From 4a8e20d5c3b48cac1b9cd8123fe27e8ab5029640 Mon Sep 17 00:00:00 2001
From: Dao McGill <77309217+daomcgill@users.noreply.github.com>
Date: Tue, 5 Nov 2024 17:06:51 -1000
Subject: [PATCH] i #314 Added exec/annotate.R and exec/query.R

- annotate.R generates the XML file
- query.R calls the query functions, depending on options

Note: may have to revisit output format after getting into the fasttext notebook

Signed-off-by: Dao McGill <dmcgill@hawaii.edu>
---
 exec/annotate.R |  4 +--
 exec/query.R    | 76 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 77 insertions(+), 3 deletions(-)
 create mode 100644 exec/query.R

diff --git a/exec/annotate.R b/exec/annotate.R
index 15d55b4e..2e9c8e19 100644
--- a/exec/annotate.R
+++ b/exec/annotate.R
@@ -35,15 +35,13 @@ OPTIONS:
   --version     Show version.
 "
 
-
-
 arguments <- docopt::docopt(doc, version = 'Kaiaulu 0.0.0.9600')
 if(arguments[["annotate"]] & arguments[["help"]]){
   cli_alert_info("Annotates source code using srcML.")
 }else if(arguments[["annotate"]]){
 
   tools_path <- arguments[["<tools.yml>"]]
-  conf_path <- arguments[["<project_conf.yml>"]]
+  conf_path <- arguments[["<maven.yml>"]]
   srcml_filepath <- arguments[["<srcml_filepath>"]]
 
   tool <- yaml::read_yaml(tools_path)
diff --git a/exec/query.R b/exec/query.R
new file mode 100644
index 00000000..2514e243
--- /dev/null
+++ b/exec/query.R
@@ -0,0 +1,76 @@
+#!/usr/local/bin/Rscript
+
+require(yaml, quietly = TRUE)
+require(cli, quietly = TRUE)
+require(docopt, quietly = TRUE)
+require(kaiaulu, quietly = TRUE)
+require(data.table, quietly = TRUE)
+require(stringi, quietly = TRUE)
+require(XML, quietly = TRUE)
+
+doc <- "
+USAGE:
+  query.R query <tools.yml> <project_conf.yml> <srcml_filepath> [--namespace | --file-docs | --class-docs | --variables | --packages | --functions | --imports]
+
+DESCRIPTION:
+  Runs a specified query on the annotated XML file and saves the output to the appropriate path.
+
+OPTIONS:
+  --namespace      Run the namespace query.
+  --file-docs      Run the file-level documentation query.
+  --class-docs     Run the class-level documentation query.
+  --variables      Run the variables query.
+  --packages       Run the packages query.
+  --functions      Run the functions query.
+  --imports        Run the imports query.
+"
+
+arguments <- docopt::docopt(doc, version = 'Kaiaulu 0.0.0.9600')
+
+tools_path <- arguments[["<tools.yml>"]]
+conf_path <- arguments[["<project_conf.yml>"]]
+srcml_filepath <- arguments[["<srcml_filepath>"]]
+
+tool <- yaml::read_yaml(tools_path)
+conf <- yaml::read_yaml(conf_path)
+srcml_path <- path.expand(tool[["srcml"]])
+
+# Set output folder path
+output_folder <- "../../analysis/maven/queries"
+if (!dir.exists(output_folder)) {
+  dir.create(output_folder, recursive = TRUE)
+}
+
+# Determine which query to run and save output
+if (arguments[["--namespace"]]) {
+  output_file <- file.path(output_folder, "namespace_output.csv")
+  query_result <- query_src_text_namespace(srcml_path, srcml_filepath)
+} else if (arguments[["--file-docs"]]) {
+  output_file <- file.path(output_folder, "file_docs_output.csv")
+  query_result <- query_src_text_file_docs(srcml_path, srcml_filepath)
+} else if (arguments[["--class-docs"]]) {
+  output_file <- file.path(output_folder, "class_docs_output.csv")
+  query_result <- query_src_text_class_docs(srcml_path, srcml_filepath)
+} else if (arguments[["--variables"]]) {
+  output_file <- file.path(output_folder, "variables_output.csv")
+  query_result <- query_src_text_variables(srcml_path, srcml_filepath)
+} else if (arguments[["--packages"]]) {
+  output_file <- file.path(output_folder, "packages_output.csv")
+  query_result <- query_src_text_packages(srcml_path, srcml_filepath)
+} else if (arguments[["--functions"]]) {
+  output_file <- file.path(output_folder, "functions_output.csv")
+  query_result <- query_src_text_functions(srcml_path, srcml_filepath)
+} else if (arguments[["--imports"]]) {
+  output_file <- file.path(output_folder, "imports_output.csv")
+  query_result <- query_src_text_imports(srcml_path, srcml_filepath)
+} else {
+  stop("No valid query option provided.")
+}
+
+# Transform each row into a single line of text
+plain_text_output <- apply(query_result, 1, function(row) paste(row, collapse = " "))
+
+# Save to file
+writeLines(plain_text_output, output_file)
+cli_alert_success(paste0("Query results saved for fastText at: ", output_file))
+