asearch.c

harvest是一个下载html网页得机器人

							free_buf(text, buffer);
							NEW_FILE = OFF;
							return 0;  
						}
						printout_end = i - D_length - 1;           
						if((text != -1) && !(lasti >= Max_record + l - 1)) {
							if (-1 == output(buffer, lasti, printout_end, j)) {free_buf(text, buffer); return -1;}
						}
						else if ((text == -1) && !(lasti >= l)) {
							if (-1 == output(buffer, lasti, printout_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 */
						}
					}
					lasti = i - D_length; /* point to starting position of D_pat */
					for(k=0; k<= D; k++) {
						B[k] = Init[0];
					}
					r1 = B[0] & r_Init1;
					A[0] = (((B[0]>>1) & CMask) | r1) & D_Mask;
					for(k=1; k<= D; k++) {
						r1 = Init1 & B[k];
						r2 = B[k-1] | (((A[k-1] | B[k-1])>>1)&r_NO_ERR);
						A[k] = (((B[k]>>1)&CMask) | r1 | r2) ;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
				c = buffer[i++];
				CurrentByteOffset ++;
				CMask = Mask[c];
				r1   = r_Init1 & A[0];
				B[0] = ((A[0] >> 1 ) & CMask) | r1;
				for(k=1; k<=D; k++) {
					r1 = r_Init1 & A[k];
					r2 = A[k-1] | (((A[k-1]|B[k-1])>>1) & r_NO_ERR);
					B[k] = ((A[k] >> 1) & CMask) | r2 | r1;
				}
				if(B[0] & endpos) {
					j++;  
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					r1 = B[D];
					if(((AND == 1) && ((r1 & endposition) == endposition)) || ((AND == 0) && (r1 & endposition)) ^ 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';
							}

							/*
							if (text == -1) {
								memcpy(buffer+end-D_length, tempbuf, D_length+1);
							}
							*/
							free_buf(text, buffer);
							NEW_FILE = OFF;
							return 0; 
						}
						printout_end = i - D_length -1 ; 
						if((text != -1) && !(lasti >= Max_record + l - 1)) {
							if (-1 == output(buffer, lasti, printout_end, j)) {free_buf(text, buffer); return -1;}
						}
						else if ((text == -1) && !(lasti >= l)) {
							if (-1 == output(buffer, lasti, printout_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 */
						}
					}
					lasti = i - D_length ;
					for(k=0; k<= D; k++) {
						A[k] = Init[0];
					}
					r1 = A[0] & r_Init1; 
					B[0] = (((A[0]>>1)&CMask) | r1) & D_Mask;
					for(k=1; k<= D; k++) {
						r1 = r_Init1 & A[k];
						r2 = A[k-1] | (((A[k-1] | B[k-1])>>1)&r_NO_ERR);
						B[k] = (((A[k]>>1)&CMask) | r1 | r2) ;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
			}

			if(l < BlockSize) {
				lasti = Max_record;
			}
			else {
				ResidueSize = Max_record + l - lasti;
				if(ResidueSize > Max_record) {
					ResidueSize = Max_record;
					TRUNCATE = ON;         
				}
				strncpy(buffer+Max_record-ResidueSize, buffer+lasti, ResidueSize);
				lasti = Max_record - ResidueSize;
				if(lasti == 0)     lasti = 1; 
			}
		}
		free_buf(text, buffer);
		return 0;
#if	AGREP_POINTER
	}
	else {
		lasti = 1;
		/* if (DELIMITER) tempbuf = (CHAR*)malloc(D_length + 1); */
		buffer = (CHAR *)agrep_inbuffer;
		l = agrep_inlen;
		end = l;
		/* buffer[end-1] = '\n';*/ /* at end of the text. */
		/* buffer[0] = '\n';*/  /* in front of the  text. */
		i = 0;
		if(DELIMITER) {
			for(k=0; k<D_length; k++) {
				if(old_D_pat[k] != buffer[k]) break;
			}
			if(k>=D_length) j--;
			/*
			memcpy(tempbuf, buffer+end, D_length+1);
			strncpy(buffer+end, old_D_pat, D_length);
			buffer[end+D_length] = '\0';
			end = end + D_length;
			*/
		}

			/* An exact copy of the above ASEARCH0_PROCESS: the while-loop below */
			while (i < end )
			{
				c = buffer[i++];
				CurrentByteOffset ++;
				CMask = Mask[c];
				r1 = B[0] & r_Init1;
				A[0] = (((B[0] >> 1)) & CMask | r1 ) ;
				for(k=1; k<=D; k++) {
					r1 = r_Init1 & B[k];
					r2 = B[k-1] | (((A[k-1]|B[k-1])>>1) & r_NO_ERR);
					A[k] = ((B[k] >> 1) & CMask) | r2 | r1;
				}
				if(A[0] & endpos) {
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					j++;  
					r1 = A[D];
					if(((AND == 1) && ((r1 & endposition) == endposition)) || ((AND == 0) && (r1 & endposition)) ^ 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';
							}

							/*
							if (text == -1) {
								memcpy(buffer+end-D_length, tempbuf, D_length+1);
							}
							*/
							free_buf(text, buffer);
							NEW_FILE = OFF;
							return 0;  
						}
						printout_end = i - D_length - 1;           
						if((text != -1) && !(lasti >= Max_record + l - 1)) {
							if (-1 == output(buffer, lasti, printout_end, j)) {free_buf(text, buffer); return -1;}
						}
						else if ((text == -1) && !(lasti >= l)) {
							if (-1 == output(buffer, lasti, printout_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 */
						}
					}
					lasti = i - D_length; /* point to starting position of D_pat */
					for(k=0; k<= D; k++) {
						B[k] = Init[0];
					}
					r1 = B[0] & r_Init1;
					A[0] = (((B[0]>>1) & CMask) | r1) & D_Mask;
					for(k=1; k<= D; k++) {
						r1 = Init1 & B[k];
						r2 = B[k-1] | (((A[k-1] | B[k-1])>>1)&r_NO_ERR);
						A[k] = (((B[k]>>1)&CMask) | r1 | r2) ;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
				c = buffer[i++];
				CurrentByteOffset ++;
				CMask = Mask[c];
				r1   = r_Init1 & A[0];
				B[0] = ((A[0] >> 1 ) & CMask) | r1;
				for(k=1; k<=D; k++) {
					r1 = r_Init1 & A[k];
					r2 = A[k-1] | (((A[k-1]|B[k-1])>>1) & r_NO_ERR);
					B[k] = ((A[k] >> 1) & CMask) | r2 | r1;
				}
				if(B[0] & endpos) {
					j++;  
					if (DELIMITER) CurrentByteOffset -= D_length;
					else CurrentByteOffset -= 1;
					r1 = B[D];
					if(((AND == 1) && ((r1 & endposition) == endposition)) || ((AND == 0) && (r1 & endposition)) ^ 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';
							}

							/*
							if (text == -1) {
								memcpy(buffer+end-D_length, tempbuf, D_length+1);
							}
							*/
							free_buf(text, buffer);
							NEW_FILE = OFF;
							return 0; 
						}
						printout_end = i - D_length -1 ; 
						if((text != -1) && !(lasti >= Max_record + l - 1)) {
							if (-1 == output(buffer, lasti, printout_end, j)) {free_buf(text, buffer); return -1;}
						}
						else if ((text == -1) && !(lasti >= l)) {
							if (-1 == output(buffer, lasti, printout_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 */
						}
					}
					lasti = i - D_length ;
					for(k=0; k<= D; k++) {
						A[k] = Init[0];
					}
					r1 = A[0] & r_Init1; 
					B[0] = (((A[0]>>1)&CMask) | r1) & D_Mask;
					for(k=1; k<= D; k++) {
						r1 = r_Init1 & A[k];
						r2 = A[k-1] | (((A[k-1] | B[k-1])>>1)&r_NO_ERR);
						B[k] = (((A[k]>>1)&CMask) | r1 | r2) ;
					}
					if (DELIMITER) CurrentByteOffset += 1*D_length;
					else CurrentByteOffset += 1*1;
				}
			}

		/*
		if (DELIMITER) {
			memcpy(buffer+end, tempbuf, D_length+1);
			free(tempbuf);
		}
		*/
		return 0;
	}
#endif	/*AGREP_POINTER*/
}