Merge pull request #15 from f-hafner/drop_education_degrees

drop graduates with ed. degrees

DESCRIPTION

@@ -8,7 +8,7 @@ Description: Functions to extract, transform and load data on PhD and their
 License: MIT + file LICENSE
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.0
+RoxygenNote: 7.2.1
 Imports: 
   DBI,
   dbplyr,

R/get_proquest.R

@@ -46,7 +46,10 @@ get_proquest <- function(conn, from, start_year = 1985, end_year = 2005, ...) {
   "
   us_universities <- dplyr::tbl(conn, dbplyr::sql(query_keep_us))
 
+  special_degrees <- c("Psy.D.", "Ed.D.", "D.Ed.") # relevant for psyschology
+
   graduates <- dplyr::tbl(conn, "pq_authors") %>%
+    dplyr::filter(!(.data$degree_level %in% special_degrees)) %>%
     dplyr::inner_join(us_universities, by = "university_id") %>%
     dplyr::filter(.data$degree_year >= start_year
                   & .data$degree_year <= end_year)

R/tools/setup_db_tests.R

@@ -56,7 +56,7 @@ capture_mockdb(production_db = db_file,
                              limit = 1,
                              lazy = FALSE)
                )
-files <- c("SELECT-db1387", "SELECT-1aa0ce")
+files <- c("SELECT-db1387", "SELECT-1aa0ce", "SELECT-1aa0ce")
 purrr::map(files,
            .f = ~copy_fixture(
              origin = mock_origin,
@@ -88,12 +88,13 @@ capture_mockdb(production_db = db_file,
 )
 
 # Note: some of these here are now deleted and do not exist anymore.. need to clean up!
-files <- c("SELECT-6db156",
-           "SELECT-f49f96",
+files <- c("SELECT_-64c4e4",
            "SELECT-276cf8",
+           "SELECT-6db156",
            "SELECT-7c7feb",
            "SELECT-dfa715",
-           "SELECT_-ad0113")
+           "SELECT-f49f96"
+           )
 
 purrr::map(files,
            .f = ~copy_fixture(
@@ -110,8 +111,7 @@ capture_mockdb(production_db = db_file,
 )
 
 # Note: some of these here are now deleted and do not exist anymore.. need to clean up!
-files <- c("SELECT-124fb3", "SELECT_-a24ff2")
-
+files <- c("SELECT-124fb3", "SELECT_-052c29")
 purrr::map(files,
            .f = ~copy_fixture(
              origin = mock_origin,

README.md

@@ -35,7 +35,7 @@ library(magutils)
 db_file <- db_example("AcademicGraph.sqlite")
 conn <- connect_to_db(db_file)
 #> The database connection is: 
-#> src:  sqlite 3.38.5 [/tmp/RtmpOOS3Cx/temp_libpath3eff04106586a/magutils/extdata/AcademicGraph.sqlite]
+#> src:  sqlite 3.39.3 [/tmp/RtmpfnS8S0/temp_libpath1e5445e211e0e/magutils/extdata/AcademicGraph.sqlite]
 #> tbls: author_coauthor, author_output, AuthorAffiliation, current_links,
 #>   current_links_advisors, FieldsOfStudy, FirstNamesGender, pq_advisors,
 #>   pq_authors, pq_fields_mag, pq_unis

data-raw/DATASET.R

@@ -96,7 +96,7 @@ purrr::map(idx_cla,
 ## 2. proquest authors / advisors
 
 qry <- paste0("
-  SELECT goid, firstname, degree_year, university_id
+  SELECT goid, firstname, degree_year, degree_level, university_id
   FROM pq_authors
   WHERE goid IN (",
   paste(pq_ids, collapse = ", "), ")"

magutils.Rproj

@@ -18,6 +18,5 @@ LineEndingConversion: Posix
 
 BuildType: Package
 PackageUseDevtools: Yes
-PackageCleanBeforeInstall: Yes
 PackageInstallArgs: --no-multiarch --with-keep.source
 PackageRoxygenize: rd,collate,namespace

tests/testthat/mock_db/SELECT_-052c29.R

@@ -0,0 +1,4 @@
+structure(list(goid = c(303763395L, 303483090L, 303690188L), 
+    position = c(0L, 0L, 0L), relationship_id = c("303763395_0", 
+    "303483090_0", "303690188_0"), gender = c("Male", "Male", 
+    "Male")), class = "data.frame", row.names = c(NA, -3L))

tests/testthat/test-example.R

@@ -2,7 +2,7 @@
 
 test_that("db_example() finds right file", {
   expect_error(db_example("randomfile"),
-               regexp = "no file found|No file found")
+               regexp = "no file found|No file found|Can't find package file")
   # R CMD check gives "no file found", test() gives "No file found"
   expect_match(db_example("AcademicGraph.sqlite"), "AcademicGraph")
 })