oarepo-runtime 1.7.2__py3-none-any.whl → 1.9.0__py3-none-any.whl

oarepo_runtime/datastreams/types.py

@@ -155,11 +155,11 @@ class StreamEntry:
     def __str__(self):
         ret = [
             f"Entry #{self.seq}: id {self.id or 'not yet set'}, filtered: {self.filtered}, deleted: {self.deleted}",
-            f"Content:",
+            "Content:",
             textwrap.indent(
                 json.dumps(self.entry, ensure_ascii=False, indent=4), "    "
             ),
-            f"Context:",
+            "Context:",
             textwrap.indent(
                 json.dumps(self.context, ensure_ascii=False, indent=4), "    "
             ),
@@ -238,14 +238,14 @@ class DataStreamCallback:
     def batch_started(self, batch):
         log.info("Batch started: %s", batch)
 
-    def batch_finished(self, batch):
-        log.info("Batch finished: %s", batch)
+    def batch_finished(self, batch: StreamBatch):
+        log.info("Batch finished: %s", batch.seq)
         for err in batch.errors:
-            log.error("Failed batch: %s: %s", err, batch)
+            log.error("Failed batch: %s: %s", err, batch.seq)
         if self.log_error_entry:
             for entry in batch.entries:
                 if entry.errors:
-                    log.error("Failed entry: %s", entry)
+                    log.error("Failed entry: %s in batch %s", entry, batch.seq)
 
     def reader_error(self, reader, exception):
         log.error("Reader error: %s: %s", reader, exception)

oarepo_runtime/records/systemfields/__init__.py

@@ -1,12 +1,27 @@
-from .icu import ICUField, ICUSortField, ICUSuggestField
+from .icu import (
+    FulltextIndexField,
+    ICUField,
+    ICUSearchField,
+    ICUSortField,
+    ICUSuggestField,
+    TermIndexField,
+)
 from .mapping import MappingSystemFieldMixin, SystemFieldDumperExt
-from .selectors import FirstItemSelector, PathSelector, Selector, FilteredSelector, MultiSelector
+from .selectors import (
+    FilteredSelector,
+    FirstItemSelector,
+    MultiSelector,
+    PathSelector,
+    Selector,
+)
 from .synthetic import SyntheticSystemField
 
 __all__ = (
     "ICUField",
     "ICUSuggestField",
     "ICUSortField",
+    "ICUSearchField",
+    "FulltextIndexField",
     "MappingSystemFieldMixin",
     "SystemFieldDumperExt",
     "SyntheticSystemField",
@@ -15,4 +30,5 @@ __all__ = (
     "FirstItemSelector",
     "FilteredSelector",
     "MultiSelector",
+    "TermIndexField",
 )

oarepo_runtime/records/systemfields/icu.py

@@ -1,3 +1,4 @@
+from abc import abstractmethod, abstractproperty
 from functools import cached_property
 from typing import Dict
 
@@ -8,17 +9,16 @@ from oarepo_runtime.records.relations.lookup import lookup_key
 from oarepo_runtime.records.systemfields.mapping import MappingSystemFieldMixin
 
 
-class ICUField(MappingSystemFieldMixin, SystemField):
+class ICUBase(MappingSystemFieldMixin, SystemField):
     """
-    A system field that acts as an opensearch "proxy" to another field.
-    It creates a top-level mapping field with the same name and copies
-    content of {another field}.language into {mapping field}.language.
-
-    The language accessor can be modified by overriding get_values method.
+    Base class for ICU system fields.
+    It provides the basic functionality for ICU fields, such as
+    getting the attribute name and handling the key.
     """
 
-    def __init__(self, *, source_field, key=None):
-        super().__init__(key)
+    def __init__(self, source_field=None, key=None):
+        super().__init__(key=key)
+        self._attr_name = key or self.__class__.__name__.lower()
         self.source_field = source_field
 
     @cached_property
@@ -37,26 +37,71 @@ class ICUField(MappingSystemFieldMixin, SystemField):
         ret = []
         for l in lookup_key(data, f"{self.source_field}"):
             if isinstance(l.value, str):
+                # take single value as being always the the language provided
                 ret.append(l.value)
             elif isinstance(l.value, dict):
+                # expected to be {"cs": "", "en": ""}
                 val = l.value.get(language)
                 if val:
                     ret.append(val)
+                elif "lang" in l.value:
+                    # for [{"lang": "", "value": ""}, ...] we get each item separately
+                    # that's why we do not iterate over l.value
+                    if l.value["lang"] == language:
+                        ret.append(l.value["value"])
         return ret
 
+    @abstractproperty
+    def mapping(self):
+        """
+        The mapping for the field. It should return a dictionary with the
+        mapping for the field, based on the current configuration of the application.
+        """
+        raise NotImplementedError("Subclasses must implement the mapping property.")
+
+    @abstractmethod
     def search_dump(self, data, record):
-        ret = {}
-        for lang in self.languages:
-            ret[lang] = self.get_values(data, lang)
-        data[self.attr_name] = ret
+        """
+        Dump custom field. This method should be implemented by subclasses
+        to provide the functionality for dumping the field data into the
+        OpenSearch data structure.
+        """
+        raise NotImplementedError("Subclasses must implement the search_dump method.")
 
     def search_load(self, data, record_cls):
+        """
+        Just remove the field from the data on load.
+        """
         data.pop(self.attr_name, None)
 
     def __get__(self, instance, owner):
         return self
 
 
+class ICUField(ICUBase):
+    """
+    A system field that acts as an opensearch "proxy" to another field.
+    It creates a top-level mapping field with the same name and copies
+    content of {another field}.language into {mapping field}.language.
+
+    The language accessor can be modified by overriding get_values method.
+    """
+
+    def __init__(self, *, source_field, key=None):
+        super().__init__(source_field=source_field, key=key)
+
+    def search_dump(self, data, record):
+        ret = {}
+        for lang in self.languages:
+            r = self.get_values(data, lang)
+            if r:
+                # if the language is not empty, add it to the result
+                # otherwise do not add it at all to safe transport
+                ret[lang] = r
+        if ret:
+            data[self.attr_name] = ret
+
+
 class ICUSortField(ICUField):
     """
     A field that adds icu sorting field
@@ -132,23 +177,24 @@ class ICUSuggestField(ICUField):
         }
 
 
-class ICUSearchField(ICUField):
-    """
-    A field that adds stemming-aware search field
-    """
+class ICUSearchAnalyzerMixin:
 
     default_stemming_analyzers = {
         "stemming_analyzer_cs": {
             "tokenizer": "standard",
-            "filter": ["stemming_filter_cs"],
+            "filter": ["stemming_filter_cs", "lowercase"],
         },
         "stemming_analyzer_en": {
             "tokenizer": "standard",
-            "filter": ["stemming_filter_en"],
+            "filter": ["stemming_filter_en", "lowercase"],
         },
         "ascii_folding_analyzer": {
             "tokenizer": "standard",
-            "filter": ["ascii_folding_filter"],
+            "filter": ["ascii_folding_filter", "lowercase"],
+        },
+        "lowercase_analyzer": {
+            "tokenizer": "standard",
+            "filter": ["lowercase"],
         },
     }
 
@@ -166,8 +212,31 @@ class ICUSearchField(ICUField):
         "ascii_folding_filter": {"type": "asciifolding", "preserve_original": True},
     }
 
-    def __init__(self, source_field, key=None):
+    @property
+    def mapping_settings(self):
+        return {
+            "analysis": {
+                "analyzer": current_app.config.get(
+                    "OAREPO_ICU_SEARCH_ANALYZERS", self.default_stemming_analyzers
+                ),
+                "filter": current_app.config.get(
+                    "OAREPO_ICU_SEARCH_FILTERS", self.default_stemming_filters
+                ),
+            }
+        }
+
+
+class ICUSearchField(ICUSearchAnalyzerMixin, ICUField):
+    """
+    A field that adds stemming-aware search field for multilingual data (
+        e.g. data that contains {"cs": "...", "en": "..."}
+        or [{"lang": "cs", "value": "..."}, ...]
+    )
+    """
+
+    def __init__(self, source_field, key=None, boost=1):
         super().__init__(source_field=source_field, key=key)
+        self.boost = boost
 
     @property
     def mapping(self):
@@ -180,17 +249,22 @@ class ICUSearchField(ICUField):
                         "search",
                         {
                             "type": "text",
-                            "boost": 1,
+                            "boost": 1 * self.boost,
                             "fields": {
                                 "stemmed": {
                                     "type": "text",
                                     "analyzer": f"stemming_analyzer_{lang}",
-                                    "boost": 0.5,
+                                    "boost": 0.5 * self.boost,
+                                },
+                                "lowercase": {
+                                    "type": "text",
+                                    "boost": 0.8 * self.boost,
+                                    "analyzer": "lowercase_analyzer",
                                 },
                                 "ascii_folded": {
                                     "type": "text",
                                     "analyzer": "ascii_folding_analyzer",
-                                    "boost": 0.3,
+                                    "boost": 0.3 * self.boost,
                                 },
                             },
                         },
@@ -200,15 +274,98 @@ class ICUSearchField(ICUField):
             },
         }
 
+    def get_values(self, data, language):
+        return super().get_values(data, language=language)
+
+
+class SingleLanguageSearchField(ICUSearchAnalyzerMixin, ICUBase):
+    """
+    A base class for single-language search fields - that is, data contain a text
+    value in a pre-defined, single language.
+    """
+
+    def __init__(self, *, source_field, key=None, language=None, boost=1):
+        super().__init__(source_field=source_field, key=key)
+        self.language = language
+        self.boost = boost
+
+    def search_dump(self, data, record):
+        """Dump custom field."""
+        ret = self.get_values(data, language=self.language)
+        if ret:
+            data[self.attr_name] = ret
+
+
+class FulltextIndexField(SingleLanguageSearchField):
+    """
+    A system field that makes the field searchable in OpenSearch,
+    regardless if it is indexed/analyzed, embedded in Nested or not.
+
+    It creates a top-level mapping field and copies
+    content of {source_field} into it. It also provides the correct mapping
+    for the field based on the current configuration of the application.
+
+    Unlike the ICU, this field is a single-language and the language should
+    be provided when initializing the field.
+    It defaults to the BABEL_DEFAULT_LOCALE if not provided.
+    """
+
     @property
-    def mapping_settings(self):
-        return {
-            "analysis": {
-                "analyzer": current_app.config.get(
-                    "OAREPO_ICU_SEARCH_ANALYZERS", self.default_stemming_analyzers
-                ),
-                "filter": current_app.config.get(
-                    "OAREPO_ICU_SEARCH_FILTERS", self.default_stemming_filters
-                ),
+    def mapping(self):
+        language = self.language or current_app.config.get("BABEL_DEFAULT_LOCALE", "en")
+        mapping_settings = self.languages.get(language, None)
+        if mapping_settings:
+            mapping_settings = mapping_settings.get("search")
+        if not mapping_settings:
+            mapping_settings = {
+                "type": "text",
+                "boost": 1 * self.boost,
+                "fields": {
+                    "stemmed": {
+                        "type": "text",
+                        "analyzer": f"stemming_analyzer_{language}",
+                        "boost": 0.5 * self.boost,
+                    },
+                    "lowercase": {
+                        "type": "text",
+                        "boost": 0.8 * self.boost,
+                        "analyzer": "lowercase_analyzer",
+                    },
+                    "ascii_folded": {
+                        "type": "text",
+                        "analyzer": "ascii_folding_analyzer",
+                        "boost": 0.3 * self.boost,
+                    },
+                },
             }
+
+        return {self.attr_name: mapping_settings}
+
+    def search_load(self, data, record_cls):
+        """Load custom field."""
+        data.pop(self.attr_name, None)
+
+
+class TermIndexField(SingleLanguageSearchField):
+    """
+    A system field that makes the field searchable in OpenSearch,
+    regardless if it is indexed/analyzed, embedded in Nested or not.
+
+    It creates a top-level mapping field and copies
+    content of {source_field} into it. It also provides the correct mapping
+    for the field based on the current configuration of the application.
+
+    Unlike the ICU, this field is a single-language and the language should
+    be provided when initializing the field.
+    It defaults to the BABEL_DEFAULT_LOCALE if not provided.
+    """
+
+    @property
+    def mapping(self):
+        mapping_settings = {
+            "type": "keyword",
+            "boost": 1 * self.boost,
+            "ignore_above": 256,
         }
+
+        return {self.attr_name: mapping_settings}

oarepo_runtime/records/systemfields/mapping.py

@@ -16,12 +16,10 @@ class MappingSystemFieldMixin:
     def dynamic_templates(self):
         return []
 
-    @classmethod
-    def search_dump(cls, data, record):
+    def search_dump(self, data, record):
         """Dump custom field."""
 
-    @classmethod
-    def search_load(cls, data, record_cls):
+    def search_load(self, data, record_cls):
         """Load custom field."""
 
 

{oarepo_runtime-1.7.2.dist-info → oarepo_runtime-1.9.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oarepo-runtime
-Version: 1.7.2
+Version: 1.9.0
 Summary: A set of runtime extensions of Invenio repository
 Description-Content-Type: text/markdown
 License-File: LICENSE

{oarepo_runtime-1.7.2.dist-info → oarepo_runtime-1.9.0.dist-info}/RECORD

@@ -31,7 +31,7 @@ oarepo_runtime/datastreams/json.py,sha256=OkIkGKUawtYoGelyS5V92DVk7Ei7SlkMMny2Ue
 oarepo_runtime/datastreams/semi_asynchronous.py,sha256=kNc6BBnV6oFoY9kHgf5l8fd1wibRfI0dwyzLtu4fmUA,2940
 oarepo_runtime/datastreams/synchronous.py,sha256=t5lfnMkLqy3jK5zMl-nIuA0HlMPiHGjwCqZ8XQP-3GM,2595
 oarepo_runtime/datastreams/transformers.py,sha256=q5KzHPl2kJg7HP1BtKJ7F_UMqg_7L1ZGDX0O7s8D6UI,521
-oarepo_runtime/datastreams/types.py,sha256=KZjblc3T_UFFW7LrMDmiR8lqVf86V484LAHj6yg05EI,9908
+oarepo_runtime/datastreams/types.py,sha256=ZeXX5rVPwCpzJ0drl5ApgN1l43G-aSjStQbt_vdJgRk,9950
 oarepo_runtime/datastreams/utils.py,sha256=w24qRRw1ZawPFbgqtxhpYnxPnghSn0pp49JLLy8wTms,4280
 oarepo_runtime/datastreams/readers/__init__.py,sha256=P1n3llZQ3AFHnSPbeT1VaCJcEtRFz9AbHfjkZv5LG7s,1103
 oarepo_runtime/datastreams/readers/attachments.py,sha256=A7EC1TqyTHG-go5DIaRotlBSOm6o9hGqAKyVVAceCRU,1956
@@ -70,11 +70,11 @@ oarepo_runtime/records/relations/base.py,sha256=ESTwj0-eT8HRTJ8QcE5fmqzjOjBFHpQq
 oarepo_runtime/records/relations/internal.py,sha256=OTp8iJqyl80sWDk0Q0AK42l6UsxZDABspVU_GwWza9o,1556
 oarepo_runtime/records/relations/lookup.py,sha256=wi3jPfOedazOmhOMrgu50PUETc1jfSdpmjK0wvOFsEM,848
 oarepo_runtime/records/relations/pid_relation.py,sha256=eojw5uIo5zXmJGge_bj6Wj2njCRY5S4o4B_h_HFyaDY,3901
-oarepo_runtime/records/systemfields/__init__.py,sha256=LL1R64RUakA_4r0IkTq9MtwqD5eV-AQaj5u96zkWa74,533
+oarepo_runtime/records/systemfields/__init__.py,sha256=fU1IiZ24AX92TQi4GZsuBjT9JvtSklk5ANXbabcRkZg,709
 oarepo_runtime/records/systemfields/featured_file.py,sha256=MbSaYR130_o5S9gEOblnChq-PVK4xGPGpSCrzwG3cwc,1720
 oarepo_runtime/records/systemfields/has_draftcheck.py,sha256=4JkMEefPLpqtPtlTgK3UT0KzTRgyw5_Qtkss2qcz5xk,1643
-oarepo_runtime/records/systemfields/icu.py,sha256=sSGAgi5WhsAY4cCBL7-7nMpvHAuctpW8Y8vRExHQUfk,6738
-oarepo_runtime/records/systemfields/mapping.py,sha256=tXOK_jkdY1pOUO7_VfChfDNB8UTi21GUXaidpugTnO8,1017
+oarepo_runtime/records/systemfields/icu.py,sha256=LqttWKdbtBw2FIIrtfDzn3-8jj-e0blm3Oz_XjFtiJw,12625
+oarepo_runtime/records/systemfields/mapping.py,sha256=_Xr9iFJg5Gu198o4rH8bD3TQkRE4AVKXfZBTHa6QHh0,985
 oarepo_runtime/records/systemfields/owner.py,sha256=dYRVBinniW7ECHuSnTAjeN6x1KhhJtNR9vxmD1KswMs,3805
 oarepo_runtime/records/systemfields/record_status.py,sha256=U3kem4-JkNsT17e0iAl3HIAZ2MvO5lY_0U757aZvTKE,935
 oarepo_runtime/records/systemfields/selectors.py,sha256=Q9jE1smSN3heT2LIpK_jB6bIRjll1kX0AW9AhTsIYiU,2830
@@ -150,13 +150,13 @@ oarepo_runtime/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hS
 oarepo_runtime/utils/functools.py,sha256=gKS9YZtlIYcDvdNA9cmYO00yjiXBYV1jg8VpcRUyQyg,1324
 oarepo_runtime/utils/index.py,sha256=ArrUUXB-KowUcUksRKqcFpmqct4bn9alO1zd_kX2tmU,292
 oarepo_runtime/utils/path.py,sha256=V1NVyk3m12_YLbj7QHYvUpE1wScO78bYsX1LOLeXDkI,3108
-oarepo_runtime-1.7.2.dist-info/licenses/LICENSE,sha256=h2uWz0OaB3EN-J1ImdGJZzc7yvfQjvHVYdUhQ-H7ypY,1064
+oarepo_runtime-1.9.0.dist-info/licenses/LICENSE,sha256=h2uWz0OaB3EN-J1ImdGJZzc7yvfQjvHVYdUhQ-H7ypY,1064
 tests/marshmallow_to_json/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 tests/marshmallow_to_json/test_datacite_ui_schema.py,sha256=82iLj8nW45lZOUewpWbLX3mpSkpa9lxo-vK-Qtv_1bU,48552
 tests/marshmallow_to_json/test_simple_schema.py,sha256=izZN9p0v6kovtSZ6AdxBYmK_c6ZOti2_z_wPT_zXIr0,1500
 tests/pkg_data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-oarepo_runtime-1.7.2.dist-info/METADATA,sha256=zgNr97N1sEuZ23n7YI99fqBLKHIRKdQdxwp9vS9DtLo,4788
-oarepo_runtime-1.7.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-oarepo_runtime-1.7.2.dist-info/entry_points.txt,sha256=k7O5LZUOGsVeSpB7ulU0txBUNp1CVQG7Q7TJIVTPbzU,491
-oarepo_runtime-1.7.2.dist-info/top_level.txt,sha256=bHhlkT1_RQC4IkfTQCqA3iN4KCB6cSFQlsXpQMSP-bE,21
-oarepo_runtime-1.7.2.dist-info/RECORD,,
+oarepo_runtime-1.9.0.dist-info/METADATA,sha256=PNBhOPkZ-HmVABJk4k9cb5up2HStMFInTy0ZSw0NEZs,4788
+oarepo_runtime-1.9.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+oarepo_runtime-1.9.0.dist-info/entry_points.txt,sha256=k7O5LZUOGsVeSpB7ulU0txBUNp1CVQG7Q7TJIVTPbzU,491
+oarepo_runtime-1.9.0.dist-info/top_level.txt,sha256=bHhlkT1_RQC4IkfTQCqA3iN4KCB6cSFQlsXpQMSP-bE,21
+oarepo_runtime-1.9.0.dist-info/RECORD,,