bug fixes, restructure, user module parsing

Makefile

@@ -18,6 +18,7 @@ info:
 	$(info $(OS)) 
 	$(info $(Threading))
 	$(info "test")
+	$(info $(CXX))
 server.bin:
 	$(CXX) server/server.cpp server/io.cpp server/table.cpp $(OS_SUPPORT) $(Threading) -flto --std=c++1z -O3 -march=native -o server.bin
 server.so:

README.md

@@ -2,7 +2,17 @@
 ## Introduction
 
 AQuery++ Database is a cross-platform, In-Memory Column-Store Database that incorporates compiled query execution.
-Compiler frontend built on top of [mo-sql-parsing](https://github.com/klahnakoski/mo-sql-parsing).
+
+## Architecture 
+### AQuery Compiler
+- The query is first processed by the AQuery Compiler which is composed of a frontend that parses the query into AST and a backend that generates target code that delivers the query.
+- Front end of AQuery++ Compiler is built on top of [mo-sql-parsing](https://github.com/klahnakoski/mo-sql-parsing) with modifications to handle AQuery dialect and extension.
+- Backend of AQuery++ Compiler generates target code dependent on the Execution Engine. It can either be the C++ code for AQuery Execution Engine or sql and C++ post-processor for Hybrid Engine or k9 for the k9 Engine.
+### Execution Engines
+- AQuery++ supports different execution engines thanks to the decoupled compiler structure.
+- AQuery Execution Engine: executes query by compiling the query plan to C++ code. Doesn't support joins and udf functions. 
+- Hybrid Execution Engine: decouples the query into two parts. The sql-compliant part is executed by an Embedded version of Monetdb and everything else is executed by a post-process module which is generated by AQuery++ Compiler in C++ and then compiled and executed.
+- K9 Execution Engine (discontinued).
   
 ## Roadmap
 - [x] SQL Parser -> AQuery Parser (Front End)
@@ -16,29 +26,39 @@ Compiler frontend built on top of [mo-sql-parsing](https://github.com/klahnakosk
       -  [x] Order by
       -  [x] Assumption
       -  [x] Flatten
-   -  [ ] Multi-table 
+   -  [x] Multi-table 
-      -  [ ] Join
+      -  [x] Join
    -  [ ] Subqueries 
 - [ ] -> Optimizing Compiler
 
 ## TODO:
-- [ ] C++ Meta-Programming: Elimilate template recursions as much as possible.
+
+- [ ] User Module load syntax parsing (fn definition/registration)
+- [ ] User Module test
+- [ ] Interval based triggers
+- [ ] C++ Meta-Programming: Eliminate template recursions as much as possible.
 - [ ] IPC: Better ways to communicate between Interpreter (Python) and Executer (C++).
   - [ ] Sockets? stdin/stdout capture?
 
 ## Requirements
-Recent version of Linux, Windows or MacOS, with recent C++ compiler that has C++17 (1z) support (e.g. gcc 6.0, MSVC 2017, clang 6.0), and python 3.6 or above.
+Recent version of Linux, Windows or MacOS, with recent C++ compiler that has C++17 (1z) support.
+- GCC: 9.0 or above (g++ 7.x, 8.x fail to handle variadic template expansion due to compiler bug)
+- Clang: 6.0 or above (Recommended)
+- MSVC: 2019 or later
 
 ## Usage
 `python3 prompt.py` will launch the interactive command prompt. The server binary will be autometically rebuilt and started.
 #### Commands:
 - `<sql statement>`: parse sql statement
 - `f <filename>`: parse all sql statements in file
+- `dbg` start debugging session 
 - `print`: printout parsed sql statements
-- `exec`: execute last parsed statement(s)
+- `exec`: execute last parsed statement(s) with AQuery Execution Engine. AQuery Execution Engine executes query by compiling it to C++ code and then executing it.
+  
+- `xexec`: execute last parsed statement(s) with Hybrid Execution Engine. Hybrid Execution Engine decouples the query into two parts. The sql-compliant part is executed by an Embedded version of Monetdb and everything else is executed by a post-process module which is generated by AQuery++ Compiler in C++ and then compiled and executed.
 - `r`: run the last generated code snippet
 - `save <OPTIONAL: filename>`: save current code snippet. will use random filename if not specified.
 - `exit`: quit the prompt
 #### Example:
    `f moving_avg.a` <br>
-   `exec`
+   `xexec`

aquery_parser/sql_parser.py

@@ -7,6 +7,7 @@
 # Contact: Kyle Lahnakoski (kyle@lahnakoski.com)
 #
 
+from sre_parse import WHITESPACE
 from mo_parsing.helpers import restOfLine
 from mo_parsing.infix import delimited_list
 from mo_parsing.whitespaces import NO_WHITESPACE, Whitespace
@@ -648,9 +649,8 @@ def parser(literal_string, ident, sqlserver=False):
             + Optional(assign("where", expr))
         ) / to_json_call
 
-        load = (
+        load_data = (
-            keyword("load")("op") 
+            keyword("data").suppress() 
-            + keyword("data").suppress() 
             + keyword("infile")("loc")  
             + literal_string ("file")
             + INTO
@@ -662,6 +662,42 @@ def parser(literal_string, ident, sqlserver=False):
                   + keyword("by").suppress() 
                   + literal_string ("term")
             )
+        )
+        
+        module_func_def = (
+            var_name("fname")
+            + LB 
+            + delimited_list(
+                (
+                    var_name("arg") 
+                    + COLON 
+                    + var_name("type")
+                )("vars")
+            ) 
+            + RB
+            + LAMBDA
+            + var_name("ret_type")
+        )
+        
+        load_module = (
+            keyword("module").suppress() 
+            + FROM 
+            + literal_string ("file")
+             + Optional(
+                keyword("FUNCTIONS").suppress()
+                 + LB
+                 + module_func_def("funcs") 
+                 + ZeroOrMore(Suppress(',') 
+                              + module_func_def("funcs"), 
+                              Whitespace()
+                            )
+                 + RB
+            )
+        )
+
+        load = (
+            keyword("load")("op") 
+            + (load_data | load_module)
         ) ("load")
 
 

engine/types.py

@@ -1,3 +1,4 @@
+from copy import deepcopy
 from engine.utils import defval
 from aquery_config import have_hge
 from typing import Dict, List
@@ -51,6 +52,12 @@ class Types:
         else:
             raise Exception(f'Illeagal cast: from {ty.name} to {self.name}.')
  
+    def __call__(self, args):
+        arg_str = ', '.join([a.__str__() for a in args])
+        ret = deepcopy(self)
+        ret.sqlname = self.sqlname + f'({arg_str})'
+        return ret
+
     def __repr__(self) -> str:
         return self.sqlname
     def __str__(self) -> str:

prompt.py

@@ -24,6 +24,36 @@ Run prompt.py without supplying with any arguments to run in interactive mode.
     parse only: parse the file and print out the AST
 '''
 
+prompt_help = '''\
+
+******** AQuery Prompt Help *********
+
+help:
+    print out this message
+help commandline:
+    print help message for AQuery Commandline
+<sql statement>: 
+    parse sql statement
+f <query file>: 
+    parse all AQuery statements in file
+script <AQuery Script file>:
+    run AQuery Script in file
+dbg: 
+    start debugging session with current context
+print: 
+    printout parsed sql statements
+exec: 
+    execute last parsed statement(s) with AQuery Execution Engine
+xexec: 
+    execute last parsed statement(s) with Hybrid Execution Engine
+r: 
+    run the last generated code snippet
+save <OPTIONAL: filename>: 
+    save current code snippet. will use timestamp as filename if not specified.
+exit or Ctrl+C: 
+    exit prompt mode
+'''
+
 if __name__ == '__main__':
     import mimetypes
     mimetypes._winreg = None
@@ -308,7 +338,13 @@ def main(running = lambda:True, next = input, state = None):
                 if subprocess.call(['make', 'snippet'], stdout = nullstream) == 0:
                     state.set_ready()
                 continue
-            
+            if q.startswith('help'):
+                qs = re.split(r'[ \t]', q)
+                if len(qs) > 1 and qs[1].startswith('c'):
+                    print(help_message)
+                else:
+                    print(prompt_help)
+                continue
             elif q == 'xexec': # generate build and run (MonetDB Engine)
                 state.cfg.backend_type = Backend_Type.BACKEND_MonetDB.value
                 cxt = xengine.exec(state.stmts, cxt, keep)

reconstruct/storage.py

@@ -1,10 +1,12 @@
 from engine.types import *
-
+from engine.utils import enlist
 class ColRef:
-    def __init__(self, _ty, cobj, table:'TableInfo', name, id, compound = False):
+    def __init__(self, _ty, cobj, table:'TableInfo', name, id, compound = False, _ty_args = None):
         self.type : Types = AnyT
         if type(_ty) is str:
             self.type = builtin_types[_ty.lower()]
+            if _ty_args:
+                self.type = self.type(enlist(_ty_args))
         elif type(_ty) is Types:
             self.type = _ty
         self.cobj = cobj
@@ -47,9 +49,13 @@ class TableInfo:
             
     def add_col(self, c, new = True, i = 0):
         _ty = c['type']
+        _ty_args = None
+        if type(_ty) is dict:
+            _ty_val = list(_ty.keys())[0]
+            _ty_args = _ty[_ty_val]
+            _ty = _ty_val
         if new:
-            _ty = _ty if type(c) is ColRef else list(_ty.keys())[0]
+            col_object = ColRef(_ty, c, self, c['name'], len(self.columns), _ty_args = _ty_args)
-            col_object =  ColRef(_ty, c, self, c['name'], len(self.columns))
         else:
             col_object = c
             c.table = self

server/server.cpp

@@ -153,8 +153,8 @@ extern "C" int __DLLEXPORT__ main(int argc, char** argv) {
 #ifdef THREADING
     auto tp = new ThreadPool();
     cxt->thread_pool = tp;
-#endif
     
+#endif
     
    const char* shmname;
    if (argc < 0)

server/threading.h

@@ -3,11 +3,7 @@
 
 #include <stdint.h>
 
-class ThreadPool{
+typedef int(*payload_fn_t)(void*);
-
-public:
-    typedef void(*payload_fn_t)(void*);
-
 struct payload_t{
     payload_fn_t f;
     void* args;
@@ -19,6 +15,10 @@ public:
     void empty() { f = nullptr; args = nullptr; }
     void operator()() { f(args); }
 };
+
+class ThreadPool{
+
+public:
     ThreadPool(uint32_t n_threads = 0);
     void enqueue_task(const payload_t& payload);
     bool busy();
@@ -39,4 +39,31 @@ private:
 
 };
 
+class Trigger{
+private:
+    void* triggers;  //min-heap by t-rem
+    virtual void tick() = 0;
+
+public:
+    Trigger(ThreadPool* tp);
+};
+
+class IntervalBasedTrigger : public Trigger{
+public:
+    struct timer{
+        uint32_t interval; // in milliseconds
+        uint32_t time_remaining;
+    };
+    void add_trigger();
+private:
+    void tick() override;
+};
+
+class CallbackBasedTrigger : public Trigger{
+public:
+    void add_trigger();
+private:
+    void tick() override;
+};
+
 #endif

tests/modules.a

@@ -1,4 +1,5 @@
 LOAD MODULE FROM "test.so" 
-FUNCTIONS (div(a:int, b:int) -> double, 
+FUNCTIONS (
+    div(a:int, b:int) -> double, 
     mulvec(a:int, b:vecfloat) -> vecfloat
 );

tests/strings.a

@@ -0,0 +1,7 @@
+CREATE TABLE types_test(names varchar(10), val real, id int)
+
+LOAD DATA INFILE "datatypes.csv"
+INTO TABLE types_test
+FIELDS TERMINATED BY ","
+
+select names, val * 10000 + id from types_test

udftest.cpp

@@ -1,14 +0,0 @@
-#include "udf.hpp"
-
-int main(){
-    vector_type _a{1,2,3,4};
-    vector_type _b{2,3,3,5};
-    ColRef<int> a("a");
-    ColRef<int> b("b");
-    a.initfrom(_a, "a");
-    b.initfrom(_b, "b");
-    ColRef<decltype(covariances2_gettype(a,b,0))> ret{4};
-    covariances2(a,b,2,4,ret);
-    
-    print(ret);
-}

udftest_compile.cpp

@@ -1,5 +0,0 @@
-#include "udf.hpp"
-
-int main(){
-	 
-}