Ecosystem/misc-research
1
0
Fork 0
mirror of https://github.com/Rucknium/misc-research.git synced 2024-12-22 11:29:22 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Piecewise processing for BTC to avoid integer overflow

Browse source
This commit is contained in:
Rucknium 2022-08-09 22:14:59 +00:00
parent c0af6a5862
commit 6445948fe8
1 changed files with 87 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

104
General-Blockchain-Age-of-Spent-Outputs/create-dataset-for-analysis.R
View file

@ -5,30 +5,99 @@ library(DBI)
# NOTE: Also need lubridate package installed, but not loading it due to # NOTE: Also need lubridate package installed, but not loading it due to
# it masking functions # it masking functions
is.btc <- FALSE
# Change to TRUE if processing BTC
data.dir <- "" data.dir <- ""
# Input data directory here, with trailing "/" # Input data directory here, with trailing "/"
con <- DBI::dbConnect(RSQLite::SQLite(), paste0(data.dir, "tx-graph-node-indices.db")) con <- DBI::dbConnect(RSQLite::SQLite(), paste0(data.dir, "tx-graph-node-indices.db"))
master.edgelist <- DBI::dbGetQuery(con,
"SELECT origin_index, destination_index,block_height,value FROM edgelist_intermediate_2")
master.edgelist.output.created <- master.edgelist[ if ( ! is.btc) {
(! is.na(master.edgelist$value)) & master.edgelist$value > 0 , c("destination_index", "block_height")]
colnames(master.edgelist.output.created) <- c("output_index", "output.created.block_height") master.edgelist.output.created <- master.edgelist[
setDT(master.edgelist.output.created) (! is.na(master.edgelist$value)) & master.edgelist$value > 0 , c("destination_index", "block_height")]
colnames(master.edgelist.output.created) <- c("output_index", "output.created.block_height")
master.edgelist.output.spent <- master.edgelist[, c("origin_index", "block_height")] setDT(master.edgelist.output.created)
colnames(master.edgelist.output.spent) <- c("output_index", "output.spent.block_height")
setDT(master.edgelist.output.spent) master.edgelist.output.spent <- master.edgelist[, c("origin_index", "block_height")]
colnames(master.edgelist.output.spent) <- c("output_index", "output.spent.block_height")
# Only include positive _value_s for output created, since that's the value of the created output setDT(master.edgelist.output.spent)
# Then, below do an "inner merge" to get the proper outputs on the spending side
# Only include positive _value_s for output created, since that's the value of the created output
rm(master.edgelist) # Then, below do an "inner merge" to get the proper outputs on the spending side
master.edgelist.output.spent <- merge(master.edgelist.output.created, master.edgelist.output.spent)
rm(master.edgelist.output.created) rm(master.edgelist)
master.edgelist.output.spent <- merge(master.edgelist.output.created, master.edgelist.output.spent)
rm(master.edgelist.output.created)
} else {
# Split data requests and processing if BTC to avoid integer overflow issue in this step of processing.
# data.table issue:
# https://github.com/Rdatatable/data.table/issues/3957
# SQLite issue: "negative length vectors are not allowed"
master.edgelist.lt.500k <- DBI::dbGetQuery(con,
"SELECT origin_index, destination_index,block_height,value FROM edgelist_intermediate_2 WHERE block_height < 500000")
master.edgelist.output.created.lt.500k <- master.edgelist.lt.500k[
(! is.na(master.edgelist.lt.500k$value)) & master.edgelist.lt.500k$value > 0 , c("destination_index", "block_height")]
colnames(master.edgelist.output.created.lt.500k) <- c("output_index", "output.created.block_height")
setDT(master.edgelist.output.created.lt.500k)
master.edgelist.output.spent.lt.500k <- master.edgelist.lt.500k[, c("origin_index", "block_height")]
colnames(master.edgelist.output.spent.lt.500k) <- c("output_index", "output.spent.block_height")
setDT(master.edgelist.output.spent.lt.500k)
rm(master.edgelist.lt.500k)
master.edgelist.gt.500k.lt.700k <- DBI::dbGetQuery(con,
"SELECT origin_index, destination_index,block_height,value FROM edgelist_intermediate_2 WHERE block_height > 499999 AND block_height < 700000")
master.edgelist.output.created.gt.500k.lt.700k <- master.edgelist.gt.500k.lt.700k[
(! is.na(master.edgelist.gt.500k.lt.700k$value) & master.edgelist.gt.500k.lt.700k$value > 0 ) , c("destination_index", "block_height")]
colnames(master.edgelist.output.created.gt.500k.lt.700k) <- c("output_index", "output.created.block_height")
setDT(master.edgelist.output.created.gt.500k.lt.700k)
master.edgelist.output.spent.gt.500k.lt.700k <- master.edgelist.gt.500k.lt.700k[, c("origin_index", "block_height")]
colnames(master.edgelist.output.spent.gt.500k.lt.700k) <- c("output_index", "output.spent.block_height")
setDT(master.edgelist.output.spent.gt.500k.lt.700k)
rm(master.edgelist.gt.500k.lt.700k)
master.edgelist.gt.700k <- DBI::dbGetQuery(con,
"SELECT origin_index, destination_index,block_height,value FROM edgelist_intermediate_2 WHERE block_height > 699999")
master.edgelist.output.created.gt.700k <- master.edgelist.gt.700k[
(! is.na(master.edgelist.gt.700k$value) & master.edgelist.gt.700k$value > 0 ), c("destination_index", "block_height")]
colnames(master.edgelist.output.created.gt.700k) <- c("output_index", "output.created.block_height")
setDT(master.edgelist.output.created.gt.700k)
master.edgelist.output.spent.gt.700k <- master.edgelist.gt.700k[, c("origin_index", "block_height")]
colnames(master.edgelist.output.spent.gt.700k) <- c("output_index", "output.spent.block_height")
setDT(master.edgelist.output.spent.gt.700k)
rm(master.edgelist.gt.700k)
master.edgelist.output.created <- rbindlist(list(master.edgelist.output.created.lt.500k,
master.edgelist.output.created.gt.500k.lt.700k, master.edgelist.output.created.gt.700k))
master.edgelist.output.spent.lt.500k <- merge(master.edgelist.output.created, master.edgelist.output.spent.lt.500k)
master.edgelist.output.spent.gt.500k.lt.700k <- merge(master.edgelist.output.created, master.edgelist.output.spent.gt.500k.lt.700k)
master.edgelist.output.spent.gt.700k <- merge(master.edgelist.output.created, master.edgelist.output.spent.gt.700k)
master.edgelist.output.spent <- rbindlist(list(master.edgelist.output.spent.lt.500k,
master.edgelist.output.spent.gt.500k.lt.700k, master.edgelist.output.spent.gt.700k))
rm(master.edgelist.output.created)
}
@ -69,5 +138,6 @@ master.edgelist.output.spent <- merge(master.edgelist.output.spent,
saveRDS(master.edgelist.output.spent, paste0(data.dir, "master_edgelist_output_spent.rds")) saveRDS(master.edgelist.output.spent, paste0(data.dir, "master_edgelist_output_spent.rds"))
saveRDS(master.edgelist.output.spent, paste0(data.dir, "master_edgelist_output_spent-uncompressed.rds"), compress = FALSE)