Merge pull request #536 from axif0/all_in_one

Feat: Translation cmd for scribe-data
scribe-org · Jan 4, 2025 · 4132f57 · 4132f57
2 parents b27af41 + f28a176
commit 4132f57
Show file tree

Hide file tree

Showing 31 changed files with 1,679 additions and 640 deletions.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -300,13 +300,18 @@ def example_function(argument: argument_type) -> return_type:
 
     Parameters
     ----------
-        argument: argument_type
-            Description of your argument.
+    argument : argument_type
+        Description of your argument.
 
     Returns
     -------
-        return_value : return_type
-            Description of your return value.
+    return_value : return_type
+        Description of your return value.
+
+    Raises
+    ------
+    ErrorType
+        Description of the error and the condition that raises it.
     """
 
     ...

diff --git a/requirements.txt b/requirements.txt
@@ -18,3 +18,4 @@ ruff>=0.3.3
 SPARQLWrapper>=2.0.0
 sphinx-rtd-theme>=3.0.0
 tqdm==4.66.4
+orjson>=3.10.12
diff --git a/src/scribe_data/check/check_project_metadata.py b/src/scribe_data/check/check_project_metadata.py
@@ -88,16 +88,16 @@ def get_missing_languages(
 
     Parameters
     ----------
-        reference_languages : dict
-            A dictionary of languages from the reference source.
+    reference_languages : dict
+        A dictionary of languages from the reference source.
 
-        target_languages : dict
-            A dictionary of languages from the target source to check for missing entries.
+    target_languages : dict
+        A dictionary of languages from the target source to check for missing entries.
 
     Returns
     -------
-        list[str]
-            A list of languages and sub-languages that are in target_languages but not in reference_languages.
+    list[str]
+        A list of languages and sub-languages that are in target_languages but not in reference_languages.
     """
     missing_languages = []
     reference_keys = reference_languages.keys()
@@ -130,17 +130,17 @@ def validate_language_properties(languages_dict: dict) -> dict:
 
     Parameters
     ----------
-        languages_dict : dict
-            A dictionary where each key is a language, and the value is another dictionary containing details about the language. If the language has sub-languages, they are stored under the 'sub_languages' key.
+    languages_dict : dict
+        A dictionary where each key is a language, and the value is another dictionary containing details about the language. If the language has sub-languages, they are stored under the 'sub_languages' key.
 
     Returns
     -------
-        dict: A dictionary with two lists:
-            - "missing_qids": Languages or sub-languages missing the 'qid' property.
-            - "missing_isos": Languages or sub-languages missing the 'iso' property.
+    dict: A dictionary with two lists:
+        - "missing_qids": Languages or sub-languages missing the 'qid' property.
+        - "missing_isos": Languages or sub-languages missing the 'iso' property.
 
-            Each entry in these lists is in the format "parent_language - sub_language" for sub-languages,
-            or simply "parent_language" for the parent languages.
+        Each entry in these lists is in the format "parent_language - sub_language" for sub-languages,
+        or simply "parent_language" for the parent languages.
     """
     missing_qids = []
     missing_isos = []

diff --git a/src/scribe_data/check/check_project_structure.py b/src/scribe_data/check/check_project_structure.py
@@ -48,24 +48,24 @@ def check_for_sparql_files(folder_path, data_type, language, subdir, missing_que
 
     Parameters
     ----------
-        folder_path : str
-            The path to the data-type folder.
+    folder_path : str
+        The path to the data-type folder.
 
-        data_type : str
-            The name of the data type being checked.
+    data_type : str
+        The name of the data type being checked.
 
-        language : str
-            The name of the language being processed.
+    language : str
+        The name of the language being processed.
 
-        subdir : str or None
-            The name of the sub-directory (for languages with sub-dialects), or None.
+    subdir : str or None
+        The name of the sub-directory (for languages with sub-dialects), or None.
 
-        missing_queries : list
-            A list to which missing SPARQL query files will be appended.
+    missing_queries : list
+        A list to which missing SPARQL query files will be appended.
 
     Returns
     -------
-        bool: True if at least one .sparql file is found, False otherwise.
+    bool: True if at least one .sparql file is found, False otherwise.
     """
     sparql_files = [f for f in os.listdir(folder_path) if f.endswith(".sparql")]
 

diff --git a/src/scribe_data/check/check_pyicu.py b/src/scribe_data/check/check_pyicu.py
@@ -27,8 +27,8 @@
 from pathlib import Path
 
 import pkg_resources
+import questionary
 import requests
-from questionary import confirm
 
 
 def check_if_pyicu_installed():
@@ -90,15 +90,15 @@ def download_wheel_file(wheel_url, output_dir):
 
     Parameters
     ----------
-        wheel_url : str
-            The URL of the wheel file to download.
+    wheel_url : str
+        The URL of the wheel file to download.
 
-        output_dir : str
-            The directory to save the downloaded file.
+    output_dir : str
+        The directory to save the downloaded file.
 
     Returns
     -------
-        str : path to the downloaded wheel file.
+    str : path to the downloaded wheel file.
     """
     response = requests.get(wheel_url)
     response.raise_for_status()  # raise an error for bad responses
@@ -118,18 +118,18 @@ def find_matching_wheel(wheels, python_version, architecture):
 
     Parameters
     ----------
-        wheels : list
-            The list of available wheels.
+    wheels : list
+        The list of available wheels.
 
-        python_version : str
-            The Python version (e.g., 'cp311').
+    python_version : str
+        The Python version (e.g., 'cp311').
 
-        architecture : str
-            The architecture type (e.g., 'win_amd64').
+    architecture : str
+        The architecture type (e.g., 'win_amd64').
 
     Returns
     -------
-        str : The download URL of the matching wheel or None if not found.
+    str : The download URL of the matching wheel or None if not found.
     """
     return next(
         (
@@ -148,8 +148,7 @@ def check_and_install_pyicu():
         # Fetch available wheels from GitHub to estimate download size.
         wheels, total_size_mb = fetch_wheel_releases()
 
-        # Use questionary to ask for user confirmation
-        user_wants_to_proceed = confirm(
+        user_wants_to_proceed = questionary.confirm(
             f"{package_name} is not installed.\nScribe-Data can install the package and the needed dependencies."
             f"\nApproximately {total_size_mb:.2f} MB will be downloaded.\nDo you want to proceed?"
         ).ask()

diff --git a/src/scribe_data/check/check_query_forms.py b/src/scribe_data/check/check_query_forms.py
@@ -57,18 +57,18 @@ def extract_forms_from_sparql(file_path: Path) -> str:
 
     Parameters
     ----------
-        file_path : Path
-            The path to the SPARQL query file from which to extract forms.
+    file_path : Path
+        The path to the SPARQL query file from which to extract forms.
 
     Returns
     -------
-        query_form_dict : dict
-            The file path with form labels of the query and their respective QIDs.
+    query_form_dict : dict
+        The file path with form labels of the query and their respective QIDs.
 
     Raises
     ------
-        FileNotFoundError
-            If the specified file does not exist.
+    FileNotFoundError
+        If the specified file does not exist.
     """
     optional_pattern = r"\s\sOPTIONAL\s*\{([^}]*)\}"
     try:
@@ -95,13 +95,13 @@ def extract_form_rep_label(form_text: str):
 
     Parameters
     ----------
-        form_text : str
-            The text that defines the form within the query.
+    form_text : str
+        The text that defines the form within the query.
 
     Returns
     -------
-        str
-            The label of the form representation.
+    str
+        The label of the form representation.
     """
     onto_rep_pattern = r"ontolex:representation .* ;"
     if line_match := re.search(pattern=onto_rep_pattern, string=form_text):
@@ -119,13 +119,13 @@ def decompose_label_features(label):
 
     Parameters
     ----------
-        label : str
-            The concatenated label string composed of several grammatical features.
+    label : str
+        The concatenated label string composed of several grammatical features.
 
     Returns
     -------
-        list
-            A list of grammatical features extracted from the label in their original order.
+    list
+        A list of grammatical features extracted from the label in their original order.
     """
     components = re.findall(r"[A-Za-z][^A-Z]*", label)
     valid_components = []
@@ -157,13 +157,13 @@ def extract_form_qids(form_text: str):
 
     Parameters
     ----------
-        form_text : str
-            The text that defines the form within the query.
+    form_text : str
+        The text that defines the form within the query.
 
     Returns
     -------
-        list[str]
-            All QIDS that make up the form.
+    list[str]
+        All QIDS that make up the form.
     """
     qids_pattern = r"wikibase:grammaticalFeature .+ \."
     if match := re.search(pattern=qids_pattern, string=form_text):
@@ -179,13 +179,13 @@ def check_form_label(form_text: str):
 
     Parameters
     ----------
-        form_text : str
-            The text that defines the form within the query.
+    form_text : str
+        The text that defines the form within the query.
 
     Returns
     -------
-        bool
-            Whether the form and its current representation label match (repForm and rep).
+    bool
+        Whether the form and its current representation label match (repForm and rep).
     """
     form_label_line_pattern = r"\?lexeme ontolex:lexicalForm .* \."
 
@@ -221,13 +221,13 @@ def check_query_formatting(form_text: str):
 
     Parameters
     ----------
-        query_text : str
-            The SPARQL query text to check.
+    query_text : str
+        The SPARQL query text to check.
 
     Returns
     -------
-        bool
-            Whether there are formatting errors with the query.
+    bool
+        Whether there are formatting errors with the query.
     """
     # Check for spaces before commas that should not exist.
     if re.search(r"\s,", form_text):
@@ -249,13 +249,13 @@ def return_correct_form_label(qids: list):
 
     Parameters
     ----------
-        qids : list[str]
-            All QIDS that make up the form.
+    qids : list[str]
+        All QIDS that make up the form.
 
     Returns
     -------
-        correct_label : str
-            The label for the representation given the QIDs.
+    correct_label : str
+        The label for the representation given the QIDs.
     """
     if not qids:
         return "Invalid query formatting found"
@@ -289,14 +289,14 @@ def validate_forms(query_text: str) -> str:
 
     Parameters
     ----------
-        query_file : str
-            The SPARQL query text as a string.
+    query_file : str
+        The SPARQL query text as a string.
 
     Returns
     -------
-        str
-            Error message if there are any issues with the order of variables or forms,
-            otherwise an empty string.
+    str
+        Error message if there are any issues with the order of variables or forms,
+        otherwise an empty string.
     """
     select_pattern = r"SELECT\s+(.*?)\s+WHERE"
 
@@ -376,13 +376,13 @@ def check_docstring(query_text: str) -> bool:
 
     Parameters
     ----------
-        query_text : str
-            The SPARQL query's text to be checked.
+    query_text : str
+        The SPARQL query's text to be checked.
 
     Returns
     -------
-        bool
-            True if the docstring is correctly formatted.
+    bool
+        True if the docstring is correctly formatted.
     """
     # Split the text into lines.
     query_lines = query_text.splitlines(keepends=True)
@@ -418,14 +418,14 @@ def check_forms_order(query_text):
 
     Parameters
     ----------
-        query_text : str
-            The SPARQL query text containing the SELECT statement with variables.
+    query_text : str
+        The SPARQL query text containing the SELECT statement with variables.
 
     Returns
     -------
-        list or bool
-            A sorted list of variables if the ordering differs from the original,
-            otherwise a boolean indicating that the order matches.
+    list or bool
+        A sorted list of variables if the ordering differs from the original,
+        otherwise a boolean indicating that the order matches.
     """
     select_pattern = r"SELECT\s+(.*?)\s+WHERE"
 
@@ -496,14 +496,14 @@ def check_optional_qid_order(query_file: str) -> str:
 
     Parameters
     ----------
-        query_file : str
-            The path to the SPARQL query file to be checked.
+    query_file : str
+        The path to the SPARQL query file to be checked.
 
     Returns
     -------
-        str
-            A formatted string with details on any order mismatches in the QIDs, or an empty
-            string if all QIDs are correctly ordered.
+    str
+        A formatted string with details on any order mismatches in the QIDs, or an empty
+        string if all QIDs are correctly ordered.
     """
     forms = extract_forms_from_sparql(query_file)
     error_messages = []