[MediaWiki-commits] [Gerrit] adds property labels to property usage table - change (wikidata...dashboard)
Christopher Johnson (WMDE) has uploaded a new change for review.
https://gerrit.wikimedia.org/r/252024
Change subject: adds property labels to property usage table
..
adds property labels to property usage table
Change-Id: I7ad226c89cc7eacde5394ac5892ff9c361af14a0
---
M src/output/server-properties.R
M src/scripts/property_usage.R
M src/utils.R
3 files changed, 25 insertions(+), 13 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/wikidata/analytics/dashboard
refs/changes/24/252024/1
diff --git a/src/output/server-properties.R b/src/output/server-properties.R
index 48cfdab..353ba28 100644
--- a/src/output/server-properties.R
+++ b/src/output/server-properties.R
@@ -6,13 +6,16 @@
box(title = "Definition", width = 6, status = "info", metric_desc)
})
output$wikidata_property_usage_count_table <- DT::renderDataTable({
- datatable(property_usage_counts, class = "display compact", colnames =
c("Property", "Count"),
+ datatable(property_usage_counts, class = "display compact", colnames =
c("Property", "Label", "Count"),
options = list(
- order = list(2, 'desc'),
+ order = list(3, 'desc'),
pageLength = 100,
columnDefs = list(
-list(className = 'dt-left', targets = c(1,2)),
+list(className = 'dt-left', targets = c(1,2,3)),
list(visible = FALSE, targets = c(0)),
+list(width='10%', targets = c(1)),
+list(width='40%', targets = c(2)),
+list(width='40%', targets = c(3)),
list(targets = c(1), render = JS(
"function(data, type, row, meta) {",
"return 'https://www.wikidata.org/wiki/Property:'+data+'\"
target=\"_blank\">'+data+''",
diff --git a/src/scripts/property_usage.R b/src/scripts/property_usage.R
index 908aaf4..d85a081 100644
--- a/src/scripts/property_usage.R
+++ b/src/scripts/property_usage.R
@@ -7,19 +7,24 @@
write_prop_usage_counts <- function() {
query <- get_property_list_query()
prefix <- get_property_label_prefixes()
- plist <- get_sparql_result_from_uri(wdmrdf_uri, prefix, query)
+ doc <- get_sparql_result(wdmrdf_uri, prefix, query)
+ plist <- get_dataframe_from_xml_result(doc, "//sq:uri")
props <- lapply(plist, function(x) gsub("http://www.wikidata.org/entity/;,
"", x))
+ labels <- get_dataframe_from_xml_result(doc, "//sq:literal")
values <- lapply(props$text, function(x)
get_estimated_card_from_prop_predicate(estcard.uri, x))
vals <- do.call(c, unlist(values, recursive=FALSE))
+ labels <- data.table(labels$text)
+ labels$id <- seq_len(nrow(labels))
prop_counts <- data.table(vals)
props <- data.table(props$text)
props$id <- seq_len(nrow(props))
prop_counts$id <- seq_len(nrow(prop_counts))
setkey(props, id)
setkey(prop_counts, id)
+ setkey(labels, id)
dt_join_prop_usage <- props[prop_counts]
- dt_join_prop_usage <- dt_join_prop_usage[,.SD,.SDcols=c(1,3)]
- dt_join_prop_usage <- setnames(dt_join_prop_usage, c("Property", "Count"))
+ dt_join_prop_usage <- dt_join_prop_usage[,.SD,.SDcols=c(1,3,4)]
+ dt_join_prop_usage <- setnames(dt_join_prop_usage,
c("Property","Label","Count"))
write.table(dt_join_prop_usage, paste0(sparql_data_uri, "prop_usage.tsv"),
sep = "\t", row.names = FALSE)
}
diff --git a/src/utils.R b/src/utils.R
index b858894..d08a766 100644
--- a/src/utils.R
+++ b/src/utils.R
@@ -230,21 +230,21 @@
}
get_property_list_query <- function(){
- query = curl_escape("SELECT ?s WHERE {?s ?p wikibase:Property}")
+ query = curl_escape("SELECT ?s ?o WHERE {?s ?p wikibase:Property .
+SERVICE wikibase:label {
+ bd:serviceParam wikibase:language \"en\" .
+ ?s rdfs:label ?o}}")
return(query)
}
get_sparql_result <- function(uri = wdqs_uri, prefix, query) {
xml_result <- readLines(curl(paste0(uri, prefix, query)))
doc = xmlParse(xml_result)
- result = xmlToDataFrame(nodes = getNodeSet(doc, "//sq:literal", c(sq =
"http://www.w3.org/2005/sparql-results#;)))
- return(result)
+ return(doc)
}
-get_sparql_result_from_uri <- function(uri = wdmrdf_uri, prefix, query) {
- xml_result <- readLines(curl(paste0(uri, prefix, query)))
- doc = xmlParse(xml_result)
- result = xmlToDataFrame(nodes = getNodeSet(doc, "//sq:uri", c(sq =
"http://www.w3.org/2005/sparql-results#;)))
+get_dataframe_from_xml_result <- function(doc, qname) {
+ result = xmlToDataFrame(nodes = getNodeSet(doc, qname, c(sq =
"http://www.w3.org/2005/sparql-results#;)))
return(result)
}
@@ -254,4 +254,8 @@
result = xpathApply(doc, "//data[@rangeCount]", xmlGetAttr, "rangeCount")
return(result)
+}
+
+join_data_frames <- function(x, y) {
+
}
\ No newline at end of file
--
To view, visit https://gerrit.wikimedia.org/r/252024
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id:
[MediaWiki-commits] [Gerrit] adds property labels to property usage table - change (wikidata...dashboard)
Christopher Johnson (WMDE) has submitted this change and it was merged.
Change subject: adds property labels to property usage table
..
adds property labels to property usage table
Change-Id: I7ad226c89cc7eacde5394ac5892ff9c361af14a0
---
M src/output/server-properties.R
M src/scripts/property_usage.R
M src/utils.R
3 files changed, 25 insertions(+), 13 deletions(-)
Approvals:
Christopher Johnson (WMDE): Verified; Looks good to me, approved
diff --git a/src/output/server-properties.R b/src/output/server-properties.R
index 48cfdab..353ba28 100644
--- a/src/output/server-properties.R
+++ b/src/output/server-properties.R
@@ -6,13 +6,16 @@
box(title = "Definition", width = 6, status = "info", metric_desc)
})
output$wikidata_property_usage_count_table <- DT::renderDataTable({
- datatable(property_usage_counts, class = "display compact", colnames =
c("Property", "Count"),
+ datatable(property_usage_counts, class = "display compact", colnames =
c("Property", "Label", "Count"),
options = list(
- order = list(2, 'desc'),
+ order = list(3, 'desc'),
pageLength = 100,
columnDefs = list(
-list(className = 'dt-left', targets = c(1,2)),
+list(className = 'dt-left', targets = c(1,2,3)),
list(visible = FALSE, targets = c(0)),
+list(width='10%', targets = c(1)),
+list(width='40%', targets = c(2)),
+list(width='40%', targets = c(3)),
list(targets = c(1), render = JS(
"function(data, type, row, meta) {",
"return 'https://www.wikidata.org/wiki/Property:'+data+'\"
target=\"_blank\">'+data+''",
diff --git a/src/scripts/property_usage.R b/src/scripts/property_usage.R
index 908aaf4..d85a081 100644
--- a/src/scripts/property_usage.R
+++ b/src/scripts/property_usage.R
@@ -7,19 +7,24 @@
write_prop_usage_counts <- function() {
query <- get_property_list_query()
prefix <- get_property_label_prefixes()
- plist <- get_sparql_result_from_uri(wdmrdf_uri, prefix, query)
+ doc <- get_sparql_result(wdmrdf_uri, prefix, query)
+ plist <- get_dataframe_from_xml_result(doc, "//sq:uri")
props <- lapply(plist, function(x) gsub("http://www.wikidata.org/entity/;,
"", x))
+ labels <- get_dataframe_from_xml_result(doc, "//sq:literal")
values <- lapply(props$text, function(x)
get_estimated_card_from_prop_predicate(estcard.uri, x))
vals <- do.call(c, unlist(values, recursive=FALSE))
+ labels <- data.table(labels$text)
+ labels$id <- seq_len(nrow(labels))
prop_counts <- data.table(vals)
props <- data.table(props$text)
props$id <- seq_len(nrow(props))
prop_counts$id <- seq_len(nrow(prop_counts))
setkey(props, id)
setkey(prop_counts, id)
+ setkey(labels, id)
dt_join_prop_usage <- props[prop_counts]
- dt_join_prop_usage <- dt_join_prop_usage[,.SD,.SDcols=c(1,3)]
- dt_join_prop_usage <- setnames(dt_join_prop_usage, c("Property", "Count"))
+ dt_join_prop_usage <- dt_join_prop_usage[,.SD,.SDcols=c(1,3,4)]
+ dt_join_prop_usage <- setnames(dt_join_prop_usage,
c("Property","Label","Count"))
write.table(dt_join_prop_usage, paste0(sparql_data_uri, "prop_usage.tsv"),
sep = "\t", row.names = FALSE)
}
diff --git a/src/utils.R b/src/utils.R
index b858894..d08a766 100644
--- a/src/utils.R
+++ b/src/utils.R
@@ -230,21 +230,21 @@
}
get_property_list_query <- function(){
- query = curl_escape("SELECT ?s WHERE {?s ?p wikibase:Property}")
+ query = curl_escape("SELECT ?s ?o WHERE {?s ?p wikibase:Property .
+SERVICE wikibase:label {
+ bd:serviceParam wikibase:language \"en\" .
+ ?s rdfs:label ?o}}")
return(query)
}
get_sparql_result <- function(uri = wdqs_uri, prefix, query) {
xml_result <- readLines(curl(paste0(uri, prefix, query)))
doc = xmlParse(xml_result)
- result = xmlToDataFrame(nodes = getNodeSet(doc, "//sq:literal", c(sq =
"http://www.w3.org/2005/sparql-results#;)))
- return(result)
+ return(doc)
}
-get_sparql_result_from_uri <- function(uri = wdmrdf_uri, prefix, query) {
- xml_result <- readLines(curl(paste0(uri, prefix, query)))
- doc = xmlParse(xml_result)
- result = xmlToDataFrame(nodes = getNodeSet(doc, "//sq:uri", c(sq =
"http://www.w3.org/2005/sparql-results#;)))
+get_dataframe_from_xml_result <- function(doc, qname) {
+ result = xmlToDataFrame(nodes = getNodeSet(doc, qname, c(sq =
"http://www.w3.org/2005/sparql-results#;)))
return(result)
}
@@ -254,4 +254,8 @@
result = xpathApply(doc, "//data[@rangeCount]", xmlGetAttr, "rangeCount")
return(result)
+}
+
+join_data_frames <- function(x, y) {
+
}
\ No newline at end of file
--
To view, visit https://gerrit.wikimedia.org/r/252024
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: I7ad226c89cc7eacde5394ac5892ff9c361af14a0
Gerrit-PatchSet:
