Merge pull request #6 from robertatakenaka/0.1.1

Ajustes para uso em scms-upload (migration)

scielo_classic_website/classic_ws.py

@@ -0,0 +1,43 @@
+from scielo_classic_website import controller, config
+from scielo_classic_website.models.issue_files import IssueFiles
+
+# Para Journal, Issue, Document devem ficar disponíveis neste módulo
+from scielo_classic_website.models.journal import Journal
+from scielo_classic_website.models.issue import Issue
+from scielo_classic_website.models.document import Document
+
+
+def get_bases_work_acron_path(cisis_path, bases_work_acron_file_path, issue_folder=None):
+    if issue_folder:
+        return controller.isis_cmd.get_documents_by_issue_folder(
+            cisis_path, bases_work_acron_file_path, issue_folder)
+    else:
+        return bases_work_acron_file_path
+
+
+def get_document_pids(from_date, to_date):
+    return controller.isis_cmd.get_document_pids(from_date, to_date)
+
+
+def get_paragraphs_records(pid):
+    id_file_path = config.get_paragraphs_id_file_path(pid)
+    return controller.pids_and_their_records(id_file_path, "artigo")
+
+
+def get_records_by_pid(pid):
+    source_path = controller.isis_cmd.get_document_isis_db(pid)
+    return controller.pids_and_their_records(source_path, "artigo")
+
+
+def get_records_by_acron(acron, id_folder_path=None):
+    source_path = controller.get_acron_db_path(acron, id_folder_path)
+    return controller.pids_and_their_records(source_path, "artigo")
+
+
+def get_records_by_source_path(db_type, source_path):
+    return controller.pids_and_their_records(source_path, db_type)
+
+
+def get_issue_files(acron, issue_folder, config):
+    issue_files = IssueFiles(acron, issue_folder, config)
+    return issue_files.files

scielo_classic_website/cli/models_builder.py

@@ -91,7 +91,7 @@ def __init__(self, class_name, data_dictionary):
     def create_base_module(self, class_file_path):
         with open(class_file_path, "w") as fp:
             fp.write("# generated by ModelBuilder\n")
-            fp.write("from scielo_migration.isisdb.meta_record import MetaRecord\n\n\n")
+            fp.write("from scielo_classic_website.isisdb.meta_record import MetaRecord\n\n\n")
             fp.write(f"{_attributes_var(self.get_attributes())}\n\n\n")
 
     def get_attributes(self):
@@ -108,7 +108,7 @@ def create_main_module(self, class_file_path, base_module_name):
         class_code = _class_init_builder(f"{self._class_name}", f"{base_class_name}")
         content = (
             "# generated by ModelBuilder",
-            f"from scielo_migration.isisdb.{base_module_name} import {base_class_name}\n",
+            f"from scielo_classic_website.isisdb.{base_module_name} import {base_class_name}\n",
             "",
             f"{class_code}",
         )

scielo_classic_website/controller.py

@@ -1,6 +1,6 @@
 from scielo_classic_website.iid2json import id2json3
 
-from scielo_classic_website import isis_cmd
+from scielo_classic_website.isisdb import isis_cmd
 
 
 def pids_and_their_records(source_file_path, db_type):

scielo_classic_website/isisdb/base_h_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.meta_record import MetaRecord
+from scielo_classic_website.isisdb.meta_record import MetaRecord
 
 
 ATTRIBUTES = (

scielo_classic_website/isisdb/base_issue_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.meta_record import MetaRecord
+from scielo_classic_website.isisdb.meta_record import MetaRecord
 
 
 ATTRIBUTES = (

scielo_classic_website/isisdb/base_p_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.meta_record import MetaRecord
+from scielo_classic_website.isisdb.meta_record import MetaRecord
 
 
 ATTRIBUTES = (

scielo_classic_website/isisdb/h_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.base_h_record import BaseDocumentRecord
+from scielo_classic_website.isisdb.base_h_record import BaseDocumentRecord
 
 
 # generated by ModelBuilder

scielo_classic_website/isisdb/isis_cmd.py

@@ -12,9 +12,7 @@
 
 def get_document_isis_db(pid):
     """
-    Consulta a base de dados ISIS artigo e retorna os pids atualizados
-    em um intervalo de datas (data de processamento do converter)
-
+    Consulta a base de dados ISIS artigo e retorna os registros do pid
     """
     BASES_ARTIGO_PATH = config.get_bases_artigo_path()
     name = date_now_as_folder_name()
@@ -179,3 +177,27 @@ def get_document_pids(from_date=None, to_date=None):
             # 1|OAITS=20210917=2675-54752021000300700
             parts = row.strip().split("=")
             yield {"updated": parts[1], "pid": "S" + parts[-1]}
+
+
+def get_documents_by_issue_folder(cisis_path, bases_work_acron_file_path, issue_folder):
+    """
+    Consulta a base de dados ISIS bases-work/acron/acron e filtra por issue_folder
+
+    """
+    name = date_now_as_folder_name()
+    finished_file_path = create_temp_file(f"{name}_{issue_folder}_finished.out")
+    output_file_path = create_temp_file(f"{name}_{issue_folder}_output")
+
+    cmds = []
+    cmds.append(
+        f'''{cisis_path}/mx {bases_work_acron_file_path} btell=0 '''
+        f'''"bool={issue_folder}" '''
+        f'''append={output_file_path} now -all'''
+    )
+    cmds.append(
+        f"echo finished > {finished_file_path}"
+    )
+    os.system(";".join(cmds))
+    while "finished" not in read_file(finished_file_path):
+        pass
+    return output_file_path

scielo_classic_website/isisdb/issue_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.base_issue_record import BaseIssueRecord
+from scielo_classic_website.isisdb.base_issue_record import BaseIssueRecord
 
 
 # generated by ModelBuilder
@@ -11,13 +11,20 @@ def __init__(
         super().__init__(
             record, multi_val_tags, data_dictionary)
 
+    @property
+    def bibliographic_strip_months(self):
+        for item in self.bibliographic_strip:
+            yield {"lang": item['language'], "text": item.get('months')}
+
     @property
     def start_month_text(self):
-        return (self.bibliographic_strip.get("month") or '').split("/")[0]
+        for item in self.bibliographic_strip_months:
+            return (item.get("text") or '').split("/")[0]
 
     @property
     def end_month_text(self):
-        return (self.bibliographic_strip.get("month") or '').split("/")[-1]
+        for item in self.bibliographic_strip_months:
+            return (item.get("text") or '').split("/")[-1]
 
     @property
     def start_month(self):

scielo_classic_website/isisdb/p_record.py

@@ -1,5 +1,5 @@
 # generated by ModelBuilder
-from scielo_migration.isisdb.base_p_record import BaseParagraphRecord
+from scielo_classic_website.isisdb.base_p_record import BaseParagraphRecord
 
 
 # generated by ModelBuilder

scielo_classic_website/models/html_body.py

@@ -1,4 +1,4 @@
-from dsm.utils.files_utils import read_file
+from scielo_classic_website.utils.files_utils import read_file
 
 
 class BodyFromISIS:

scielo_classic_website/models/issue.py

@@ -1,3 +1,5 @@
+import logging
+
 from scielo_classic_website.isisdb.issue_record import IssueRecord
 
 
@@ -11,12 +13,16 @@ def __getattr__(self, name):
         # fica menos acoplado
         if hasattr(self.issue_record, name):
             return getattr(self.issue_record, name)
-        raise AttributeError(name)
+        raise AttributeError(f"classic_website.Issue has no attribute {name}")
 
     @property
     def record(self):
         return self._record
 
+    @property
+    def publication_year(self):
+        return self.issue_record.publication_date[:4]
+
     @property
     def supplement(self):
         return self.supplement_volume or self.supplement_number
@@ -33,3 +39,23 @@ def order(self):
     def pid(self):
         # 0001-371419980003
         return self.journal + self.order
+
+    @property
+    def isis_created_date(self):
+        return self.issue_record.creation_date
+
+    @property
+    def isis_updated_date(self):
+        return self.issue_record.update_date
+
+    @property
+    def issue_label(self):
+        pr = self.is_press_release or ''
+        if self.number in ("ahead", "review"):
+            return self.publication_year + "n" + self.number + pr
+        return "".join([
+            k + v
+            for k, v in zip(("v", "n", "s", ""), (self.volume, self.number, self.suppl, pr))
+            if v
+            ])
+

scielo_classic_website/models/issue_files.py

@@ -7,14 +7,15 @@
 
 class IssueFiles:
 
-    def __init__(self, acron, issue_folder):
+    def __init__(self, acron, issue_folder, config):
         self.acron = acron
         self.issue_folder = issue_folder
         self._subdir_acron_issue = os.path.join(acron, issue_folder)
         self._htdocs_img_revistas_files = None
         self._bases_translation_files = None
         self._bases_pdf_files = None
         self._bases_xml_files = None
+        self._config = config
 
     @property
     def files(self):
@@ -51,7 +52,7 @@ def bases_translation_files(self):
 
             paths = glob.glob(
                 os.path.join(
-                    config.BASES_TRANSLATION_PATH,
+                    self._config["BASES_TRANSLATION_PATH"],
                     self._subdir_acron_issue, "*")
             )
             files = []
@@ -93,7 +94,7 @@ def bases_pdf_files(self):
 
             paths = glob.glob(
                 os.path.join(
-                    config.BASES_PDF_PATH,
+                    self._config["BASES_PDF_PATH"],
                     self._subdir_acron_issue, "*")
             )
             files = []
@@ -137,7 +138,7 @@ def htdocs_img_revistas_files(self):
         if self._htdocs_img_revistas_files is None:
             paths = glob.glob(
                 os.path.join(
-                    config.HTDOCS_IMG_REVISTAS_PATH,
+                    self._config["HTDOCS_IMG_REVISTAS_PATH"],
                     self._subdir_acron_issue,
                     "*"
                 )
@@ -163,10 +164,12 @@ def htdocs_img_revistas_files(self):
     @property
     def bases_xml_files(self):
         if self._bases_xml_files is None:
-            paths = os.path.join(
-                config.BASES_XML_PATH,
-                self._subdir_acron_issue,
-                "*.xml"
+            paths = glob.glob(
+                os.path.join(
+                    self._config["BASES_XML_PATH"],
+                    self._subdir_acron_issue,
+                    "*.xml"
+                )
             )
             files = []
             for path in paths:

scielo_classic_website/models/journal.py

@@ -1,16 +1,18 @@
 from scielo_classic_website.isisdb.meta_record import MetaRecord
+from scielo_classic_website.isisdb.journal_record import JournalRecord
 
 
 class Journal:
     def __init__(self, journal_record):
-        self.journal_record = journal_record
+        self.journal_record = JournalRecord(journal_record)
 
     def __getattr__(self, name):
         # desta forma Journal não precisa herdar de JournalRecord
         # fica menos acoplado
         if hasattr(self.journal_record, name):
             return getattr(self.journal_record, name)
-        raise AttributeError(name)
+        raise AttributeError(
+            f"classic_website.Journal has no attribute {name}")
 
     @property
     def record(self):
@@ -57,13 +59,6 @@ def online_submission_url(self):
     def publication_status(self):
         return self.journal_record.current_status
 
-    @property
-    def mission(self):
-        return {
-            m['l']: m['_']
-            for m in self.journal_record.mission
-        }
-
     @property
     def index_at(self):
         return self.journal_record.index_coverage

scielo_classic_website/spsxml/sps_xml_article_meta.py

@@ -1,7 +1,7 @@
 import plumber
 from lxml import etree as ET
 
-from scielo_migration.spsxml.sps_xml_attributes import (
+from scielo_classic_website.spsxml.sps_xml_attributes import (
     CONTRIB_ROLES,
     get_attribute_value,
 )

scielo_classic_website/spsxml/sps_xml_attributes.py

@@ -5,7 +5,7 @@
     ATTRIBUTES_PATH,
 )
 
-from scielo_migration.attr_values import AttrValues
+from scielo_classic_website.attr_values import AttrValues
 
 
 def _read_csv_file(file_path):

scielo_classic_website/spsxml/sps_xml_pipes.py

@@ -7,7 +7,7 @@
     COUNTRY_ITEMS,
 )
 
-from scielo_migration.spsxml.sps_xml_article_meta import (
+from scielo_classic_website.spsxml.sps_xml_article_meta import (
     XMLArticleMetaSciELOArticleIdPipe,
     XMLArticleMetaArticleIdDOIPipe,
     XMLArticleMetaArticleCategoriesPipe,