From 0a1b8781978926a9a3a8edcaa91dfcfa774bef40 Mon Sep 17 00:00:00 2001
From: Beaudette <dylan.beaudette@usda.gov>
Date: Tue, 20 Feb 2024 09:00:13 -0800
Subject: [PATCH] updates

---
 MIR/art-from-MIR-spectral-DB.R       |  4 ++--
 MIR/build-compact-spectral-library.R |  6 +++---
 MIR/eval-errors.R                    | 14 ++++++++++++++
 MIR/main.R                           | 11 +++++++----
 4 files changed, 26 insertions(+), 9 deletions(-)
 create mode 100644 MIR/eval-errors.R

diff --git a/MIR/art-from-MIR-spectral-DB.R b/MIR/art-from-MIR-spectral-DB.R
index 286ee78..9d3fa73 100644
--- a/MIR/art-from-MIR-spectral-DB.R
+++ b/MIR/art-from-MIR-spectral-DB.R
@@ -41,13 +41,13 @@ parseSpectra <- function(.txt, compressed = TRUE) {
 base.path <- 'E:/MIR'
 
 # compressed spectra
-# db.file <- file.path(base.path, 'MIR-compact-gz.sqlite')
+db.file <- file.path(base.path, 'MIR-compact-compressed.sqlite')
 
 # plain-text spectra
 # db.file <- file.path(base.path, 'MIR-compact-text.sqlite')
 
 # full
-db.file <- file.path(base.path, 'MIR-compact.sqlite')
+# db.file <- file.path(base.path, 'MIR-compact.sqlite')
 
 
 db <- dbConnect(RSQLite::SQLite(), db.file)
diff --git a/MIR/build-compact-spectral-library.R b/MIR/build-compact-spectral-library.R
index 4267e0e..37d9c4c 100644
--- a/MIR/build-compact-spectral-library.R
+++ b/MIR/build-compact-spectral-library.R
@@ -17,7 +17,7 @@ library(purrr)
 ## flag for embedding compressed spectra
 # useful for stand-alone SQLite versions, not for SDA
 # requires a little more work when reading data out of the SQLite DB
-compressSpec <- TRUE
+compressSpec <- FALSE
 
 # functions waiting for an R package
 source('../code/snapshot-preparation/snapshot-functions.R')
@@ -169,8 +169,8 @@ dbDisconnect(db)
 
 
 ## SQLite file stats
-# uncompressed / txt:     9.5GB
-# gzipped / txt:          4.2GB
+# uncompressed / txt:     5.5GB
+# gzipped / txt:          GB
 
 # uncompressed / gz(txt): 2.6GB
 # gzipped / gz(txt):      GB
diff --git a/MIR/eval-errors.R b/MIR/eval-errors.R
new file mode 100644
index 0000000..43ecc61
--- /dev/null
+++ b/MIR/eval-errors.R
@@ -0,0 +1,14 @@
+library(opusreader2)
+library(purrr)
+
+# all collections
+x <- readRDS('errors.rds')
+
+# find just the collections with errors
+idx <- which(! sapply(x, is.null))
+x <- x[idx]
+
+
+## 2024-02-14 errors:
+## C2019USNJ085
+
diff --git a/MIR/main.R b/MIR/main.R
index b49f2ab..f8c4d91 100644
--- a/MIR/main.R
+++ b/MIR/main.R
@@ -10,6 +10,9 @@ setwd('MIR')
 # save to temporary folder for later use
 source('pre-process-collections.R')
 
+# evaluate errors
+# eval-errors.R
+
 # create collection/sample/integer wn-sequence table
 # create wn-sequence metadata table
 # flag samples with odd wn-sequence
@@ -20,8 +23,8 @@ source('build-compact-spectral-library.R')
 
 
 ## TODO:
-## 1. create spectra -- sample LUT
-## 2. index
-## 3. simplify code if possible
-## 4. wrapper / helper functions
+# 1. create spectra -- sample LUT
+# 2. index
+# 3. simplify code if possible
+# 4. wrapper / helper functions