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 "svarlang.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 'e':
        linebuff[i] = 0x1B; /* ESC code, using hex because '\e' is not ANSI C */
        break;
      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 OUT.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 1248	Rev 1250
1	`/*`	1	`/*`
2	`* Copyright (C) 2021-2023 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 "svarlang.h"`	31	`#include "svarlang.h"`
32		32
33		33
34	`struct bitmap {`	34	`struct bitmap {`
35	`unsigned char bits[8192];`	35	`unsigned char bits[8192];`
36	`};`	36	`};`
37		37
38	`static void bitmap_set(struct bitmap *b, unsigned short id) {`	38	`static void bitmap_set(struct bitmap *b, unsigned short id) {`
39	`b->bits[id >> 3] \|= 1 << (id & 7);`	39	`b->bits[id >> 3] \|= 1 << (id & 7);`
40	`}`	40	`}`
41		41
42	`static int bitmap_get(const struct bitmap *b, unsigned short id) {`	42	`static int bitmap_get(const struct bitmap *b, unsigned short id) {`
43	`return(b->bits[id >> 3] & (1 << (id & 7)));`	43	`return(b->bits[id >> 3] & (1 << (id & 7)));`
44	`}`	44	`}`
45		45
46	`static void bitmap_init(struct bitmap *b) {`	46	`static void bitmap_init(struct bitmap *b) {`
47	`bzero(b, sizeof(struct bitmap));`	47	`bzero(b, sizeof(struct bitmap));`
48	`}`	48	`}`
49		49
50		50
51		51
52	`/* 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`
53	`* ending CR/LF is trimmed, as well as any trailing spaces */`	53	`* ending CR/LF is trimmed, as well as any trailing spaces */`
54	`static unsigned short readl(char dst, size_t dstsz, FILE fd) {`	54	`static unsigned short readl(char dst, size_t dstsz, FILE fd) {`
55	`unsigned short l, lastnonspace = 0;`	55	`unsigned short l, lastnonspace = 0;`
56		56
57	`if (fgets(dst, dstsz, fd) == NULL) return(0xffff); /* EOF */`	57	`if (fgets(dst, dstsz, fd) == NULL) return(0xffff); /* EOF */`
58	`/* trim at first CR or LF and return len */`	58	`/* trim at first CR or LF and return len */`
59	`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++) {`
60	`if (dst[l] != ' ') lastnonspace = l;`	60	`if (dst[l] != ' ') lastnonspace = l;`
61	`}`	61	`}`
62		62
63	`if (lastnonspace < l) l = lastnonspace + 1; /* rtrim */`	63	`if (lastnonspace < l) l = lastnonspace + 1; /* rtrim */`
64	`dst[l] = 0;`	64	`dst[l] = 0;`
65		65
66	`return(l);`	66	`return(l);`
67	`}`	67	`}`
68		68
69		69
70	`/* 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`
71	`* the actual string part on success, or NULL on error */`	71	`* the actual string part on success, or NULL on error */`
72	`static const char parseline(unsigned short id, const char *s) {`	72	`static const char parseline(unsigned short id, const char *s) {`
73	`int i;`	73	`int i;`
74	`int dotpos = 0, colpos = 0, gotdigits = 0;`	74	`int dotpos = 0, colpos = 0, gotdigits = 0;`
75		75
76	`/* strings prefixed by '?' are flagged as "dirty": ignore this flag here */`	76	`/* strings prefixed by '?' are flagged as "dirty": ignore this flag here */`
77	`if (*s == '?') s++;`	77	`if (*s == '?') s++;`
78		78
79	`/* I must have a . and a : in the first 9 bytes */`	79	`/* I must have a . and a : in the first 9 bytes */`
80	`for (i = 0;; i++) {`	80	`for (i = 0;; i++) {`
81	`if (s[i] == '.') {`	81	`if (s[i] == '.') {`
82	`if ((dotpos != 0) \|\| (gotdigits == 0)) break;`	82	`if ((dotpos != 0) \|\| (gotdigits == 0)) break;`
83	`dotpos = i;`	83	`dotpos = i;`
84	`gotdigits = 0;`	84	`gotdigits = 0;`
85	`} else if (s[i] == ':') {`	85	`} else if (s[i] == ':') {`
86	`if (gotdigits != 0) colpos = i;`	86	`if (gotdigits != 0) colpos = i;`
87	`break;`	87	`break;`
88	`} else if ((s[i] < '0') \|\| (s[i] > '9')) {`	88	`} else if ((s[i] < '0') \|\| (s[i] > '9')) {`
89	`break;`	89	`break;`
90	`}`	90	`}`
91	`gotdigits++;`	91	`gotdigits++;`
92	`}`	92	`}`
93	`/* did I collect everything? */`	93	`/* did I collect everything? */`
94	`if ((dotpos == 0) \|\| (colpos == 0)) return(NULL);`	94	`if ((dotpos == 0) \|\| (colpos == 0)) return(NULL);`
95	`if (s[colpos + 1] == 0) return(NULL);`	95	`if (s[colpos + 1] == 0) return(NULL);`
96		96
97	`*id = atoi(s);`	97	`*id = atoi(s);`
98	`*id <<= 8;`	98	`*id <<= 8;`
99	`*id \|= atoi(s + dotpos + 1);`	99	`*id \|= atoi(s + dotpos + 1);`
100		100
101	`/* printf("parseline(): %04X = '%s'\r\n", id, s + colpos + 1); /`	101	`/* printf("parseline(): %04X = '%s'\r\n", id, s + colpos + 1); /`
102		102
103	`return(s + colpos + 1);`	103	`return(s + colpos + 1);`
104	`}`	104	`}`
105		105
106		106
107	`/* converts escape sequences like "\n" or "\t" into actual bytes, returns`	107	`/* converts escape sequences like "\n" or "\t" into actual bytes, returns`
108	`* the new length of the string. */`	108	`* the new length of the string. */`
109	`static unsigned short unesc_string(char *linebuff) {`	109	`static unsigned short unesc_string(char *linebuff) {`
110	`unsigned short i;`	110	`unsigned short i;`
111	`for (i = 0; linebuff[i] != 0; i++) {`	111	`for (i = 0; linebuff[i] != 0; i++) {`
112	`if (linebuff[i] != '\\') continue;`	112	`if (linebuff[i] != '\\') continue;`
113	`strcpy(linebuff + i, linebuff + i + 1);`	113	`strcpy(linebuff + i, linebuff + i + 1);`
114	`if (linebuff[i] == 0) break;`	114	`if (linebuff[i] == 0) break;`
115	`switch (linebuff[i]) {`	115	`switch (linebuff[i]) {`
116	`case 'e':`	116	`case 'e':`
117	`linebuff[i] = 0x1B; /* ESC code, using hex because '\e' is not ANSI C */`	117	`linebuff[i] = 0x1B; /* ESC code, using hex because '\e' is not ANSI C */`
118	`break;`	118	`break;`
119	`case 'n':`	119	`case 'n':`
120	`linebuff[i] = '\n';`	120	`linebuff[i] = '\n';`
121	`break;`	121	`break;`
122	`case 'r':`	122	`case 'r':`
123	`linebuff[i] = '\r';`	123	`linebuff[i] = '\r';`
124	`break;`	124	`break;`
125	`case 't':`	125	`case 't':`
126	`linebuff[i] = '\t';`	126	`linebuff[i] = '\t';`
127	`break;`	127	`break;`
128	`}`	128	`}`
129	`}`	129	`}`
130	`return(i);`	130	`return(i);`
131	`}`	131	`}`
132		132
133		133
134	`/* opens a CATS-style file and compiles it into a ressources lang block`	134	`/* opens a CATS-style file and compiles it into a ressources lang block`
135	`* returns 0 on error, or the size of the generated data block otherwise */`	135	`* returns 0 on error, or the size of the generated data block otherwise */`
136	`static unsigned short gen_langstrings(unsigned char buff, const char langid, struct bitmap b, const struct bitmap refb, const unsigned char *refblock) {`	136	`static unsigned short gen_langstrings(unsigned char buff, const char langid, struct bitmap b, const struct bitmap refb, const unsigned char *refblock) {`
137	`unsigned short len = 0, linelen;`	137	`unsigned short len = 0, linelen;`
138	`FILE *fd;`	138	`FILE *fd;`
139	`char fname[] = "XX.TXT";`	139	`char fname[] = "XX.TXT";`
140	`static char linebuf[8192];`	140	`static char linebuf[8192];`
141	`const char *ptr;`	141	`const char *ptr;`
142	`unsigned short id, linecount;`	142	`unsigned short id, linecount;`
143		143
144	`bitmap_init(b);`	144	`bitmap_init(b);`
145		145
146	`memcpy(fname + strlen(fname) - 6, langid, 2);`	146	`memcpy(fname + strlen(fname) - 6, langid, 2);`
147		147
148	`fd = fopen(fname, "rb");`	148	`fd = fopen(fname, "rb");`
149	`if (fd == NULL) {`	149	`if (fd == NULL) {`
150	`printf("ERROR: FAILED TO OPEN '%s'\r\n", fname);`	150	`printf("ERROR: FAILED TO OPEN '%s'\r\n", fname);`
151	`return(0);`	151	`return(0);`
152	`}`	152	`}`
153		153
154	`for (linecount = 1;; linecount++) {`	154	`for (linecount = 1;; linecount++) {`
155		155
156	`linelen = readl(linebuf, sizeof(linebuf), fd);`	156	`linelen = readl(linebuf, sizeof(linebuf), fd);`
157	`if (linelen == 0xffff) break; /* EOF */`	157	`if (linelen == 0xffff) break; /* EOF */`
158	`if ((linelen == 0) \|\| (linebuf[0] == '#')) continue;`	158	`if ((linelen == 0) \|\| (linebuf[0] == '#')) continue;`
159		159
160	`/* convert escaped chars to actual bytes (\n -> newline, etc) */`	160	`/* convert escaped chars to actual bytes (\n -> newline, etc) */`
161	`linelen = unesc_string(linebuf);`	161	`linelen = unesc_string(linebuf);`
162		162
163	`/* read id and get ptr to actual string ("1.15:string") */`	163	`/* read id and get ptr to actual string ("1.15:string") */`
164	`ptr = parseline(&id, linebuf);`	164	`ptr = parseline(&id, linebuf);`
165	`if (ptr == NULL) {`	165	`if (ptr == NULL) {`
166	`printf("ERROR: line #%u of %s is malformed (linelen = %u):\r\n", linecount, fname, linelen);`	166	`printf("ERROR: line #%u of %s is malformed (linelen = %u):\r\n", linecount, fname, linelen);`
167	`puts(linebuf);`	167	`puts(linebuf);`
168	`len = 0;`	168	`len = 0;`
169	`break;`	169	`break;`
170	`}`	170	`}`
171		171
172	`/* warn about dirty lines */`	172	`/* warn about dirty lines */`
173	`if (linebuf[0] == '?') {`	173	`if (linebuf[0] == '?') {`
174	`printf("WARNING: %s[#%u] string id %u.%u is flagged as 'dirty'\r\n", fname, linecount, id >> 8, id & 0xff);`	174	`printf("WARNING: %s[#%u] string id %u.%u is flagged as 'dirty'\r\n", fname, linecount, id >> 8, id & 0xff);`
175	`}`	175	`}`
176		176
177	`/* write string into block (II LL S) */`	177	`/* write string into block (II LL S) */`
178	`memcpy(buff + len, &id, 2);`	178	`memcpy(buff + len, &id, 2);`
179	`len += 2;`	179	`len += 2;`
180	`{`	180	`{`
181	`unsigned short slen = strlen(ptr) + 1;`	181	`unsigned short slen = strlen(ptr) + 1;`
182	`memcpy(buff + len, &slen, 2);`	182	`memcpy(buff + len, &slen, 2);`
183	`len += 2;`	183	`len += 2;`
184	`memcpy(buff + len, ptr, slen);`	184	`memcpy(buff + len, ptr, slen);`
185	`len += slen;`	185	`len += slen;`
186	`}`	186	`}`
187		187
188	`/* if reference bitmap provided: check that the id is valid */`	188	`/* if reference bitmap provided: check that the id is valid */`
189	`if ((refb != NULL) && (bitmap_get(refb, id) == 0)) {`	189	`if ((refb != NULL) && (bitmap_get(refb, id) == 0)) {`
190	`printf("WARNING: %s[#%u] has an invalid id (%u.%u not present in ref lang)\r\n", fname, linecount, id >> 8, id & 0xff);`	190	`printf("WARNING: %s[#%u] has an invalid id (%u.%u not present in ref lang)\r\n", fname, linecount, id >> 8, id & 0xff);`
191	`}`	191	`}`
192		192
193	`/* make sure this id is not already present */`	193	`/* make sure this id is not already present */`
194	`if (bitmap_get(b, id) == 0) {`	194	`if (bitmap_get(b, id) == 0) {`
195	`/* set bit in bitmap to remember I have this string */`	195	`/* set bit in bitmap to remember I have this string */`
196	`bitmap_set(b, id);`	196	`bitmap_set(b, id);`
197	`} else {`	197	`} else {`
198	`printf("WARNING: %s[#%u] has a duplicated id (%u.%u)\r\n", fname, linecount, id >> 8, id & 0xff);`	198	`printf("WARNING: %s[#%u] has a duplicated id (%u.%u)\r\n", fname, linecount, id >> 8, id & 0xff);`
199	`}`	199	`}`
200	`}`	200	`}`
201		201
202	`fclose(fd);`	202	`fclose(fd);`
203		203
204	`/* if refblock provided, pull missing strings from it */`	204	`/* if refblock provided, pull missing strings from it */`
205	`if (refblock != NULL) {`	205	`if (refblock != NULL) {`
206	`for (;;) {`	206	`for (;;) {`
207	`unsigned short slen;`	207	`unsigned short slen;`
208	`id = ((unsigned short *)refblock)[0];`	208	`id = ((unsigned short *)refblock)[0];`
209	`slen = ((unsigned short *)refblock)[1];`	209	`slen = ((unsigned short *)refblock)[1];`
210	`if ((id == 0) && (slen == 0)) break;`	210	`if ((id == 0) && (slen == 0)) break;`
211	`if (bitmap_get(b, id) == 0) {`	211	`if (bitmap_get(b, id) == 0) {`
212	`printf("WARNING: %s is missing string %u.%u (pulled from ref lang)\r\n", fname, id >> 8, id & 0xff);`	212	`printf("WARNING: %s is missing string %u.%u (pulled from ref lang)\r\n", fname, id >> 8, id & 0xff);`
213	`/* copy missing string from refblock */`	213	`/* copy missing string from refblock */`
214	`memcpy(buff + len, refblock, slen + 4);`	214	`memcpy(buff + len, refblock, slen + 4);`
215	`len += slen + 4;`	215	`len += slen + 4;`
216	`}`	216	`}`
217	`refblock += slen + 4;`	217	`refblock += slen + 4;`
218	`}`	218	`}`
219	`}`	219	`}`
220		220
221	`/* write the block terminator (0-long string) */`	221	`/* write the block terminator (0-long string) */`
222	`buff[len++] = 0; /* id */`	222	`buff[len++] = 0; /* id */`
223	`buff[len++] = 0; /* id */`	223	`buff[len++] = 0; /* id */`
224	`buff[len++] = 0; /* len */`	224	`buff[len++] = 0; /* len */`
225	`buff[len++] = 0; /* len */`	225	`buff[len++] = 0; /* len */`
226	`buff[len++] = 0; /* empty string */`	226	`buff[len++] = 0; /* empty string */`
227		227
228	`return(len);`	228	`return(len);`
229	`}`	229	`}`
230		230
231		231
232	`#define MEMBLOCKSZ 65000`	232	`#define MEMBLOCKSZ 65000`
233		233
234	`int main(int argc, char **argv) {`	234	`int main(int argc, char **argv) {`
235	`FILE *fd;`	235	`FILE *fd;`
236	`int ecode = 0;`	236	`int ecode = 0;`
237	`char buff, refblock;`	237	`char buff, refblock;`
238	`unsigned short refblocksz = 0;`	238	`unsigned short refblocksz = 0;`
239	`static struct bitmap bufbitmap;`	239	`static struct bitmap bufbitmap;`
240	`static struct bitmap refbitmap;`	240	`static struct bitmap refbitmap;`
241	`unsigned short i;`	241	`unsigned short i;`
242	`unsigned short biggest_langsz = 0;`	242	`unsigned short biggest_langsz = 0;`
243		243
244	`if (argc < 2) {`	244	`if (argc < 2) {`
245	`puts("tlumacz ver " SVARLANGVER " - this tool is part of the SvarLANG project.");`	245	`puts("tlumacz ver " SVARLANGVER " - this tool is part of the SvarLANG project.");`
246	`puts("converts a set of CATS-style translations in files EN.TXT, PL.TXT, etc");`	246	`puts("converts a set of CATS-style translations in files EN.TXT, PL.TXT, etc");`
247	`puts("into a single resource file (OUT.LNG).");`	247	`puts("into a single resource file (OUT.LNG).");`
248	`puts("");`	248	`puts("");`
249	`puts("usage: tlumacz en fr pl ...");`	249	`puts("usage: tlumacz en fr pl ...");`
250	`return(1);`	250	`return(1);`
251	`}`	251	`}`
252		252
253	`buff = malloc(MEMBLOCKSZ);`	253	`buff = malloc(MEMBLOCKSZ);`
254	`refblock = malloc(MEMBLOCKSZ);`	254	`refblock = malloc(MEMBLOCKSZ);`
255	`if ((buff == NULL) \|\| (refblock == NULL)) {`	255	`if ((buff == NULL) \|\| (refblock == NULL)) {`
256	`puts("out of memory");`	256	`puts("out of memory");`
257	`return(1);`	257	`return(1);`
258	`}`	258	`}`
259		259
260	`fd = fopen("out.lng", "wb");`	260	`fd = fopen("out.lng", "wb");`
261	`if (fd == NULL) {`	261	`if (fd == NULL) {`
262	`puts("ERR: failed to open or create SVARCOM.LNG");`	262	`puts("ERR: failed to open or create OUT.LNG");`
263	`return(1);`	263	`return(1);`
264	`}`	264	`}`
265		265
266	`/* write sig */`	266	`/* write sig */`
267	`fwrite("SvL\x1b", 1, 4, fd);`	267	`fwrite("SvL\x1b", 1, 4, fd);`
268		268
269	`/* write lang blocks */`	269	`/* write lang blocks */`
270	`for (i = 1; i < argc; i++) {`	270	`for (i = 1; i < argc; i++) {`
271	`unsigned short sz;`	271	`unsigned short sz;`
272	`char id[3];`	272	`char id[3];`
273		273
274	`if (strlen(argv[i]) != 2) {`	274	`if (strlen(argv[i]) != 2) {`
275	`printf("INVALID LANG SPECIFIED: %s\r\n", argv[i]);`	275	`printf("INVALID LANG SPECIFIED: %s\r\n", argv[i]);`
276	`ecode = 1;`	276	`ecode = 1;`
277	`break;`	277	`break;`
278	`}`	278	`}`
279		279
280	`id[0] = argv[i][0];`	280	`id[0] = argv[i][0];`
281	`id[1] = argv[i][1];`	281	`id[1] = argv[i][1];`
282	`id[2] = 0;`	282	`id[2] = 0;`
283	`if (id[0] >= 'a') id[0] -= 'a' - 'A';`	283	`if (id[0] >= 'a') id[0] -= 'a' - 'A';`
284	`if (id[1] >= 'a') id[1] -= 'a' - 'A';`	284	`if (id[1] >= 'a') id[1] -= 'a' - 'A';`
285		285
286	`sz = gen_langstrings(buff, id, &bufbitmap, (i != 1)?&refbitmap:NULL, (i != 1)?refblock:NULL);`	286	`sz = gen_langstrings(buff, id, &bufbitmap, (i != 1)?&refbitmap:NULL, (i != 1)?refblock:NULL);`
287	`if (sz == 0) {`	287	`if (sz == 0) {`
288	`printf("ERROR COMPUTING LANG '%s'\r\n", id);`	288	`printf("ERROR COMPUTING LANG '%s'\r\n", id);`
289	`ecode = 1;`	289	`ecode = 1;`
290	`break;`	290	`break;`
291	`} else {`	291	`} else {`
292	`printf("computed %s lang block of %u bytes\r\n", id, sz);`	292	`printf("computed %s lang block of %u bytes\r\n", id, sz);`
293	`if (sz > biggest_langsz) biggest_langsz = sz;`	293	`if (sz > biggest_langsz) biggest_langsz = sz;`
294	`}`	294	`}`
295	`/* write lang ID to file, followed by block size and then the actual block */`	295	`/* write lang ID to file, followed by block size and then the actual block */`
296	`if ((fwrite(id, 1, 2, fd) != 2) \|\|`	296	`if ((fwrite(id, 1, 2, fd) != 2) \|\|`
297	`(fwrite(&sz, 1, 2, fd) != 2) \|\|`	297	`(fwrite(&sz, 1, 2, fd) != 2) \|\|`
298	`(fwrite(buff, 1, sz, fd) != sz)) {`	298	`(fwrite(buff, 1, sz, fd) != sz)) {`
299	`printf("ERROR WRITING TO OUTPUT FILE\r\n");`	299	`printf("ERROR WRITING TO OUTPUT FILE\r\n");`
300	`ecode = 1;`	300	`ecode = 1;`
301	`break;`	301	`break;`
302	`}`	302	`}`
303	`/* remember reference data for other languages */`	303	`/* remember reference data for other languages */`
304	`if (i == 1) {`	304	`if (i == 1) {`
305	`refblocksz = sz;`	305	`refblocksz = sz;`
306	`memcpy(refblock, buff, MEMBLOCKSZ);`	306	`memcpy(refblock, buff, MEMBLOCKSZ);`
307	`memcpy(&refbitmap, &bufbitmap, sizeof(struct bitmap));`	307	`memcpy(&refbitmap, &bufbitmap, sizeof(struct bitmap));`
308	`}`	308	`}`
309	`}`	309	`}`
310		310
311	`fclose(fd);`	311	`fclose(fd);`
312		312
313	`/* compute the deflang.c file containing a dump of the reference block */`	313	`/* compute the deflang.c file containing a dump of the reference block */`
314	`fd = fopen("DEFLANG.C", "wb");`	314	`fd = fopen("DEFLANG.C", "wb");`
315	`if (fd == NULL) {`	315	`if (fd == NULL) {`
316	`puts("ERROR: FAILED TO OPEN OR CREATE DEFLANG.C");`	316	`puts("ERROR: FAILED TO OPEN OR CREATE DEFLANG.C");`
317	`ecode = 1;`	317	`ecode = 1;`
318	`} else {`	318	`} else {`
319	`unsigned short allocsz = biggest_langsz + (biggest_langsz / 20);`	319	`unsigned short allocsz = biggest_langsz + (biggest_langsz / 20);`
320	`printf("biggest lang block is %u bytes -> allocating a %u bytes buffe (5%% safety margin)r\n", biggest_langsz, allocsz);`	320	`printf("biggest lang block is %u bytes -> allocating a %u bytes buffe (5%% safety margin)r\n", biggest_langsz, allocsz);`
321	`fprintf(fd, "/* THIS FILE HAS BEEN GENERATED BY TLUMACZ (PART OF THE SVARLANG LIBRARY) */\r\n");`	321	`fprintf(fd, "/* THIS FILE HAS BEEN GENERATED BY TLUMACZ (PART OF THE SVARLANG LIBRARY) */\r\n");`
322	`fprintf(fd, "const unsigned short svarlang_memsz = %uu;\r\n", allocsz);`	322	`fprintf(fd, "const unsigned short svarlang_memsz = %uu;\r\n", allocsz);`
323	`fprintf(fd, "char svarlang_mem[%u] = {\r\n", allocsz);`	323	`fprintf(fd, "char svarlang_mem[%u] = {\r\n", allocsz);`
324	`for (i = 0; i < refblocksz; i++) {`	324	`for (i = 0; i < refblocksz; i++) {`
325	`fprintf(fd, "%u", refblock[i]);`	325	`fprintf(fd, "%u", refblock[i]);`
326	`if (i + 1 < refblocksz) fprintf(fd, ",");`	326	`if (i + 1 < refblocksz) fprintf(fd, ",");`
327	`if ((i & 15) == 15) fprintf(fd, "\r\n");`	327	`if ((i & 15) == 15) fprintf(fd, "\r\n");`
328	`}`	328	`}`
329	`fprintf(fd, "};\r\n");`	329	`fprintf(fd, "};\r\n");`
330	`fclose(fd);`	330	`fclose(fd);`
331	`}`	331	`}`
332		332
333	`return(ecode);`	333	`return(ecode);`
334	`}`	334	`}`
335		335

Subversion Repositories SvarDOS

(root)/svarlang.lib/trunk/tlumacz.c – Rev 1248 → 1250