rara-tools 0.0.12__tar.gz → 0.1.0__tar.gz

{rara_tools-0.0.12/rara_tools.egg-info → rara_tools-0.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: rara-tools
-Version: 0.0.12
+Version: 0.1.0
 Summary: Tools to support Kata's work.
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
@@ -18,6 +18,7 @@ Requires-Dist: iso639-lang
 Provides-Extra: testing
 Requires-Dist: pytest>=8.0; extra == "testing"
 Requires-Dist: pytest-order; extra == "testing"
+Dynamic: license-file
 
 # RaRa Tools
 

rara_tools-0.1.0/VERSION

@@ -0,0 +1 @@
+0.1.0

{rara_tools-0.0.12 → rara_tools-0.1.0}/rara_tools/constants/digitizer.py

@@ -1,4 +1,12 @@
+COMPONENT_KEY = "digitizer"
+
+
+class ModelTypes:
+    IMAGE_PROCESSOR = "image_processor"
+
+
 class StatusKeys:
+    DOWNLOAD_MODELS = "digitizer_download_models"
     CLEAN_UP = "digitizer_clean_up"
     ELASTICSEARCH_UPLOAD = "digitizer_elasticsearch_upload"
     UPLOAD = "s3_upload"
@@ -11,3 +19,7 @@ class Queue:
     DOWNLOAD = "download"
     FINISH = "finish"
     OCR = "ocr"
+
+
+class Tasks:
+    MODEL_UPDATE = "component_model_update"

{rara_tools-0.0.12 → rara_tools-0.1.0}/rara_tools/constants/general.py

@@ -1,4 +1,5 @@
 class Status:
+    SKIPPED = "SKIPPED"
     FAILED = "FAILED"
     PENDING = "PENDING"
     RUNNING = "RUNNING"

rara_tools-0.1.0/rara_tools/converters.py

@@ -0,0 +1,81 @@
+from .exceptions import SierraResponseConverterException
+
+
+class SierraResponseConverter:
+    """Converts a JSON response from the Sierra API to MARC-in-JSON format."""
+    
+    def __init__(self, response: dict):
+        if not isinstance(response, dict):
+            raise SierraResponseConverterException("Please provide a valid JSON response.")
+        self.response = response
+    
+    def _map_control_fields(self, field: dict) -> dict:
+        # for tags < 010, no subfields, instead one str value in "value" 
+        return {field["tag"]: field["value"]}
+    
+    def _map_data_fields(self, field: dict) -> dict:
+        """ Maps marc fields > 010. 
+
+        Args:
+            field (dict): Contains the marc tag and list with indicators and subfields.
+
+        Returns:
+            dict: standardised marc-in-json format.
+        """
+        
+        data = field["data"]
+        
+        # Order matters ind1, in2, subfields
+        field_data = {
+            "ind1": data.get("ind1", " "),
+            "ind2": data.get("ind2", " "),
+            "subfields": data.get("subfields", [])
+        }
+        
+        return {field["tag"]: field_data}
+    
+    def _is_marc21structured(self, field: dict) -> bool:
+        """Checks if the field is already structured according to MARC21 in JSON"""
+        return any(key.isdigit() for key in field.keys())
+        
+    
+    def _handle_field_type(self, field: dict) -> dict:
+        
+        if self._is_marc21structured(field):
+            return field
+        
+        if field.get("data"):
+            return self._map_data_fields(field)
+        
+        tag = field.get("tag")
+        
+        if not tag:
+            raise SierraResponseConverterException("Field is missing MARC21 tag.")
+        
+        if tag < "010":
+            return self._map_control_fields(field)
+        else:
+            return self._map_data_fields(field)
+        
+    def _convert_response(self) -> list:
+        entries = self.response.get("entries")
+        if not entries:
+            raise SierraResponseConverterException("No entries found in the response.")
+        
+        try:
+            return {"fields": [
+                {e["id"]: [
+                    self._handle_field_type(f) for f in e["marc"]["fields"]
+                    ]}
+                for e in entries
+            ]}
+            
+        except KeyError as e:
+            raise SierraResponseConverterException(f"Malformed response: missing key {e}")
+        
+    
+    def convert(self) -> list:
+        try:
+            return self._convert_response()
+        except Exception as e:
+            raise SierraResponseConverterException(f"An unexpected error occurred: {e}")

{rara_tools-0.0.12 → rara_tools-0.1.0}/rara_tools/exceptions.py

@@ -7,6 +7,10 @@ class S3InitException(Exception):
 class S3ConnectionException(Exception):
     """Raised S3 Bucket/Connection Error."""
 
+class S3DownloadException(Exception):
+    """Raised S3 Download Error."""
+
+
 class ElasticsearchException(Exception):
     """Raised Elasticsearch Error."""
 

{rara_tools-0.0.12 → rara_tools-0.1.0}/rara_tools/s3.py

@@ -1,11 +1,20 @@
+import logging
 import os
+import pathlib
+import time
 import uuid
 from typing import Any, Generator, List, Optional
 
-from minio import Minio
+from minio import Minio, S3Error
 
-from .exceptions import (S3ConnectionException, S3InitException,
-                         S3InputException)
+from .exceptions import (
+    S3ConnectionException,
+    S3InitException,
+    S3InputException,
+    S3DownloadException
+)
+
+logger = logging.getLogger("tools.s3")
 
 
 class S3Files:
@@ -76,9 +85,49 @@ class S3Files:
         list_of_objects = list(self.minio_client.list_objects(self.bucket, prefix=path, recursive=True))
         for minio_object in list_of_objects:
             full_path = os.path.join(download_dir, minio_object.object_name)
-            self.minio_client.fget_object(self.bucket, minio_object.object_name, full_path)
+            self._download_file(minio_object.object_name, full_path)
             yield full_path
 
+    def _download_file(self, path, download_dir=".", max_retries=3) -> str:
+        """Download a single file with retry and resume support."""
+        attempts = 0
+
+        while attempts < max_retries:
+            try:
+                stat = self.minio_client.stat_object(self.bucket, path)
+                file_size = stat.size
+                temp_path = download_dir + ".part"
+                pathlib.Path(temp_path).parent.mkdir(parents=True, exist_ok=True)
+
+                # Check if a partial file exists
+                downloaded_size = os.path.getsize(temp_path) if os.path.exists(temp_path) else 0
+
+                if downloaded_size >= file_size:
+                    os.rename(temp_path, download_dir)  # Rename to final filename
+                    logger.info(f"Completed: {path}")
+                    return str(pathlib.Path(download_dir) / path)
+
+                logger.info(f"Downloading {path} ({downloaded_size}/{file_size} bytes)...")
+
+                # Open file in append mode to resume download
+                with open(temp_path, "ab") as f:
+                    response = self.minio_client.get_object(self.bucket, path, offset=downloaded_size)
+                    for data in response.stream(32 * 1024):  # 32KB chunks
+                        f.write(data)
+                    response.close()
+                    response.release_conn()
+
+                os.rename(temp_path, download_dir)  # Rename temp to final
+                logger.info(f"Downloaded: {path}")
+                return str(pathlib.Path(download_dir) / path)
+
+            except S3Error as e:
+                logger.info(f"Error downloading {path}, attempt {attempts + 1}: {e}")
+                attempts += 1
+                time.sleep(2 ** attempts)  # Exponential backoff
+
+        raise S3DownloadException(f"Failed to download {path} after {max_retries} attempts.")
+
     def upload(self, path: str, prefix: Optional[str] = "") -> str:
         """Uploads file or folder to S3 bucket.
         :param: path str: Path to the file to upload in local file system.

{rara_tools-0.0.12 → rara_tools-0.1.0/rara_tools.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: rara-tools
-Version: 0.0.12
+Version: 0.1.0
 Summary: Tools to support Kata's work.
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
@@ -18,6 +18,7 @@ Requires-Dist: iso639-lang
 Provides-Extra: testing
 Requires-Dist: pytest>=8.0; extra == "testing"
 Requires-Dist: pytest-order; extra == "testing"
+Dynamic: license-file
 
 # RaRa Tools
 

{rara_tools-0.0.12 → rara_tools-0.1.0}/tests/test_converters.py

@@ -5,12 +5,22 @@ import pytest
 from rara_tools.converters import SierraResponseConverter
 from rara_tools.exceptions import SierraResponseConverterException
 
+import json
+
 root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 
 SIERRA_TEST_DATA_DIR = os.path.join(root, "tests", "test_data", "sierra")
 INPUT_DIR = os.path.join(SIERRA_TEST_DATA_DIR, "input")
 OUTPUT_DIR = os.path.join(SIERRA_TEST_DATA_DIR, "output")
 
+def compare_results(expected, converted):
+    return json.dumps(expected) == json.dumps(converted)
+    
+def read_json_file(file_path):
+    with open(file_path, "r") as f:
+        data = f.read()
+        return json.loads(data)
+
 example_res =  {
             "total": 100,
             "start": 50000,
@@ -27,6 +37,8 @@ example_res =  {
                             {
                                 # "tag": "100",
                                 "data": {
+                                    "ind1": "1",
+                                    "ind2": " ",
                                     "subfields": [
                                         {
                                             "code": "a",
@@ -36,18 +48,13 @@ example_res =  {
                                             "code": "d",
                                             "data": "1975-"
                                         }
-                                    ],
-                                    "ind1": "1",
-                                    "ind2": " "
+                                    ]
                                 }
                             },
             ]}}]}
 
 
-def read_json_file(file_path):
-    with open(file_path, "r") as f:
-        data = f.read()
-        return json.loads(data)
+
         
 def test_convert_bibs_response():
     response = read_json_file(os.path.join(INPUT_DIR, "bibs.json"))	
@@ -55,8 +62,9 @@ def test_convert_bibs_response():
     converter = SierraResponseConverter(response)
     data = converter.convert()
     
-    expected = read_json_file(os.path.join(OUTPUT_DIR, "bibs.json"))    
-    assert data == expected 
+    expected = read_json_file(os.path.join(OUTPUT_DIR, "bibs.json"))
+    
+    assert compare_results(expected, data)
     
     
 def test_convert_keywords_response():
@@ -67,9 +75,10 @@ def test_convert_keywords_response():
     converter = SierraResponseConverter(response)
     data = converter.convert()
     
+
     expected = read_json_file(os.path.join(OUTPUT_DIR, "keywords.json"))
     
-    assert data == expected
+    assert compare_results(expected, data)
     
     
 def test_convert_authorities_response():
@@ -82,7 +91,20 @@ def test_convert_authorities_response():
     
     expected = read_json_file(os.path.join(OUTPUT_DIR, "authorities.json"))
     
-    assert data == expected
+    assert compare_results(expected, data)
+    
+def test_converter_handles_marc_in_json_response():
+    """ Gracefully handle entries already in MARC-in-JSON format """
+    with open(os.path.join(INPUT_DIR, "bibsmarc.json"), "r") as f:
+        response = f.read()
+        response = json.loads(response)
+        
+    converter = SierraResponseConverter(response)
+    data = converter.convert()
+    
+    expected = read_json_file(os.path.join(OUTPUT_DIR, "bibsmarc.json"))
+    
+    assert compare_results(expected, data)
     
 def test_convert_with_wrong_format():
     with pytest.raises(SierraResponseConverterException):

{rara_tools-0.0.12 → rara_tools-0.1.0}/tests/test_elastic_vector_and_search_operations.py

@@ -15,7 +15,7 @@ TEST_DOCUMENTS = load_json("./tests/test_data/elastic_vectorized_docs.json")
 TEST_VECTOR_DATA = load_json("./tests/test_data/test_vector_data.json")
 TEST_VECTOR = TEST_VECTOR_DATA.get("vector")
     
-es_url = os.getenv("ELASTIC_TEST_URL", "http://rara-elastic.texta.ee:9200")#http://localhost:9200")
+es_url = os.getenv("ELASTIC_TEST_URL", "http://localhost:9200")
 ELASTIC = KataElastic(es_url)
 
 TEST_KNN_INDEX_NAME = "tools_knn_testing_index"

rara_tools-0.0.12/VERSION

@@ -1 +0,0 @@
-0.0.12

rara_tools-0.0.12/rara_tools/converters.py

@@ -1,41 +0,0 @@
-from .exceptions import SierraResponseConverterException
-
-
-class SierraResponseConverter:
-    """  Takes a JSON response from the Sierra API (https://tester.ester.ee/iii/sierra-api/swagger/index.html)
-    and converts it to MARC-in-JSON format.
-    
-    """
-    
-    def __init__(self, response: dict):
-        if not isinstance(response, dict):
-            raise SierraResponseConverterException("Please provide a valid JSON response.")
-        self.response = response
-        
-    def _map_field_data(self, field):
-        tag = field.get("tag")
-        if not tag:
-            raise SierraResponseConverterException("Field is missing a valid 'tag'.")
-        data = field.get("data", {})
-        return {tag: data}
-        
-    def _convert_response(self):
-        response = self.response
-        
-        entries = response.get("entries")
-        if not entries:
-            raise SierraResponseConverterException("No entries found in the response.")
-        
-        try:
-            fields = [self._map_field_data(f) for e in entries for f in e["marc"]["fields"]]
-        except KeyError as e:
-            raise SierraResponseConverterException(f"Missing expected MARC fields in the response: {e}")
-        
-        return {"fields": fields}
-            
-    def convert(self):
-        """Runner method, converts the response to MARC-in-JSON format with error handling."""
-        try:
-            return self._convert_response()
-        except Exception as e:
-            raise SierraResponseConverterException(f"An unexpected error occurred during conversion: {e}")