Add initial version of source code

- Only supports simple queries with projection, ie. "select a, b from
foobar.csv"

.gitignore

@@ -17,4 +17,4 @@ _build/
 
 # oasis generated files
 setup.data
-setup.log
+setup.log

.merlin

@@ -0,0 +1,5 @@
+B _build
+S *.ml
+PKG mparser
+PKG ounit
+PKG str

Makefile

@@ -0,0 +1,10 @@
+.DEFAULT_GOAL := csv_query.native
+
+test : QueryParserTest.native
+	./QueryParserTest.native
+
+QueryParserTest.native : QueryParserTest.ml QueryParser.ml Query.ml
+	ocamlbuild -use-ocamlfind 'QueryParserTest.native'
+
+csv_query.native : csv_query.ml Query.ml QueryParser.ml QueryExecutor.ml
+	ocamlbuild -use-ocamlfind 'csv_query.native'

Query.ml

@@ -0,0 +1,40 @@
+type file_name = string
+
+type column_name = string
+
+(*type row = 'a 'b map*)
+
+type table =
+    Table of file_name
+
+type column =
+    Column of column_name
+
+type projection =
+    | All
+    | Columns of column list
+
+type query =
+    Select of (projection * table)
+
+(*
+example query: select e.name, d.name from employees.csv as e join departments.csv as d where d.name = accounting
+*)
+
+let print_table = function
+    | Table table_name -> "(Table " ^ table_name ^ ")"
+
+let print_column = function
+    | Column column_name -> "column_name"
+
+let print_projection = function
+    | All -> "All"
+    | Columns columns ->
+        let column_names = List.map print_column columns in
+        "[" ^ String.concat ", " column_names ^ "]"
+
+let print_query = function
+    | Select (projection, table) ->
+        let projection_string = print_projection projection in
+        let table_string = print_table table in
+        "Select (" ^ projection_string ^ ", " ^ table_string ^ ")"

QueryExecutor.ml

@@ -0,0 +1,56 @@
+open Query
+
+type indexed_projection =
+    | AllIndices
+    | ColumnIndices of int list
+
+let deoptionalize l =
+  let rec deopt acc = function
+    | [] -> List.rev acc
+    | None::tl -> deopt acc tl
+    | Some x::tl -> deopt (x::acc) tl
+  in 
+  deopt [] l
+
+let find_index list element =
+    let rec go i l e =
+        match l with
+            | [] -> -1
+            | h :: t -> if h = e then i else go (i + 1) t e
+    in go 0 list element
+
+let optional_find list i =
+    try
+        Some (List.nth list i)
+    with
+        | _ -> None
+
+let get_all_indices indices list =
+   let found_values = List.map (fun i -> optional_find list i) indices in
+   deoptionalize found_values
+
+let project values = function
+    | AllIndices -> values
+    | ColumnIndices indices ->
+        get_all_indices indices values
+
+let get_indexed_projection header_row = function
+    | All -> AllIndices
+    | Columns columns ->
+        ColumnIndices (List.map (fun (Column projection_col_name) -> find_index header_row projection_col_name) columns)
+
+let execute_query = function
+    | Select (projection, Table table_file) ->
+        let in_channel = open_in table_file in
+        let header_line = input_line in_channel in
+        let header_row = Str.split (Str.regexp ",") header_line in
+        let indexed_projection = get_indexed_projection header_row projection in
+        print_endline (String.concat "," (project header_row indexed_projection));
+        try
+          while true do
+            let line = input_line in_channel in
+            let values = Str.split (Str.regexp ",") line in
+            print_endline (String.concat "," (project values indexed_projection))
+          done
+        with End_of_file ->
+          close_in in_channel

QueryParser.ml

@@ -0,0 +1,34 @@
+open MParser
+open Query
+
+let whitespace = blank <|> newline
+
+let identifier_char = alphanum
+
+let tablename = many1_chars (identifier_char <|> char '.')
+
+let column_spec = (many1_chars identifier_char) <|> (string "*")
+
+let csv p =
+    p >>= fun parsed_head ->
+    many (skip_char ',' >> whitespace >> p) >>= fun parsed_rest ->
+    return (parsed_head :: parsed_rest)
+
+let strings_to_projection names =
+    if names = ["*"]
+    then All
+    else Columns (List.map (fun n -> Column n) names)
+
+let query_parser =
+    skip_string "select" >>
+    whitespace >>
+    csv column_spec >>= fun column_names ->
+    whitespace >>
+    skip_string "from" >>
+    whitespace >>
+    tablename >>= fun tn ->
+    let projection = strings_to_projection column_names in
+    return (Query.Select (projection, Query.Table tn))
+
+let parse_query string =
+    parse_string query_parser string ()

QueryParser.mli

@@ -0,0 +1,6 @@
+open MParser
+open Query
+
+val query_parser : (query, unit) MParser.t
+
+val parse_query : string -> query MParser.result

QueryParserTest.ml

@@ -0,0 +1,30 @@
+open MParser
+open OUnit2
+open Query
+open QueryParser
+
+exception Parse_failure of string
+
+let unwrap_result = function
+    | Success a -> a
+    | Failed (message, _) -> assert_failure message
+
+let parse_test query_string expected_query test_ctxt =
+    let actual_query = unwrap_result (parse_string query_parser query_string ()) in
+    let error_msg = "Expected query: " ^ (print_query expected_query) ^ " was not equal to actual query: " ^ (print_query actual_query) in
+    assert_equal ~msg:error_msg actual_query expected_query
+
+let parse_simple_all_projection_query =
+    parse_test "select * from helloworld.csv" (Select (All, Table "helloworld.csv"))
+
+let parse_simple_projection_query =
+    parse_test "select hello, world from helloworld.csv" (Select (Columns [Column "hello"; Column "world"], Table "helloworld.csv"))
+
+let suite = "select query suite" >:::
+    [
+        "parsing a simple * query" >:: parse_simple_all_projection_query;
+        "parsing a simple query with projection" >:: parse_simple_projection_query
+    ]
+
+let () =
+    run_test_tt_main suite

README.md

@@ -1,2 +1,45 @@
-# csv-query
-A command line tool for performing SQL queries on CSV files
+# csv_query - A command line tool for performing SQL queries on CSV files
+
+Simple usage:
+
+```
+make
+./csv_query.native 'select * from employees.csv'
+```
+
+## How to build
+
+Run `make` to run the default target which builds the csv_query.native binary.
+
+###
+
+1. mparser
+2. ounit (for testing)
+3. str
+
+I still need to figure out how to extract the dependency specificaiton to a file and what command to use to automatically install them. I installed them manually using opam, eg. `opam install mparser` on my own machine.
+
+## How to test
+
+Run `make test` to compile and run the test suite.
+
+## How does it work?
+
+The query string is parsed using a simple monadic parser constructed using [MParser](https://github.com/cakeplus/mparser). The code for the parser is located in QueryParser.ml. The parser returns a `query` data structure which can be found in Query.ml. The `query` can be executed by the `execute_query` function in QueryExecutor.ml to produce output based on the given SQL query.
+
+## Supported SQL features
+
+- SELECT *
+- SELECT a, b
+- csv files as table names, eg. FROM employees.csv
+
+## Future work
+
+- Add more SQL features (WHERE, JOIN, aggregate functions)
+- Add method for fetching dependencies easier
+- Add property based testing which would automatically generate a Query data value, print it as a string and then parse it back and check that the parsed query matches the original randomly generated one
+- Improve build process (I'm terrible at writing Makefiles)
+
+## Dev environment
+
+I use VSCode with the OCaml plugin and Merlin. The code is built using `ocamlbuild` and `ocamlfind`

_tags

@@ -0,0 +1 @@
+true: package(ounit), package(mparser), package(str)

csv_query.ml

@@ -0,0 +1,11 @@
+open MParser
+open QueryParser
+open QueryExecutor
+
+let () =
+    let query_string = Sys.argv.(1) in
+    match parse_query query_string with
+        | Success parsed_query ->
+            execute_query parsed_query
+        | Failed (message, _) ->
+            print_string message

samples/employees.csv

@@ -0,0 +1,3 @@
+name,salary,office_id
+alice,2000000,1
+bob,1000000,1