minimized setup (#9)

DESCRIPTION

@@ -0,0 +1,10 @@
+Type: PACTA
+Description: Run PACTA.
+Depends:
+  dplyr,
+  jsonlite,
+  pacta.portfolio.allocate,
+  pacta.portfolio.audit,
+  pacta.portfolio.import,
+  pacta.portfolio.utils,
+  readr

Dockerfile

@@ -3,20 +3,26 @@
 # https://rocker-project.org/images/versioned/r-ver.html
 #
 # sets CRAN repo to use Posit Package Manager to freeze R package versions to
-# those available on 2023-06-30
+# those available on 2023-09-29
 # https://packagemanager.posit.co/client/#/repos/2/overview
-# https://packagemanager.posit.co/cran/__linux__/jammy/2023-06-30
+# https://packagemanager.posit.co/cran/__linux__/jammy/2023-09-29
 
+# set proper base image
 ARG PLATFORM="linux/amd64"
 ARG R_VERS="4.3.1"
 FROM --platform=$PLATFORM rocker/r-ver:$R_VERS
 
+# set Docker image labels
 LABEL org.opencontainers.image.source=https://github.com/RMI-PACTA/workflow.pacta
 LABEL org.opencontainers.image.description="Docker image to run PACTA"
 LABEL org.opencontainers.image.licenses=MIT
-
-ARG CRAN_REPO="https://packagemanager.posit.co/cran/__linux__/jammy/2023-06-30"
-RUN echo "options(repos = c(CRAN = '$CRAN_REPO'))" >> "${R_HOME}/etc/Rprofile.site"
+LABEL org.opencontainers.image.title=""
+LABEL org.opencontainers.image.revision=""
+LABEL org.opencontainers.image.version=""
+LABEL org.opencontainers.image.vendor=""
+LABEL org.opencontainers.image.base.name=""
+LABEL org.opencontainers.image.ref.name=""
+LABEL org.opencontainers.image.authors=""
 
 # set apt-get to noninteractive mode
 ARG DEBIAN_FRONTEND="noninteractive"
@@ -35,73 +41,42 @@ RUN apt-get update \
     && chmod -R a+rwX /root \
     && rm -rf /var/lib/apt/lists/*
 
-# install system dependencies for R packages
-ARG R_PKG_SYS_DEPS="\
-    libfontconfig1-dev \
-    libfreetype6-dev \
-    libfribidi-dev \
-    libgit2-dev \
-    libharfbuzz-dev \
-    libicu-dev \
-    libjpeg-dev \
-    libpng-dev \
-    libtiff-dev \
-    libxml2-dev \
-    libxt6 \
-    make \
-    pandoc \
-    perl \
-    zlib1g-dev \
-    "
-RUN apt-get update \
-    && apt-get install -y --no-install-recommends $R_PKG_SYS_DEPS \
-    && rm -rf /var/lib/apt/lists/*
+# set frozen CRAN repo
+ARG CRAN_REPO="https://packagemanager.posit.co/cran/__linux__/jammy/2023-09-29"
+RUN echo "options(repos = c(CRAN = '$CRAN_REPO'))" >> "${R_HOME}/etc/Rprofile.site"
 
-# install pak for R package installation
+# install packages for dependency resolution and installation
 RUN Rscript -e "install.packages('pak')"
+RUN Rscript -e "pak::pkg_install('renv')"
 
-# copy in scripts from this repo
-ARG WORKFLOW_DIR="/workflow.pacta"
-COPY . $WORKFLOW_DIR
-WORKDIR $WORKFLOW_DIR
+# copy in everything from this repo
+COPY . /
 
 # PACTA R package tags
-ARG summary_tag="/tree/main"
 ARG allocate_tag="/tree/main"
 ARG audit_tag="/tree/main"
 ARG import_tag="/tree/main"
-ARG report_tag="/tree/main"
 ARG utils_tag="/tree/main"
 
-ARG summary_url="https://github.com/rmi-pacta/pacta.executive.summary"
 ARG allocate_url="https://github.com/rmi-pacta/pacta.portfolio.allocate"
 ARG audit_url="https://github.com/rmi-pacta/pacta.portfolio.audit"
 ARG import_url="https://github.com/rmi-pacta/pacta.portfolio.import"
-ARG report_url="https://github.com/rmi-pacta/pacta.portfolio.report"
 ARG utils_url="https://github.com/rmi-pacta/pacta.portfolio.utils"
 
 # install R package dependencies
 RUN Rscript -e "\
   gh_pkgs <- \
     c( \
-      paste0('$summary_url', '$summary_tag'), \
       paste0('$allocate_url', '$allocate_tag'), \
       paste0('$audit_url', '$audit_tag'), \
       paste0('$import_url', '$import_tag'), \
-      paste0('$report_url', '$report_tag'), \
       paste0('$utils_url', '$utils_tag') \
     ); \
-  source('pkg_deps.R'); \
-  pak::pak(c(gh_pkgs, pkg_deps)); \
+  workflow_pkgs <- renv::dependencies('DESCRIPTION')[['Package']]; \
+  workflow_pkgs <- grep('^pacta[.]', workflow_pkgs, value = TRUE, invert = TRUE); \
+  pak::pak(c(gh_pkgs, workflow_pkgs)); \
   "
 
-# set permissions for PACTA repos that need local content
-RUN chmod -R a+rwX $WORKFLOW_DIR
-
-# set the build_version environment variable
-ARG image_tag
-ENV build_version=$image_tag
-ARG head_hashes
-ENV head_hashes=$head_hashes
-
-CMD ["/bin/sh"]
+# set default run behavior
+ENTRYPOINT ["/run-pacta.sh"]
+CMD ["input_dir/default_config.json"]

input_dir/default_config.json

@@ -0,0 +1,12 @@
+{
+  "data_dir": "../pacta-data/2021Q4",
+  "portfolio_path": "input_dir/default_portfolio.csv",
+  "output_dir": "output_dir",
+  "start_year": 2021,
+  "equity_market_list": ["GlobalMarket", "DevelopedMarket", "EmergingMarket"],
+  "scenario_sources_list": ["GECO2022", "IPR2021", "ISF2021", "WEO2022"],
+  "scenario_geographies_list": ["Global", "GlobalAggregate", "NonOECD", "OECD"],
+  "sector_list": ["Power", "Automotive", "Oil&Gas", "Coal", "Steel", "Aviation", "Cement"],
+  "has_map": false,
+  "project_code": "GENERAL"
+}

input_dir/default_portfolio.csv

@@ -0,0 +1,91 @@
+isin,market_value,currency
+XS1297557172,6821206,USD
+DE000A1R07L4,4724723,USD
+US345370BV11,5619252,USD
+CND10001QD60,1124438,USD
+XS2356040357,3076756,USD
+US31428XAY22,8235607,USD
+FR0012650281,641235,USD
+FR001400F2Q0,4190307,USD
+USQ94606AE22,4118654,USD
+US911312BB15,5414864,USD
+INE331A07273,2972779,USD
+INE481G07166,3244497,USD
+INE010A08065,4488737,USD
+US86614RAF47,5793655,USD
+FR0124261621,1581767,USD
+US60937CAA36,900439,USD
+US93317QAD79,9993273,USD
+TH0148034703,8657460,USD
+XS2325566847,9459870,USD
+US02076XAC65,3952496,USD
+US880779BA01,3506143,USD
+USU88125AC04,3401211,USD
+INE208A07281,8670483,USD
+US69373UAA51,267878,USD
+US12594KAA07,4094599,USD
+US205677AB38,3309593,USD
+US69370RAK32,3519158,USD
+US707887A@47,1577541,USD
+US166764AA86,6575752,USD
+US71654QCP54,7084677,USD
+INE206D08162,6210214,USD
+US745310AB85,709698,USD
+CA44882ZXZ12,2809584,USD
+JP360540AR10,9641731,USD
+CA44885ZEQ96,9683822,USD
+INE017A08235,3910673,USD
+XS1555576641,2598631,USD
+XS1381693248,2117715,USD
+NO0010892870,8899756,USD
+NO0010806912,7280862,USD
+USU85795AQ29,2560264,USD
+JP338603AA35,6806168,USD
+DE000A2GESQ1,1567613,USD
+INE114A07802,8605761,USD
+US03938LAA26,2698943,USD
+KR7005380001,9338795,USD
+CNE100000Q35,1068423,USD
+KYG3777B1032,9637388,USD
+US5787871038,5467105,USD
+SE0016844831,4168401,USD
+TW0002646007,6371201,USD
+INE646L01027,8355228,USD
+US38045R2067,7952384,USD
+PHY1234G1032,180886,USD
+US00910M1009,6809079,USD
+INE070A01015,1013567,USD
+CNE000000XV4,8873793,USD
+PK0059501012,5829360,USD
+INE583C01021,6059486,USD
+MYL5000OO004,6433976,USD
+AU000000BCB5,9186193,USD
+US03940R1077,265189,USD
+ID1000162902,1712519,USD
+US03485P3001,4712749,USD
+CNE0000017V2,5989615,USD
+NL0010545661,3343735,USD
+CNE000000FH0,8915246,USD
+US6937181088,9800223,USD
+US9288541082,7226287,USD
+CA62910L1022,9449577,USD
+IL0011419699,7746343,USD
+SE0020180917,1788562,USD
+AU000000BUY9,1578448,USD
+KR7096770003,860209,USD
+US45325E2019,4071578,USD
+CA45790B7088,8510107,USD
+US0188021085,7030638,USD
+JP3981200003,344741,USD
+GB00B1VNSX38,6493591,USD
+BRENGICDAM16,9257903,USD
+MHY271836006,1571628,USD
+MHY2687W1241,2627572,USD
+TH0363010R10,8711798,USD
+JP3247600004,4447848,USD
+IL0065100930,1121308,USD
+QA000A0KD6K3,6405719,USD
+US92023R4074,8204370,USD
+JP3342000001,8564047,USD
+US88031M1099,149263,USD
+TW0009957001,4538558,USD

pacta_01.R

@@ -0,0 +1,90 @@
+suppressPackageStartupMessages({
+  library(pacta.portfolio.utils)
+  library(pacta.portfolio.import)
+  library(pacta.portfolio.audit)
+  library(dplyr)
+  library(readr)
+  library(jsonlite)
+})
+
+
+# -------------------------------------------------------------------------
+
+cfg_path <- commandArgs(trailingOnly = TRUE)
+if (length(cfg_path) == 0 || cfg_path == "") { cfg_path <- "input_dir/default_config.json" }
+cfg <- fromJSON(cfg_path)
+
+
+# load necessary input data ----------------------------------------------------
+
+currencies <- readRDS(file.path(cfg$data_dir, "currencies.rds"))
+
+fund_data <- readRDS(file.path(cfg$data_dir, "fund_data.rds"))
+total_fund_list <- readRDS(file.path(cfg$data_dir, "total_fund_list.rds"))
+isin_to_fund_table <- readRDS(file.path(cfg$data_dir, "isin_to_fund_table.rds"))
+
+fin_data <- readRDS(file.path(cfg$data_dir, "financial_data.rds"))
+
+entity_info <- get_entity_info(dir = cfg$data_dir)
+
+abcd_flags_equity <- readRDS(file.path(cfg$data_dir, "abcd_flags_equity.rds"))
+abcd_flags_bonds <- readRDS(file.path(cfg$data_dir, "abcd_flags_bonds.rds"))
+
+entity_emission_intensities <- readRDS(file.path(cfg$data_dir, "iss_entity_emission_intensities.rds"))
+average_sector_emission_intensities <- readRDS(file.path(cfg$data_dir, "iss_average_sector_emission_intensities.rds"))
+
+
+# Portfolios -------------------------------------------------------------------
+
+portfolio_raw <- read_portfolio_csv(cfg$portfolio_path)
+
+portfolio <- process_raw_portfolio(
+  portfolio_raw = portfolio_raw,
+  fin_data = fin_data,
+  fund_data = fund_data,
+  entity_info = entity_info,
+  currencies = currencies,
+  total_fund_list = total_fund_list,
+  isin_to_fund_table = isin_to_fund_table
+)
+
+# FIXME: this is necessary because pacta.portfolio.allocate::add_revenue_split()
+#  was removed in #142, but later we realized that it had a sort of hidden
+#  behavior where if there is no revenue data it maps the security_mapped_sector
+#  column of the portfolio data to financial_sector, which is necessary later
+portfolio <-
+  portfolio %>%
+  mutate(
+    has_revenue_data = FALSE,
+    financial_sector = .data$security_mapped_sector
+  )
+
+portfolio <- create_ald_flag(portfolio, comp_fin_data = abcd_flags_equity, debt_fin_data = abcd_flags_bonds)
+
+portfolio_total <- add_portfolio_flags(portfolio)
+
+portfolio_overview <- portfolio_summary(portfolio_total)
+
+audit_file <- create_audit_file(portfolio_total, has_revenue = FALSE)
+
+emissions_totals <- calculate_portfolio_financed_emissions(
+  portfolio_total,
+  entity_info,
+  entity_emission_intensities,
+  average_sector_emission_intensities
+)
+
+
+# Saving -----------------------------------------------------------------------
+
+export_audit_information_data(
+  audit_file_ = audit_file,
+  portfolio_total_ = portfolio_total,
+  folder_path = cfg$output_dir
+)
+
+saveRDS(portfolio_total, file.path(cfg$output_dir, "total_portfolio.rds"))
+saveRDS(portfolio_overview, file.path(cfg$output_dir, "overview_portfolio.rds"))
+saveRDS(audit_file, file.path(cfg$output_dir, "audit_file.rds"))
+write_csv(audit_file, file.path(cfg$output_dir, "audit_file.csv"))
+saveRDS(emissions_totals, file.path(cfg$output_dir, "emissions.rds"))