Add first data pulling function

DESCRIPTION

@@ -30,3 +30,12 @@ License: MIT + file LICENSE
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.2.3
+Imports: 
+    DBI,
+    dbplyr,
+    dplyr,
+    logger,
+    RPostgres,
+    withr
+Suggests: 
+    rstudioapi

Dockerfile

@@ -33,23 +33,19 @@ RUN groupadd -r runner-workflow-factset \
       && chown -R runner-workflow-factset /home/runner-workflow-factset
 WORKDIR /home/runner-workflow-factset
 
-# # install system dependencies
-# RUN apt-get update \
-#     && apt-get install -y --no-install-recommends \
-#       git=1:2.34.* \
-#       libcurl4-openssl-dev=7.81.* \
-#       libicu-dev=70.* \
-#       libssl-dev=3.0.* \
-#       openssh-client=1:8.* \
-#       wget=1.21.* \
-#     && chmod -R a+rwX /root \
-#     && rm -rf /var/lib/apt/lists/*
+# install system dependencies
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+      libicu-dev=70.* \
+      libpq-dev=14.* \
+    && chmod -R a+rwX /root \
+    && rm -rf /var/lib/apt/lists/*
 
 # set frozen CRAN repo
 ARG CRAN_REPO="https://packagemanager.posit.co/cran/__linux__/jammy/2023-10-30"
 RUN echo "options(repos = c(CRAN = '$CRAN_REPO'), pkg.sysreqs = FALSE)" >> "${R_HOME}/etc/Rprofile.site" \
       # install packages for dependency resolution and installation
-      && Rscript -e "install.packages('pak')"
+      && Rscript -e "install.packages(c('pak', 'jsonlite'))"
 
 # copy in everything from this repo
 COPY . /workflow.factset

NAMESPACE

@@ -1,2 +1,4 @@
 # Generated by roxygen2: do not edit by hand
 
+export(get_factset_entity_info)
+importFrom(dplyr,"%>%")

R/connect_factset_db.R

@@ -0,0 +1,98 @@
+# Connection function
+
+connect_factset_db <-
+  function(
+      dbname = "delta",
+      host = "data-eval-db.postgres.database.azure.com",
+      port = 5432L,
+      options = "-c search_path=fds",
+      username = Sys.getenv("R_DATABASE_USER"),
+      password = Sys.getenv("R_DATABASE_PASSWORD"),
+      keyring_service_name = "2dii_factset_database") {
+
+    if (username == "") {
+      logger::log_error("No database username could be found. Please set the username as an environment variable")
+    }
+
+    if (password == "") {
+      # if password not defined in .env, look in systems keyring
+      if (requireNamespace("keyring", quietly = TRUE)) {
+        if (!username %in% keyring::key_list(service = keyring_service_name)$username) {
+          keyring::key_set(
+            service = keyring_service_name,
+            username = username,
+            prompt = "Enter password for the FactSet database (it will be stored in your system's keyring): "
+          )
+        }
+        password <- keyring::key_get(
+          service = keyring_service_name,
+          username = username
+        )
+      } else if (interactive() && requireNamespace("rstudioapi", quietly = TRUE)) {
+        password <- rstudioapi::askForPassword(
+          prompt = "Please enter the FactSet database password:"
+        )
+      } else {
+        logger::log_error(
+          "No database password could be found. Please set the password
+          as an environment variable"
+        )
+      }
+    }
+
+    logger::log_trace(
+      "Connecting to database {dbname} on {host}:{port} as {username}"
+    )
+    conn <-
+      DBI::dbConnect(
+        drv = RPostgres::Postgres(),
+        dbname = dbname,
+        host = host,
+        port = port,
+        user = username,
+        password = password,
+        options = options
+      )
+
+    reg_conn_finalizer(conn, DBI::dbDisconnect, parent.frame())
+  }
+
+# connection finalizer to ensure connection is closed --------------------------
+# adapted from: https://shrektan.com/post/2019/07/26/create-a-database-connection-that-can-be-disconnected-automatically/
+
+reg_conn_finalizer <- function(conn, close_fun, envir) {
+  is_parent_global <- identical(.GlobalEnv, envir)
+
+  if (isTRUE(is_parent_global)) {
+    env_finalizer <- new.env(parent = emptyenv())
+    env_finalizer$conn <- conn
+    attr(conn, "env_finalizer") <- env_finalizer
+
+    reg.finalizer(env_finalizer, function(e) {
+      if (DBI::dbIsValid(e$conn)) {
+        logger::log_warn("Warning: A database connection was closed automatically because the connection object was removed or the R session was closed.")
+        try(close_fun(e$conn))
+      }
+    }, onexit = TRUE)
+  } else {
+    withr::defer(
+      {
+        if (DBI::dbIsValid(conn)) {
+          dbname <- DBI::dbGetInfo(conn)$dbname
+          host <- DBI::dbGetInfo(conn)$host
+
+          logger::log_warn(
+            "The database connection to {dbname} on {host} was
+        closed automatically because the calling environment was closed."
+          )
+          try(close_fun(conn))
+        }
+      },
+      envir = envir,
+      priority = "last"
+    )
+  }
+
+  logger::log_trace("Database connection registered for finalization")
+  return(conn)
+}

R/get_factset_entity_info.R

@@ -0,0 +1,113 @@
+#' Get the entity info data from the FactSet database and prepare the
+#' `factset_entity_info` tibble
+#'
+#' @param ... Arguments to be passed to the `connect_factset_db()` function (for
+#'   specifying database connection parameters)
+#'
+#' @return A tibble properly prepared to be saved as the
+#'   `factset_entity_info.rds` output file
+#'
+#' @export
+
+get_factset_entity_info <-
+  function(...) {
+    # build connection to database ---------------------------------------------
+
+    factset_db <- connect_factset_db(...)
+
+    logger::log_debug("Extracting entity info from database.")
+
+    # company_name -------------------------------------------------------------
+
+    logger::log_trace("Accessing entity proper names.")
+    factset_entity_id__entity_proper_name <-
+      dplyr::tbl(factset_db, "sym_v1_sym_entity") %>%
+      dplyr::select("factset_entity_id", "entity_proper_name")
+
+
+    # country_of_domicile ------------------------------------------------------
+
+    logger::log_trace("Accessing entity country of domicile.")
+    factset_entity_id__iso_country <-
+      dplyr::tbl(factset_db, "sym_v1_sym_entity") %>%
+      dplyr::select("factset_entity_id", "iso_country")
+
+
+    # sector -------------------------------------------------------------------
+
+    logger::log_trace("Accessing entity sector.")
+    factset_entity_id__sector_code <-
+      dplyr::tbl(factset_db, "sym_v1_sym_entity_sector") %>%
+      dplyr::select("factset_entity_id", "sector_code")
+
+    factset_sector_code__factset_sector_desc <-
+      dplyr::tbl(factset_db, "ref_v2_factset_sector_map") %>%
+      dplyr::select(.data$factset_sector_code, .data$factset_sector_desc)
+
+    factset_entity_id__factset_sector_desc <-
+      factset_entity_id__sector_code %>%
+      dplyr::left_join(factset_sector_code__factset_sector_desc, by = c("sector_code" = "factset_sector_code")) %>%
+      dplyr::select("factset_entity_id", "sector_code", "factset_sector_desc")
+
+
+    # sub-sector/industry ------------------------------------------------------
+
+    logger::log_trace("Accessing entity industry/sector/subsector.")
+    factset_entity_id__industry_code <-
+      dplyr::tbl(factset_db, "sym_v1_sym_entity_sector") %>%
+      dplyr::select("factset_entity_id", "industry_code")
+
+    factset_industry_code_factset_industry_desc <-
+      dplyr::tbl(factset_db, "ref_v2_factset_industry_map") %>%
+      dplyr::select("factset_industry_code", "factset_industry_desc")
+
+    factset_entity_id__factset_industry_desc <-
+      factset_entity_id__industry_code %>%
+      dplyr::left_join(factset_industry_code_factset_industry_desc, by = c("industry_code" = "factset_industry_code")) %>%
+      dplyr::select("factset_entity_id", "industry_code", "factset_industry_desc")
+
+
+    # credit risk parent -------------------------------------------------------
+
+    logger::log_trace("Accessing entity credit risk parent.")
+    ent_v1_ent_entity_affiliates <- dplyr::tbl(factset_db, "ent_v1_ent_entity_affiliates")
+    ref_v2_affiliate_type_map <- dplyr::tbl(factset_db, "ref_v2_affiliate_type_map")
+
+    ent_entity_affiliates_last_update <-
+      dplyr::tbl(factset_db, "fds_fds_file_history") %>%
+      dplyr::filter(.data$table_name == "ent_entity_affiliates") %>%
+      dplyr::filter(.data$begin_time == max(.data$begin_time, na.rm = TRUE)) %>%
+      dplyr::pull("begin_time")
+
+    factset_entity_id__credit_parent_id <-
+      ent_v1_ent_entity_affiliates %>%
+      dplyr::left_join(ref_v2_affiliate_type_map, by = "aff_type_code") %>%
+      dplyr::filter(.data$aff_type_desc == "Credit Risk Parent") %>%
+      dplyr::select(
+        factset_entity_id = "factset_affiliated_entity_id",
+        credit_parent_id = "factset_entity_id"
+      ) %>%
+      dplyr::mutate(ent_entity_affiliates_last_update = .env$ent_entity_affiliates_last_update)
+
+
+    # merge and collect --------------------------------------------------------
+
+    logger::log_trace("Merging entity info.")
+    entity_info <-
+      factset_entity_id__entity_proper_name %>%
+      dplyr::left_join(factset_entity_id__iso_country, by = "factset_entity_id") %>%
+      dplyr::left_join(factset_entity_id__factset_sector_desc, by = "factset_entity_id") %>%
+      dplyr::left_join(factset_entity_id__factset_industry_desc, by = "factset_entity_id") %>%
+      dplyr::left_join(factset_entity_id__credit_parent_id, by = "factset_entity_id")
+
+    logger::log_trace("Downloading merged entity info from database.")
+    entity_info <- dplyr::collect(entity_info)
+    logger::log_trace("Download complete.")
+
+    logger::log_trace("Disconnecting from database.")
+    DBI::dbDisconnect(factset_db)
+
+
+    # return prepared data -----------------------------------------------------
+    return(entity_info)
+  }

R/workflow.factset-package.R

@@ -0,0 +1,7 @@
+#' @keywords internal
+"_PACKAGE"
+
+## usethis namespace: start
+#' @importFrom dplyr %>%
+## usethis namespace: end
+NULL