123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238 |
- #include <string.h>
- #include <stdlib.h>
- #include "msg.h"
- #include "bigram.h"
- #include "lcd_buf_draw.h"
- #ifdef WIKIPCF
- extern void showMsg(int currentLevel, char *format, ...);
- #include <assert.h>
- char aBigram[128][2];
- #else
- #include "file-io.h"
- #include "wiki_info.h"
- extern int nCurrentWiki;
- char aBigram[MAX_WIKIS_PER_DEVICE][128][2];
- #endif
- int32_t aCharIdx[128];
- #ifdef WIKIPCF
- void init_bigram(FILE *fd)
- {
- init_char_idx();
- ssize_t n = fread(aBigram, 1, sizeof(aBigram), fd);
- assert(sizeof(aBigram) == n);
- }
- #else
- void init_bigram(int fd)
- {
- init_char_idx();
- wl_read(fd, aBigram[nCurrentWiki], sizeof(aBigram) / MAX_WIKIS_PER_DEVICE);
- }
- #endif
- void init_char_idx()
- {
- char c;
- int i;
- int idx = 1;
- static int inited = 0;
-
- if (inited)
- return;
- inited = 1;
- memset(aCharIdx, 0, sizeof(aCharIdx));
- for (i = 0; i < 128; i++)
- {
- c = (char)i;
- if (is_supported_search_char(c))
- {
- if ('a' <= c && c <= 'z')
- aCharIdx[i] = aCharIdx[(int)'A' + (c - 'a')];
- else
- aCharIdx[i] = idx++;
- }
- }
- }
- int bigram_char_idx(char c)
- {
- return aCharIdx[(int)c];
- }
- #ifdef WIKIPCF
- void bigram_encode(char *outStr, char *inStr)
- {
- int i;
- int len;
- int rc;
- char c;
- int idxMatchedBigram;
- *outStr = '\0';
- len = strlen(inStr);
- while (len >= 2)
- {
- idxMatchedBigram = -1;
- for (i=0; i < 128; i++)
- {
- if ((rc = memcmp(inStr, &aBigram[i][0], 2)) == 0)
- {
- idxMatchedBigram = i;
- break;
- }
- else if (rc < 0)
- break;
- }
- if (idxMatchedBigram >= 0)
- {
- c = (char)i;
- c |= 0x80;
- *outStr = c;
- outStr++;
- inStr += 2;
- len -= 2;
- }
- else
- {
- *outStr = *inStr;
- outStr++;
- inStr++;
- len--;
- }
- }
- while (len > 0)
- {
- *outStr = *inStr;
- outStr++;
- inStr++;
- len--;
- }
- *outStr = '\0';
- }
- void bigram_decode(char *outStr, char *inStr, int lenMax)
- {
- unsigned char c;
- while (lenMax > 1 && (c = *inStr++) != '\0')
- {
- if (c >= 128 && aBigram[c-128][0])
- {
- *outStr = aBigram[c-128][0];
- outStr++;
- lenMax--;
- if (lenMax > 1)
- {
- *outStr = aBigram[c-128][1];
- outStr++;
- lenMax--;
- }
- }
- else
- {
- *outStr = c;
- outStr++;
- lenMax--;
- }
- }
- *outStr = '\0';
- }
- #else
- void bigram_decode(char *outStr, char *inStr, int lenMax)
- {
- unsigned char c;
- while (lenMax > 1 && (c = *inStr++) != '\0')
- {
- if (c >= 128 && aBigram[nCurrentWiki][c-128][0])
- {
- *outStr = aBigram[nCurrentWiki][c-128][0];
- outStr++;
- lenMax--;
- if (lenMax > 1)
- {
- *outStr = aBigram[nCurrentWiki][c-128][1];
- outStr++;
- lenMax--;
- }
- }
- else
- {
- *outStr = c;
- outStr++;
- lenMax--;
- }
- }
- *outStr = '\0';
- }
- #endif
- int is_supported_search_char(char c)
- {
- if (c && (strchr(SUPPORTED_SEARCH_CHARS, c) || ('A' <= c && c <= 'Z')))
- return 1;
- else
- return 0;
- }
- int search_string_cmp(char *title, char *search, int len)
- {
- int rc = 0;
- char c = 0;
- #if 0
- #ifdef WIKIPCF
- char temp[512];
- memcpy(temp, search, len);
- temp[len] = '\0';
- showMsg(3, "[%s][%s]\n", title, temp);
- #endif
- #endif
- while (!rc && len > 0)
- {
- c = *title;
- if (c && !is_supported_search_char(c))
- {
- title++;
- }
- else
- {
- if ('A' <= c && c <= 'Z')
- c += 32;
- if (c == *search)
- {
- title++;
- search++;
- len--;
- }
- else if (c > *search)
- rc = 1;
- else
- rc = -1;
- }
- }
- return rc;
- }
|