build_in.c

linux下阅读源码的好工具

				    onechar[1] = (x&0x0000ff00)>>8;
				    onechar[2] = x&0x000000ff;
				    write(f_out, onechar, 3);
#endif
				}
				x = ip->index[j];	/* for next round */
#if	0
				printf("#######x=%d ", x);
#endif	/*0*/
				y = 0;
			    }

			    diff = ip->offset[j] - y;
			    y = ip->offset[j];
			    if (diff < MaxNum1BPartition) {
				temp = encode8b(diff);
#if	!CRAZYSEGV
				putc(temp, f_out);
#else
			        onechar[0] = temp;
			        write(f_out, onechar, 1);
#endif
			    }
			    else if (diff < MaxNum2BPartition) {
				temp = encode8b((diff/MaxNum8bPartition) | 0x40);
#if	!CRAZYSEGV
				putc(temp, f_out);
#else
			        onechar[0] = temp;
			        write(f_out, onechar, 1);
#endif
				temp = encode8b(diff % MaxNum8bPartition);
#if	!CRAZYSEGV
				putc(temp, f_out);
#else
			        onechar[0] = temp;
			        write(f_out, onechar, 1);
#endif
			    }
			    else if (diff < MaxNum3BPartition) {
				temp = encode8b((diff/MaxNum16bPartition) | 0x80);
#if	!CRAZYSEGV
				putc(temp, f_out);
#else
			        onechar[0] = temp;
			        write(f_out, onechar, 1);
#endif
				temp = encode16b(diff % MaxNum16bPartition);
#if	!CRAZYSEGV
				putc((temp & 0x0000ff00) >> 8, f_out);
				putc(temp & 0x000000ff, f_out);
#else
			        onechar[0] = (temp & 0x0000ff00) >> 8;
			        onechar[1] = temp & 0x000000ff;
			        write(f_out, onechar, 2);
#endif
			    }
			    else {
				temp = encode8b((diff/MaxNum24bPartition) | 0xc0);
#if	!CRAZYSEGV
				putc(temp, f_out);
#else
				onechar[0] = temp;
				write(f_out, onechar, 1);
#endif
				temp = encode24b(diff % MaxNum24bPartition);
#if	!CRAZYSEGV
				putc((temp & 0x00ff0000) >> 16, f_out);
				putc((temp & 0x0000ff00) >> 8, f_out);
				putc(temp & 0x000000ff, f_out);
#else
			        onechar[0] = (temp & 0x00ff0000) >> 16;
				onechar[1] = (temp & 0x0000ff00) >> 8;
				onechar[2] = temp & 0x000000ff;
			        write(f_out, onechar, 3);
#endif
			    }
			    numonline ++;
			}
		    } /* ByteLevelIndex */
		    else if (OneFilePerBlock) {
			if (file_num <= MaxNum8bPartition) {
			    for(j=0; j < INDEX_SET_SIZE; j++) {
				if (ip->index[j] == INDEX_ELEM_FREE) continue;
#if	!CRAZYSEGV
				putc(encode8b(ip->index[j]), f_out);
#else
			        onechar[0] = encode8b(ip->index[j]);
			        write(f_out, onechar, 1);
#endif

			    }
			}
			else if (file_num <= MaxNum12bPartition) {
			  for(j=0; j < INDEX_SET_SIZE; j++) {
			    if (ip->index[j] == INDEX_ELEM_FREE) continue;
			    x = encode12b(ip->index[j]);
			    if (even_words) {
#if	!CRAZYSEGV
				putc(x & 0x000000ff, f_out);	/* lsb */
#else
			        onechar[0] = x & 0x000000ff;
			        write(f_out, onechar, 1);
#endif
				y = (x & 0x00000f00)>>8;	/* msb */
				even_words = 0;
			    }
			    else {	/* odd number of words so far */
			        y |= (x&0x00000f00)>>4; /* msb of x into msb of y */
#if	!CRAZYSEGV
				putc(y, f_out);
				putc(x&0x000000ff, f_out);
#else
			        onechar[0] = y;
			        onechar[1] = x&0x000000ff;
			        write(f_out, onechar, 2);
#endif
				even_words = 1;
			    }
			  }
			}
			else if (file_num <= MaxNum16bPartition) {
			    for(j=0; j < INDEX_SET_SIZE; j++) {
				if (ip->index[j] == INDEX_ELEM_FREE) continue;
				x = encode16b(ip->index[j]);
#if	!CRAZYSEGV
				putc((x&0x0000ff00)>>8, f_out);
				putc(x&0x000000ff, f_out);
#else
				onechar[0] = (x&0x0000ff00)>>8;
				onechar[1] = x&0x000000ff;
			        write(f_out, onechar, 2);
#endif
			    }
			}
			else {
			    for(j=0; j < INDEX_SET_SIZE; j++) {
				if (ip->index[j] == INDEX_ELEM_FREE) continue;
				x = encode24b(ip->index[j]);
#if	!CRAZYSEGV
				putc((x&0x00ff0000)>>16, f_out);
				putc((x&0x0000ff00)>>8, f_out);
				putc(x&0x000000ff, f_out);
#else
				onechar[0] = (x&0x00ff0000)>>16;
				onechar[1] = (x&0x0000ff00)>>8;
				onechar[2] = x&0x000000ff;
			        write(f_out, onechar, 3);
#endif
			    }
			}
		    } /* OneFilePerBlock */
		    else {	/* normal partitions */
			for(j=0; j < INDEX_SET_SIZE; j++) {
			    if (ip->index[j] == INDEX_ELEM_FREE) continue;
#if	!CRAZYSEGV
			    putc(ip->index[j], f_out);
#else
			    onechar[0] = ip->index[j];
			    write(f_out, onechar, 1);
#endif
			}
		    }
		}
                ip = ip->next_i;   /* go to next indices */
                indicesfree(ip_old, sizeof(struct indices));
                ip_old = ip;
            }
	    if (!ByteLevelIndex && OneFilePerBlock && !even_words && (file_num > MaxNum8bPartition) && (file_num <= MaxNum12bPartition)) {
#if	!CRAZYSEGV
		putc(y, f_out);
#else
		onechar[0] = y;
		write(f_out, onechar, 1);
#endif

	    }

next_token:

#if	!CRAZYSEGV
	    if (putc('\n', f_out) == EOF) {
#else
	    onechar[0] = '\n';
	    if (write(f_out, onechar, 1) <= 0) {
#endif
		fprintf(stderr, "Error: write failed at %s:%d\n", __FILE__, __LINE__); 
		exit(2);
	    }
            tp = tp->next_t;    /* go to next token */
#if	0
	    fprintf(stderr, "numelements=%d\n", numelements);
#endif	/*0*/
#if	BG_DEBUG
	    memory_usage -= (strlen(tp_old->word) + 1);
#endif	/*BG_DEBUG*/
	    wordfree(tp_old->word, 0);
            tokenfree(tp_old, sizeof(struct token));
            tp_old = tp;
	    numseencount ++;
        }
    }
    tokenallfree();
    indicesallfree();
    wordallfree();
#if	BG_DEBUG
    fprintf(stderr, "out of traverse(): saved/freed %d tokens: new usage: %d\n", numseencount, memory_usage);
#endif
#if	!CRAZYSEGV
    fflush(f_out);
    fclose(f_out);
#else
    close(f_out);
#endif
}

traverse1()
{
    FILE *i1, *i2, *i3;
    int ret;
    char s[MAX_LINE_LEN], es1[MAX_LINE_LEN], es2[MAX_LINE_LEN], es3[MAX_LINE_LEN];
    char s1[MAX_LINE_LEN];
    extern int errno;
    static int maxsortlinelen = 0;
    int	i;

    if (maxsortlinelen <= 0) {
	if (file_num < MaxNum8bPartition)
	    maxsortlinelen = round((MaxNum8bPartition * sizeof(int) + MAX_NAME_SIZE), MAX_LINE_LEN) * MAX_LINE_LEN;
	else if (file_num < MaxNum12bPartition)
	    maxsortlinelen = round((MaxNum12bPartition * sizeof(int) + MAX_NAME_SIZE), MAX_LINE_LEN) * MAX_LINE_LEN;
	else maxsortlinelen = MAX_SORTLINE_LEN;
    }

    traverse();	/* will produce .i2 and my_free allocated memory */

#if	USESORT_Z_OPTION
#if	DONTUSESORT_T_OPTION || SFS_COMPAT
    sprintf(s, "exec %s -z %d '%s/%s' > '%s/%s'\n", SYSTEM_SORT, maxsortlinelen, escapesinglequote(INDEX_DIR, es1), I2, escapesinglequote(INDEX_DIR, es2), O2);
#else
    sprintf(s, "exec %s -T '%s' -z %d '%s/%s' > '%s/%s'\n", SYSTEM_SORT, escapesinglequote(INDEX_DIR, es1), maxsortlinelen, escapesinglequote(INDEX_DIR, es2), I2, escapesinglequote(INDEX_DIR, es3), O2);
#endif
#else
#if	DONTUSESORT_T_OPTION || SFS_COMPAT
    sprintf(s, "exec %s '%s/%s' > '%s/%s'\n", SYSTEM_SORT, escapesinglequote(INDEX_DIR, es1), I2, escapesinglequote(INDEX_DIR, es2), O2);
#else
    sprintf(s, "exec %s -T '%s' '%s/%s' > '%s/%s'\n", SYSTEM_SORT, escapesinglequote(INDEX_DIR, es1), escapesinglequote(INDEX_DIR, es2), I2, escapesinglequote(INDEX_DIR, es3), O2);
#endif
#endif

#ifdef SW_DEBUG
    printf("%s", s);
#endif
    if((ret=system(s)) != 0) {
	sprintf(s1, "system('%s') failed at:\n\t File=%s, Line=%d, Errno=%d", s, __FILE__, __LINE__, errno);
	perror(s1);
	fprintf(stderr, "Please try to run the program again\n(If there's no memory, increase the swap area / don't use -M and -B options)\n");
	exit(2);
    }

#ifdef SW_DEBUG
    printf("mv .o2 .i2\n"); fflush(stdout);
#endif
#if SFS_COMPAT
    sprintf(s, "%s/%s", INDEX_DIR, O2);
    sprintf(s1, "%s/%s", INDEX_DIR, I2);
    rename(s, s1);
#else
    sprintf(s, "exec %s '%s/%s' '%s/%s'\n", SYSTEM_MV, escapesinglequote(INDEX_DIR, es1), O2, escapesinglequote(INDEX_DIR, es2), I2);
    system(s);
#endif
    system(sync_path);	/* sync() has a bug */
#if	0
    printf("traversed\n");
    sprintf(s, "exec %s -10 '%s/%s'\n", SYSTEM_HEAD, escapesinglequote(INDEX_DIR, es1), I2);
    system(s);
#endif	/*0*/

    /*
     * This flag is set from outside iff build-fast | build-addto option is set.
     */
    if(FirstTraverse1) {
	/* Mention whether numbers are indexed */
	if(IndexNumber) sprintf(s, "exec %s %%1234567890 > '%s/%s'\n", SYSTEM_ECHO, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	else sprintf(s, "exec %s %% > '%s/%s'\n", SYSTEM_ECHO, escapesinglequote(INDEX_DIR,es1), INDEX_FILE);
	system(s);

	/* Put the magic number: 0 if not 1file/blk, numfiles otherwise */
	if (OneFilePerBlock) {
		if (ByteLevelIndex) sprintf(s, "exec %s %%-%d >> '%s/%s'\n", SYSTEM_ECHO, file_num, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
		else sprintf(s, "exec %s %%%d >> '%s/%s'\n", SYSTEM_ECHO, file_num, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	}
	else sprintf(s, "exec %s %%0 >> '%s/%s'\n", SYSTEM_ECHO, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	system(s);

	/* Put the magic number: 0 if not structured index, 1 if so */
	if (StructuredIndex) sprintf(s, "exec %s %%%d >> '%s/%s'\n", SYSTEM_ECHO, attr_num, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	else if (RecordLevelIndex) sprintf(s, "exec %s %%-2 %s >> '%s/%s'\n", SYSTEM_ECHO, old_rdelim, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	else sprintf(s, "exec %s %%0 >> '%s/%s'\n", SYSTEM_ECHO, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
	system(s);

#ifdef	SW_DEBUG
	sprintf(s, "exec %s -l %s/.glimpse*\n", SYSTEM_LS, escapesinglequote(INDEX_DIR, es1));
	system(s);
#endif
        sprintf(s, "exec %s '%s/%s' >> '%s/%s'\n", SYSTEM_CAT, escapesinglequote(INDEX_DIR, es1), I2, escapesinglequote(INDEX_DIR, es2), INDEX_FILE);
        system(s);

#if	SFS_COMPAT
	sprintf(s, "%s/%s", INDEX_DIR, I2);
	unlink(s);
#else
	sprintf(s, "exec %s '%s/%s'\n", SYSTEM_RM, escapesinglequote(INDEX_DIR, es1), I2);
	system(s);
#endif

#ifdef	SW_DEBUG
	sprintf(s, "exec %s -l %s/.glimpse*\n", SYSTEM_LS, escapesinglequote(INDEX_DIR, es1));
	system(s);
#endif
#if	0
    printf("catted\n");
    sprintf(s, "exec %s -10 '%s/%s'\n", SYSTEM_HEAD, escapesinglequote(INDEX_DIR, es1), INDEX_FILE);
    system(s);
#endif	/*0*/
        FirstTraverse1 = 0;
	system(sync_path);	/* sync() has a bug */
        return;
    }
    /* else not first-traverse */

    sprintf(s, "%s/%s", INDEX_DIR, INDEX_FILE);
    if((i1 = fopen(s, "r")) == NULL) {	/* new stuff */
        fprintf(stderr, "can't open %s for reading\n", s);
        exit(2);
    }
    sprintf(s, "%s/%s", INDEX_DIR, I2);
    if((i2 = fopen(s, "r")) == NULL) {	/* old stuff */
        fprintf(stderr, "can't open %s for reading\n", s);
        exit(2);
    }
    sprintf(s, "%s/%s", INDEX_DIR, I3);
    if((i3 = fopen(s, "w")) == NULL) {	/* result */
        fprintf(stderr, "can't open %s for writing\n", s);
        exit(2);
    }

    /* Copy the 3 option fields (indexnumber, onefileperblock, structuredqueries) */
    fgets(s, 256, i1);
    s[255] = '\0';
    fputs(s, i3);
    fgets(s, 256, i1);
    s[255] = '\0';
    fputs(s, i3);
    fgets(s, 256, i1);
    s[255] = '\0';
    fputs(s, i3);

    merge_in(i2, i1, i3);
    /* merge_in(i1, i2, i3); */
#ifdef	BG_DEBUG
    fprintf(stderr, "out of merge_in()\n");
#endif	/*BG_DEBUG*/
    fclose(i1);
    fflush(i2);
    fclose(i2);
    fflush(i3);
    fclose(i3);
    system(sync_path);	/* sync() has a bug */
#ifdef SW_DEBUG
    printf("mv .i3 %s\n", INDEX_FILE); fflush(stdout);
#endif
#if	SFS_COMPAT
    sprintf(s, "%s/%s", INDEX_DIR, I3);
    sprintf(s1, "%s/%s", INDEX_DIR, INDEX_FILE);
    rename(s, s1);
#else
    sprintf(s, "exec %s '%s/%s' '%s/%s'", SYSTEM_MV, escapesinglequote(INDEX_DIR, es1), I3, escapesinglequote(INDEX_DIR, es2), INDEX_FILE);
    system(s);
#endif
/* #ifdef SW_DEBUG */
#if 0
    printf("ls -l .i2 %s\n", INDEX_FILE); fflush(stdout);
    sprintf(s, "exec %s -l %s/.glimpse*", SYSTEM_LS, escapesinglequote(INDEX_DIR, es1));
    printf("%d\n", system(s));