usage: tlumacz en fr pl etc`	4	`* usage: tlumacz en fr pl etc`
5	`*`	5	`*`
6	`* computes an out.lng file that contains all language ressources.`	6	`* computes an out.lng file that contains all language ressources.`
7	`*`	7	`*`
8	`* DAT format:`	8	`* DAT format:`
9	`*`	9	`*`
10	`* 4-bytes signature:`	10	`* 4-bytes signature:`
11	`* "SvL\x1b"`	11	`* "SvL\x1b"`
12	`*`	12	`*`
13	`* Then "LANG BLOCKS" follow. Each LANG BLOCK is prefixed with 4 bytes:`	13	`* Then "LANG BLOCKS" follow. Each LANG BLOCK is prefixed with 4 bytes:`
14	`* II LL - II is the LANG identifier ("EN", "PL", etc) and LL is the size`	14	`* II LL - II is the LANG identifier ("EN", "PL", etc) and LL is the size`
15	`* of the block (65535 bytes max).`	15	`* of the block (65535 bytes max).`
16	`*`	16	`*`
17	`* Inside a LANG BLOCK is a set of strings:`	17	`* Inside a LANG BLOCK is a set of strings:`
18	`*`	18	`*`
19	`* II LL S where II is the string's 16-bit identifier, LL is its length`	19	`* II LL S where II is the string's 16-bit identifier, LL is its length`
20	`* (1-65535) and S is the actual string. All strings are ASCIIZ (ie.`	20	`* (1-65535) and S is the actual string. All strings are ASCIIZ (ie.`
21	`* they end with a NULL terminator).`	21	`* they end with a NULL terminator).`
22	`*`	22	`*`
23	`* The list of strings ends with a single 0-long string.`	23	`* The list of strings ends with a single 0-long string.`
24	`*/`	24	`*/`
25		25
26		26
27	`#include <stdio.h>`	27	`#include <stdio.h>`
28	`#include <stdlib.h>`	28	`#include <stdlib.h>`
29	`#include <string.h>`	29	`#include <string.h>`
30		30
-		31	`#include "version.h"`
31		32
32		33
33	`struct bitmap {`	34	`struct bitmap {`
34	`unsigned char bits[8192];`	35	`unsigned char bits[8192];`
35	`};`	36	`};`
36		37
37	`static void bitmap_set(struct bitmap *b, unsigned short id) {`	38	`static void bitmap_set(struct bitmap *b, unsigned short id) {`
38	`b->bits[id >> 3] \|= 1 << (id & 7);`	39	`b->bits[id >> 3] \|= 1 << (id & 7);`
39	`}`	40	`}`
40		41
41	`static int bitmap_get(const struct bitmap *b, unsigned short id) {`	42	`static int bitmap_get(const struct bitmap *b, unsigned short id) {`
42	`return(b->bits[id >> 3] & (1 << (id & 7)));`	43	`return(b->bits[id >> 3] & (1 << (id & 7)));`
43	`}`	44	`}`
44		45
45	`static void bitmap_init(struct bitmap *b) {`	46	`static void bitmap_init(struct bitmap *b) {`
46	`bzero(b, sizeof(struct bitmap));`	47	`bzero(b, sizeof(struct bitmap));`
47	`}`	48	`}`
48		49
49		50
50		51
51	`/* read a single line from fd and fills it into dst, returns line length`	52	`/* read a single line from fd and fills it into dst, returns line length`
52	`* ending CR/LF is trimmed, as well as any trailing spaces */`	53	`* ending CR/LF is trimmed, as well as any trailing spaces */`
53	`static unsigned short readl(char dst, size_t dstsz, FILE fd) {`	54	`static unsigned short readl(char dst, size_t dstsz, FILE fd) {`
54	`unsigned short l, lastnonspace = 0;`	55	`unsigned short l, lastnonspace = 0;`
55		56
56	`if (fgets(dst, dstsz, fd) == NULL) return(0xffff); /* EOF */`	57	`if (fgets(dst, dstsz, fd) == NULL) return(0xffff); /* EOF */`
57	`/* trim at first CR or LF and return len */`	58	`/* trim at first CR or LF and return len */`
58	`for (l = 0; (dst[l] != 0) && (dst[l] != '\r') && (dst[l] != '\n'); l++) {`	59	`for (l = 0; (dst[l] != 0) && (dst[l] != '\r') && (dst[l] != '\n'); l++) {`
59	`if (dst[l] != ' ') lastnonspace = l;`	60	`if (dst[l] != ' ') lastnonspace = l;`
60	`}`	61	`}`
61		62
62	`if (lastnonspace < l) l = lastnonspace + 1; /* rtrim */`	63	`if (lastnonspace < l) l = lastnonspace + 1; /* rtrim */`
63	`dst[l] = 0;`	64	`dst[l] = 0;`
64		65
65	`return(l);`	66	`return(l);`
66	`}`	67	`}`
67		68
68		69
69	`/* parse a line in format "[?]1.50:somestring". fills id and returns a pointer to`	70	`/* parse a line in format "[?]1.50:somestring". fills id and returns a pointer to`
70	`* the actual string part on success, or NULL on error */`	71	`* the actual string part on success, or NULL on error */`
71	`static const char parseline(unsigned short id, const char *s) {`	72	`static const char parseline(unsigned short id, const char *s) {`
72	`int i;`	73	`int i;`
73	`int dotpos = 0, colpos = 0, gotdigits = 0;`	74	`int dotpos = 0, colpos = 0, gotdigits = 0;`
74		75
75	`/* strings prefixed by '?' are flagged as "dirty": ignore this flag here */`	76	`/* strings prefixed by '?' are flagged as "dirty": ignore this flag here */`
76	`if (*s == '?') s++;`	77	`if (*s == '?') s++;`
77		78
78	`/* I must have a . and a : in the first 9 bytes */`	79	`/* I must have a . and a : in the first 9 bytes */`
79	`for (i = 0;; i++) {`	80	`for (i = 0;; i++) {`
80	`if (s[i] == '.') {`	81	`if (s[i] == '.') {`
81	`if ((dotpos != 0) \|\| (gotdigits == 0)) break;`	82	`if ((dotpos != 0) \|\| (gotdigits == 0)) break;`
82	`dotpos = i;`	83	`dotpos = i;`
83	`gotdigits = 0;`	84	`gotdigits = 0;`
84	`} else if (s[i] == ':') {`	85	`} else if (s[i] == ':') {`
85	`if (gotdigits != 0) colpos = i;`	86	`if (gotdigits != 0) colpos = i;`
86	`break;`	87	`break;`
87	`} else if ((s[i] < '0') \|\| (s[i] > '9')) {`	88	`} else if ((s[i] < '0') \|\| (s[i] > '9')) {`
88	`break;`	89	`break;`
89	`}`	90	`}`
90	`gotdigits++;`	91	`gotdigits++;`
91	`}`	92	`}`
92	`/* did I collect everything? */`	93	`/* did I collect everything? */`
93	`if ((dotpos == 0) \|\| (colpos == 0)) return(NULL);`	94	`if ((dotpos == 0) \|\| (colpos == 0)) return(NULL);`
94	`if (s[colpos + 1] == 0) return(NULL);`	95	`if (s[colpos + 1] == 0) return(NULL);`
95		96
96	`*id = atoi(s);`	97	`*id = atoi(s);`
97	`*id <<= 8;`	98	`*id <<= 8;`
98	`*id \|= atoi(s + dotpos + 1);`	99	`*id \|= atoi(s + dotpos + 1);`
99		100
100	`/* printf("parseline(): %04X = '%s'\r\n", id, s + colpos + 1); /`	101	`/* printf("parseline(): %04X = '%s'\r\n", id, s + colpos + 1); /`
101		102
102	`return(s + colpos + 1);`	103	`return(s + colpos + 1);`
103	`}`	104	`}`
104		105
105		106
106	`/* converts escape sequences like "\n" or "\t" into actual bytes, returns`	107	`/* converts escape sequences like "\n" or "\t" into actual bytes, returns`
107	`* the new length of the string. */`	108	`* the new length of the string. */`
108	`static unsigned short unesc_string(char *linebuff) {`	109	`static unsigned short unesc_string(char *linebuff) {`
109	`unsigned short i;`	110	`unsigned short i;`
110	`for (i = 0; linebuff[i] != 0; i++) {`	111	`for (i = 0; linebuff[i] != 0; i++) {`
111	`if (linebuff[i] != '\\') continue;`	112	`if (linebuff[i] != '\\') continue;`
112	`strcpy(linebuff + i, linebuff + i + 1);`	113	`strcpy(linebuff + i, linebuff + i + 1);`
113	`if (linebuff[i] == 0) break;`	114	`if (linebuff[i] == 0) break;`
114	`switch (linebuff[i]) {`	115	`switch (linebuff[i]) {`
115	`case 'n':`	116	`case 'n':`
116	`linebuff[i] = '\n';`	117	`linebuff[i] = '\n';`
117	`break;`	118	`break;`
118	`case 'r':`	119	`case 'r':`
119	`linebuff[i] = '\r';`	120	`linebuff[i] = '\r';`
120	`break;`	121	`break;`
121	`case 't':`	122	`case 't':`
122	`linebuff[i] = '\t';`	123	`linebuff[i] = '\t';`
123	`break;`	124	`break;`
124	`}`	125	`}`
125	`}`	126	`}`
126	`return(i);`	127	`return(i);`
127	`}`	128	`}`
128		129
129		130
130	`/* opens a CATS-style file and compiles it into a ressources lang block`	131	`/* opens a CATS-style file and compiles it into a ressources lang block`
131	`* returns 0 on error, or the size of the generated data block otherwise */`	132	`* returns 0 on error, or the size of the generated data block otherwise */`
132	`static unsigned short gen_langstrings(unsigned char buff, const char langid, struct bitmap b, const struct bitmap refb, const unsigned char *refblock) {`	133	`static unsigned short gen_langstrings(unsigned char buff, const char langid, struct bitmap b, const struct bitmap refb, const unsigned char *refblock) {`
133	`unsigned short len = 0, linelen;`	134	`unsigned short len = 0, linelen;`
134	`FILE *fd;`	135	`FILE *fd;`
135	`char fname[] = "XX.TXT";`	136	`char fname[] = "XX.TXT";`
136	`static char linebuf[8192];`	137	`static char linebuf[8192];`
137	`const char *ptr;`	138	`const char *ptr;`
138	`unsigned short id, linecount;`	139	`unsigned short id, linecount;`
139		140
140	`bitmap_init(b);`	141	`bitmap_init(b);`
141		142
142	`memcpy(fname + strlen(fname) - 6, langid, 2);`	143	`memcpy(fname + strlen(fname) - 6, langid, 2);`
143		144
144	`fd = fopen(fname, "rb");`	145	`fd = fopen(fname, "rb");`
145	`if (fd == NULL) {`	146	`if (fd == NULL) {`
146	`printf("ERROR: FAILED TO OPEN '%s'\r\n", fname);`	147	`printf("ERROR: FAILED TO OPEN '%s'\r\n", fname);`
147	`return(0);`	148	`return(0);`
148	`}`	149	`}`
149		150
150	`for (linecount = 1;; linecount++) {`	151	`for (linecount = 1;; linecount++) {`
151		152
152	`linelen = readl(linebuf, sizeof(linebuf), fd);`	153	`linelen = readl(linebuf, sizeof(linebuf), fd);`
153	`if (linelen == 0xffff) break; /* EOF */`	154	`if (linelen == 0xffff) break; /* EOF */`
154	`if ((linelen == 0) \|\| (linebuf[0] == '#')) continue;`	155	`if ((linelen == 0) \|\| (linebuf[0] == '#')) continue;`
155		156
156	`/* convert escaped chars to actual bytes (\n -> newline, etc) */`	157	`/* convert escaped chars to actual bytes (\n -> newline, etc) */`
157	`linelen = unesc_string(linebuf);`	158	`linelen = unesc_string(linebuf);`
158		159
159	`/* read id and get ptr to actual string ("1.15:string") */`	160	`/* read id and get ptr to actual string ("1.15:string") */`
160	`ptr = parseline(&id, linebuf);`	161	`ptr = parseline(&id, linebuf);`
161	`if (ptr == NULL) {`	162	`if (ptr == NULL) {`
162	`printf("ERROR: line #%u of %s is malformed (linelen = %u):\r\n", linecount, fname, linelen);`	163	`printf("ERROR: line #%u of %s is malformed (linelen = %u):\r\n", linecount, fname, linelen);`
163	`puts(linebuf);`	164	`puts(linebuf);`
164	`len = 0;`	165	`len = 0;`
165	`break;`	166	`break;`
166	`}`	167	`}`
167		168
168	`/* warn about dirty lines */`	169	`/* warn about dirty lines */`
169	`if (linebuf[0] == '?') {`	170	`if (linebuf[0] == '?') {`
170	`printf("WARNING: %s[#%u] string id %u.%u is flagged as 'dirty'\r\n", fname, linecount, id >> 8, id & 0xff);`	171	`printf("WARNING: %s[#%u] string id %u.%u is flagged as 'dirty'\r\n", fname, linecount, id >> 8, id & 0xff);`
171	`}`	172	`}`
172		173
173	`/* write string into block (II LL S) */`	174	`/* write string into block (II LL S) */`
174	`memcpy(buff + len, &id, 2);`	175	`memcpy(buff + len, &id, 2);`
175	`len += 2;`	176	`len += 2;`
176	`{`	177	`{`
177	`unsigned short slen = strlen(ptr) + 1;`	178	`unsigned short slen = strlen(ptr) + 1;`
178	`memcpy(buff + len, &slen, 2);`	179	`memcpy(buff + len, &slen, 2);`
179	`len += 2;`	180	`len += 2;`
180	`memcpy(buff + len, ptr, slen);`	181	`memcpy(buff + len, ptr, slen);`
181	`len += slen;`	182	`len += slen;`
182	`}`	183	`}`
183		184
184	`/* if reference bitmap provided: check that the id is valid */`	185	`/* if reference bitmap provided: check that the id is valid */`
185	`if ((refb != NULL) && (bitmap_get(refb, id) == 0)) {`	186	`if ((refb != NULL) && (bitmap_get(refb, id) == 0)) {`
186	`printf("WARNING: %s[#%u] has an invalid id (%u.%u not present in ref lang)\r\n", fname, linecount, id >> 8, id & 0xff);`	187	`printf("WARNING: %s[#%u] has an invalid id (%u.%u not present in ref lang)\r\n", fname, linecount, id >> 8, id & 0xff);`
187	`}`	188	`}`
188		189
189	`/* make sure this id is not already present */`	190	`/* make sure this id is not already present */`
190	`if (bitmap_get(b, id) == 0) {`	191	`if (bitmap_get(b, id) == 0) {`
191	`/* set bit in bitmap to remember I have this string */`	192	`/* set bit in bitmap to remember I have this string */`
192	`bitmap_set(b, id);`	193	`bitmap_set(b, id);`
193	`} else {`	194	`} else {`
194	`printf("WARNING: %s[#%u] has a duplicated id (%u.%u)\r\n", fname, linecount, id >> 8, id & 0xff);`	195	`printf("WARNING: %s[#%u] has a duplicated id (%u.%u)\r\n", fname, linecount, id >> 8, id & 0xff);`
195	`}`	196	`}`
196	`}`	197	`}`
197		198
198	`fclose(fd);`	199	`fclose(fd);`
199		200
200	`/* if refblock provided, pull missing strings from it */`	201	`/* if refblock provided, pull missing strings from it */`
201	`if (refblock != NULL) {`	202	`if (refblock != NULL) {`
202	`for (;;) {`	203	`for (;;) {`
203	`unsigned short slen;`	204	`unsigned short slen;`
204	`id = ((unsigned short *)refblock)[0];`	205	`id = ((unsigned short *)refblock)[0];`
205	`slen = ((unsigned short *)refblock)[1];`	206	`slen = ((unsigned short *)refblock)[1];`
206	`if ((id == 0) && (slen == 0)) break;`	207	`if ((id == 0) && (slen == 0)) break;`
207	`if (bitmap_get(b, id) == 0) {`	208	`if (bitmap_get(b, id) == 0) {`
208	`printf("WARNING: %s is missing string %u.%u (pulled from ref lang)\r\n", fname, id >> 8, id & 0xff);`	209	`printf("WARNING: %s is missing string %u.%u (pulled from ref lang)\r\n", fname, id >> 8, id & 0xff);`
209	`/* copy missing string from refblock */`	210	`/* copy missing string from refblock */`
210	`memcpy(buff + len, refblock, slen + 4);`	211	`memcpy(buff + len, refblock, slen + 4);`
211	`len += slen + 4;`	212	`len += slen + 4;`
212	`}`	213	`}`
213	`refblock += slen + 4;`	214	`refblock += slen + 4;`
214	`}`	215	`}`
215	`}`	216	`}`
216		217
217	`/* write the block terminator (0-long string) */`	218	`/* write the block terminator (0-long string) */`
218	`buff[len++] = 0; /* id */`	219	`buff[len++] = 0; /* id */`
219	`buff[len++] = 0; /* id */`	220	`buff[len++] = 0; /* id */`
220	`buff[len++] = 0; /* len */`	221	`buff[len++] = 0; /* len */`
221	`buff[len++] = 0; /* len */`	222	`buff[len++] = 0; /* len */`
222	`buff[len++] = 0; /* empty string */`	223	`buff[len++] = 0; /* empty string */`
223		224
224	`return(len);`	225	`return(len);`
225	`}`	226	`}`
226		227
227		228
228	`#define MEMBLOCKSZ 65000`	229	`#define MEMBLOCKSZ 65000`
229		230
230	`int main(int argc, char **argv) {`	231	`int main(int argc, char **argv) {`
231	`FILE *fd;`	232	`FILE *fd;`
232	`int ecode = 0;`	233	`int ecode = 0;`
233	`char buff, refblock;`	234	`char buff, refblock;`
234	`unsigned short refblocksz = 0;`	235	`unsigned short refblocksz = 0;`
235	`static struct bitmap bufbitmap;`	236	`static struct bitmap bufbitmap;`
236	`static struct bitmap refbitmap;`	237	`static struct bitmap refbitmap;`
237	`unsigned short i;`	238	`unsigned short i;`
238	`unsigned short biggest_langsz = 0;`	239	`unsigned short biggest_langsz = 0;`
239		240
240	`if (argc < 2) {`	241	`if (argc < 2) {`
-		242	`puts("tlumacz ver " SVARLANGVER " - this tool is part of the SvarLANG project.");`
-		243	`puts("converts a set of CATS-style translations in files EN.TXT, PL.TXT, etc");`
-		244	`puts("into a single resource file (OUT.LNG).");`
-		245	`puts("");`
241	`puts("usage: tlumacz en fr pl etc");`	246	`puts("usage: tlumacz en fr pl ...");`
242	`return(1);`	247	`return(1);`
243	`}`	248	`}`
244		249
245	`buff = malloc(MEMBLOCKSZ);`	250	`buff = malloc(MEMBLOCKSZ);`
246	`refblock = malloc(MEMBLOCKSZ);`	251	`refblock = malloc(MEMBLOCKSZ);`
247	`if ((buff == NULL) \|\| (refblock == NULL)) {`	252	`if ((buff == NULL) \|\| (refblock == NULL)) {`
248	`puts("out of memory");`	253	`puts("out of memory");`
249	`return(1);`	254	`return(1);`
250	`}`	255	`}`
251		256
252	`fd = fopen("out.lng", "wb");`	257	`fd = fopen("out.lng", "wb");`
253	`if (fd == NULL) {`	258	`if (fd == NULL) {`
254	`puts("ERR: failed to open or create SVARCOM.LNG");`	259	`puts("ERR: failed to open or create SVARCOM.LNG");`
255	`return(1);`	260	`return(1);`
256	`}`	261	`}`
257		262
258	`/* write sig */`	263	`/* write sig */`
259	`fwrite("SvL\x1b", 1, 4, fd);`	264	`fwrite("SvL\x1b", 1, 4, fd);`
260		265
261	`/* write lang blocks */`	266	`/* write lang blocks */`
262	`for (i = 1; i < argc; i++) {`	267	`for (i = 1; i < argc; i++) {`
263	`unsigned short sz;`	268	`unsigned short sz;`
264	`char id[3];`	269	`char id[3];`
265		270
266	`if (strlen(argv[i]) != 2) {`	271	`if (strlen(argv[i]) != 2) {`
267	`printf("INVALID LANG SPECIFIED: %s\r\n", argv[i]);`	272	`printf("INVALID LANG SPECIFIED: %s\r\n", argv[i]);`
268	`ecode = 1;`	273	`ecode = 1;`
269	`break;`	274	`break;`
270	`}`	275	`}`
271		276
272	`id[0] = argv[i][0];`	277	`id[0] = argv[i][0];`
273	`id[1] = argv[i][1];`	278	`id[1] = argv[i][1];`
274	`id[2] = 0;`	279	`id[2] = 0;`
275	`if (id[0] >= 'a') id[0] -= 'a' - 'A';`	280	`if (id[0] >= 'a') id[0] -= 'a' - 'A';`
276	`if (id[1] >= 'a') id[1] -= 'a' - 'A';`	281	`if (id[1] >= 'a') id[1] -= 'a' - 'A';`
277		282
278	`sz = gen_langstrings(buff, id, &bufbitmap, (i != 1)?&refbitmap:NULL, (i != 1)?refblock:NULL);`	283	`sz = gen_langstrings(buff, id, &bufbitmap, (i != 1)?&refbitmap:NULL, (i != 1)?refblock:NULL);`
279	`if (sz == 0) {`	284	`if (sz == 0) {`
280	`printf("ERROR COMPUTING LANG '%s'\r\n", id);`	285	`printf("ERROR COMPUTING LANG '%s'\r\n", id);`
281	`ecode = 1;`	286	`ecode = 1;`
282	`break;`	287	`break;`
283	`} else {`	288	`} else {`
284	`printf("computed %s lang block of %u bytes\r\n", id, sz);`	289	`printf("computed %s lang block of %u bytes\r\n", id, sz);`
285	`if (sz > biggest_langsz) biggest_langsz = sz;`	290	`if (sz > biggest_langsz) biggest_langsz = sz;`
286	`}`	291	`}`
287	`/* write lang ID to file, followed by block size and then the actual block */`	292	`/* write lang ID to file, followed by block size and then the actual block */`
288	`if ((fwrite(id, 1, 2, fd) != 2) \|\|`	293	`if ((fwrite(id, 1, 2, fd) != 2) \|\|`
289	`(fwrite(&sz, 1, 2, fd) != 2) \|\|`	294	`(fwrite(&sz, 1, 2, fd) != 2) \|\|`
290	`(fwrite(buff, 1, sz, fd) != sz)) {`	295	`(fwrite(buff, 1, sz, fd) != sz)) {`
291	`printf("ERROR WRITING TO OUTPUT FILE\r\n");`	296	`printf("ERROR WRITING TO OUTPUT FILE\r\n");`
292	`ecode = 1;`	297	`ecode = 1;`
293	`break;`	298	`break;`
294	`}`	299	`}`
295	`/* remember reference data for other languages */`	300	`/* remember reference data for other languages */`
296	`if (i == 1) {`	301	`if (i == 1) {`
297	`refblocksz = sz;`	302	`refblocksz = sz;`
298	`memcpy(refblock, buff, MEMBLOCKSZ);`	303	`memcpy(refblock, buff, MEMBLOCKSZ);`
299	`memcpy(&refbitmap, &bufbitmap, sizeof(struct bitmap));`	304	`memcpy(&refbitmap, &bufbitmap, sizeof(struct bitmap));`
300	`}`	305	`}`
301	`}`	306	`}`
302		307
303	`fclose(fd);`	308	`fclose(fd);`
304		309
305	`/* compute the deflang.c file containing a dump of the reference block */`	310	`/* compute the deflang.c file containing a dump of the reference block */`
306	`fd = fopen("DEFLANG.C", "wb");`	311	`fd = fopen("DEFLANG.C", "wb");`
307	`if (fd == NULL) {`	312	`if (fd == NULL) {`
308	`puts("ERROR: FAILED TO OPEN OR CREATE DEFLANG.C");`	313	`puts("ERROR: FAILED TO OPEN OR CREATE DEFLANG.C");`
309	`ecode = 1;`	314	`ecode = 1;`
310	`} else {`	315	`} else {`
311	`unsigned short allocsz = biggest_langsz + (biggest_langsz / 20);`	316	`unsigned short allocsz = biggest_langsz + (biggest_langsz / 20);`
312	`printf("biggest lang block is %u bytes -> allocating a %u bytes buffer\n", biggest_langsz, allocsz);`	317	`printf("biggest lang block is %u bytes -> allocating a %u bytes buffe (5%% safety margin)r\n", biggest_langsz, allocsz);`
313	`fprintf(fd, "/* THIS FILE HAS BEEN GENERATED BY TLUMACZ (PART OF THE SVARLANG LIBRARY) */\r\n");`	318	`fprintf(fd, "/* THIS FILE HAS BEEN GENERATED BY TLUMACZ (PART OF THE SVARLANG LIBRARY) */\r\n");`
314	`fprintf(fd, "const unsigned short svarlang_memsz = %uu;\r\n", allocsz);`	319	`fprintf(fd, "const unsigned short svarlang_memsz = %uu;\r\n", allocsz);`
315	`fprintf(fd, "char svarlang_mem[%u] = {\r\n", allocsz);`	320	`fprintf(fd, "char svarlang_mem[%u] = {\r\n", allocsz);`
316	`for (i = 0; i < refblocksz; i++) {`	321	`for (i = 0; i < refblocksz; i++) {`
317	`fprintf(fd, "%u", refblock[i]);`	322	`fprintf(fd, "%u", refblock[i]);`
318	`if (i + 1 < refblocksz) fprintf(fd, ",");`	323	`if (i + 1 < refblocksz) fprintf(fd, ",");`
319	`if ((i & 15) == 15) fprintf(fd, "\r\n");`	324	`if ((i & 15) == 15) fprintf(fd, "\r\n");`
320	`}`	325	`}`
321	`fprintf(fd, "};\r\n");`	326	`fprintf(fd, "};\r\n");`
322	`fclose(fd);`	327	`fclose(fd);`
323	`}`	328	`}`
324		329
325	`return(ecode);`	330	`return(ecode);`
326	`}`	331	`}`
327		332

Subversion Repositories SvarDOS

(root)/svarlang.lib/trunk/tlumacz.c @ 1114 – Rev 1114 → 1247