Snakecase

README.md

@@ -14,7 +14,7 @@ For the time being, there are two sets of annotation generators:
 
 ## Adding an annotation
 
-- Add the command in the dictionary in (https://github.com/binpash/annotations/blob/main/pash_annotations/annotation_generation/AnnotationGeneration.py#L13)
+- Add the command in the dictionary in (https://github.com/binpash/annotations/blob/main/pash_annotations/annotation_generation/AnnotationGenerator().py#L13)
 - Add a json file with the command flags in (https://github.com/binpash/annotations/tree/main/pash_annotations/parser/command_flag_option_info/data). This could be used to generate a first version of it: (https://github.com/binpash/annotations/blob/main/pash_annotations/parser/command_flag_option_info/manpage-to-json.sh).
 - Add an `InputOutputInfoGeneratorXXX.py` in (https://github.com/binpash/annotations/tree/main/pash_annotations/annotation_generation/annotation_generators)
 - (Optionally) add a `ParallelizabilityInfoGeneratorXXX.py` in (https://github.com/binpash/annotations/tree/main/pash_annotations/annotation_generation/annotation_generators)

pash_annotations/annotation_cli.py

@@ -2,28 +2,40 @@
 from typing import Optional
 
 # create parser
-from annotation_generation.AnnotationGeneration import get_input_output_info_from_cmd_invocation, \
-    get_parallelizability_info_from_cmd_invocation
+from pash_annotations.annotation_generation.annotation_generation import (
+    get_input_output_info_from_cmd_invocation,
+    get_parallelizability_info_from_cmd_invocation,
+)
 
-from annotation_generation.datatypes.InputOutputInfo import InputOutputInfo
-from annotation_generation.datatypes.ParallelizabilityInfo import ParallelizabilityInfo
-from datatypes.CommandInvocationInitial import CommandInvocationInitial
-from parser.parser import parse
+from pash_annotations.annotation_generation.datatypes.input_output_info import InputOutputInfo
+from pash_annotations.annotation_generation.datatypes.parallelizability_info import ParallelizabilityInfo
+from pash_annotations.datatypes.command_invocation_initial import CommandInvocationInitial
+from pash_annotations.parser.parser import parse
 
 parser = argparse.ArgumentParser()
 
 # add arguments to the parser
-parser.add_argument('--command_invocation', metavar='STRING', type=str, required=True,
-                    help='specifies the command invocation to check (enclosed by \")')
-parser.add_argument('--save_to', metavar='FILE', type=str, default=None,
-                    help='store output in file (relative to where the script is called from); '
-                         'will not overwrite existing files but then print instead')
+parser.add_argument(
+    "--command_invocation",
+    metavar="STRING",
+    type=str,
+    required=True,
+    help='specifies the command invocation to check (enclosed by ")',
+)
+parser.add_argument(
+    "--save_to",
+    metavar="FILE",
+    type=str,
+    default=None,
+    help="store output in file (relative to where the script is called from); "
+    "will not overwrite existing files but then print instead",
+)
 
 # parse the arguments
 options = parser.parse_args()
 
 script_path = __file__
-split_path = script_path.rpartition('/')
+split_path = script_path.rpartition("/")
 script_prefix = "".join(split_path[:2])
 
 where_to_save = options.save_to
@@ -38,7 +50,9 @@
     except IOError:
         shall_we_write_to_file = False
     if not shall_we_write_to_file:
-        print("There exists a file on the provided path so the result will be output (only). ")
+        print(
+            "There exists a file on the provided path so the result will be output (only). "
+        )
 
 
 result = ""
@@ -49,18 +63,26 @@
 result += str(command_invocation) + "\n"
 
 result += ">>> INPUT-OUTPUT INFORMATION (applied to command invocation if possible): \n"
-io_info: Optional[InputOutputInfo] = get_input_output_info_from_cmd_invocation(command_invocation)
+io_info: Optional[InputOutputInfo] = get_input_output_info_from_cmd_invocation(
+    command_invocation
+)
 if io_info is None:
     result += f"Information not provided so considered side-effectful."
 elif io_info.has_other_outputs():
     result += f"Provided command has outputs other than streaming."
 else:
-    command_invocation_with_io = io_info.apply_input_output_info_to_command_invocation(command_invocation)
+    command_invocation_with_io = io_info.apply_input_output_info_to_command_invocation(
+        command_invocation
+    )
     result += str(command_invocation_with_io)
 result += "\n"
-para_info: Optional[ParallelizabilityInfo] = get_parallelizability_info_from_cmd_invocation(command_invocation)
+para_info: Optional[
+    ParallelizabilityInfo
+] = get_parallelizability_info_from_cmd_invocation(command_invocation)
 if para_info is None:
-    para_info = ParallelizabilityInfo() # defaults to no parallelizer's and all properties False
+    para_info = (
+        ParallelizabilityInfo()
+    )  # defaults to no parallelizer's and all properties False
 result += ">>> PARALLELIZABILITY INFORMATION: \n"
 # TODO: change representation when we move commutativity into parallelizers
 result += str(para_info)

pash_annotations/annotation_generation/annotation_generation.py

@@ -0,0 +1,110 @@
+import os
+import importlib
+from typing import Optional
+
+from collections import namedtuple
+
+from pash_annotations.datatypes.command_invocation_initial import (
+    CommandInvocationInitial,
+)
+from pash_annotations.annotation_generation.datatypes.input_output_info import (
+    InputOutputInfo,
+)
+from pash_annotations.annotation_generation.datatypes.parallelizability_info import (
+    ParallelizabilityInfo,
+)
+
+### directory paths
+ANNOTATION_GENERATORS = "pash_annotations.annotation_generation.annotation_generators"
+CMD_NAMES = [
+    "alt_bigrams_aux",
+    "alt_bigram_aux_reduce",
+    "awk",
+    "bigrams_aux",
+    "bigram_aux_map",
+    "bigram_aux_reduce",
+    "cat",
+    "col",
+    "comm",
+    "custom_sort",
+    "custom_tr",
+    "cut",
+    "diff",
+    "grep",
+    "head",
+    "mkfifo",
+    "mv",
+    "rm",
+    "sed",
+    "set_diff",
+    "seq",
+    "sort",
+    "tail",
+    "tee",
+    "test_one",
+    "test_two",
+    "tr",
+    "uniq",
+    "xarg",
+    "wc",
+]
+
+IO_INFO_PREFIX = "input_output_info_generator"
+PAR_INFO_PREFIX = "parallelizability_info_generator"
+
+FileModulePair = namedtuple("FileModulePair", ["file", "module"])
+
+
+class AnnotationGenerator:
+    # cannot be merged due to types
+    def get_input_output_info_from_cmd_invocation(
+        self,
+        cmd_invocation: CommandInvocationInitial,
+    ) -> Optional[InputOutputInfo]:
+        try:
+            # Get the Generator, info_generator_class_for_cmd_repr, info_generator_class_for_cmd_repr
+            cmd = cmd_invocation.cmd_name
+            module = importlib.import_module(
+                f"{ANNOTATION_GENERATORS}.{IO_INFO_PREFIX}_{cmd}"
+            )
+            info_gen_class = getattr(
+                module, self.to_pascal_case(IO_INFO_PREFIX) + self.to_pascal_case(cmd)
+            )
+            # Initialize the info generator object
+            info_generator_object = info_gen_class(cmd_invocation)
+            # Generate info
+            info_generator_object.generate_info()
+            return info_generator_object.get_info()
+        except:
+            return None
+
+    def get_parallelizability_info_from_cmd_invocation(
+        self,
+        cmd_invocation: CommandInvocationInitial,
+    ) -> Optional[ParallelizabilityInfo]:
+        # Get the Generator
+        try:
+            cmd = cmd_invocation.cmd_name
+            module = importlib.import_module(
+                f"{ANNOTATION_GENERATORS}.{PAR_INFO_PREFIX}_{cmd}"
+            )
+            print(module)
+            info_gen_class = getattr(
+                module, self.to_pascal_case(PAR_INFO_PREFIX) + self.to_pascal_case(cmd)
+            )
+            # Initialize the info generator object
+            info_generator_object = info_gen_class(cmd_invocation)
+            # Initialize the info generator object
+            # TODO: be more rigorous and allow empty parallelization annotation: return ParallelizabilityInfo with [] as default
+            # Generate info
+            info_generator_object.generate_info()
+            return info_generator_object.get_info()
+        except:
+            return None
+
+    @staticmethod
+    def to_pascal_case(string: str) -> str:
+        """
+        Turns a snake_case string to PascalCase
+        """
+        return string.replace("_", " ").title().replace(" ", "")