agrep.c

harvest是一个下载html网页得机器人

				CurrentByteOffset --;
				FIRST_TIME = 0;
			}

			/* RE1_PROCESS_WHEN_DNOTZERO: the while loop below */
			while (i < end )
			{
				c = buffer[i];
				CMask = Mask[c];
				if(c !=  Newline)
				{
					if(CMask != 0) {  
						r2 = B[0];
						r1 = Init1 & r2;
						A[0] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
						r3 = B[1];
						r1 = Init1 & r3;
						r0 = r2 | A[0];     /* A[0] | B[0] */
						A[1] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | (( r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextcharfile;
						r2 = B[2];
						r1 = Init1 & r2;
						r0 = r3 | A[1];
						A[2] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextcharfile;
						r3 = B[3];
						r1 = Init1 & r3;
						r0 = r2 | A[2];
						A[3] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextcharfile;
						r2 = B[4];
						r1 = Init1 & r2;
						r0 = r3 | A[3];
						A[4] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4)  goto Nextcharfile;
					}  /* if(CMask) */
					else  {
						r2 = B[0];
						A[0] = r2 & Init1; 
						r3 = B[1];
						r1 = Init1 & r3;
						r0 = r2 | A[0];
						A[1] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextcharfile;
						r2 = B[2];
						r1 = Init1 & r2;
						r0 = r3 | A[1];
						A[2] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextcharfile;
						r3 = B[3];
						r1 = Init1 & r3;
						r0 = r2 | A[2];
						A[3] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextcharfile;
						r2 = B[4];
						r1 = Init1 & r2;
						r0 = r3 | A[3];
						A[4] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4) goto Nextcharfile;
					}
				}
				else {  
					j++;
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					r1 = Init1 & B[D];            /* match against endofline */
					A[D] = ((Next[B[D]>>hh] | Next1[B[D]&LL]) & CMask) | r1;
					if(TAIL) A[D] = ( Next[A[D]>>hh] | Next1[A[D]&LL] ) | A[D]; 
					/* epsilon move */
					if(( A[D] & 1 ) ^ INVERSE) {
						if(FILENAMEONLY && (NEW_FILE || !POST_FILTER)) {
							num_of_matched++;

							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "%s", CurrentFileName);
							else {
								int outindex;
								for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
										(CurrentFileName[outindex] != '\0'); outindex++) {
									agrep_outbuffer[agrep_outpointer+outindex] = CurrentFileName[outindex];
								}
								if ((CurrentFileName[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								agrep_outpointer += outindex;
							}
							if (PRINTFILETIME) {
								char *s = aprint_file_time(CurrentFileTime);
								if (agrep_finalfp != NULL)
									fprintf(agrep_finalfp, "%s", s);
								else {
									int outindex;
									for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
											(s[outindex] != '\0'); outindex++) {
										agrep_outbuffer[agrep_outpointer+outindex] = s[outindex];
									}
									if ((s[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
										OUTPUT_OVERFLOW;
										free_buf(Text, buffer);
										return -1;
									}
									agrep_outpointer += outindex;
								}
							}
							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "\n");
							else {
								if (agrep_outpointer+1>=agrep_outlen) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								else agrep_outbuffer[agrep_outpointer++] = '\n';
							}

							free_buf(Text, buffer);
							NEW_FILE = OFF;
							return 0;
						} 
						if (-1 == r_output(buffer, i, end, j)) {free_buf(Text, buffer); return -1;}
						if (((LIMITOUTPUT > 0) && (LIMITOUTPUT <= num_of_matched)) ||
						    ((LIMITPERFILE > 0) && (LIMITPERFILE <= num_of_matched - prev_num_of_matched))) {
							free_buf(Text, buffer);
							return 0;	/* done */
						}
					}
					for(k=0; k<=D; k++)  B[k] = Init[0];
					r1 = Init1 & B[0];
					A[0] = (( Next[B[0]>>hh] | Next1[B[0]&LL]) & CMask) | r1;
					for(k=1; k<=D; k++) {
						r3 = B[k];
						r1 = Init1 & r3;
						r2 = A[k-1] | B[k-1];
						A[k] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((B[k-1] | Next[r2>>hh] | Next1[r2&LL]) & r_NO_ERR) | r1;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
	Nextcharfile: 
				i=i+1;
				CurrentByteOffset ++;
				c = buffer[i];
				CMask = Mask[c];
				if(c != Newline)
				{
					if(CMask != 0) {  
						r2 = A[0];
						r1 = Init1 & r2;
						B[0] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
						r3 = A[1];
						r1 = Init1 & r3;
						r0 = B[0] | r2;
						B[1] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL]) & r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextchar1file;
						r2 = A[2];
						r1 = Init1 & r2;
						r0 = B[1] | r3;
						B[2] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextchar1file;
						r3 = A[3];
						r1 = Init1 & r3;
						r0 = B[2] | r2;
						B[3] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextchar1file;
						r2 = A[4];
						r1 = Init1 & r2;
						r0 = B[3] | r3;
						B[4] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4)   goto Nextchar1file;
					}  /* if(CMask) */
					else  {
						r2 = A[0];
						B[0] = r2 & Init1; 
						r3 = A[1];
						r1 = Init1 & r3;
						r0 = B[0] | r2;
						B[1] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextchar1file;
						r2 = A[2];
						r1 = Init1 & r2;
						r0 = B[1] | r3;
						B[2] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextchar1file;
						r3 = A[3];
						r1 = Init1 & r3;
						r0 = B[2] | r2;
						B[3] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextchar1file;
						r2 = A[4];
						r1 = Init1 & r2;
						r0 = B[3] | r3;
						B[4] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4) goto Nextchar1file;
					}
				} /* if(NOT Newline) */
				else {  
					j++;
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					r1 = Init1 & A[D];            /* match against endofline */
					B[D] = ((Next[A[D]>>hh] | Next1[A[D]&LL]) & CMask) | r1;
					if(TAIL) B[D] = ( Next[B[D]>>hh] | Next1[B[D]&LL] ) | B[D]; 
					/* epsilon move */
					if(( B[D] & 1 ) ^ INVERSE) {
						if(FILENAMEONLY && (NEW_FILE || !POST_FILTER)) {
							num_of_matched++;

							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "%s", CurrentFileName);
							else {
								int outindex;
								for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
										(CurrentFileName[outindex] != '\0'); outindex++) {
									agrep_outbuffer[agrep_outpointer+outindex] = CurrentFileName[outindex];
								}
								if ((CurrentFileName[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								agrep_outpointer += outindex;
							}
							if (PRINTFILETIME) {
								char *s = aprint_file_time(CurrentFileTime);
								if (agrep_finalfp != NULL)
									fprintf(agrep_finalfp, "%s", s);
								else {
									int outindex;
									for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
											(s[outindex] != '\0'); outindex++) {
										agrep_outbuffer[agrep_outpointer+outindex] = s[outindex];
									}
									if ((s[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
										OUTPUT_OVERFLOW;
										free_buf(Text, buffer);
										return -1;
									}
									agrep_outpointer += outindex;
								}
							}
							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "\n");
							else {
								if (agrep_outpointer+1>=agrep_outlen) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								else agrep_outbuffer[agrep_outpointer++] = '\n';
							}

							free_buf(Text, buffer);
							NEW_FILE = OFF;
							return 0;
						} 
						if (-1 == r_output(buffer, i, end, j)) {free_buf(Text, buffer); return -1;}
						if (((LIMITOUTPUT > 0) && (LIMITOUTPUT <= num_of_matched)) ||
						    ((LIMITPERFILE > 0) && (LIMITPERFILE <= num_of_matched - prev_num_of_matched))) {
							free_buf(Text, buffer);
							return 0;	/* done */
						}
					}
					for(k=0; k<=D; k++) A[k] = Init0; 
					r1 = Init1 & A[0];
					B[0] = ((Next[A[0]>>hh] | Next1[A[0]&LL]) & CMask) | r1;
					for(k=1; k<=D; k++) {
						r3 = A[k];
						r1 = Init1 & r3;
						r2 = A[k-1] | B[k-1];
						B[k] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((A[k-1] | Next[r2>>hh] | Next1[r2&LL]) & r_NO_ERR) | r1;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
	Nextchar1file: 
				i=i+1;
				CurrentByteOffset ++;
			} /* while i < end */

			strncpy(buffer, buffer+num_read, Maxline);
		} /* while fill_buf... */
		free_buf(Text, buffer);
		return 0;
#if	AGREP_POINTER
	}
	else {	/* within the memory buffer: assume it starts with a newline at position 0, the actual pattern follows that, and it ends with a '\n' */
		num_read = agrep_inlen;
		buffer = (CHAR *)agrep_inbuffer;
		end = num_read;
		/* buffer[end-1] = '\n';*/ /* at end of the text. */
		/* buffer[0] = '\n';*/  /* in front of the  text. */
		i = 0;

			/* An exact copy of the above RE1_PROCESS_WHEN_DNOTZERO: the while loop below */
			while (i < end )
			{
				c = buffer[i];
				CMask = Mask[c];
				if(c !=  Newline)
				{
					if(CMask != 0) {  
						r2 = B[0];
						r1 = Init1 & r2;
						A[0] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | r1;
						r3 = B[1];
						r1 = Init1 & r3;
						r0 = r2 | A[0];     /* A[0] | B[0] */
						A[1] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | (( r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextcharmem;
						r2 = B[2];
						r1 = Init1 & r2;
						r0 = r3 | A[1];
						A[2] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextcharmem;
						r3 = B[3];
						r1 = Init1 & r3;
						r0 = r2 | A[2];
						A[3] = ((Next[r3>>hh] | Next1[r3&LL]) & CMask) | ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextcharmem;
						r2 = B[4];
						r1 = Init1 & r2;
						r0 = r3 | A[3];
						A[4] = ((Next[r2>>hh] | Next1[r2&LL]) & CMask) | ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4)  goto Nextcharmem;
					}  /* if(CMask) */
					else  {
						r2 = B[0];
						A[0] = r2 & Init1; 
						r3 = B[1];
						r1 = Init1 & r3;
						r0 = r2 | A[0];
						A[1] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 1) goto Nextcharmem;
						r2 = B[2];
						r1 = Init1 & r2;
						r0 = r3 | A[1];
						A[2] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 2) goto Nextcharmem;
						r3 = B[3];
						r1 = Init1 & r3;
						r0 = r2 | A[2];
						A[3] = ((r2 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 3) goto Nextcharmem;
						r2 = B[4];
						r1 = Init1 & r2;
						r0 = r3 | A[3];
						A[4] = ((r3 | Next[r0>>hh] | Next1[r0&LL])&r_NO_ERR) | r1 ;  
						if(D == 4) goto Nextcharmem;
					}
				}
				else {  
					j++;
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					r1 = Init1 & B[D];            /* match against endofline */
					A[D] = ((Next[B[D]>>hh] | Next1[B[D]&LL]) & CMask) | r1;
					if(TAIL) A[D] = ( Next[A[D]>>hh] | Next1[A[D]&LL] ) | A[D]; 
					/* epsilon move */
					if(( A[D] & 1 ) ^ INVERSE) {
						if(FILENAMEONLY && (NEW_FILE || !POST_FILTER)) {
							num_of_matched++;

							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "%s", CurrentFileName);
							else {
								int outindex;
								for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
										(CurrentFileName[outindex] != '\0'); outindex++) {
									agrep_outbuffer[agrep_outpointer+outindex] = CurrentFileName[outindex];
								}
								if ((CurrentFileName[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								agrep_outpointer += outindex;
							}
							if (PRINTFILETIME) {
								char *s = aprint_file_time(CurrentFileTime);
								if (agrep_finalfp != NULL)
									fprintf(agrep_finalfp, "%s", s);
								else {
									int outindex;
									for(outindex=0; (outindex+agrep_outpointer<agrep_outlen) &&
											(s[outindex] != '\0'); outindex++) {
										agrep_outbuffer[agrep_outpointer+outindex] = s[outindex];
									}
									if ((s[outindex] != '\0') || (outindex+agrep_outpointer>=agrep_outlen)) {
										OUTPUT_OVERFLOW;
										free_buf(Text, buffer);
										return -1;
									}
									agrep_outpointer += outindex;
								}
							}
							if (agrep_finalfp != NULL)
								fprintf(agrep_finalfp, "\n");
							else {
								if (agrep_outpointer+1>=agrep_outlen) {
									OUTPUT_OVERFLOW;
									free_buf(Text, buffer);
									return -1;
								}
								else agrep_outbuffer[agrep_outpointer++] = '\n';
							}

							free_buf(Text, buffer);
							NEW_FILE = OFF;
							return 0;
						}