NLPPlus 0.2.1.dev0__cp312-cp312-win_amd64.whl

NLPPlus/__init__.py

@@ -0,0 +1,204 @@
+"""Python extension for NLP++ text analysis engine.
+
+Basic usage:
+
+    import NLPPlus
+    xml = NLPPlus.analyze("This is some text to be parsed")
+    print(xml)
+
+"""
+
+
+# start delvewheel patch
+def _delvewheel_patch_1_10_0():
+    import os
+    if os.path.isdir(libs_dir := os.path.abspath(os.path.join(os.path.dirname(__file__), os.pardir, 'nlpplus.libs'))):
+        os.add_dll_directory(libs_dir)
+
+
+_delvewheel_patch_1_10_0()
+del _delvewheel_patch_1_10_0
+# end delvewheel patch
+
+import json
+import logging
+from shutil import copytree, rmtree
+from tempfile import TemporaryDirectory
+from os import PathLike, getcwd
+from pathlib import Path
+from typing import Optional, Any
+import os
+import glob
+
+from .bindings import NLP_ENGINE  # type: ignore
+
+LOGGER = logging.getLogger("NLPPlus")
+
+
+def maybe_readfile(path: Path) -> Optional[str]:
+    """Bogus utility function to maybe read a file."""
+    if not path.exists():
+        return None
+    with open(path, "rt") as infh:
+        return infh.read()
+
+
+class EngineException(BaseException):
+    pass
+
+
+class Results:
+    """Various results produced by the NLP++ analyzer."""
+
+    def __init__(self, outtext: str, outdir: PathLike):
+        LOGGER.info("Reading output from %s", outdir)
+        self.output_text = outtext
+        self.outdir = Path(outdir)
+
+    @property
+    def final_tree(self) -> Optional[str]:
+        """The final parse tree, if any was produced."""
+        return maybe_readfile(self.outdir / "final.tree")
+
+    @property
+    def output_json(self) -> Optional[str]:
+        """The output JSON text, if any was produced."""
+        return maybe_readfile(self.outdir / "output.json")
+
+    @property
+    def output(self) -> Optional[Any]:
+        """The parsed output Json, if any was produced"""
+        output_json = self.output_json
+        if output_json is not None:
+            return json.loads(output_json)
+        return None
+
+
+class Engine:
+    """NLP++ Engine for a given working folder.
+
+    Args:
+
+      working_folder(optional, PathLike): Working folder for this
+           instance.  If None, a temporary directory will be created
+           and initialized with the default analyzers.  Otherwise,
+           this must contain an `analyzers` and a `data` folder,
+           unless `initialize` is `True`.
+      verbose(optional, bool): Be more verbose.
+      initialize(optional, bool): Initialize `working_folder` with
+           the default analyzers.
+    """
+    def __init__(
+        self,
+        working_folder: Optional[PathLike] = None,
+        analyzer_path: str = None,
+        verbose: bool = False,
+        initialize: bool = False,
+    ):
+        if working_folder is None:
+            self.tmpdir = TemporaryDirectory(prefix="NLPPlus-")
+            self.working_folder = Path(self.tmpdir.name)
+            initialize = True
+        else:
+            self.tmpdir = None
+            self.working_folder = Path(working_folder)
+        self.analyzer_path = None
+        if initialize:
+            copytree(
+                Path(__file__).parent / "analyzers", self.working_folder / "analyzers"
+            )
+            copytree(Path(__file__).parent / "data", self.working_folder / "data")
+            LOGGER.info("Initialized working folder in %s", self.working_folder)
+        if not (self.working_folder / "analyzers").is_dir():
+            raise EngineException(
+                f"analyzers directory not found in folder '{working_folder}'"
+            )
+        if not (self.working_folder / "data").is_dir():
+            raise EngineException(
+                f"data directory not found in folder '{working_folder}'"
+            )
+        self.engine = NLP_ENGINE(str(self.working_folder), silent=not verbose)
+
+    def analyze(self, text: str, analyzer_name: str, develop: bool = False) -> Results:
+        """Analyze text with the named analyzer."""
+        analyzer_name = Path(analyzer_name)
+        outdir = self.working_folder / "analyzers" / analyzer_name / "output"
+        if self.analyzer_path:
+            analyzer_name = Path(self.analyzer_path) / analyzer_name
+            outdir = Path(self.analyzer_path) / "analyzers" / analyzer_name / "output"
+        # Delete all files in the outdir
+        file_list = glob.glob(str(outdir / "*"))
+        for file_path in file_list:
+            os.remove(file_path)
+        outtext = self.engine.analyze(str(analyzer_name), text, develop)
+        return Results(outtext, outdir)
+    
+    def input_text(self, analyzer_name: str, file_name: str) -> str:
+        """Return the text from a file in the input directory."""
+        file_path = Path(self.analyzer_path) / analyzer_name / "input" / file_name
+        if not file_path.is_file():
+            raise EngineException(
+                f"File not found in input directory '{file_path}'"
+            )
+        with open(file_path, "rt", encoding="utf-8") as file:
+            text = file.read()
+        return text
+    
+    def set_analyzers_folder(self, analyzer_name: str):
+        """Set analyzers directory path."""
+        self.analyzer_path = analyzer_name
+
+    def copy_library_analyzers(self, to_dir: str, overwrite: bool=True):
+        """Copy the library files to a directory."""
+        copy_it = True
+
+        if os.path.exists(to_dir):
+            if overwrite:
+                rmtree(to_dir)
+            else:
+                copy_it = False
+
+        if copy_it:
+            copytree(
+                Path(__file__).parent / "analyzers", Path(to_dir)
+            )
+        self.analyzer_path = str(to_dir)
+
+
+engine = Engine()
+
+
+def set_working_folder(working_folder: Optional[str] = None, initialize: bool = False):
+    """Reinitialize the NLP++ engine with a different working folder.
+
+    Args:
+
+      working_folder(str): Working folder to use, or `None` to use the
+                           current working directory.
+      initialize(bool): Initialize the new working folder with the built-in
+                        analyzers and data.  (Optional, default=False)
+    """
+    global engine
+    if working_folder is None:
+        working_folder = getcwd()
+    engine = Engine(Path(working_folder), initialize=initialize)
+
+
+def copy_library_analyzers(analyzer_folder_path: str, overwrite=True):
+    """Run the analyzer named on the input string."""
+    engine.copy_library_analyzers(analyzer_folder_path, overwrite)
+
+
+def set_analyzers_folder(analyzer_folder_path: str):
+    """Run the analyzer named on the input string."""
+    engine.set_analyzers_folder(analyzer_folder_path)
+
+
+def analyze(text: str, parser: str = "parse-en-us", develop: bool = False) -> str:
+    """Run the analyzer named on the input string."""
+    return engine.analyze(text, parser, develop).output_text
+
+
+def input_text(analyzer_name: str, file_name: str):
+    """Return the text from a file in the input directory."""
+    return engine.intput_text(analyzer_name, file_name)

NLPPlus/_version.py

@@ -0,0 +1,21 @@
+# file generated by setuptools-scm
+# don't change, don't track in version control
+
+__all__ = ["__version__", "__version_tuple__", "version", "version_tuple"]
+
+TYPE_CHECKING = False
+if TYPE_CHECKING:
+    from typing import Tuple
+    from typing import Union
+
+    VERSION_TUPLE = Tuple[Union[int, str], ...]
+else:
+    VERSION_TUPLE = object
+
+version: str
+__version__: str
+__version_tuple__: VERSION_TUPLE
+version_tuple: VERSION_TUPLE
+
+__version__ = version = '0.2.1.dev0'
+__version_tuple__ = version_tuple = (0, 2, 1, 'dev0')

NLPPlus/analyzers/address-parser/README.md

@@ -0,0 +1,12 @@
+# ADDRESS PARSER  
+
+This is an NLP++ Analyzer that parses through text and identifies addresses(in USPS formats) listed anywhere in the text. 
+
+From the addresses, the following information is extracted(based on the type of address all or some of this information is extracted):
+House number, Street Number, Street Name, Street Suffix(like ST- for street),Street type(like Lane or Road), city, state, pincode, type of address.
+
+Additionally Rural Route and Highway Contract addresses are also parsed form which the information of Higway contract number/Rural Route and the Post box number is extracted.
+
+All this information is made available in the json format. A sample output can be found(for the sample text) in input/text.text_log/output.json.
+
+To run the analyzer, create a text file in input folder consisting of the text to be parsed.

NLPPlus/analyzers/address-parser/input/text.txt

@@ -0,0 +1,24 @@
+The address is 123 Main Street,345 Wyoming chanel Cityville, 64775 Country .
+Located at 456 Oak Street, Townsville,  24512 USA.
+John Smith resides in Apartment 789, Pine Towers,76 sde 45114 Canada.
+She's living at 987 Elm Lane, Riverside, Canada 14521.
+Dwelling at 654 Cedar Road, Hamletsville, 15451 India.
+My home address is 876 Pine Circle, Mountainside, 51873 Brazil.
+Here is my address - 234 Willow Street, Hilltop, 48929 South Africa.
+Find me at 543 Redwood Lane, Lakeside,38849 Mexico .
+His residence is 321 Maple Avenue, Villageton, 56234 UK .
+We stay at 789 Birch Place, Meadowville, New Zealand 34245.
+12 1ST ST NW
+HAMPTON IA 50441-1902
+BG LIGHT CO
+HC 72 BOX 293 A
+DULUTH MN 55811-9702
+ATTN MS LUCY MARTIN
+XYZ CORPORATION
+4321 OAK ST
+OAKTON MD 12345-6789
+You can reach me at 456 Maple Road, Suburbia, France 34245.
+Come to 987 Oak Lane, Riverside, Canada 34235.
+Check out 876 Pine Circle, Mountainside, Brazil 32345.
+Living in 543 Redwood Lane, Lakeside, Mexico 34415.
+At this location, you'll find 654 Cedar Road, Hamletsville, Australia 34213.

NLPPlus/analyzers/address-parser/kb/user/Country.dict

@@ -0,0 +1,196 @@
+Afghanistan country=1
+Albania country=1
+Algeria country=1
+Andorra country=1
+Angola country=1
+Antigua country=1
+Argentina country=1
+Armenia country=1
+Australia country=1
+Austria country=1
+Azerbaijan country=1
+Bahamas country=1
+Bahrain country=1
+Bangladesh country=1
+Barbados country=1
+Belarus country=1
+Belgium country=1
+Belize country=1
+Benin country=1
+Bhutan country=1
+Bolivia country=1
+Bosnia country=1
+Botswana country=1
+Brazil country=1
+Brunei country=1
+Bulgaria country=1
+Burkina country=1
+Burundi country=1
+Cabo country=1
+Cambodia country=1
+Cameroon country=1
+Canada country=1
+Central country=1
+Chad country=1
+Chile country=1
+China country=1
+Colombia country=1
+Comoros country=1
+Congo country=1
+Costa country=1
+Croatia country=1
+Cuba country=1
+Cyprus country=1
+Czechia country=1
+Democratic country=1
+Denmark country=1
+Djibouti country=1
+Dominica country=1
+Dominican country=1
+East country=1
+Ecuador country=1
+Egypt country=1
+El country=1
+Equatorial country=1
+Eritrea country=1
+Estonia country=1
+Eswatini country=1
+Ethiopia country=1
+Fiji country=1
+Finland country=1
+France country=1
+Gabon country=1
+Gambia country=1
+Georgia country=1
+Germany country=1
+Ghana country=1
+Greece country=1
+Grenada country=1
+Guatemala country=1
+Guinea country=1
+Guinea-Bissau country=1
+Guyana country=1
+Haiti country=1
+Honduras country=1
+Hungary country=1
+Iceland country=1
+India country=1
+Indonesia country=1
+Iran country=1
+Iraq country=1
+Ireland country=1
+Israel country=1
+Italy country=1
+Ivory country=1
+Jamaica country=1
+Japan country=1
+Jordan country=1
+Kazakhstan country=1
+Kenya country=1
+Kiribati country=1
+Kosovo country=1
+Kuwait country=1
+Kyrgyzstan country=1
+Laos country=1
+Latvia country=1
+Lebanon country=1
+Lesotho country=1
+Liberia country=1
+Libya country=1
+Liechtenstein country=1
+Lithuania country=1
+Luxembourg country=1
+Madagascar country=1
+Malawi country=1
+Malaysia country=1
+Maldives country=1
+Mali country=1
+Malta country=1
+Marshall country=1
+Mauritania country=1
+Mauritius country=1
+Mexico country=1
+Micronesia country=1
+Moldova country=1
+Monaco country=1
+Mongolia country=1
+Montenegro country=1
+Morocco country=1
+Mozambique country=1
+Myanmar country=1
+Namibia country=1
+Nauru country=1
+Nepal country=1
+Netherlands country=1
+New country=1
+Nicaragua country=1
+Niger country=1
+Nigeria country=1
+North country=1
+North country=1
+Norway country=1
+Oman country=1
+Pakistan country=1
+Palau country=1
+Panama country=1
+Papua country=1
+Paraguay country=1
+Peru country=1
+Philippines country=1
+Poland country=1
+Portugal country=1
+Qatar country=1
+Romania country=1
+Russia country=1
+Rwanda country=1
+Saint country=1
+Saint country=1
+Saint country=1
+Samoa country=1
+San country=1
+Sao country=1
+Saudi country=1
+Senegal country=1
+Serbia country=1
+Seychelles country=1
+Sierra country=1
+Singapore country=1
+Slovakia country=1
+Slovenia country=1
+Solomon country=1
+Somalia country=1
+South country=1
+South country=1
+South country=1
+Spain country=1
+Sri country=1
+Sudan country=1
+Suriname country=1
+Sweden country=1
+Switzerland country=1
+Syria country=1
+Taiwan country=1
+Tajikistan country=1
+Tanzania country=1
+Thailand country=1
+Togo country=1
+Tonga country=1
+Trinidad country=1
+Tunisia country=1
+Turkey country=1
+Turkmenistan country=1
+Tuvalu country=1
+Uganda country=1
+Ukraine country=1
+UnitedStates country=1
+US country=1
+USA country=1
+Uruguay country=1
+Uzbekistan country=1
+Vanuatu country=1
+Vatican country=1
+Venezuela country=1
+Vietnam country=1
+Yemen country=1
+Zambia country=1
+Zimbabwe country=1

NLPPlus/analyzers/address-parser/kb/user/address-synonym.dict

@@ -0,0 +1,14 @@
+address address=1
+located address=1
+resides address=1
+living address=1
+dwelling address=1
+dwells address=1
+lives address=1
+resides address=1
+residence address=1
+stay address=1
+staying address=1
+location address=1
+stayed address=1
+resided address=1