Opasnet Base Connection for R: Difference between revisions

From Opasnet
Jump to navigation Jump to search
 
(75 intermediate revisions by 3 users not shown)
Line 4: Line 4:
[[Category:SQL tool]]
[[Category:SQL tool]]
[[Category:Open assessment]]
[[Category:Open assessment]]
[[Heande:Opasnet Base Connection for R]]
[[Category:Code under inspection]]
{{tool|moderator=Teemu R}}
{{tool|moderator=Teemu R}}
Code for R for the purpose of interacting with the [[Opasnet Base]] is collected on this page. To use it, copy paste the code to R; this defines the functions, after which they can be called for in R.  
{{comment|# |Should we merge this page with [[OpasnetBaseUtils]]?|--[[User:Jouni|Jouni]] 20:18, 28 December 2011 (EET)}}
 
Code for [[R]] for the purpose of interacting with the [[Opasnet Base]] is collected on this page. To use it, copy paste the code you need to the R console; this defines the functions, after which they can be called for in that R session. Or alternatively install the [[OpasnetBaseUtils]] package.  


==Package dependencies==
==Package dependencies==


These packages are required for most of the code to work. Load these first (copy-pasteable).
These packages are required for most of the code to work. To install: from the top bar menu Packages>Install. To load: copy-paste.  


library(utils)
  library(RODBC)
  library(RODBC)


==Setup==
==Downloading data==
 
Establishes the connection to the database. Copy paste first.


db <- odbcConnect("opasnet_base")
'''op_baseGetData


*Assuming "opasnet_base" is a correctly defined DSN.
<rcode name="op_baseGetData">
**Note that uploading requires writing permissions.
op_baseGetData <- function(dsn, ident, include = NULL, exclude = NULL, series_id = NULL, iterations = NULL, use.utf8 = TRUE, apply.utf8 = TRUE) {
if (use.utf8) db <- odbcConnect(dsn, DBMSencoding = "UTF-8") else db <- odbcConnect(dsn)
obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep=''))[1,1]
if (length(series_id) == 0) {series_id <- sqlQuery(db, paste('SELECT series_id FROM actobj WHERE obj_id = ', obj_id,
' ORDER BY series_id DESC LIMIT 1', sep = ''))[1,1]}
sliced <- FALSE
locations <- NULL
x <- 1
basequery <- paste('SELECT loccell.cell_id FROM actobj LEFT JOIN cell ON actobj.id = cell.actobj_id LEFT JOIN',
' loccell ON cell.id = loccell.cell_id WHERE actobj.obj_id = ', obj_id, ' AND actobj.series_id = ',
series_id, ' AND loccell.loc_id IN(', sep = '')
if (length(include) != 0) {
sliced <- TRUE
locations[x] <- paste("IN(", basequery, paste(include, collapse = ","), ")", sep = "")
x <- x + 1
}
if (length(exclude) != 0) {
sliced <- TRUE
locations[x] <- paste("NOT IN(", basequery, paste(exclude, collapse = ","), ")", sep = "")
}
if (sliced == FALSE) {
Data <- sqlQuery(db, paste('SELECT cell.id, res.obs, obj.ident AS ind, loc.location AS loc, res.result,',
' res.restext FROM actobj LEFT JOIN cell ON actobj.id = cell.actobj_id LEFT JOIN res ON cell.id =',
' res.cell_id LEFT JOIN loccell ON cell.id = loccell.cell_id LEFT JOIN loc ON loccell.loc_id',
' = loc.id LEFT JOIN obj ON loc.obj_id_i = obj.id WHERE actobj.obj_id = ', obj_id,
' AND actobj.series_id = ', series_id, if(length(iterations)==1){paste(" AND obs <= ", iterations,
sep = "")}, sep = '')) } else {
Data <- sqlQuery(db, paste('SELECT cell.id, res.obs, obj.ident AS ind, loc.location AS loc, res.result,',
' res.restext FROM actobj LEFT JOIN cell ON actobj.id = cell.actobj_id LEFT JOIN res ON cell.id =',
' res.cell_id LEFT JOIN loccell ON cell.id = loccell.cell_id LEFT JOIN loc ON loccell.loc_id',
' = loc.id LEFT JOIN obj ON loc.obj_id_i = obj.id WHERE actobj.obj_id = ', obj_id,
' AND actobj.series_id = ', series_id, if(length(iterations)==1){paste(" AND obs <= ", iterations,
sep = "")}, ' AND (cell.id ', paste(locations, collapse = ') AND cell.id '),
'))', sep = ''))
}
odbcClose(db)
Data <- Data[order(Data[,1], Data[,2], Data[,3]),]
nind <- length(levels(Data[,3]))
nres <- nrow(Data)/nind
dataframe <- Data[1:nres*nind, c(1,2)]
for (i in 1:nind) {
dataframe[,2 + i] <- factor(Data[1:nres*nind - (nind - i), 4])
levels(dataframe[,2 + i]) <- gsub(" *$", "",gsub("^ *", "", levels(dataframe[,2 + i])))
colnames(dataframe)[2 + i] <- as.character(Data[i, 3])
if(apply.utf8) Encoding(levels(dataframe[,2 + i])) <- "UTF-8"
}
dataframe[,1:2 + 2 + nind] <- Data[1:nres*nind, 5:6]
colnames(dataframe)[1:2 + 2 + nind] <- c("Result", "Result.Text")
if(apply.utf8) {if(is.factor(dataframe[,"Result.Text"])) {Encoding(levels(dataframe[,"Result.Text"])) <- "UTF-8"} else if(is.character(dataframe[,"Result.Text"])) {
Encoding(dataframe[,"Result.Text"]) <- "UTF-8"}}
rownames(dataframe) <- 1:nres
return(dataframe)
}
</rcode>


==Downloading data==
===Usage===


===Functions===
variable <- op_baseGetData("opasnet_base", "page identifier", include = vector_of_loc_ids, exclude = vector_of_loc_ids)


*Now simplified, a lot faster. Copy paste second.
*Assuming "opasnet_base" is a correctly defined DSN (Data Service Name; in Windows XP: Control Panel\Administrative tools\Data Sources (ODBC)).
*Include and exclude are optional.
**Include picks all cells in the locations given.
***The clearest case is when all the included locations belong to the same index: Any cells in the non-included locations of the index will be left out.
***In case given locations are in multiple indices: The effect produced will be the same as picking separately for each index and removing duplicates.
**Exclude unpicks any cells which are indexed by the locations given. Slower than include.
**They can be used in unison.
*Result will be in a table format with columns: id, obs, ind1, ind2 ... indn, Result.
*series_id is an optional parameter, if it is not given the most current upload of the data will be downloaded.


op_baseGetData <- function(db, ident) {
===Finding index data===
<nowiki>obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep=''))[1,1]</nowiki>
series_id <- sqlQuery(db, paste('SELECT series_id FROM actobj WHERE obj_id = ', obj_id,
<nowiki>' ORDER BY series_id DESC LIMIT 1'))[1,1]</nowiki>
Data <- sqlQuery(db, paste('SELECT cell.id, res.obs, obj.name AS ind, loc.location AS loc, res.result',
' FROM actobj LEFT JOIN cell ON actobj.id = cell.actobj_id LEFT JOIN res ON cell.id =',
' res.cell_id LEFT JOIN loccell ON cell.id = loccell.cell_id LEFT JOIN loc ON loccell.loc_id',
' = loc.id LEFT JOIN obj ON loc.obj_id_i = obj.id WHERE actobj.obj_id = ', obj_id,
<nowiki>' AND actobj.series_id = ', series_id, ' ORDER BY id, obs, ind', sep = ''))</nowiki>
nind <- length(levels(Data[,3]))
nres <- nrow(Data)/nind
dataframe <- Data[1:nres*nind, c(1,2)]
for (i in nind:1) {
dataframe[,2 + nind - i + 1] <- Data[1:nres*nind - i + 1, 4]
colnames(dataframe)[2 + nind - i + 1] <- as.character(Data[nind - i + 1, 3])
}
dataframe[,3 + nind] <- Data[1:nres*nind, 5]
colnames(dataframe)[3 + nind] <- "Result"
rownames(dataframe) <- 1:nres
dataframe
}


====Usage====
'''op_baseGetLocs


variable <- op_baseGetData(db, "page identifier")
<rcode name="op_baseGetLocs">
op_baseGetLocs <- function(dsn, ident, series_id = NULL, use.utf8 = TRUE, apply.utf8 = TRUE) {
if (use.utf8) db <- odbcConnect(dsn, DBMSencoding = "UTF-8") else db <- odbcConnect(dsn)
obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep=''))[1,1]
if (length(series_id) == 0) {series_id <- sqlQuery(db, paste('SELECT series_id FROM actobj WHERE obj_id = ', obj_id,
' ORDER BY series_id DESC LIMIT 1', sep = ''))[1,1]}
Locs <- sqlQuery(db, paste("SELECT DISTINCT obj.ident AS ind, loc.location AS loc, loc.id AS loc_id",
" FROM actobj LEFT JOIN actloc ON actobj.id = actloc.actobj_id LEFT JOIN loc ON actloc.loc_id = loc.id",
" LEFT JOIN obj ON loc.obj_id_i = obj.id WHERE actobj.obj_id = ", obj_id, " AND actobj.series_id = ",
series_id, sep = ""))
odbcClose(db)
Locs <- Locs[order(Locs$ind, Locs$loc_id),]
rownames(Locs) <- 1:nrow(Locs)
if(apply.utf8) {Encoding(levels(Locs$ind)) <- "UTF-8"; Encoding(levels(Locs$loc)) <- "UTF-8"}
return(Locs)
}
</rcode>


*Result will be in a table format with columns: id, obs, ind1, ind2 ... indn, Result.
*Returns all indexes and locations and their ids in a table of format: ind, loc, loc_id.


==Manipulating data==
==Manipulating data==


===Functions===
'''DataframeToArray


DataframeToArray <- function(dataframe) {
<rcode name="DataframeToArray">
DimNames <- rep(vector("list", 1), ncol(dataframe) - 2)
DataframeToArray <- function(dataframe, rescol = NULL) {
names(DimNames) <- c(colnames(dataframe)[2:(ncol(dataframe) - 1)])
ColNames <- colnames(dataframe[(colnames(dataframe) == "id") == FALSE])
nind <- length(names(DimNames))
if (length(ColNames[(ColNames == "obs")])>0) {if(length(levels(factor(dataframe[,"obs"]))) == 1) {ColNames <- ColNames[
indlengths <- 0
(ColNames == "obs") == FALSE]} else {dataframe[,"obs"] <- factor(as.character(dataframe[,"obs"]))}}
for (i in 1:nind) {
if (length(rescol)==0) {
<nowiki>DimNames[[i]] <- levels(factor(dataframe[,1 + i]))</nowiki>
rescol <- colnames(dataframe) == "Freq"
<nowiki>indlengths[i] <- length(DimNames[[i]])</nowiki>
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "Freq" else {
}
rescol <- colnames(dataframe) == "Result"
array <- rep(NA, prod(indlengths))
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "Result" else {
dim(array) <- indlengths
rescol <- colnames(dataframe) == "result"
dimnames(array) <- DimNames
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "result" else stop("No result column found")
array[as.matrix(dataframe[,c(1:nind + 1)])] <- dataframe[,ncol(dataframe)]
}
array
}
}
} else {ColNames <- ColNames[(ColNames == rescol) == FALSE]}
ColNames <- ColNames[(ColNames == "Result") == FALSE]
ColNames <- ColNames[(ColNames == "result") == FALSE]
ColNames <- ColNames[(ColNames == "Freq") == FALSE]
nind <- length(ColNames)
DimNames <- rep(vector("list", 1), nind)
names(DimNames) <- ColNames
indlengths <- 0
for (i in 1:nind) {
DimNames[[i]] <- levels(factor(dataframe[,ColNames[i]]))
indlengths[i] <- length(DimNames[[i]])
}
array <- rep(NA, prod(indlengths))
dim(array) <- indlengths
dimnames(array) <- DimNames
array[as.matrix(dataframe[,ColNames])] <- dataframe[,rescol]
return(array)
}
</rcode>


====Usage====
===Usage===


  variable2 <- DataframeToArray(variable1)
  variable2 <- DataframeToArray(variable1, rescol = NULL)


*variable1 must be in similar format as the result when downloading.
*variable1 must be in similar format as the result when downloading.
**first column (id) is ignored
*Columns named "id" and various versions of "Result" are ignored for dimension creation.
*variable1 <- function(variable1) is allowed in R
**"obs" column will also be ignored if there's only one.
*The column containing the values may be defined in the parameters, otherwise it is assumed to be either "Freq", "Result" or "result" in that order.


===Other useful stuff===
==Uploading data==
 
as.data.frame(as.table(array))
dataframe[is.na(dataframe[,ncol(dataframe)])==FALSE,]
 
*First line returns an array from a data frame.
*Second line returns rows without empty values.
 
dataframe[grep("location", dataframe$index, ignore.case = TRUE),]


*Returns all rows of a data frame where column "index" value is "location".
'''op_baseWrite


dataframe[order(dataframe[,"col1"],dataframe[,"col2"], ... ,dataframe[,"coln"]),]
<rcode name="op_baseWrite">
op_baseWrite <- function(dsn, input, ident = NULL, name = NULL, unit = NULL, objtype_id = NULL, who = NULL, acttype = NULL,
rescol = NULL, n.obs.const = FALSE, maxrows = 50000, use.utf8 = TRUE, use.utf8.read = TRUE, latin1.2.utf8.conv.write = TRUE, utf8.2.latin1.conv.read = TRUE) {
# Coerce input into a data frame if it isn't one already; get rid of empty cells
if (is.array(input)) dataframe <- as.data.frame(as.table(input)) else dataframe <- input
if (is.null(rescol)) {
rescol <- colnames(dataframe) == "Freq"
if (sum(rescol) == 1) rescol <- "Freq" else {
rescol <- colnames(dataframe) == "Result"
if (sum(rescol) == 1) rescol <- "Result" else {
rescol <- colnames(dataframe) == "result"
if (sum(rescol) == 1) rescol <- "result"
}
}}
dataframe <- dataframe[is.na(dataframe[,rescol]) == FALSE,]
ColNames <- colnames(dataframe)[!(colnames(dataframe)%in%c(rescol, "id", "obs"))]
for (i in ColNames) {
dataframe[,i] <- factor(dataframe[,i])
levels(dataframe[,i]) <- gsub(" *$", "",gsub("^ *", "", levels(dataframe[,i])))
if(latin1.2.utf8.conv.write) if(sum(Encoding(levels(dataframe[,i]))=="latin1")!=0) levels(dataframe[,i]) <- iconv(levels(dataframe[,i]), "latin1", "UTF-8")
}
#if(!is.numeric(dataframe[,rescol]))
# Open database connection
if(use.utf8) db <- odbcConnect(dsn, DBMSencoding = "UTF-8") else db <- odbcConnect(dsn)
if(!use.utf8.read) db2 <- odbcConnect(dsn)
# Add page to database (if it doesn't already exist)
if (is.null(ident)) if (interactive()) ident <- readline(paste("What is the identifier of this object?",
"\n", sep = "")) else stop("indentifier of object no given")
obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep = ''))[1,1]
if (is.na(obj_id)) {
# Wiki id
if (substr(ident, 1,5)=="Op_en") {wiki_id <- 1; page <- substr(ident, 6, nchar(ident))} else {
if (substr(ident, 1,5)=="Op_fi") {wiki_id <- 2; page <- substr(ident, 6, nchar(ident))} else {
if (substr(ident, 1,6)=="Heande") {wiki_id <- 6; page <- substr(ident, 7, nchar(ident))} else {
if (substr(ident, 1,4)=="Erac") {wiki_id <- 6; page <- substr(ident, 5, nchar(ident))} else {
wiki_id <- 0; page <- 0; warning("No wiki id found in ident, writing zero.")}}}}
page <- as.numeric(page)
if (is.na(page)) stop("could not convert characters following the wiki ident into a page number")
# Name etc.
if (is.null(name)) if (interactive()) name <- readline(paste("What is the name of this object?",
"\n", sep = "")) else stop("object name not given")
if (is.null(objtype_id)) if (interactive()) objtype_id <- readline(paste("What type of object is",
" this (id)?", paste(paste(sqlQuery(db, "SELECT id FROM objtype")[,1], sqlQuery(db, paste("SELECT objtype",
" FROM objtype", sep = ""))[,1], sep = " - "), collapse = ", "), "\n", collapse = " ")) else {
stop("object type not given")}
sqlQuery(db, paste('INSERT INTO obj (ident, name, objtype_id, page, wiki_id) VALUES ("', paste(ident,
name, sep = '","'), '",', paste(objtype_id, page, wiki_id, sep = ','), ')', sep = ''))
obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep = ''))[1,1]
}
# Write act and actobj
if (is.null(who)==TRUE) if (interactive()) {who <- readline(paste("What is the name of the uploader?", "\n", sep = ""))
} else stop("uploader name not given")
series_id <- sqlQuery(db, paste("SELECT series_id FROM actobj WHERE obj_id = ", obj_id, " ORDER BY series_id DESC LIMIT 1",
sep = ""))[1,1]
if (is.na(series_id)==FALSE) {if (is.null(acttype)==TRUE) {if (interactive()) {acttype <- readline(paste("What type of upload",
" is this? 4 - new data to replace any existing, 5 - new data to be appended to existing data (must have the same",
" indices).", "\n", sep = ""))
} else acttype <- 4
}} else acttype <- 4
if (!(acttype%in%c(4,5))) stop ("proper acttype not given")
sqlQuery(db, paste('INSERT INTO act (acttype_id, who, comments) VALUES (', acttype, ',"', who, '","R upload")', sep = ''))
act_id <- sqlQuery(db, paste('SELECT id FROM act WHERE who = "', who,'" AND comments = "R upload" ORDER BY id DESC LIMIT 1',
sep = ''))[1,1]
if (acttype == 4) series_id <- act_id
if (is.null(unit)) if (interactive()) unit <- readline(paste("What is the unit of this object?",
"\n", sep = "")) else stop("unit not given")
sqlQuery(db, paste('INSERT INTO actobj (act_id, obj_id, series_id, unit) VALUES (', paste(act_id, obj_id, series_id, sep = ','),
',"', unit, '")', sep = ''))
actobj_id <- sqlQuery(db, paste('SELECT id FROM actobj WHERE act_id = ', act_id, sep = ''))[1,1]
#Write indexes
for (i in ColNames) {
sqlQuery(db, paste('INSERT IGNORE INTO obj (ident, name, objtype_id) VALUES ("', gsub(' ', '_', i), '","',
i, '", 6)', sep = ''))
}
IndIds <- sqlQuery((if(use.utf8.read) db else db2), paste('SELECT id, ident FROM obj WHERE ident IN("', paste(gsub(" ", "_", ColNames),
collapse = '","'), '")', sep = ''))
if(utf8.2.latin1.conv.read) levels(IndIds$ident) <- iconv(levels(IndIds$ident), "UTF-8", "latin1")
IndIdMap <- IndIds$id
names(IndIdMap) <- tolower(IndIds$ident)
ColIds <- as.character(IndIdMap[tolower(gsub(" ", "_", ColNames))])
colnames(dataframe)[colnames(dataframe)%in%ColNames] <- ColIds
#Write locations
for (i in ColIds) {
for (j in levels(dataframe[, i])) {
sqlQuery(db, paste('INSERT IGNORE INTO loc (obj_id_i, location) VALUES (', i, ',"', j, '")',
sep = ''))
}
}
LocIds <- sqlQuery((if(use.utf8.read) db else db2), paste('SELECT id, obj_id_i, location FROM loc WHERE obj_id_i IN("', paste(ColIds, collapse = '","'),
'")', sep = ''))
if(utf8.2.latin1.conv.read) levels(LocIds$location) <- iconv(levels(LocIds$location), "UTF-8", "latin1")
for (i in ColIds) {
LocIdMap <- LocIds[LocIds$obj_id_i == i, 1]
names(LocIdMap) <- gsub(" *$", "",gsub("^ *", "", tolower(LocIds[LocIds$obj_id_i == i, 3])))
levels(dataframe[, i]) <- LocIdMap[tolower(levels(dataframe[, i]))]
if (sum(is.na(levels(dataframe[, i]))) != 0) stop("Faulty location matching. Usually caused by special characters.")
#Writing actloc
sqlQuery(db, paste("INSERT INTO actloc (actobj_id, loc_id) VALUES (", paste(actobj_id, levels(dataframe[, i]),
sep = ",", collapse = "),("), ")", sep = ""))
}
#Writing cell
n <- tapply(dataframe[,rescol], dataframe[,ColIds], length)
ncell <- sum(!is.na(n))
if (is.numeric(dataframe[,rescol])) means <- tapply(dataframe[,rescol], dataframe[,ColIds], mean) else means <- rep(0, ncell)
if (is.numeric(dataframe[,rescol])) {
sds <- tapply(dataframe[,rescol], dataframe[,ColIds], sd); sds[] <- ifelse(n == 1, 0, sds)} else sds <- rep(0, ncell)
cellQuery <- paste(actobj_id, means[!is.na(means)], sds[!is.na(sds)], n[!is.na(n)], sep = ",")
i <- 1
while (length(cellQuery) >= (i + maxrows - 1)) {
sqlQuery(db, paste('INSERT INTO cell (actobj_id, mean, sd, n) VALUES (', paste(cellQuery[i:(i + maxrows - 1)],
collapse = '),('), ')', sep = ''))
i <- i + maxrows
}
if (length(cellQuery) %% maxrows != 0) {
sqlQuery(db, paste('INSERT INTO cell (actobj_id, mean, sd, n) VALUES (',
paste(cellQuery[i:length(cellQuery)], collapse = '),('), ')', sep = ''))
}
#Writing res
cell_id <- sqlQuery(db, paste("SELECT id FROM cell WHERE actobj_id = ", actobj_id, " ORDER BY ID", sep = ""))[,1]
if (length(cell_id) != ncell) stop("number of written cells differs from given data")
if (is.numeric(dataframe[,rescol])) ids <- means else ids <- n
ids[!is.na(ids)] <- cell_id
dataframe[, ncol(dataframe) + 1] <- ids[as.matrix(dataframe[,ColIds])]
colnames(dataframe)[ncol(dataframe)] <- "cell_id"
resQuery <- paste(dataframe[,"cell_id"], ',', if(sum(colnames(dataframe) == "obs") == 0) 1 else dataframe[,"obs"], ',',
if (!is.numeric(dataframe[,rescol])) '"', dataframe[,rescol], if (!is.numeric(dataframe[,rescol])) '"', sep = "")
i <- 1
while (length(resQuery) >= (i + maxrows - 1)) {
sqlQuery(db, paste('INSERT INTO res (cell_id, obs, ', ifelse(is.numeric(dataframe[,rescol]), "result", "restext"), ') VALUES (',
paste(resQuery[i:(i + maxrows - 1)], collapse = '),('), ')', sep = ''))
i <- i + maxrows
}
if (length(resQuery) %% maxrows != 0) {
sqlQuery(db, paste('INSERT INTO res (cell_id, obs, ', ifelse(is.numeric(dataframe[,rescol]), "result", "restext"), ') VALUES (',
paste(resQuery[i:length(resQuery)], collapse = '),('), ')', sep = ''))
}
#Writing loccell
ids <- as.data.frame(as.table(ids))
ids <- ids[!is.na(ids$Freq),]
loccellQuery <- paste(ids$Freq, unlist(ids[,-ncol(ids)]), sep = ",")
i <- 1
while (length(loccellQuery) >= (i + maxrows - 1)) {
sqlQuery(db, paste('INSERT INTO loccell (cell_id, loc_id) VALUES (', paste(loccellQuery[i:(i + maxrows - 1)], collapse = '),('), ')',
sep = ''))
i <- i + maxrows
}
if (length(loccellQuery) %% maxrows != 0) {
sqlQuery(db, paste('INSERT INTO loccell (cell_id, loc_id) VALUES (', paste(loccellQuery[i:length(loccellQuery)], collapse = '),('), ')',
sep = ''))
}
#Close database connection
odbcClose(db)
cat("Successful\n")
return(character())
}
</rcode>


*Returns dataframe ordered by col1, col2, ... , coln.
===Usage===


  read.csv("table.csv", sep = ",")
  op_baseWrite(dsn, input, ident = NULL, name = NULL, unit = NULL, objtype_id = NULL, who = NULL, acttype = NULL, rescol = NULL)


*Returns a data.frame from a .csv file, sep is the separator used in the file.
*dsn and input must be defined, the rest of the object and act parameters if not defined are prompted for by the function as needed.
 
**For uploading the DSN defined must have writers permissions.
==Uploading data==
*rescol defines the column from which the values are chosen from, both numerical and textual data are allowed, if left undefined the function will check column matches for "Freq", "Result" and "result" in that order.


*Tested and works.
====Restrictions====
**Very efficient.
*Probabilistic data supported, albeit with following weaknesses:
**Varying numbers of iterations between cells not supported.
**Probabilistic data must also follow a certain sorting order (indices first, obs last).
*Uploading indices and locations relies on updating a dummy column when a duplicate is encountered.
**Those dummy columns are "newest" and "roww" for indices and locations respectively, which are obsolete anyway.
**Using a dummy column is probably faster and simpler than some other workaround.


===Functions===
*Input may only be given in either array or data.frame form.
**Indexes used may not exceed the character limit of 20.
***Indexes should preferably match an earlier entry: [[Special:OpasnetBaseIndices]].
***Indexes are treated as identifiers for indexes in the database, spaces in the indexes are converted to _. This ensures maximum compatibility and ease in operations in which data is downloaded and uploaded again. Names and more specific details can be edited into the indexes separately.


op_baseWrite <- function(db, input, ident = NULL, name = NULL, unit = NULL, objtype_id = NULL, who = NULL, acttype = NULL) {
== Regarding special characters and character encoding ==
#Coerce input into a data frame if it isn't one already; get rid of empty cells
if (is.array(input)) dataframe <- as.data.frame(as.table(input)) else dataframe <- input
rescol <- colnames(dataframe) == "Freq"
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "Freq" else {
rescol <- colnames(dataframe) == "Result"
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "Result" else {
rescol <- colnames(dataframe) == "result"
if (length(rescol[rescol==TRUE]) >= 1) rescol <- "result"
}
}
values <- dataframe[,rescol]
cond <- (is.na(values) == FALSE)
dataframe <- dataframe[cond,]
#Add page to database (if it doesn't already exist)
if (is.null(ident)==TRUE) if (interactive()) ident <- readline(paste("What is the identifier of this object?",
"\n", sep = "")) else stop("indentifier of object no given")
wiki_name <- substring(ident, 1, 5)
if (wiki_name=="Op_en") {wiki_id <- 1; page <- substring(ident, 6, nchar(ident))} else if (wiki_name=="Op_fi") {
wiki_id <- 2; page <- substring(ident, 6, nchar(ident))} else {wiki_id <- 0; page <- 0}
<nowiki>obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep = ''))[1,1]</nowiki>
if (is.na(obj_id)==TRUE) {
if (is.null(name)==TRUE) if (interactive()) name <- readline(paste("What is the name of this object?",
"\n", sep = "")) else stop("object name not given")
if (is.null(unit)==TRUE) if (interactive()) unit <- readline(paste("What is the unit of this object?",
"\n", sep = "")) else stop("unit not given")
if (is.null(objtype_id)==TRUE) if (interactive()) objtype_id <- readline(paste("What type of object is",
" this (id)?", paste(paste(sqlQuery(db, "SELECT id FROM objtype")[,1], sqlQuery(db, "SELECT objtype",
" FROM objtype")[,1], sep = " - "), collapse = ", "), "\n", collapse = " ")) else {
stop("object type not given")}
sqlQuery(db, paste('INSERT INTO obj (ident, name, unit, objtype_id, page, wiki_id) VALUES ("', paste(ident,
<nowiki>name, unit, sep = '","'), '",', paste(objtype_id, page, wiki_id, sep = ','), ')', sep = ''))</nowiki>
<nowiki>obj_id <- sqlQuery(db, paste('SELECT id FROM obj WHERE ident = "', ident, '"', sep = ''))[1,1]</nowiki>
}
#Write act and actobj
if (is.null(who)==TRUE) if (interactive()) {who <- readline(paste("What is the name of the uploader?", "\n", sep = ""))
} else stop("uploader name not given")
series_id <- sqlQuery(db, paste("SELECT series_id FROM actobj WHERE obj_id = ", obj_id, " ORDER BY series_id DESC LIMIT 1",
sep = ""))[1,1]
if (is.na(series_id)==FALSE) {if (is.null(acttype)==TRUE) if (interactive()) {acttype <- readline(paste("What type of upload",
" is this? 4 - new data to replace any existing, 5 - new data to be appended to existing data (must have the same",
" indices).", "\n", sep = ""))
} else acctype <- 4
} else acctype <- 4
if (acttype != 4 & acttype != 5) stop ("proper acttype not given")
<nowiki>sqlQuery(db, paste('INSERT INTO act (acttype_id, who, comments) VALUES (', acttype, ',"', who, '","R upload")', sep = ''))</nowiki>
act_id <- sqlQuery(db, "SELECT id FROM act ORDER BY id DESC LIMIT 1")[1,1]
if (acttype == 4) series_id <- act_id
sqlQuery(db, paste('INSERT INTO actobj (act_id, obj_id, series_id) VALUES (', paste(act_id, obj_id, series_id, sep = ','),
<nowiki>')', sep = ''))</nowiki>
<nowiki>actobj_id <- sqlQuery(db, paste('SELECT id FROM actobj WHERE act_id = ', act_id, sep = ''))[1,1]</nowiki>
#Write indexes
ColNames <- colnames(dataframe[(colnames(dataframe) == "id") == FALSE])
ColNames <- ColNames[(ColNames == "obs") == FALSE]
ColNames <- ColNames[(ColNames == "Result") == FALSE]
ColNames <- ColNames[(ColNames == "result") == FALSE]
ColNames <- ColNames[(ColNames == "Freq") == FALSE]
nind <- length(ColNames)
DimNames <- rep(vector("list", 1), nind)
names(DimNames) <- ColNames
indlengths <- 0
for (i in 1:nind) {
<nowiki>DimNames[[i]] <- levels(factor(dataframe[, ColNames[i]]))</nowiki>
<nowiki>indlengths[i] <- length(DimNames[[i]])</nowiki>
}
indQuery <- NA
indQuery[1:nind*5 - 4] <- gsub(" ", "_", ColNames)
indQuery[1:nind*5 - 3] <- '","'
indQuery[1:nind*5 - 2] <- ColNames
indQuery[1:nind*5 - 1] <- '",6'
indQuery[1:(nind - 1)*5] <- '),("'
<nowiki>sqlQuery(db, paste('INSERT INTO obj (ident, name, objtype_id) VALUES ("', paste(indQuery, collapse = ''), ')', </nowiki>
<nowiki>' ON DUPLICATE KEY UPDATE newest = 0', sep = ''))</nowiki>
IndIds <- sqlQuery(db, paste('SELECT id, ident FROM obj WHERE ident IN("', paste(gsub(" ", "_", ColNames),
<nowiki>collapse = '","'), '")', sep = ''))</nowiki>
DimIds <- DimNames
DimN <- 1:nind
names(DimN) <- gsub(" ", "_", ColNames)
for (i in 1:nrow(IndIds)) {
names(DimIds)[DimN[IndIds[i,2]]] <- IndIds[grep(gsub(" ", "_", ColNames[i]), IndIds$ident), 1]
}
#Write locations
y <- 1
LocNames <- as.data.frame(matrix(rep(NA, 2*sum(indlengths)), sum(indlengths), 2))
for (i in 1:nind) {
LocNames[y:(y + indlengths[i] - 1),1:2] <- matrix(c(rep(ColNames[i], indlengths[i]),
levels(factor(dataframe[,ColNames[i]]))), indlengths[i], 2)
y <- y + indlengths[i]
}
LocNames[,1] <- names(DimIds)[DimN[LocNames[,1]]]
nloc <- nrow(LocNames)
locQuery <- NA
locQuery[1:nloc*4 - 3] <- LocNames[,1]
locQuery[1:nloc*4 - 2] <- ',"'
locQuery[1:nloc*4 - 1] <- LocNames[,2]
locQuery[1:(nloc - 1)*4] <- '"),('
<nowiki>sqlQuery(db, paste('INSERT INTO loc (obj_id_i, location) VALUES (', paste(locQuery, collapse = ''), '")', </nowiki>
<nowiki>' ON DUPLICATE KEY UPDATE roww = 0', sep = ''))</nowiki>
LocIds <- sqlQuery(db, paste('SELECT id, obj_id_i, location FROM loc WHERE location IN("', paste(LocNames[,2],
<nowiki>collapse = '","'), '") AND obj_id_i IN(', paste(names(DimIds), collapse = ','), ')', sep = ''))</nowiki>
LocMap <- NA
y <- 1
for (i in 1:nind) {
LocMap <- LocIds[grep(names(DimIds)[i], LocIds$obj_id_i), 1]
names(LocMap) <- LocIds[grep(names(DimIds)[i], LocIds$obj_id_i), 3]
<nowiki>DimIds[[i]] <- LocMap[DimNames[[i]]]</nowiki>
}
#Writing actloc
sqlQuery(db, paste("INSERT INTO actloc (actobj_id, loc_id) VALUES (", paste(actobj_id, LocIds, sep = ",",
collapse = "),("), ")", sep = ""))
#Changing location names in table into ids
for (i in 1:nind) {
dataframe[,ColNames[i]] <- factor(dataframe[,ColNames[i]])
<nowiki>levels(dataframe[,ColNames[i]]) <- DimIds[[i]]</nowiki>
}
#A hidden parameter for adjusting query packet sizes, the higher the faster, though crash becomes likelier
maxrows <- 50000
#Writing cell
obscol <- colnames(dataframe) == "obs"
if (length(obscol[obscol==TRUE]) >= 1) {
obscol <- TRUE
n <- length(levels(factor(dataframe[,"obs"])))} else {
obscol <- FALSE
n <- 1
}
ncell <- nrow(dataframe)/n
cellQuery <- NA
cellQuery[1:ncell*6-5] <- actobj_id
cellQuery[1:ncell*6-4] <- ","
if (n == 1) {
cellQuery[1:ncell*6-3] <- dataframe[,rescol]} else {
cellQuery[1:ncell*6-3] <- apply(matrix(dataframe[,rescol], n, ncell), 2, mean)
}
cellQuery[1:ncell*6-2] <- ","
cellQuery[1:ncell*6-1] <- n
cellQuery[1:(ncell-1)*6] <- "),("
if (ncell >= maxrows) {
for (i in 1:(ncell%/%maxrows)) {
sqlQuery(db, paste('INSERT INTO cell (actobj_id, mean, n) VALUES (', paste(cellQuery[
<nowiki>((i-1)*6*maxrows+1):(i*6*maxrows-1)], collapse = ''), ')', sep = ''))</nowiki>
}
}
if (ncell%%maxrows != 0) sqlQuery(db, paste('INSERT INTO cell (actobj_id, mean, n) VALUES (', paste(cellQuery[(ncell%/%
<nowiki>maxrows*6*maxrows+1):length(cellQuery)], collapse = ''), ')', sep = ''))</nowiki>
#Writing res
cell_id <- sqlQuery(db, paste("SELECT id FROM cell WHERE actobj_id = ", actobj_id, " ORDER BY ID", sep = ""))[,1]
if (ncell != length(cell_id)) stop("Number of cells written does not match with expected value")
resQuery <- NA
resQuery[1:nrow(dataframe)*6-5] <- rep(cell_id, each = n)
resQuery[1:nrow(dataframe)*6-4] <- ","
if (n==1) resQuery[1:nrow(dataframe)*6-3] <- "0" else resQuery[1:nrow(dataframe)*6-3] <- dataframe[,"obs"]
resQuery[1:nrow(dataframe)*6-2] <- ","
resQuery[1:nrow(dataframe)*6-1] <- dataframe[,rescol]
resQuery[1:(nrow(dataframe)-1)*6] <- "),("
if (nrow(dataframe) >= maxrows) {
for (i in 1:(nrow(dataframe)%/%maxrows)) {
sqlQuery(db, paste('INSERT INTO res (cell_id, obs, result) VALUES (', paste(resQuery[((i-1)
<nowiki>*4*maxrows+1):(i*4*maxrows-1)], collapse = ''), ')', sep = ''))</nowiki>
}
}
if (nrow(dataframe)%%maxrows != 0) sqlQuery(db, paste('INSERT INTO res (cell_id, obs, result) VALUES (', paste(resQuery[
<nowiki>(nrow(dataframe)%/%maxrows*4*maxrows+1):length(resQuery)], collapse = ''), ')', sep = ''))</nowiki>
#Writing loccell
locidmatrix <- rep(NA, nind*ncell)
dim(locidmatrix) <- c(nind,ncell)
for (i in 1:nind) {
locidmatrix[i,] <- as.character(dataframe[1:ncell*n - n + 1,ColNames[i]])
}
loccellQuery <- rep(NA, 4*nind*ncell)
dim(loccellQuery) <- c(4*nind, ncell)
loccellQuery[1:nind*4-3, 1:ncell] <- rep(cell_id, each = nind)
loccellQuery[1:nind*4-2, 1:ncell] <- ","
loccellQuery[1:nind*4-1, 1:ncell] <- locidmatrix
loccellQuery[1:nind*4, 1:ncell] <- "),("
loccellQuery[nind*4, ncell] <- ""
if (ncell >= maxrows) {
for (i in 1:(ncell%/%maxrows)) {
loccellQuery[nind*4, i*maxrows] <- ""
sqlQuery(db, paste('INSERT INTO loccell (cell_id, loc_id) VALUES (', paste(loccellQuery[,
<nowiki>((i-1)*maxrows+1):(i*maxrows)], collapse = ''), ')', sep = ''))</nowiki>
}
}
if (ncell%%maxrows != 0) sqlQuery(db, paste('INSERT INTO loccell (cell_id, loc_id) VALUES (', paste(loccellQuery[,
<nowiki>(ncell%/%maxrows*maxrows+1):ncell], collapse = ''), ')', sep = ''))</nowiki>
}


====Usage====
Using special characters like ä and ö when our database is encoded in latin1 while wiki is in UTF-8 is a bit complicated. New parameters -- for forcing the odbc connection to use UTF-8 and in-R conversion of one encoding to the other prior to writing or after reading -- have been created. The defaults have been made to work with opasnet_base. For use with heande_base, forcing UTF-8 on the odbc connection when reading should be disabled, meaning that ''use.utf8'' should be set to ''FALSE'' when using ''op_baseGetData'' and ''use.utf8.read'' should be set to ''FALSE'' when using ''op_baseWrite''.


op_baseWrite(db, input, ident = NULL, name = NULL, unit = NULL, objtype_id = NULL, who = NULL, acttype = NULL)
== See also ==


*Db and input must be defined, the rest if not defined is prompted for by the function as needed.
{{Opasnet Base}}


=====Restrictions=====
*[[A Tutorial on R]]
*Input may be given in either array or data.frame form.
**Input in data.frame form must contain a "Freq", "Result" or "result" column, this is where the numerical value is read.
**Probabilistic input must contain a dimension (in case of array) or a column (in case of data.frame) named "obs". Also: all cells must have values for each "obs" value. And the sorting order must be: indices first, obs last; for arrays this means that the first index must be "obs".

Latest revision as of 20:29, 10 April 2015


----#: . Should we merge this page with OpasnetBaseUtils? --Jouni 20:18, 28 December 2011 (EET) (type: truth; paradigms: science: comment)

Code for R for the purpose of interacting with the Opasnet Base is collected on this page. To use it, copy paste the code you need to the R console; this defines the functions, after which they can be called for in that R session. Or alternatively install the OpasnetBaseUtils package.

Package dependencies

These packages are required for most of the code to work. To install: from the top bar menu Packages>Install. To load: copy-paste. 

library(RODBC)

Downloading data

op_baseGetData

+ Show code

Usage

variable <- op_baseGetData("opasnet_base", "page identifier", include = vector_of_loc_ids, exclude = vector_of_loc_ids)
  • Assuming "opasnet_base" is a correctly defined DSN (Data Service Name; in Windows XP: Control Panel\Administrative tools\Data Sources (ODBC)).
  • Include and exclude are optional.
    • Include picks all cells in the locations given.
      • The clearest case is when all the included locations belong to the same index: Any cells in the non-included locations of the index will be left out.
      • In case given locations are in multiple indices: The effect produced will be the same as picking separately for each index and removing duplicates.
    • Exclude unpicks any cells which are indexed by the locations given. Slower than include.
    • They can be used in unison.
  • Result will be in a table format with columns: id, obs, ind1, ind2 ... indn, Result.
  • series_id is an optional parameter, if it is not given the most current upload of the data will be downloaded.

Finding index data

op_baseGetLocs

+ Show code

  • Returns all indexes and locations and their ids in a table of format: ind, loc, loc_id.

Manipulating data

DataframeToArray

+ Show code

Usage

variable2 <- DataframeToArray(variable1, rescol = NULL)
  • variable1 must be in similar format as the result when downloading.
  • Columns named "id" and various versions of "Result" are ignored for dimension creation.
    • "obs" column will also be ignored if there's only one.
  • The column containing the values may be defined in the parameters, otherwise it is assumed to be either "Freq", "Result" or "result" in that order.

Uploading data

op_baseWrite

+ Show code

Usage

op_baseWrite(dsn, input, ident = NULL, name = NULL, unit = NULL, objtype_id = NULL, who = NULL, acttype = NULL, rescol = NULL)
  • dsn and input must be defined, the rest of the object and act parameters if not defined are prompted for by the function as needed.
    • For uploading the DSN defined must have writers permissions.
  • rescol defines the column from which the values are chosen from, both numerical and textual data are allowed, if left undefined the function will check column matches for "Freq", "Result" and "result" in that order.

Restrictions

  • Input may only be given in either array or data.frame form.
    • Indexes used may not exceed the character limit of 20.
      • Indexes should preferably match an earlier entry: Special:OpasnetBaseIndices.
      • Indexes are treated as identifiers for indexes in the database, spaces in the indexes are converted to _. This ensures maximum compatibility and ease in operations in which data is downloaded and uploaded again. Names and more specific details can be edited into the indexes separately.

Regarding special characters and character encoding

Using special characters like ä and ö when our database is encoded in latin1 while wiki is in UTF-8 is a bit complicated. New parameters -- for forcing the odbc connection to use UTF-8 and in-R conversion of one encoding to the other prior to writing or after reading -- have been created. The defaults have been made to work with opasnet_base. For use with heande_base, forcing UTF-8 on the odbc connection when reading should be disabled, meaning that use.utf8 should be set to FALSE when using op_baseGetData and use.utf8.read should be set to FALSE when using op_baseWrite.

See also

Pages related to Opasnet Base

Opasnet Base · Uploading to Opasnet Base · Data structures in Opasnet · Opasnet Base UI · Modelling in Opasnet · Special:Opasnet Base Import · Opasnet Base Connection for R (needs updating) · Converting KOPRA data into Opasnet Base · Poll · Working with sensitive data · Saved R objects

Pages related to the 2008-2011 version of Opasnet Base

Opasnet base connection for Analytica · Opasnet base structure · Related Analytica file (old version File:Transferring to result database.ANA) · Analytica Web Player · Removed pages and other links · Standard run · OpasnetBaseUtils

Retrieved from "https://en.opasnet.org/index.php?title=Opasnet_Base_Connection_for_R&oldid=35364"