From 304bd3e53083eeed30aa2531787d860fb3499cc0 Mon Sep 17 00:00:00 2001
From: rhijmans <r.hijmans@gmail.com>
Date: Tue, 21 May 2024 08:07:26 -0700
Subject: [PATCH] m

---
 source/_R/contributors.rmd | 68 +++++++++++++++++++++++++++-----------
 source/contributors.rst    |  3 +-
 source/standard.rst        |  5 ++-
 3 files changed, 52 insertions(+), 24 deletions(-)

diff --git a/source/_R/contributors.rmd b/source/_R/contributors.rmd
index 8d69fbf..25dc39e 100644
--- a/source/_R/contributors.rmd
+++ b/source/_R/contributors.rmd
@@ -25,30 +25,58 @@ groups <- gsub("^carob_", "", groups)
 
 d <- lapply(f, read.csv)
 
-p <- lapply(1:length(d), \(i) {
-			n <- d[[i]]$carob_contributor
-			n <- unlist(strsplit(n, " and "))
-			n <- unlist(strsplit(n, ", "))
-			n <- unlist(strsplit(n, " & "))
-			data.frame(id=i, p=n)
-		}
-	)
-p <- do.call(rbind, p)
-#groups <- gsub("conservation_agriculture", "cons_ag", groups)
-#groups <- gsub("_trials", "", groups)
-#groups <- gsub("_samples", "", groups)
-#p[,1] <- groups[p[,1]]
-#tab <- table(p[,2], p[,1])
-tab <- as.data.frame(table(p$p))
-colnames(tab) <- c("name", "datasets")
-```
+get_tab <- function(d, var) {
+	p <- lapply(1:length(d), \(i) {
+				n <- d[[i]][[var]]
+				n <- unlist(strsplit(n, ";"))
+				data.frame(id=i, p=trimws(n))
+			}
+		)
+	p <- do.call(rbind, p)
+	p$p <- gsub(",", ", ", p$p)
+	p$p <- gsub(",  ", ", ", p$p)
+	p$p <- gsub("-, ", "", p$p) # "-, RHoMIS"
+	tab <- as.data.frame(table(p$p))
+	colnames(tab) <- c("name", "datasets")
+	tab 
+}
+
+cartab <- get_tab(d, "carob_contributor")
+autab <- get_tab(d, "authors")
+intab <- get_tab(d, "data_institute")
+i <- tolower(autab$name) %in% c("", "not applicable")
+autab <- autab[!i, ]
 
+```
+<br>
 
 <br>
 
-```{r grouptable, echo=FALSE}
-kableExtra::kbl(tab) |> kable_classic(full_width = FALSE) |>
-   kable_styling(bootstrap_options = c("striped", "hover"))
+### Script authors
+
+```{r catable, echo=FALSE}
+out = paste0(paste0(cartab[,1], " (", cartab[,2], ")"), collapse="; ")
+#kableExtra::kbl(cartab) |> kable_classic(full_width = FALSE) |>
+#   kable_styling(bootstrap_options = c("striped", "hover"))
+```
+`r out`
+
+### Dataset institutes
+
+```{r instable, echo=FALSE}
+out = paste0(paste0(intab[,1], " (", intab[,2], ")"), collapse="; ")
+#kableExtra::kbl(intab) |> kable_classic(full_width = FALSE) |>
+#   kable_styling(bootstrap_options = c("striped", "hover"))
+```
+`r out`
+
+### Dataset authors
+
+```{r autable, echo=FALSE}
+out <- paste0(paste0(autab[,1], " (", autab[,2], ")"), collapse="; ")
+#DT::datatable(autab, escape=FALSE, rownames=FALSE)
 ```
+`r out`
+
 
 <br>
diff --git a/source/contributors.rst b/source/contributors.rst
index 90b0a35..56219f8 100644
--- a/source/contributors.rst
+++ b/source/contributors.rst
@@ -1,7 +1,8 @@
 Contributors
 ============
 
-Here is a table with the names of *Carob* contributors and the number of datasets that they have contributed.
+Here are lists the contributors of Carob scripts and of the datasets processed. Names are followed by the number of scripts/datasets. Note that dataset author names are not standardized.
 
 .. raw:: html
    :file: _R/contributors.html
+
diff --git a/source/standard.rst b/source/standard.rst
index 4998f31..9d370ef 100644
--- a/source/standard.rst
+++ b/source/standard.rst
@@ -1,11 +1,10 @@
 The standard
 ============
 
-*Carob* uses the *terminag* standard that defines a controlled vocabulary of variable names, their units, and acceptable (ranges of) values. 
+*Carob* uses the **terminag** standard that defines a controlled vocabulary of variable names, their units, and acceptable (ranges of) values. 
 
 The *terminag* standard can be used "stand-alone" for your own data, and as part of the data compilation done through the Carob project.
 
-The standard is defined in a number of tables that are available on the [termiag](https://github.com/reagro/termiag) github site and via the R package carobiner. The standard is not an ontology in the sense that, with very few exceptions, it does not semantically relate variables to each other.
-
+The standard is defined in a number of tables that are available on the `termiag <https://github.com/reagro/terminag>`__ github site and via the R package `carobiner <https://github.com/reagro/carobiner>`__. The terminag standard is not an ontology in the sense that, with very few exceptions, it does not semantically relate variables to each other.