From fa0460915de51890aae8d0353f1feae55b03d758 Mon Sep 17 00:00:00 2001
From: Frankie Robertson <frankie@robertson.name>
Date: Mon, 29 Aug 2022 17:53:34 +0300
Subject: [PATCH 1/3] Streamline adding a new dataset

 * Add instructions to README for adding a new dataset
 * Add scripts to update the dataset metadata
 * Add update_doc method to only add a single dataset
 * Add HTML documentation generation to update_doc
 * Change update_doc to correctly round trip quotes in the metadata CSV
---
 README.md                 |  17 +++
 scripts/update_doc_all.sh |   4 +
 scripts/update_doc_one.sh |   4 +
 src/update_doc.r          | 211 +++++++++++++++++++++++---------------
 4 files changed, 154 insertions(+), 82 deletions(-)
 create mode 100755 scripts/update_doc_all.sh
 create mode 100755 scripts/update_doc_one.sh

diff --git a/README.md b/README.md
index ce0927bd..2003f5b2 100644
--- a/README.md
+++ b/README.md
@@ -74,6 +74,23 @@ mlmRev|guImmun|Immunization in Guatemala|2159|13
 mlmRev|guPrenat|Prenatal care in Guatemala|2449|15
 mlmRev|star|Student Teacher Achievement Ratio (STAR) project data|26796|18
 
+# How to add datasets from a new package
+
+**Step 1: add the data from the package**
+
+ 1. In your clone of this repo `mkdir -p data/$PKG`
+ 2. Go to CRAN
+ 3. Download the *source package*
+ 4. Extract one or more of the datasets in the `data` directory into the new directory
+
+**Step 2: add the metadata**
+
+Run the script:
+
+     $ scripts/update_doc_one.sh $PKG
+
+Now it's ready for you to submit your pull request.
+
 # Licensing and Intellectual Property
 
 Following Vincent's lead, we have assumed that all of the data sets in this repository can be made available under the GPL-3 license. If you know that one of the datasets released here should not be released publicly or if you know that a data set can only be released under a different license, please contact me so that I can remove the data set from this repository.
diff --git a/scripts/update_doc_all.sh b/scripts/update_doc_all.sh
new file mode 100755
index 00000000..0d8ff1a1
--- /dev/null
+++ b/scripts/update_doc_all.sh
@@ -0,0 +1,4 @@
+R --no-save <<END
+source("src/update_doc.r")
+update_docs(".")
+END
diff --git a/scripts/update_doc_one.sh b/scripts/update_doc_one.sh
new file mode 100755
index 00000000..cbebb97a
--- /dev/null
+++ b/scripts/update_doc_one.sh
@@ -0,0 +1,4 @@
+R --no-save <<END
+source("src/update_doc.r")
+update_package_doc(".", "$1")
+END
diff --git a/src/update_doc.r b/src/update_doc.r
index 1ac023af..c1b3ef65 100644
--- a/src/update_doc.r
+++ b/src/update_doc.r
@@ -1,3 +1,129 @@
+install_packages <- function(packages) {
+    # Pick a decent mirror if none set
+    r <- getOption("repos")
+    if (r["CRAN"] == "@CRAN@") {
+        r <- "http://cran.rstudio.com/"
+    }
+    suppressWarnings({install.packages(packages, repos = r)})
+}
+
+install_packages(c("R2HTML"))
+library(R2HTML)
+
+write_doc = function(package, dataset) {
+    help.ref <- try(help(eval(dataset), package=eval(package)), silent = TRUE)
+    doc <- try(utils:::.getHelpFile(help.ref), silent = TRUE)
+    try(dir.create(paste0('doc/', package)), silent = TRUE)
+    fn_doc <- paste0('doc/', package, '/', dataset, '.html')
+    tools::Rd2HTML(doc, out = fn_doc)
+}
+
+do_package_update <- function(data_dir, package_df, old_dataset_df, dataset_df, mismatched_dims_df, package) {
+    suppressWarnings({library(package, character.only = TRUE)})
+
+    # Get package description
+    pdesc <- packageDescription(package)
+    new_row <- as.data.frame(pdesc[c("Package", "Title")],
+                             stringsAsFactors = FALSE)
+    package_df <- rbind(package_df, new_row)
+
+    pdat <- data(package=package)$results
+
+    datasets <- dir(path = file.path(data_dir, package))
+
+    # Trim filenames to dataset names
+    r <- "(.+)\\.(csv\\.gz|rda)$"
+    format_recognized <- grepl(r, datasets)
+    if (!(all(format_recognized))) {
+        stop("Unrecognized formats:\n",
+             cat(datasets[!format_recognized], sep = "\n"))
+    }
+    datanames <- sub(r, "\\1", datasets)
+
+    for (dataname in datanames) {
+        evaltext = paste0("data(", dataname, ", package=package)")
+        eval(parse(text = evaltext))
+        ds <- get(dataname)
+
+        write_doc(package, dataname)
+
+        # Get dataset description
+        title <- unique(pdat[, "Title"][pdat[, "Item"] == dataname])
+        if (length(title) != 1) {
+            stop(package, "/", dataname, " had ", length(title), " descriptions.")
+        }
+
+        # Old dims to fall back on
+        old_row = subset(old_dataset_df,
+                         Dataset == dataname & Package == package)
+        nr <- if (nrow(old_row)) old_row$Rows[[1]] else NA
+        nc <- if (nrow(old_row)) old_row$Columns[[1]] else NA
+
+        # Check against new dims when simple
+        new_nr <- NROW(ds)
+        new_nc <- NCOL(ds)
+        if (!(any(c("table", "ltraj") %in% class(ds))) &&
+            class(ds) != "list" &&
+            is.numeric(new_nr) &&
+            is.numeric(new_nc)) {
+
+            expected_cols <- c(nc, nc - 1) # row.names sometimes included
+            if (!is.numeric(nr) || !is.numeric(nc)) {
+                nr <- new_nr
+                nc <- new_nc
+            } else if (new_nr != nr || !(new_nc %in% expected_cols)) {
+
+                new_row <- data.frame(Package = package,
+                                      Dataset = dataname,
+                                      Class = class(ds),
+                                      OldRows = nr,
+                                      OldColumns = nc,
+                                      NewRows = new_nr,
+                                      NewColumns = new_nc)
+
+                mismatched_dims_df <- rbind(mismatched_dims_df, new_row)
+
+            }
+        }
+
+        new_row <- data.frame(Package = package,
+                              Dataset = dataname,
+                              Title = title,
+                              Rows = nr,
+                              Columns = nc,
+                              stringsAsFactors = FALSE)
+
+        dataset_df <- rbind(dataset_df, new_row)
+    }
+    return(list(package_df = package_df, dataset_df = dataset_df, mismatched_dims_df = mismatched_dims_df))
+}
+
+update_package_doc <- function(pkg_dir, package) {
+    data_dir <- file.path(pkg_dir, "data")
+    doc_dir <- file.path(pkg_dir, "doc")
+
+    package_fn <- file.path(doc_dir, "packages.csv")
+    dataset_fn <- file.path(doc_dir, "datasets.csv")
+
+    package_df <- read.csv(package_fn)
+    dataset_df <- read.csv(dataset_fn)
+
+    install_packages(c(package))
+
+    mismatched_dims_df <- data.frame()
+    dfs <- do_package_update(data_dir, package_df, dataset_df, dataset_df, mismatched_dims_df, package)
+    package_df <- dfs$package_df
+    dataset_df <- dfs$dataset_df
+    mismatched_dims_df <- dfs$mismatched_dims_df
+
+    package_df <- sort_upper_first(clean(package_df), c("Package"))
+    dataset_df <- sort_upper_first(clean(dataset_df), c("Package", "Dataset"))
+
+    write(package_df, package_fn)
+    write(dataset_df, dataset_fn)
+    return(mismatched_dims_df)
+}
+
 update_docs <- function(pkg_dir) {
     data_dir <- file.path(pkg_dir, "data")
     doc_dir <- file.path(pkg_dir, "doc")
@@ -16,90 +142,11 @@ update_docs <- function(pkg_dir) {
     # Install any missing packages
     new_packages <- packages[!(packages %in% installed.packages()[, "Package"])]
     if (length(new_packages)) {
-        # Pick a decent mirror if none set
-        r <- getOption("repos")
-        if (r["CRAN"] == "@CRAN@") {
-            r <- "http://cran.rstudio.com/"
-        }
-        suppressWarnings({install.packages(new_packages, repos = r)})
+        install_packages(new_packages)
     }
 
     for (package in packages) {
-        suppressWarnings({library(package, character.only = TRUE)})
-
-        # Get package description
-        pdesc <- packageDescription(package)
-        new_row <- as.data.frame(pdesc[c("Package", "Title")],
-                                 stringsAsFactors = FALSE)
-        package_df <- rbind(package_df, new_row)
-
-        pdat <- data(package=package)$results
-
-        datasets <- dir(path = file.path(data_dir, package))
-
-        # Trim filenames to dataset names
-        r <- "(.+)\\.(csv\\.gz|rda)$"
-        format_recognized <- grepl(r, datasets)
-        if (!(all(format_recognized))) {
-            stop("Unrecognized formats:\n",
-                 cat(datasets[!format_recognized], sep = "\n"))
-        }
-        datanames <- sub(r, "\\1", datasets)
-
-        for (dataname in datanames) {
-            eval(parse(text = paste0("data(", dataname, ", package=package)")))
-            ds <- get(dataname)
-
-            # TODO: Write rst and html doc per dataset
-
-            # Get dataset description
-            title <- unique(pdat[, "Title"][pdat[, "Item"] == dataname])
-            if (length(title) != 1) {
-                stop(package, "/", title, " had ", length(title), " descriptions.")
-            }
-
-            # Old dims to fall back on
-            old_row = subset(old_dataset_df,
-                             Dataset == dataname & Package == package)
-            nr <- if (nrow(old_row)) old_row$Rows[[1]] else NA
-            nc <- if (nrow(old_row)) old_row$Columns[[1]] else NA
-
-            # Check against new dims when simple
-            new_nr <- NROW(ds)
-            new_nc <- NCOL(ds)
-            if (!(any(c("table", "ltraj") %in% class(ds))) &&
-                class(ds) != "list" &&
-                is.numeric(new_nr) &&
-                is.numeric(new_nc)) {
-
-                expected_cols <- c(nc, nc - 1) # row.names sometimes included
-                if (!is.numeric(nr) || !is.numeric(nc)) {
-                    nr <- new_nr
-                    nc <- new_nc
-                } else if (new_nr != nr || !(new_nc %in% expected_cols)) {
-
-                    new_row <- data.frame(Package = package,
-                                          Dataset = dataname,
-                                          Class = class(ds),
-                                          OldRows = nr,
-                                          OldColumns = nc,
-                                          NewRows = new_nr,
-                                          NewColumns = new_nc)
-
-                    mismatched_dims_df <- rbind(mismatched_dims_df, new_row)
-
-                }
-            }
-
-            new_row <- data.frame(Package = package,
-                                  Dataset = dataname,
-                                  Title = title,
-                                  Rows = nr,
-                                  Columns = nc,
-                                  stringsAsFactors = FALSE)
-
-            dataset_df <- rbind(dataset_df, new_row)
-        }
+        do_package_update(data_dir, package_df, old_dataset_df, dataset_df, mismatched_dims_df, package)
     }
 
     stopifnot(nrow(dataset_df) > 0)
@@ -114,7 +161,7 @@ update_docs <- function(pkg_dir) {
 }
 
 write <- function(df, fn) {
-    write.table(df, file = fn, sep = ",", qmethod = "escape", row.names = FALSE)
+    write.table(df, file = fn, sep = ",", qmethod = "double", row.names = FALSE)
 }
 
 clean <- function(df) {

From 0c5b9e3fa5c8285c077f610dc90dc8ec54bac5db Mon Sep 17 00:00:00 2001
From: Frankie Robertson <frankie@robertson.name>
Date: Mon, 29 Aug 2022 18:00:18 +0300
Subject: [PATCH 2/3] Sort datasets CSV

---
 doc/datasets.csv | 60 ++++++++++++++++++++++++------------------------
 1 file changed, 30 insertions(+), 30 deletions(-)

diff --git a/doc/datasets.csv b/doc/datasets.csv
index b8081be0..8de3ed78 100644
--- a/doc/datasets.csv
+++ b/doc/datasets.csv
@@ -506,6 +506,36 @@
 "datasets","volcano","Topographic Information on Auckland's Maunga Whau Volcano",87,61
 "datasets","warpbreaks","The Number of Breaks in Yarn during Weaving",54,3
 "datasets","women","Average Heights and Weights for American Women",15,2
+"gamair","aral","aral",488,4
+"gamair","aral.bnd","aral.bnd",107,3
+"gamair","bird","bird",25100,7
+"gamair","blowfly","blowfly",180,3
+"gamair","bone","bone",23,4
+"gamair","brain","brain",1567,6
+"gamair","cairo","cairo",3780,7
+"gamair","chicago","chicago",5114,8
+"gamair","chl","chl",13840,7
+"gamair","co2s","co2s",507,4
+"gamair","coast","coast",2091,3
+"gamair","engine","engine",19,3
+"gamair","gas","gas",60,804
+"gamair","harrier","harrier",37,3
+"gamair","hubble","hubble",24,4
+"gamair","ipo","ipo",156,7
+"gamair","mack","mack",634,17
+"gamair","mackp","mackp",1162,9
+"gamair","med","med",1476,25
+"gamair","meh","meh",1476,24
+"gamair","mpg","mpg",205,27
+"gamair","prostate","prostate",654,530
+"gamair","sitka","sitka",1027,6
+"gamair","sole","sole",1575,8
+"gamair","sperm.comp1","sperm.comp1",15,5
+"gamair","sperm.comp2","sperm.comp2",24,11
+"gamair","stomata","stomata",24,4
+"gamair","swer","swer",2196,10
+"gamair","wesdr","wesdr",669,5
+"gamair","wine","wine",47,8
 "gap","PD","A study of Parkinson's disease and APOE, LRRK2, SNCA makers",825,22
 "gap","aldh2","ALDH2 markers and Alcoholism",263,18
 "gap","apoeapoc","APOE/APOC1 markers and Alzheimer's",353,8
@@ -732,33 +762,3 @@
 "vcd","VonBort","Von Bortkiewicz Horse Kicks Data",280,4
 "vcd","WeldonDice","Weldon's Dice Data",11,2
 "vcd","WomenQueue","Women in Queues",11,2
-"gamair","aral.bnd","aral.bnd",107,3
-"gamair","aral","aral",488,4
-"gamair","bird","bird",25100,7
-"gamair","blowfly","blowfly",180,3
-"gamair","bone","bone",23,4
-"gamair","brain","brain",1567,6
-"gamair","cairo","cairo",3780,7
-"gamair","chicago","chicago",5114,8
-"gamair","chl","chl",13840,7
-"gamair","co2s","co2s",507,4
-"gamair","coast","coast",2091,3
-"gamair","engine","engine",19,3
-"gamair","gas","gas",60,804
-"gamair","harrier","harrier",37,3
-"gamair","hubble","hubble",24,4
-"gamair","ipo","ipo",156,7
-"gamair","mack","mack",634,17
-"gamair","mackp","mackp",1162,9
-"gamair","med","med",1476,25
-"gamair","meh","meh",1476,24
-"gamair","mpg","mpg",205,27
-"gamair","prostate","prostate",654,530
-"gamair","sitka","sitka",1027,6
-"gamair","sole","sole",1575,8
-"gamair","sperm.comp1","sperm.comp1",15,5
-"gamair","sperm.comp2","sperm.comp2",24,11
-"gamair","stomata","stomata",24,4
-"gamair","swer","swer",2196,10
-"gamair","wesdr","wesdr",669,5
-"gamair","wine","wine",47,8

From ebcbfef06d38fbd322d6dd0bb9a58d17f65fd799 Mon Sep 17 00:00:00 2001
From: Frankie Robertson <frankie@robertson.name>
Date: Tue, 30 Aug 2022 12:36:40 +0300
Subject: [PATCH 3/3] Allow datasets with a .RData extension as well as .rda

---
 src/dataset.jl   | 5 +++++
 src/update_doc.r | 2 +-
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/src/dataset.jl b/src/dataset.jl
index 494fe00c..8df80232 100644
--- a/src/dataset.jl
+++ b/src/dataset.jl
@@ -7,6 +7,11 @@ const Dataset_typedetect_rows = Dict{Tuple{String, String}, Union{Vector,Dict}}(
 function dataset(package_name::AbstractString, dataset_name::AbstractString)
     basename = joinpath(@__DIR__, "..", "data", package_name)
 
+    rdataname = joinpath(basename, string(dataset_name, ".RData"))
+    if isfile(rdataname)
+        return load(rdataname)[dataset_name]
+    end
+
     rdaname = joinpath(basename, string(dataset_name, ".rda"))
     if isfile(rdaname)
         return load(rdaname)[dataset_name]
diff --git a/src/update_doc.r b/src/update_doc.r
index c1b3ef65..aecda680 100644
--- a/src/update_doc.r
+++ b/src/update_doc.r
@@ -32,7 +32,7 @@ do_package_update <- function(data_dir, package_df, old_dataset_df, dataset_df,
     datasets <- dir(path = file.path(data_dir, package))
 
     # Trim filenames to dataset names
-    r <- "(.+)\\.(csv\\.gz|rda)$"
+    r <- "(.+)\\.(csv\\.gz|rda|RData)$"
     format_recognized <- grepl(r, datasets)
     if (!(all(format_recognized))) {
         stop("Unrecognized formats:\n",