Merge pull request #3 from ArgLab/main

Resource fix (#2)

Author: duckduckdoof

.gitignore

@@ -0,0 +1,5 @@
+# Ignore those pesky files
+
+__pycache__/
+*.egg-info/
+build/

README.md

@@ -15,16 +15,18 @@ cleanup work for all clones/forks with unmerged changes.
 ## Installation
 
 1. Clone repo from Github
-1. `cd AWE_LanguageTool`
-1. `pip install -e .`
-1. NOTE: should this just be included in AWE Components? That's what we use to interface with the other NLP stuff
+2. `cd AWE_LanguageTool`
+3. `pip install -e .`
 
-## Running
+## LanguageTool Configuration & Running
 
-The system is ran using a client-server model.
-The server runs the relevant Java command to start the `languagetool-server.jar` file.
-This files comes from directly from the original Language Tool.
-The client handles wrapping the output and adding in additional error classification categories.
+The system is run using a client-server model. The server runs the relevant Java command to start the `languagetool-server.jar` file (which comes from LanaguageTool). The client handles wrapping the output and adding in additional error classification categories.
+
+Before running LanguageTool, *you must first copy/write a server config file into* `awe_languagetool/LanguageTool5_5/`. We've provided a sample config file called `languagetool.tmp.cfg`. If you copy this file, you must rename it to `languagetool.cfg`.
+
+By default, LT runs pretty slow with too many incoming requests; you can modify the server settings for LT in `awe_languagetool/LanguageTool5_5/languagetool.cfg`. See [this forum post](https://forum.languagetool.org/t/too-many-parallel-requests/8290/3) on a decent server config file. For a full description of all server config options, see [LT5.5 Source Code](https://github.com/languagetool-org/languagetool/blob/c6321ab5837a9e1ae5501d746f947f5706b4b274/languagetool-server/src/main/java/org/languagetool/server/HTTPServerConfig.java).
+
+With the python LT wrapper, this can be run from anywhere in the project. However, if you decide to run the java command directly (see below), this needs to be run within the `awe_languagetool/LanguageTool5_5/` directory.
 
 1. Start the server
 
@@ -36,18 +38,21 @@ languagetoolServer.runServer()
 This can also be ran using directly using the Java command.
 Note that this command has not been fully tested with which directory it needs to be run from.
 If running this does not work, see the `languagetoolServer.py` file for more information about how the system is started.
+
 ```bash
-java -cp languagetool-server.jar org.languagetool.server.HTTPServer --port {port} --allow-origin "*"
+java -cp languagetool-server.jar org.languagetool.server.HTTPServer --config languagetool.cfg --port {port} --allow-origin "*"
 ```
 
 1. Connect the client (requires another terminal)
 
 ```python
 from awe_languagetool import languagetoolClient
+import asyncio
+
 client = languagetoolClient.languagetoolClient()
 text_to_process = '...'
-output = client.summarizeText(text_to_process)
-# Example of output
+output = asyncio.run(client.summarizeText(text_to_process))
+# Example output
 # {
 #     'wordcounts': {
 #         'tokens': 0,

awe_languagetool/languagetoolClient.py

@@ -56,31 +56,39 @@ class languagetoolClient:
     def __init__(self, port=8081):
 
         self.port = port
-        self.MAPPING_PATH = \
-            resources.path('awe_languagetool',
-                           'languagetool_rulemapping.json')
-        print(self.MAPPING_PATH)
 
+        # As of python 3.11, resources.path() is deprecated, and the following
+        # code is the 'equivalent' replacement.
+        path_context = resources.as_file(
+            resources.files('awe_languagetool').joinpath('languagetool_rulemapping.json')
+        )
+        with path_context as out_path:
+            self.MAPPING_PATH = str(out_path)
+        print(str(self.MAPPING_PATH))
+
+        # The importlib.resources objects behave differently than standard paths
+        # so it is necessary to adjust the code to use the resource context functions
+        # rather than standard file functions.
         if not os.path.exists(self.MAPPING_PATH):
             raise mappingPathError(
                 "Trying to load AWE Workbench Lexicon Module \
                  without supporting datafiles"
             )
-        fo = open(self.MAPPING_PATH, "r")
-        jsonContent = fo.read()
-        self.ruleInfo = json.loads(jsonContent)
-        fo.close()
-        for rule in self.ruleInfo:
-            for subrule in self.ruleInfo[rule]:
-                [cat, subcat] = self.ruleInfo[rule][subrule]
-                if cleanstring(cat).title() not in self.categoryList:
-                    self.categoryList.append(cleanstring(cat).title())
-                subcat_name = cleanstring(cat).title() + ': ' + cleanstring(subcat).title()
-                if cleanstring(subcat_name) not in self.subcategoryList:
-                    self.subcategoryList.append(cleanstring(subcat_name))
-
-        self.categoryList.sort()
-        self.subcategoryList.sort()
+        
+        with open(self.MAPPING_PATH, "r") as fo:
+            jsonContent = fo.read()
+            self.ruleInfo = json.loads(jsonContent)
+            for rule in self.ruleInfo:
+                for subrule in self.ruleInfo[rule]:
+                    [cat, subcat] = self.ruleInfo[rule][subrule]
+                    if cleanstring(cat).title() not in self.categoryList:
+                        self.categoryList.append(cleanstring(cat).title())
+                    subcat_name = cleanstring(cat).title() + ': ' + cleanstring(subcat).title()
+                    if cleanstring(subcat_name) not in self.subcategoryList:
+                        self.subcategoryList.append(cleanstring(subcat_name))
+
+            self.categoryList.sort()
+            self.subcategoryList.sort()
 
     def make_printable(self, s):
         """Replace non-printable characters in a string."""

awe_languagetool/languagetoolServer.py

@@ -16,39 +16,34 @@
 from importlib import resources
 
 
-def runServer(fileName=None, port=8081):
+def runServer(fileName=None, port=8081, config_file="languagetool.cfg"):
     '''
     Runs the LanguageTool server, using `importlib.resources` to find the
     jar file.
     '''
-    # In order for python 3.9 to work we have to make a slight hack to the
-    # language tools module in order to ensure that this works.  To do that
-    # we first import the tool and then set the origin explicitly.  
-    #
-    # This cheap hack does just that by using the submodule_search_location
-    # value which *does* seem to be set by default to supply the location
-    # for origin.  Having done that we can then go about the rest of it
-    # without error.
-    import platform
-    if (platform.python_version()[0:3] == "3.9"):
-        import awe_languagetool.LanguageTool5_5
-        LTSpec = awe_languagetool.LanguageTool5_5.__spec__
-        LTSpec.origin = LTSpec.submodule_search_locations[0]
-
-    with resources.path('awe_languagetool.LanguageTool5_5',
-                        'languagetool-server.jar') as LANGUAGE_TOOL_PATH:
+    # NOTE: after playing with python3.9, it does not like the 'package.sub' string.
+    # So, I added multiple 'joinpaths'; this worked for both 3.9 and 3.11
+    with resources.as_file(
+            resources.files('awe_languagetool').joinpath('LanguageTool5_5').joinpath('languagetool-server.jar')
+        ) as LANGUAGE_TOOL_PATH:
+        print("Setting Language Path:",  LANGUAGE_TOOL_PATH)
         MAPPING_PATH = os.path.dirname(LANGUAGE_TOOL_PATH)
 
-        
     try:
         os.chdir(MAPPING_PATH)
+        print("Changed Dir to {}".format(MAPPING_PATH))
     except FileNotFoundError:
         print("Path not found starting LanguageTool: ", MAPPING_PATH)
         raise
 
-    language_tool_command = f"java -cp languagetool-server.jar \
-              org.languagetool.server.HTTPServer \
-              --port {port} --allow-origin \"*\""
+    if not config_file:
+        language_tool_command = f"java -cp languagetool-server.jar \
+            org.languagetool.server.HTTPServer \
+            --port {port} --allow-origin \"*\""
+    else:
+        language_tool_command = f"java -cp languagetool-server.jar \
+            org.languagetool.server.HTTPServer \
+            --config {config_file} --port {port} --allow-origin \"*\""
 
     runner = subprocess.Popen(language_tool_command, shell=True)
     if not runner:

awe_languagetool/setup/data.py

@@ -47,8 +47,9 @@ def extra_build_commands(develop=False, install=False):
         # approach.                                                     #
         #################################################################
 
-        dir_name = \
-            resources.path('awe_languagetool', '')
+        # Since python 3.11, we cannot use 'path', but this:
+        with resources.as_file(resources.files('awe_languagetool')) as path:
+            dir_name = str(path)
 
         extension = ".zip"
         os.makedirs(dir_name, exist_ok=True)

languagetool.tmp.cfg

@@ -0,0 +1,13 @@
+# This config file comes from LT5.5 forums https://forum.languagetool.org/t/languagetool-5-5-slower-than-4-5/7396
+# The full list of variables which you can modify in this config file is described in their github repo:
+#  languagetool/languagetool-server/src/main/java/org/languagetool/server/HTTPServerConfig.java
+maxTextLength=50000
+cacheSize=3000
+pipelineCaching=true
+maxPipelinePoolSize=500
+pipelineExpireTimeInSeconds=3600
+pipelinePrewarming=true
+
+# maxCheckThreads=6 and maxWorkQueueSize=100 are good values for a server with 8 cores. For more cores, increase the values accordingly.
+maxCheckThreads=3
+maxWorkQueueSize=50

tests/__init__.py

@@ -0,0 +1,146 @@
+# --- [ Test: languagetoolClient.py ] ----------------------------------------------
+# 
+# Set of "sanity tests" for languagetool client. Ideally, these should be run on
+# a fresh install to ensure that methods like 'path()' are working for the 
+# current version of python.
+# 
+# Author: Caleb Scott (cwscott3@ncsu.edu)
+# ----------------------------------------------------------------------------------
+
+# --- [ IMPORTS ] ------------------------------------------------------------------
+
+import unittest
+import asyncio
+import time
+
+from awe_languagetool import languagetoolClient
+from awe_languagetool.languagetoolClient import mappingPathError
+
+# --- [ CONSTS ] -------------------------------------------------------------------
+
+# This describes the desired mapping path pointing to a rulemapping.json file
+EXPECTED_RULEMAP_PATH = "awe_languagetool/languagetool_rulemapping.json"
+
+TEXT_BASE = "test_texts/"
+LION_TEXT = TEXT_BASE + "lion.txt"
+
+CENSORSHIP_TEXTS = [
+    "censorship1.txt",
+    "censorship2.txt",
+    "censorship3.txt",
+    "censorship4.txt",
+    "censorship5.txt",
+    "censorship6.txt",
+    "censorship7.txt",
+    "censorship8.txt",
+    "censorship9.txt",
+    "censorship10.txt",
+    "censorship11.txt",
+    "censorship12.txt",
+    "censorship13.txt",
+    "censorship14.txt",
+    "censorship15.txt",
+    "censorship16.txt",
+    "censorship17.txt",
+    "censorship18.txt",
+    "censorship19.txt",
+    "censorship20.txt",
+    "censorship21.txt",
+    "censorship22.txt",
+    "censorship23.txt",
+    "censorship24.txt",
+    "censorship25.txt"
+]
+
+# --- [ SETUP ] --------------------------------------------------------------------
+
+# Make sure you are running the server before running client tests.
+# NOTE: you must ensure that the java server is not already running; otherwise,
+# this setup step will fail saying that the port is already in use.
+
+# languagetoolServer.runServer()
+
+# --- [ CLASSES ] ------------------------------------------------------------------
+
+class LanguageToolClientTest(unittest.TestCase):
+
+    def test_client_init(self):
+        """
+        Basic test to see if the client properly aligns its paths 
+        for files in its package.
+        """
+        try:
+            new_client = languagetoolClient.languagetoolClient()
+            self.assertTrue(
+                new_client.MAPPING_PATH.endswith(EXPECTED_RULEMAP_PATH)
+            )
+        except mappingPathError as e:
+            self.fail()
+
+    def test_client_timing_java_single(self):
+        """
+        Attempt to start up a client, and pass in a sample text.
+        We are measuring runtime of how well the server responds to a 
+        single request.
+
+        Precondition: languagetoolServer is running.
+        """
+        try:
+            client = languagetoolClient.languagetoolClient()
+            with open(LION_TEXT, 'r') as text_file:
+
+                # Grab the text
+                sample_text = text_file.read()
+
+                # Start timing benchmark
+                start = time.time()
+                output = asyncio.run(client.summarizeText(sample_text))
+                end = time.time()
+
+                # Show results
+                print()
+                print("---------[ TIMING BENCHMARK ]---------")
+                print(f"Time Elapsed: {end - start}")
+                print("--------------------------------------")
+        except mappingPathError as e:
+            self.fail()
+
+    def test_client_timing_java_multiple_sequential(self):
+        """
+        Attempt to start up a client, and pass in a sample text.
+        We are measuring runtime of how well the server responds to
+        many sequential requests.
+
+        Precondition: languagetoolServer is running.
+        """
+        try:
+            client = languagetoolClient.languagetoolClient()
+
+            # Pre-load all texts
+            all_texts = {}
+            index = 0
+            for text_filename in CENSORSHIP_TEXTS:
+                with open(TEXT_BASE + text_filename) as censorship_file:
+                    all_texts[index] = censorship_file.read()
+                    index = index + 1
+
+            # Make multiple requests
+            start = time.time()
+            output = asyncio.run(
+                client.summarizeMultipleTexts(
+                    list(all_texts.keys()), 
+                    list(all_texts.values())
+                )
+            )
+            end = time.time()
+
+            # Show results
+            print()
+            print("---------[ TIMING BENCHMARK ]---------")
+            print(f"Time Elapsed: {end - start}")
+            print("--------------------------------------")
+
+        except mappingPathError as e:
+            self.fail()
+
+# --- [ END ] ----------------------------------------------------------------------

tests/test_languagetoolClient.py

@@ -0,0 +1,38 @@
+# --- [ Test: languagetoolServer.py ] ----------------------------------------------
+# 
+# Set of "sanity tests" for languagetool server. Ideally, these should be run on
+# a fresh install to ensure that methods like 'path()' are working for the 
+# current version of python.
+# 
+# Author: Caleb Scott (cwscott3@ncsu.edu)
+# ----------------------------------------------------------------------------------
+
+# --- [ IMPORTS ] ------------------------------------------------------------------
+
+import unittest
+
+from awe_languagetool import languagetoolServer
+
+# --- [ CONSTS ] -------------------------------------------------------------------
+
+# This describes the desired mapping path directory (where languagetool jar is)
+EXPECTED_MAPPING_PATH = "awe_languagetool/LanguageTool5_5"
+
+# This describes the full path to the language tool jar file
+EXPECTED_LANGTOOL_PATH = "awe_languagetool/LanguageTool5_5/languagetool-server.jar"
+
+# --- [ SETUP ] --------------------------------------------------------------------
+
+# --- [ CLASSES ] ------------------------------------------------------------------
+
+class LanguageToolServerTest(unittest.TestCase):
+
+    def test_client_init(self):
+        try:
+            languagetoolServer.runServer()
+        except FileNotFoundError as e:
+            self.fail()
+        except ChildProcessError as e:
+            self.fail()
+
+# --- [ END ] ----------------------------------------------------------------------

tests/test_languagetoolServer.py

@@ -0,0 +1,7 @@
+I do not believe that censorship should be an option for people that find a book, magazine, or movie offensive. I believe that if a person decides to take a book off the shelf because he/she is offended, they are obviously paranoid. If a few cuss words are found in a book and a person doesn't want their children to read that book, then don't allow them to get it. Don't make all of the other people suffer. Censorship is an unnecessary solution to something that isn't even a problem. You know what they say, 'If it ain't broken, don't fix it.'
+
+      First of all, people have a variety of opinions. Some people think that violence is ok and some people don't think it's ok. If one person decides to put a book off the shelf because they find it offensive, what will the next person think? They may find violence perfectly acceptable and they might allow their kids to read the book. Then it would create a big conflict.
+      
+     Lastly, parents need to allow their kids to grow up and mature. They also need their kids to know the difference between right and wrong. If they don't read books or movies with violence or cuss words when they're young, what will they do when they discover it when they're older? As a child, it is a time to learn life lessons and begin to form clear distinctions between right and wrong. Shielding your children from it will only hinder their progress in figuring things out and learning to think for themselves as adults (Paterson, 2009).
+
+     Censorship is the same thing that Hitler did to prevent people to think for themselves and discover that his tactics were a great evil to society. In other words, censorship is part of fascism and should not even be considered in American society. The citizens of the U.S. have the responsibility of exposing the truth and teaching our children the differences between right and wrong. It's one of the ideals that help make this country great