Downloads: 1
Diff of
/scripts/template1.R
[000000]
..
[c09aa8]
Switch to unified view
| a | b/scripts/template1.R | ||
|---|---|---|---|
| 1 | require(data.table) |
||
| 2 | |||
| 3 | cbindPad <- function(...){ |
||
| 4 | args <- list(...) |
||
| 5 | n <- sapply(args,nrow) |
||
| 6 | mx <- max(n) |
||
| 7 | pad <- function(x, mx){ |
||
| 8 | if (nrow(x) < mx){ |
||
| 9 | nms <- colnames(x) |
||
| 10 | padTemp <- matrix(NA, mx - nrow(x), ncol(x)) |
||
| 11 | colnames(padTemp) <- nms |
||
| 12 | if (ncol(x)==0) { |
||
| 13 | return(padTemp) |
||
| 14 | } else { |
||
| 15 | return(rbind(x,padTemp)) |
||
| 16 | } |
||
| 17 | } |
||
| 18 | else{ |
||
| 19 | return(x) |
||
| 20 | } |
||
| 21 | } |
||
| 22 | rs <- lapply(args,pad,mx) |
||
| 23 | return(do.call(cbind,rs)) |
||
| 24 | } |
||
| 25 | |||
| 26 | numbo <- 3 |
||
| 27 | |||
| 28 | |||
| 29 | if (numbo <= 0) { |
||
| 30 | str2 <- paste("clust_", numbo, ".txt", sep = "") |
||
| 31 | } |
||
| 32 | setwd("/media/hdd0/unraiddisk1/student/neilm/primestmp/clusters") |
||
| 33 | |||
| 34 | res <- try(read.table(str2),silent = TRUE) |
||
| 35 | if (!inherits(res, 'try-error')) { |
||
| 36 | b <- read.table(str2) |
||
| 37 | b <- b[,2, drop = F] |
||
| 38 | b[] <- lapply(b, as.character) |
||
| 39 | b[] <- lapply(b, tolower) |
||
| 40 | |||
| 41 | l <- as.data.frame(matrix(0, ncol = 1, nrow = (nrow(b)))) |
||
| 42 | |||
| 43 | |||
| 44 | for (i in 1:nrow(b)) { |
||
| 45 | if (length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))) > ncol(l)) { |
||
| 46 | l <- as.data.frame(matrix(0, ncol = length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))), nrow = (nrow(b)))) |
||
| 47 | } |
||
| 48 | |||
| 49 | } |
||
| 50 | n <- 1 |
||
| 51 | for (i in 1:nrow(b)) { |
||
| 52 | |||
| 53 | n <- as.numeric(as.numeric(ncol(l)) - as.numeric(length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))))) |
||
| 54 | l[i,] <- c( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE)), rep("IGNORE", times = as.numeric(n))) |
||
| 55 | } |
||
| 56 | |||
| 57 | l <- cbind(l, rep(0, times = nrow(l))) |
||
| 58 | names(l)[ncol(l)] <- "wc" |
||
| 59 | for (i in 1:nrow(l)) { |
||
| 60 | l[i,ncol(l)] <- length(grep("IGNORE", l[i,])) |
||
| 61 | } |
||
| 62 | |||
| 63 | |||
| 64 | row1 <- unlist(strsplit(unlist(strsplit(b[which.max(l$wc),1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE)) |
||
| 65 | |||
| 66 | l <- l[1:(ncol(l)-1)] |
||
| 67 | d <- as.data.frame(matrix(.Machine$double.xmax, ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 68 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 69 | |||
| 70 | results <- rep('IGNORE', times = ncol(l)) |
||
| 71 | n <- 1 |
||
| 72 | k <- 1 |
||
| 73 | g <- 1 |
||
| 74 | p <- 0 |
||
| 75 | s <- 1 |
||
| 76 | check <- 0 |
||
| 77 | ignorecheck <- 0 |
||
| 78 | switch01 <- 0 |
||
| 79 | while (n <= length(row1)) { |
||
| 80 | while (k <= length(row1)) { |
||
| 81 | p <- 0 |
||
| 82 | if (n > 0) { |
||
| 83 | if (k > 0) { |
||
| 84 | for (i in 1:nrow(l)) { |
||
| 85 | if (i > 0) { |
||
| 86 | for (v in 1:ncol(l)) { |
||
| 87 | for (w in 1:ncol(l)) { |
||
| 88 | if (s == 1) { |
||
| 89 | if ( ((paste(l[i, v:w], collapse = ' ')) == (paste(row1[n:k], collapse = ' '))) ){ |
||
| 90 | if (w < listo[i]) { |
||
| 91 | p <- p + 1 |
||
| 92 | listo[i] <- w |
||
| 93 | if (n == 1 & v > 1) { |
||
| 94 | check <- check + 1 |
||
| 95 | } |
||
| 96 | } |
||
| 97 | } |
||
| 98 | }else { |
||
| 99 | if ( ((paste(l[i, v:w], collapse = ' ')) == (paste(row1[n:k], collapse = ' '))) ){ |
||
| 100 | if (v > d[i,s-1]) { |
||
| 101 | if (w < listo[i]) { |
||
| 102 | p <- p + 1 |
||
| 103 | listo[i] <- w |
||
| 104 | } |
||
| 105 | } |
||
| 106 | } |
||
| 107 | } |
||
| 108 | } |
||
| 109 | } |
||
| 110 | } |
||
| 111 | } |
||
| 112 | } |
||
| 113 | } |
||
| 114 | if (p >= nrow(b)) { |
||
| 115 | results[g] <- paste(row1[n:k], collapse = ' ') |
||
| 116 | if (k == length(row1)) { |
||
| 117 | n <- k |
||
| 118 | } |
||
| 119 | if (k < length(row1)) { |
||
| 120 | k <- k + 1 |
||
| 121 | } |
||
| 122 | d[,s] <- listo |
||
| 123 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 124 | ignorecheck <- ignorecheck + 1 |
||
| 125 | switch01 <- 1 |
||
| 126 | } |
||
| 127 | if ((p < nrow(b))) { |
||
| 128 | if (n < length(row1)) { |
||
| 129 | if (switch01 == 0) { |
||
| 130 | n <- k + 1 |
||
| 131 | k <- n |
||
| 132 | } |
||
| 133 | if (switch01 == 1) { |
||
| 134 | n <- k |
||
| 135 | } |
||
| 136 | g <- g + 2 |
||
| 137 | if (d[1,s] != .Machine$double.xmax) { |
||
| 138 | s <- s + 1 |
||
| 139 | } |
||
| 140 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 141 | switch01 <- 0 |
||
| 142 | } |
||
| 143 | } |
||
| 144 | if (n == length(row1)) { |
||
| 145 | n <- n + 1 |
||
| 146 | k <- k + 1 |
||
| 147 | } |
||
| 148 | } |
||
| 149 | } |
||
| 150 | if (ignorecheck > 0) { |
||
| 151 | results <- unlist(strsplit(results, " ")) |
||
| 152 | |||
| 153 | results <- as.data.frame((results)) |
||
| 154 | results <- cbind(results, rep("f", times = nrow(results))) |
||
| 155 | colnames(results)[ncol(results)] <- "tf" |
||
| 156 | results[] <- lapply(results, as.character) |
||
| 157 | results <- results[complete.cases(results),] |
||
| 158 | |||
| 159 | for (i in 2:nrow(results)) { |
||
| 160 | if (results[i,1] == "IGNORE") { |
||
| 161 | if (results[i,1] == results[(i-1),1]) { |
||
| 162 | results[i,2] <- "t" |
||
| 163 | } |
||
| 164 | } |
||
| 165 | } |
||
| 166 | |||
| 167 | igcheck <- 0 |
||
| 168 | results <- subset(results, tf == "f") |
||
| 169 | results <- as.data.frame(results[,-(ncol(results))]) |
||
| 170 | colnames(results)[1] <- "Template" |
||
| 171 | results <- as.character(results[,1]) |
||
| 172 | if (length(results) < ncol(l)) { |
||
| 173 | for (i in (length(results)+1):(ncol(l))) { |
||
| 174 | results[i] <- "IGNORE" |
||
| 175 | igcheck <- 1 |
||
| 176 | } |
||
| 177 | } |
||
| 178 | |||
| 179 | if (check > 0) { |
||
| 180 | results <- c("IGNORE", results) |
||
| 181 | } |
||
| 182 | |||
| 183 | seq <- as.data.frame(t(as.data.frame(results))) |
||
| 184 | seq <- seq[,(1:(ncol(l)))] |
||
| 185 | seq [] <- lapply(seq, as.character) |
||
| 186 | l[] <- lapply(l, as.character) |
||
| 187 | |||
| 188 | p <- list(as.character(print(seq[1,]))) |
||
| 189 | p <- as.data.frame(p) |
||
| 190 | colnames(p)[1] <- "col1" |
||
| 191 | p <- cbind(p, rep("f", times =nrow(p))) |
||
| 192 | colnames(p)[2] <- "tf" |
||
| 193 | p[,2] <- sapply(p[,2], as.character) |
||
| 194 | for (i in 2:nrow(p)) { |
||
| 195 | if ((trimws(p[i,1], which = "both")) == "IGNORE") { |
||
| 196 | if (p[i,1] == p[(i-1),1]) |
||
| 197 | p[i,2] <- "t" |
||
| 198 | } |
||
| 199 | } |
||
| 200 | |||
| 201 | |||
| 202 | p <- subset(p, tf == "f") |
||
| 203 | p <- as.character(print(p[,1])) |
||
| 204 | p <- paste(p,collapse=" ") |
||
| 205 | w <- strsplit(p, "IGNORE") |
||
| 206 | w <- as.data.frame(w[[1]]) |
||
| 207 | colnames(w)[1] <- "col1" |
||
| 208 | w <- subset(w, col1 != " ") |
||
| 209 | w <- subset(w, col1 != "") |
||
| 210 | w <- as.character(w[,1]) |
||
| 211 | w <- gsub(" ","", w) |
||
| 212 | |||
| 213 | |||
| 214 | non2 <- as.data.frame(matrix(0, ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 215 | d2 <- as.data.frame(matrix("f", ncol = length(w), nrow = nrow(l)), stringsAsFactors = F) |
||
| 216 | |||
| 217 | non2[] <- lapply(non2, as.numeric) |
||
| 218 | |||
| 219 | g <- 1 |
||
| 220 | |||
| 221 | for (j in 1:nrow(l)) { |
||
| 222 | if (j > 0) { |
||
| 223 | for (i in 1:length(w)) { |
||
| 224 | if (i > 0) { |
||
| 225 | for (k in 1:ncol(l)) { |
||
| 226 | for (m in 1:ncol(l)) { |
||
| 227 | if ( (gsub(" ","",paste(l[j,k:m],collapse=" ")) == w[i]) & (d2[j,i] == "f") ){ |
||
| 228 | d2[j,i] <- "t" |
||
| 229 | if (g == 1) { |
||
| 230 | non2[j, g] <- k |
||
| 231 | non2[j, g+1] <- m |
||
| 232 | g <- g + 2 |
||
| 233 | }else if ((k > as.numeric(non2[j, (g-1)])) & (m > as.numeric(non2[j, (g-1)]))) { |
||
| 234 | non2[j, g] <- k |
||
| 235 | non2[j, g+1] <- m |
||
| 236 | g <- g + 2 |
||
| 237 | } |
||
| 238 | } |
||
| 239 | } |
||
| 240 | } |
||
| 241 | } |
||
| 242 | } |
||
| 243 | } |
||
| 244 | g <- 1 |
||
| 245 | } |
||
| 246 | |||
| 247 | non3 <- as.data.frame(matrix("IGNORE", ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 248 | |||
| 249 | g <- rep(1, times = nrow(non2)) |
||
| 250 | f <- 1 |
||
| 251 | |||
| 252 | |||
| 253 | for (m in 1:nrow(non2)) { |
||
| 254 | q <- as.character(print(non2[m,])) |
||
| 255 | q <- paste(q,collapse=" ") |
||
| 256 | q <- unlist(strsplit(q, " ")) |
||
| 257 | x <- 1 |
||
| 258 | while (x <= length(q)) { |
||
| 259 | if (q[x] == "0") { |
||
| 260 | q <- q[-x] |
||
| 261 | }else { |
||
| 262 | x <- x + 1 |
||
| 263 | } |
||
| 264 | } |
||
| 265 | q <- list(q) |
||
| 266 | q <- as.data.frame(q[[1]]) |
||
| 267 | colnames(q)[1] <- "col1" |
||
| 268 | q <- subset(q, col1 != "") |
||
| 269 | q[,1] <- sapply(q[,1], as.character) |
||
| 270 | q[,1] <- sapply(q[,1], as.numeric) |
||
| 271 | i <- 1 |
||
| 272 | while (i <= nrow(q)) { |
||
| 273 | if (i == 1) { |
||
| 274 | if (as.numeric(q[i,1]-1) >=1) { |
||
| 275 | non3[m,g[m]] <- paste(as.character(l[m,1:(as.numeric(q[i,1]-1))]), collapse = " ") |
||
| 276 | g[m] <- g[m] + 1 |
||
| 277 | }else { |
||
| 278 | g[m] <- 2 |
||
| 279 | } |
||
| 280 | } |
||
| 281 | if ((i %% 2 == 0) & (i != nrow(q)) & ((as.numeric(q[i+1,1]-1)) >= (as.numeric(q[i,1]+1)))) { |
||
| 282 | non3[m,g[m]] <- paste(as.character(l[m,(as.numeric(q[i,1]+1)):(as.numeric(q[i+1,1]-1))]), collapse = " ") |
||
| 283 | g[m] <- g[m] + 1 |
||
| 284 | } |
||
| 285 | if ((i %% 2 == 0) & (i != nrow(q)) & ((as.numeric(q[i+1,1]-1)) < (as.numeric(q[i,1]+1)))) { |
||
| 286 | g[m] <- g[m] + 1 |
||
| 287 | } |
||
| 288 | if ((i %% 2 == 0) & (i == nrow(q)) & (as.numeric(q[i,1]+1) <= as.numeric(ncol(l)))) { |
||
| 289 | non3[m,g[m]] <- paste(as.character(l[m,(as.numeric(q[i,1]+1)):(as.numeric(ncol(l)))]), collapse = " ") |
||
| 290 | g[m] <- g[m] + 1 |
||
| 291 | } |
||
| 292 | if (i < nrow(q)) { |
||
| 293 | i <- i + 1 |
||
| 294 | }else { |
||
| 295 | break |
||
| 296 | } |
||
| 297 | } |
||
| 298 | } |
||
| 299 | |||
| 300 | |||
| 301 | for (i in 1:ncol(seq)) { |
||
| 302 | if (seq[1,i] == "IGNORE") { |
||
| 303 | seq[1,i] <- "BLANK" |
||
| 304 | } |
||
| 305 | } |
||
| 306 | |||
| 307 | non3 <- as.data.frame(t(non3)) |
||
| 308 | non3 <- cbind(non3, rep("f", times = nrow(non3))) |
||
| 309 | colnames(non3)[ncol(non3)] <- "tf" |
||
| 310 | non3[] <- lapply(non3, as.character) |
||
| 311 | |||
| 312 | for (i in 1:nrow(non3)) { |
||
| 313 | if (paste(as.character(print(non3[i,1:(ncol(non3)-1)])), collapse = " ") == paste(as.character(rep("IGNORE", times = ncol(non3)-1)), collapse = " ")) { |
||
| 314 | non3[i,ncol(non3)] <- "t" |
||
| 315 | } |
||
| 316 | } |
||
| 317 | |||
| 318 | non3 <- subset(non3, tf == "f") |
||
| 319 | non3 <- non3[,-(ncol(non3))] |
||
| 320 | non3 <- as.data.frame(t(non3)) |
||
| 321 | |||
| 322 | seq <- as.data.frame(t(seq)) |
||
| 323 | seq <- cbind(seq, rep("f", times = nrow(seq))) |
||
| 324 | colnames(seq)[ncol(seq)] <- "tf" |
||
| 325 | seq[] <- lapply(seq, as.character) |
||
| 326 | |||
| 327 | for (i in 2:nrow(seq)) { |
||
| 328 | if (seq[i,1] == "BLANK") { |
||
| 329 | if (seq[i,1] == seq[(i-1),1]) { |
||
| 330 | seq[i,2] <- "t" |
||
| 331 | } |
||
| 332 | } |
||
| 333 | } |
||
| 334 | |||
| 335 | if (seq[nrow(seq),1] == "BLANK") { |
||
| 336 | if (igcheck == 1) { |
||
| 337 | seq[nrow(seq),2] <- "t" |
||
| 338 | } |
||
| 339 | } |
||
| 340 | |||
| 341 | seq <- subset(seq, tf == "f") |
||
| 342 | seq <- as.data.frame(seq[,-(ncol(seq))]) |
||
| 343 | |||
| 344 | |||
| 345 | |||
| 346 | |||
| 347 | colnames(seq)[1] <- "Template" |
||
| 348 | seq <- as.data.frame(t(seq)) |
||
| 349 | |||
| 350 | |||
| 351 | template <- seq |
||
| 352 | |||
| 353 | |||
| 354 | |||
| 355 | fillins <- non3 |
||
| 356 | |||
| 357 | if (ncol(non3) > 0) { |
||
| 358 | |||
| 359 | for (j in 1:ncol(fillins)) { |
||
| 360 | fillinstest <- as.character(fillins[,j]) |
||
| 361 | for (i in 1:length(fillinstest)) { |
||
| 362 | fillinstest[i] <- gsub("IGNORE", "", fillinstest[i]) |
||
| 363 | } |
||
| 364 | fillins[,j] <- fillinstest |
||
| 365 | } |
||
| 366 | |||
| 367 | |||
| 368 | |||
| 369 | i <- 1 |
||
| 370 | while (i <= ncol(fillins)) { |
||
| 371 | fillins <- as.data.frame(cbind (fillins[,c(1:i)], rep(1, times = nrow(fillins)), fillins[,-c(1:i)])) |
||
| 372 | i <- i+ 2 |
||
| 373 | } |
||
| 374 | |||
| 375 | for (i in 1:ncol(fillins)) { |
||
| 376 | if (i %% 2 == 0) { |
||
| 377 | colnames(fillins)[i] <- "Frequency" |
||
| 378 | fillins[,i] <- sapply(fillins[,i], as.character) |
||
| 379 | fillins[,i] <- sapply(fillins[,i], as.numeric) |
||
| 380 | } |
||
| 381 | } |
||
| 382 | for (i in 1:ncol(fillins)) { |
||
| 383 | if (i %% 2 == 1) { |
||
| 384 | colnames(fillins)[i] <- paste("Blank", (i+1)/2, collapse ="") |
||
| 385 | fillins[,i] <- sapply(fillins[,i], as.character) |
||
| 386 | } |
||
| 387 | } |
||
| 388 | for (j in 1:ncol(fillins)) { |
||
| 389 | for (i in 1:nrow(fillins)) { |
||
| 390 | if (j %% 2 == 1) { |
||
| 391 | for (k in 1:i) { |
||
| 392 | if (k < i) { |
||
| 393 | if (trimws(fillins[i, j], which = "both") == trimws(fillins[k,j], which = "both")) { |
||
| 394 | fillins[k, j+1] <- fillins[k, j+1] + 1 |
||
| 395 | fillins[i, j] <- paste("IGNORE", i, collapse ="") |
||
| 396 | fillins[i, j + 1] <- 0 |
||
| 397 | } |
||
| 398 | } |
||
| 399 | } |
||
| 400 | } |
||
| 401 | } |
||
| 402 | } |
||
| 403 | |||
| 404 | } |
||
| 405 | |||
| 406 | } else { |
||
| 407 | template <- fillins <- as.data.frame(matrix("", ncol = 1, nrow = 1)) |
||
| 408 | } |
||
| 409 | |||
| 410 | |||
| 411 | template <- cbindPad(template, b) |
||
| 412 | fillins <- cbindPad(fillins, b) |
||
| 413 | names(template)[ncol(template)] <- "Cluster Text" |
||
| 414 | names(fillins)[ncol(fillins)] <- "Cluster Text" |
||
| 415 | setwd("/media/hdd0/unraiddisk1/student/neilm/primestmp/clusters") |
||
| 416 | numbochar <- as.character(numbo) |
||
| 417 | numbotempstr <- paste("template", numbochar, ".csv", sep = "") |
||
| 418 | numbofillinsstr <- paste("fillins", numbochar, ".csv", sep = "") |
||
| 419 | write.csv(template, file = numbotempstr, row.names = F) |
||
| 420 | write.csv(fillins, file = numbofillinsstr, col.names = F) |
||
| 421 | |||
| 422 | } |
||
| 423 | numbo <- numbo + 1 |
||
| 424 | |||
| 425 | |||
| 426 | |||
| 427 | |||
| 428 | |||
| 429 | while (numbo <= (750)) { |
||
| 430 | |||
| 431 | if (numbo <= 750) { |
||
| 432 | str2 <- paste("clust_", numbo, ".txt", sep = "") |
||
| 433 | } |
||
| 434 | setwd("/media/hdd0/unraiddisk1/student/neilm/primestmp/clusters") |
||
| 435 | |||
| 436 | res <- try(read.table(str2),silent = TRUE) |
||
| 437 | |||
| 438 | if (!inherits(res, 'try-error')) { |
||
| 439 | b <- read.table(str2) |
||
| 440 | b <- b[,2, drop = F] |
||
| 441 | b[] <- lapply(b, as.character) |
||
| 442 | b[] <- lapply(b, tolower) |
||
| 443 | |||
| 444 | l <- as.data.frame(matrix(0, ncol = 1, nrow = (nrow(b)))) |
||
| 445 | |||
| 446 | |||
| 447 | for (i in 1:nrow(b)) { |
||
| 448 | if (length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))) > ncol(l)) { |
||
| 449 | l <- as.data.frame(matrix(0, ncol = length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))), nrow = (nrow(b)))) |
||
| 450 | } |
||
| 451 | |||
| 452 | } |
||
| 453 | n <- 1 |
||
| 454 | for (i in 1:nrow(b)) { |
||
| 455 | |||
| 456 | n <- as.numeric(as.numeric(ncol(l)) - as.numeric(length( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE))))) |
||
| 457 | l[i,] <- c( unlist(strsplit(unlist(strsplit (b[i,1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE)), rep("IGNORE", times = as.numeric(n))) |
||
| 458 | } |
||
| 459 | |||
| 460 | l <- cbind(l, rep(0, times = nrow(l))) |
||
| 461 | names(l)[ncol(l)] <- "wc" |
||
| 462 | for (i in 1:nrow(l)) { |
||
| 463 | l[i,ncol(l)] <- length(grep("IGNORE", l[i,])) |
||
| 464 | } |
||
| 465 | |||
| 466 | |||
| 467 | row1 <- unlist(strsplit(unlist(strsplit(b[which.max(l$wc),1], "[^[:alnum:]<>=./]")), "(?=[<>=./])", perl = TRUE)) |
||
| 468 | |||
| 469 | l <- l[1:(ncol(l)-1)] |
||
| 470 | d <- as.data.frame(matrix(.Machine$double.xmax, ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 471 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 472 | |||
| 473 | results <- rep('IGNORE', times = ncol(l)) |
||
| 474 | n <- 1 |
||
| 475 | k <- 1 |
||
| 476 | g <- 1 |
||
| 477 | p <- 0 |
||
| 478 | s <- 1 |
||
| 479 | check <- 0 |
||
| 480 | ignorecheck <- 0 |
||
| 481 | switch01 <- 0 |
||
| 482 | while (n <= length(row1)) { |
||
| 483 | while (k <= length(row1)) { |
||
| 484 | p <- 0 |
||
| 485 | if (n > 0) { |
||
| 486 | if (k > 0) { |
||
| 487 | for (i in 1:nrow(l)) { |
||
| 488 | if (i > 0) { |
||
| 489 | for (v in 1:ncol(l)) { |
||
| 490 | for (w in 1:ncol(l)) { |
||
| 491 | if (s == 1) { |
||
| 492 | if ( ((paste(l[i, v:w], collapse = ' ')) == (paste(row1[n:k], collapse = ' '))) ){ |
||
| 493 | if (w < listo[i]) { |
||
| 494 | p <- p + 1 |
||
| 495 | listo[i] <- w |
||
| 496 | if (n == 1 & v > 1) { |
||
| 497 | check <- check + 1 |
||
| 498 | } |
||
| 499 | } |
||
| 500 | } |
||
| 501 | }else { |
||
| 502 | if ( ((paste(l[i, v:w], collapse = ' ')) == (paste(row1[n:k], collapse = ' '))) ){ |
||
| 503 | if (v > d[i,s-1]) { |
||
| 504 | if (w < listo[i]) { |
||
| 505 | p <- p + 1 |
||
| 506 | listo[i] <- w |
||
| 507 | } |
||
| 508 | } |
||
| 509 | } |
||
| 510 | } |
||
| 511 | } |
||
| 512 | } |
||
| 513 | } |
||
| 514 | } |
||
| 515 | } |
||
| 516 | } |
||
| 517 | if (p >= nrow(b)) { |
||
| 518 | results[g] <- paste(row1[n:k], collapse = ' ') |
||
| 519 | if (k == length(row1)) { |
||
| 520 | n <- k |
||
| 521 | } |
||
| 522 | if (k < length(row1)) { |
||
| 523 | k <- k + 1 |
||
| 524 | } |
||
| 525 | d[,s] <- listo |
||
| 526 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 527 | ignorecheck <- ignorecheck + 1 |
||
| 528 | switch01 <- 1 |
||
| 529 | } |
||
| 530 | if ((p < nrow(b))) { |
||
| 531 | if (n < length(row1)) { |
||
| 532 | if (switch01 == 0) { |
||
| 533 | n <- k + 1 |
||
| 534 | k <- n |
||
| 535 | } |
||
| 536 | if (switch01 == 1) { |
||
| 537 | n <- k |
||
| 538 | } |
||
| 539 | g <- g + 2 |
||
| 540 | if (d[1,s] != .Machine$double.xmax) { |
||
| 541 | s <- s + 1 |
||
| 542 | } |
||
| 543 | listo <- rep(.Machine$double.xmax, times = nrow(l)) |
||
| 544 | switch01 <- 0 |
||
| 545 | } |
||
| 546 | } |
||
| 547 | if (n == length(row1)) { |
||
| 548 | n <- n + 1 |
||
| 549 | k <- k + 1 |
||
| 550 | } |
||
| 551 | } |
||
| 552 | } |
||
| 553 | if (ignorecheck > 0) { |
||
| 554 | results <- unlist(strsplit(results, " ")) |
||
| 555 | |||
| 556 | results <- as.data.frame((results)) |
||
| 557 | results <- cbind(results, rep("f", times = nrow(results))) |
||
| 558 | colnames(results)[ncol(results)] <- "tf" |
||
| 559 | results[] <- lapply(results, as.character) |
||
| 560 | results <- results[complete.cases(results),] |
||
| 561 | |||
| 562 | for (i in 2:nrow(results)) { |
||
| 563 | if (results[i,1] == "IGNORE") { |
||
| 564 | if (results[i,1] == results[(i-1),1]) { |
||
| 565 | results[i,2] <- "t" |
||
| 566 | } |
||
| 567 | } |
||
| 568 | } |
||
| 569 | |||
| 570 | igcheck <- 0 |
||
| 571 | results <- subset(results, tf == "f") |
||
| 572 | results <- as.data.frame(results[,-(ncol(results))]) |
||
| 573 | colnames(results)[1] <- "Template" |
||
| 574 | results <- as.character(results[,1]) |
||
| 575 | if (length(results) < ncol(l)) { |
||
| 576 | for (i in (length(results)+1):(ncol(l))) { |
||
| 577 | results[i] <- "IGNORE" |
||
| 578 | igcheck <- 1 |
||
| 579 | } |
||
| 580 | } |
||
| 581 | |||
| 582 | if (check > 0) { |
||
| 583 | results <- c("IGNORE", results) |
||
| 584 | } |
||
| 585 | |||
| 586 | seq <- as.data.frame(t(as.data.frame(results))) |
||
| 587 | seq <- seq[,(1:(ncol(l)))] |
||
| 588 | seq [] <- lapply(seq, as.character) |
||
| 589 | l[] <- lapply(l, as.character) |
||
| 590 | |||
| 591 | p <- list(as.character(print(seq[1,]))) |
||
| 592 | p <- as.data.frame(p) |
||
| 593 | colnames(p)[1] <- "col1" |
||
| 594 | p <- cbind(p, rep("f", times =nrow(p))) |
||
| 595 | colnames(p)[2] <- "tf" |
||
| 596 | p[,2] <- sapply(p[,2], as.character) |
||
| 597 | for (i in 2:nrow(p)) { |
||
| 598 | if ((trimws(p[i,1], which = "both")) == "IGNORE") { |
||
| 599 | if (p[i,1] == p[(i-1),1]) |
||
| 600 | p[i,2] <- "t" |
||
| 601 | } |
||
| 602 | } |
||
| 603 | |||
| 604 | |||
| 605 | p <- subset(p, tf == "f") |
||
| 606 | p <- as.character(print(p[,1])) |
||
| 607 | p <- paste(p,collapse=" ") |
||
| 608 | w <- strsplit(p, "IGNORE") |
||
| 609 | w <- as.data.frame(w[[1]]) |
||
| 610 | colnames(w)[1] <- "col1" |
||
| 611 | w <- subset(w, col1 != " ") |
||
| 612 | w <- subset(w, col1 != "") |
||
| 613 | w <- as.character(w[,1]) |
||
| 614 | w <- gsub(" ","", w) |
||
| 615 | |||
| 616 | |||
| 617 | non2 <- as.data.frame(matrix(0, ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 618 | d2 <- as.data.frame(matrix("f", ncol = length(w), nrow = nrow(l)), stringsAsFactors = F) |
||
| 619 | |||
| 620 | non2[] <- lapply(non2, as.numeric) |
||
| 621 | |||
| 622 | g <- 1 |
||
| 623 | |||
| 624 | for (j in 1:nrow(l)) { |
||
| 625 | if (j > 0) { |
||
| 626 | for (i in 1:length(w)) { |
||
| 627 | if (i > 0) { |
||
| 628 | for (k in 1:ncol(l)) { |
||
| 629 | for (m in 1:ncol(l)) { |
||
| 630 | if ( (gsub(" ","",paste(l[j,k:m],collapse=" ")) == w[i]) & (d2[j,i] == "f") ){ |
||
| 631 | d2[j,i] <- "t" |
||
| 632 | if (g == 1) { |
||
| 633 | non2[j, g] <- k |
||
| 634 | non2[j, g+1] <- m |
||
| 635 | g <- g + 2 |
||
| 636 | }else if ((k > as.numeric(non2[j, (g-1)])) & (m > as.numeric(non2[j, (g-1)]))) { |
||
| 637 | non2[j, g] <- k |
||
| 638 | non2[j, g+1] <- m |
||
| 639 | g <- g + 2 |
||
| 640 | } |
||
| 641 | } |
||
| 642 | } |
||
| 643 | } |
||
| 644 | } |
||
| 645 | } |
||
| 646 | } |
||
| 647 | g <- 1 |
||
| 648 | } |
||
| 649 | |||
| 650 | non3 <- as.data.frame(matrix("IGNORE", ncol = ncol(l), nrow = nrow(l)), stringsAsFactors = F) |
||
| 651 | |||
| 652 | g <- rep(1, times = nrow(non2)) |
||
| 653 | f <- 1 |
||
| 654 | |||
| 655 | |||
| 656 | for (m in 1:nrow(non2)) { |
||
| 657 | q <- as.character(print(non2[m,])) |
||
| 658 | q <- paste(q,collapse=" ") |
||
| 659 | q <- unlist(strsplit(q, " ")) |
||
| 660 | x <- 1 |
||
| 661 | while (x <= length(q)) { |
||
| 662 | if (q[x] == "0") { |
||
| 663 | q <- q[-x] |
||
| 664 | }else { |
||
| 665 | x <- x + 1 |
||
| 666 | } |
||
| 667 | } |
||
| 668 | q <- list(q) |
||
| 669 | q <- as.data.frame(q[[1]]) |
||
| 670 | colnames(q)[1] <- "col1" |
||
| 671 | q <- subset(q, col1 != "") |
||
| 672 | q[,1] <- sapply(q[,1], as.character) |
||
| 673 | q[,1] <- sapply(q[,1], as.numeric) |
||
| 674 | i <- 1 |
||
| 675 | while (i <= nrow(q)) { |
||
| 676 | if (i == 1) { |
||
| 677 | if (as.numeric(q[i,1]-1) >=1) { |
||
| 678 | non3[m,g[m]] <- paste(as.character(l[m,1:(as.numeric(q[i,1]-1))]), collapse = " ") |
||
| 679 | g[m] <- g[m] + 1 |
||
| 680 | }else { |
||
| 681 | g[m] <- 2 |
||
| 682 | } |
||
| 683 | } |
||
| 684 | if ((i %% 2 == 0) & (i != nrow(q)) & ((as.numeric(q[i+1,1]-1)) >= (as.numeric(q[i,1]+1)))) { |
||
| 685 | non3[m,g[m]] <- paste(as.character(l[m,(as.numeric(q[i,1]+1)):(as.numeric(q[i+1,1]-1))]), collapse = " ") |
||
| 686 | g[m] <- g[m] + 1 |
||
| 687 | } |
||
| 688 | if ((i %% 2 == 0) & (i != nrow(q)) & ((as.numeric(q[i+1,1]-1)) < (as.numeric(q[i,1]+1)))) { |
||
| 689 | g[m] <- g[m] + 1 |
||
| 690 | } |
||
| 691 | if ((i %% 2 == 0) & (i == nrow(q)) & (as.numeric(q[i,1]+1) <= as.numeric(ncol(l)))) { |
||
| 692 | non3[m,g[m]] <- paste(as.character(l[m,(as.numeric(q[i,1]+1)):(as.numeric(ncol(l)))]), collapse = " ") |
||
| 693 | g[m] <- g[m] + 1 |
||
| 694 | } |
||
| 695 | if (i < nrow(q)) { |
||
| 696 | i <- i + 1 |
||
| 697 | }else { |
||
| 698 | break |
||
| 699 | } |
||
| 700 | } |
||
| 701 | } |
||
| 702 | |||
| 703 | |||
| 704 | for (i in 1:ncol(seq)) { |
||
| 705 | if (seq[1,i] == "IGNORE") { |
||
| 706 | seq[1,i] <- "BLANK" |
||
| 707 | } |
||
| 708 | } |
||
| 709 | |||
| 710 | non3 <- as.data.frame(t(non3)) |
||
| 711 | non3 <- cbind(non3, rep("f", times = nrow(non3))) |
||
| 712 | colnames(non3)[ncol(non3)] <- "tf" |
||
| 713 | non3[] <- lapply(non3, as.character) |
||
| 714 | |||
| 715 | for (i in 1:nrow(non3)) { |
||
| 716 | if (paste(as.character(print(non3[i,1:(ncol(non3)-1)])), collapse = " ") == paste(as.character(rep("IGNORE", times = ncol(non3)-1)), collapse = " ")) { |
||
| 717 | non3[i,ncol(non3)] <- "t" |
||
| 718 | } |
||
| 719 | } |
||
| 720 | |||
| 721 | non3 <- subset(non3, tf == "f") |
||
| 722 | non3 <- non3[,-(ncol(non3))] |
||
| 723 | non3 <- as.data.frame(t(non3)) |
||
| 724 | |||
| 725 | seq <- as.data.frame(t(seq)) |
||
| 726 | seq <- cbind(seq, rep("f", times = nrow(seq))) |
||
| 727 | colnames(seq)[ncol(seq)] <- "tf" |
||
| 728 | seq[] <- lapply(seq, as.character) |
||
| 729 | |||
| 730 | for (i in 2:nrow(seq)) { |
||
| 731 | if (seq[i,1] == "BLANK") { |
||
| 732 | if (seq[i,1] == seq[(i-1),1]) { |
||
| 733 | seq[i,2] <- "t" |
||
| 734 | } |
||
| 735 | } |
||
| 736 | } |
||
| 737 | |||
| 738 | if (seq[nrow(seq),1] == "BLANK") { |
||
| 739 | if (igcheck == 1) { |
||
| 740 | seq[nrow(seq),2] <- "t" |
||
| 741 | } |
||
| 742 | } |
||
| 743 | |||
| 744 | seq <- subset(seq, tf == "f") |
||
| 745 | seq <- as.data.frame(seq[,-(ncol(seq))]) |
||
| 746 | |||
| 747 | |||
| 748 | |||
| 749 | |||
| 750 | colnames(seq)[1] <- "Template" |
||
| 751 | seq <- as.data.frame(t(seq)) |
||
| 752 | |||
| 753 | |||
| 754 | template <- seq |
||
| 755 | |||
| 756 | |||
| 757 | |||
| 758 | fillins <- non3 |
||
| 759 | |||
| 760 | if (ncol(non3) > 0) { |
||
| 761 | |||
| 762 | for (j in 1:ncol(fillins)) { |
||
| 763 | fillinstest <- as.character(fillins[,j]) |
||
| 764 | for (i in 1:length(fillinstest)) { |
||
| 765 | fillinstest[i] <- gsub("IGNORE", "", fillinstest[i]) |
||
| 766 | } |
||
| 767 | fillins[,j] <- fillinstest |
||
| 768 | } |
||
| 769 | |||
| 770 | |||
| 771 | |||
| 772 | i <- 1 |
||
| 773 | while (i <= ncol(fillins)) { |
||
| 774 | fillins <- as.data.frame(cbind (fillins[,c(1:i)], rep(1, times = nrow(fillins)), fillins[,-c(1:i)])) |
||
| 775 | i <- i+ 2 |
||
| 776 | } |
||
| 777 | |||
| 778 | for (i in 1:ncol(fillins)) { |
||
| 779 | if (i %% 2 == 0) { |
||
| 780 | colnames(fillins)[i] <- "Frequency" |
||
| 781 | fillins[,i] <- sapply(fillins[,i], as.character) |
||
| 782 | fillins[,i] <- sapply(fillins[,i], as.numeric) |
||
| 783 | } |
||
| 784 | } |
||
| 785 | for (i in 1:ncol(fillins)) { |
||
| 786 | if (i %% 2 == 1) { |
||
| 787 | colnames(fillins)[i] <- paste("Blank", (i+1)/2, collapse ="") |
||
| 788 | fillins[,i] <- sapply(fillins[,i], as.character) |
||
| 789 | } |
||
| 790 | } |
||
| 791 | for (j in 1:ncol(fillins)) { |
||
| 792 | for (i in 1:nrow(fillins)) { |
||
| 793 | if (j %% 2 == 1) { |
||
| 794 | for (k in 1:i) { |
||
| 795 | if (k < i) { |
||
| 796 | if (trimws(fillins[i, j], which = "both") == trimws(fillins[k,j], which = "both")) { |
||
| 797 | fillins[k, j+1] <- fillins[k, j+1] + 1 |
||
| 798 | fillins[i, j] <- paste("IGNORE", i, collapse ="") |
||
| 799 | fillins[i, j + 1] <- 0 |
||
| 800 | } |
||
| 801 | } |
||
| 802 | } |
||
| 803 | } |
||
| 804 | } |
||
| 805 | } |
||
| 806 | |||
| 807 | } |
||
| 808 | |||
| 809 | } else { |
||
| 810 | template <- fillins <- as.data.frame(matrix("", ncol = 1, nrow = 1)) |
||
| 811 | } |
||
| 812 | |||
| 813 | |||
| 814 | template <- cbindPad(template, b) |
||
| 815 | fillins <- cbindPad(fillins, b) |
||
| 816 | names(template)[ncol(template)] <- "Cluster Text" |
||
| 817 | names(fillins)[ncol(fillins)] <- "Cluster Text" |
||
| 818 | setwd("C:/Users/AI/Downloads/PipelineTestRun4PLAY") |
||
| 819 | numbochar <- as.character(numbo) |
||
| 820 | numbotempstr <- paste("template", numbochar, ".csv", sep = "") |
||
| 821 | numbofillinsstr <- paste("fillins", numbochar, ".csv", sep = "") |
||
| 822 | write.csv(template, file = numbotempstr, row.names = F) |
||
| 823 | write.csv(fillins, file = numbofillinsstr, col.names = F) |
||
| 824 | |||
| 825 | } |
||
| 826 | numbo <- numbo + 1 |
||
| 827 | |||
| 828 | } |