Refactor: criar packtools/sps/validation/models/ separando modelos exclusivos de validação

packtools/sps/models/accessibility_data.py

@@ -1,65 +1,6 @@
 import re
 
-
-class XMLAccessibilityData:
-
-    def __init__(self, xmltree):
-        self.xmltree = xmltree
-
-    @property
-    def transcripts(self):
-        if not hasattr(self, '_transcripts') or not self._transcripts:
-            self._transcripts = {}
-            for item in self.xmltree.xpath(".//sec[@sec-type='transcript']"):
-                transcript = Transcript(item)
-                self._transcripts[transcript.transcript_id] = transcript.data
-        return self._transcripts
-
-    @property
-    def misplaced_alt_texts(self):
-        """
-        Detecta elementos <alt-text> que estão fora dos elementos permitidos.
-        Segundo SPS 1.9/1.10, <alt-text> só pode estar dentro de:
-        <graphic>, <inline-graphic>, <media>, <inline-media>
-        """
-        valid_parents = ("graphic", "inline-graphic", "media", "inline-media")
-        misplaced = []
-
-        # Procura TODOS os <alt-text> no documento
-        for alt_text in self.xmltree.xpath(".//alt-text"):
-            parent = alt_text.getparent()
-            if parent is not None and parent.tag not in valid_parents:
-                # <alt-text> está no lugar errado!
-                misplaced.append({
-                    "tag": "alt-text",
-                    "parent_tag": parent.tag,
-                    "parent_id": parent.get("id"),
-                    "alt_text": alt_text.text,
-                    "alt_text_length": len(alt_text.text or ""),
-                    "expected_location": valid_parents,
-                    "current_location": parent.tag,
-                })
-
-        return misplaced
-
-    @property
-    def data(self):
-        # CORREÇÃO: XPath simplificado para evitar duplicatas
-        # Captura apenas os elementos base que podem conter dados de acessibilidade
-        xpaths = [
-            ".//graphic",
-            ".//inline-graphic",
-            ".//media",
-            ".//inline-media",
-        ]
-        for item in self.xmltree.xpath("|".join(xpaths)):
-            model = AccessibilityData(item)
-            model.transcript_data = self.transcripts.get(model.xref_sec_rid)
-            yield model.data
-
-        # NOVO: Dados dos elementos mal posicionados
-        for misplaced in self.misplaced_alt_texts:
-            yield misplaced
+import warnings as _warnings
 
 
 class AccessibilityData:
@@ -211,3 +152,20 @@ def data(self):
             "speakers": self.speaker_data,
             "tag": self.node.tag,
         }
+
+
+def __getattr__(name):
+    _moved = {
+        "XMLAccessibilityData": "packtools.sps.validation.models.accessibility_data",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.accessibility_data is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/app_group.py

@@ -1,3 +1,5 @@
+import warnings as _warnings
+
 from packtools.sps.models.article_and_subarticles import Fulltext
 from packtools.sps.models.label_and_caption import LabelAndCaption
 from packtools.sps.models.media import Media
@@ -31,15 +33,18 @@ def data(self):
         }
 
 
-class XmlAppGroup:
-    def __init__(self, xml_tree):
-        self.xml_tree = xml_tree
-
-    @property
-    def data(self):
-        for node in self.xml_tree.xpath(".|.//sub-article"):
-            full_text = Fulltext(node)
-
-            for app_node in node.xpath("./back//app"):
-                app_data = App(app_node).data
-                yield {**app_data, **full_text.attribs_parent_prefixed}
+def __getattr__(name):
+    _moved = {
+        "XmlAppGroup": "packtools.sps.validation.models.app_group",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.app_group is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/article_data_availability.py

@@ -21,36 +21,21 @@
 from packtools.sps.models.article_and_subarticles import Fulltext
 
 
-class DataAvailability:
-    def __init__(self, xmltree):
-        self.xmltree = xmltree
+import warnings as _warnings
 
-    @property
-    def items(self):
-        xpath_query = './body//sec[@sec-type="data-availability"] | ./body//fn[@fn-type="data-availability"] | ./back//sec[@sec-type="data-availability"] | ./back//fn[@fn-type="data-availability"]'
 
-        for node in self.xmltree.xpath(
-            ". | ./sub-article[@article-type='translation']"
-        ):
-            fulltext = Fulltext(node)
-            items = fulltext.node.xpath(xpath_query)
-            if len(items) == 0:
-                yield fulltext.attribs_parent_prefixed
-            else:
-                for item in items:
-                    data = {
-                        "tag": item.tag,
-                        "specific_use": item.get("specific-use"),
-                        "label": item.findtext("label"),
-                        "text": " ".join(item.xpath("./p//text()")),
-                    }
-                    data.update(fulltext.attribs_parent_prefixed)
-                    yield data
-
-    @property
-    def items_by_lang(self):
-        d = {}
-        for item in self.items:
-            d.setdefault(item["parent_lang"], [])
-            d[item["parent_lang"]].append(item)
-        return d
+def __getattr__(name):
+    _moved = {
+        "DataAvailability": "packtools.sps.validation.models.article_data_availability",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.article_data_availability is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/author_notes.py

@@ -44,15 +44,21 @@ def corresp_data(self):
             yield data
 
 
-class XMLAuthorNotes:
-    def __init__(self, xml_tree):
-        self.xml_tree = xml_tree
-
-    def article_author_notes(self):
-        group = FulltextAuthorNotes(self.xml_tree.find("."))
-        return {"corresp_data": list(group.corresp_data), "fns": list(group.items)}
-
-    def sub_article_author_notes(self):
-        for sub_article in self.xml_tree.xpath(".//sub-article"):
-            group = FulltextAuthorNotes(sub_article)
-            yield {"corresp_data": list(group.corresp_data), "fns": list(group.items)}
+import warnings as _warnings
+
+
+def __getattr__(name):
+    _moved = {
+        "XMLAuthorNotes": "packtools.sps.validation.models.author_notes",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.author_notes is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/fig.py

@@ -196,26 +196,21 @@ def figs(self):
             yield put_parent_context(data, self.lang, self.article_type, self.parent, self.parent_id)
 
 
-class ArticleFigs:
-    def __init__(self, xml_tree):
-        self.xml_tree = xml_tree
-
-    @property
-    def get_all_figs(self):
-        yield from self.get_article_figs
-        yield from self.get_sub_article_translation_figs
-        yield from self.get_sub_article_non_translation_figs
-
-    @property
-    def get_article_figs(self):
-        yield from Figs(self.xml_tree.find(".")).figs()
-
-    @property
-    def get_sub_article_translation_figs(self):
-        for node in self.xml_tree.xpath(".//sub-article[@article-type='translation']"):
-            yield from Figs(node).figs()
-
-    @property
-    def get_sub_article_non_translation_figs(self):
-        for node in self.xml_tree.xpath(".//sub-article[@article-type!='translation']"):
-            yield from Figs(node).figs()
+import warnings as _warnings
+
+
+def __getattr__(name):
+    _moved = {
+        "ArticleFigs": "packtools.sps.validation.models.fig",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.fig is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/fn.py

@@ -26,22 +26,21 @@ def __init__(self, node):
         super().__init__(node, "fn-group", FnGroup)
 
 
-class XMLFns:
-    def __init__(self, xml_tree):
-        self.xml_tree = xml_tree
-
-    def article_fn_groups_notes(self):
-        yield from FulltextFnGroups(self.xml_tree.find(".")).items
-
-    def sub_article_fn_groups_notes(self):
-        for sub_article in self.xml_tree.xpath(".//sub-article"):
-            yield from FulltextFnGroups(sub_article).items
-
-    @property
-    def fn_edited_by(self):
-        for item in self.xml_tree.xpath(". | .//sub-article"):
-            fulltext = Fulltext(item)
-            for node in fulltext.node.xpath("*//fn[@fn-type='edited-by']"):
-                data = fulltext.attribs_parent_prefixed
-                data.update(Fn(node).data)
-                yield data
+import warnings as _warnings
+
+
+def __getattr__(name):
+    _moved = {
+        "XMLFns": "packtools.sps.validation.models.fn",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.fn is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

packtools/sps/models/formula.py

@@ -182,81 +182,21 @@ def data(self):
         }
 
 
-class ArticleFormulas:
-    """
-    Represents an article with its associated formulas, grouped by language.
-
-    Parameters:
-        xml_tree (xml.etree.ElementTree.ElementTree): The parsed XML document representing the article.
-    """
-
-    def __init__(self, xml_tree):
-        """
-        Initializes an ArticleFormulas object.
-
-        **Parameters:**
-            xml_tree (xml.etree.ElementTree.ElementTree): The parsed XML document representing the article.
-        """
-        self.xml_tree = xml_tree
-
-    @property
-    def disp_formula_items(self):
-        """
-        Generator that yields formulas with their respective parent context.
-
-        Yields:
-            dict: A dictionary containing the formula data along with its parent context,
-                  including language and article type details.
-        """
-        for node, lang, article_type, parent, parent_id in get_parent_context(self.xml_tree):
-            for item in node.xpath(".//disp-formula"):
-                formula = Formula(item)
-                data = formula.data
-                parent_data = put_parent_context(data, lang, article_type, parent, parent_id)
-                yield parent_data
-
-    @property
-    def inline_formula_items(self):
-        """
-        Generator that yields inline formulas with their respective parent context.
-
-        Yields:
-            dict: A dictionary containing the formula data along with its parent context,
-                  including language and article type details.
-        """
-        for node, lang, article_type, parent, parent_id in get_parent_context(self.xml_tree):
-            for item in node.xpath(".//inline-formula"):
-                formula = Formula(item)
-                data = formula.data
-                parent_data = put_parent_context(data, lang, article_type, parent, parent_id)
-                yield parent_data
-
-    @property
-    def disp_formula_items_by_lang(self):
-        """
-        Returns a dictionary of formulas grouped by language.
-
-        Returns:
-            dict: A dictionary where keys are language codes (str) and values are
-                  dictionaries with formula data within that language context.
-        """
-        langs = {}
-        for item in self.disp_formula_items:
-            lang = item.get("parent_lang")
-            langs[lang] = item
-        return langs
-
-    @property
-    def inline_formula_items_by_lang(self):
-        """
-        Returns a dictionary of inline formulas grouped by language.
-
-        Returns:
-            dict: A dictionary where keys are language codes (str) and values are
-                  dictionaries with formula data within that language context.
-        """
-        langs = {}
-        for item in self.inline_formula_items:
-            lang = item.get("parent_lang")
-            langs[lang] = item
-        return langs
+import warnings as _warnings
+
+
+def __getattr__(name):
+    _moved = {
+        "ArticleFormulas": "packtools.sps.validation.models.formula",
+    }
+    if name in _moved:
+        import importlib
+        _warnings.warn(
+            f"{name} has moved to {_moved[name]}. "
+            f"Importing from packtools.sps.models.formula is deprecated.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        mod = importlib.import_module(_moved[name])
+        return getattr(mod, name)
+    raise AttributeError(f"module {__name__!r} has no attribute {name!r}")