Cria parte do utilitário para incluir DOI nos registros h e f de uma data bases

proc/scielo_doi/app/cisis.py

@@ -0,0 +1,71 @@
+# coding=utf-8
+import os
+
+
+CISIS = "cisis/mx"
+
+
+def update_command(db, mfn, count, proc):
+    cmd = ("{CISIS} {db} from={mfn} count={count} "
+           "\"proc={proc}\""
+           " copy={db} now -all").format(
+            CISIS=CISIS,
+            db=db,
+            mfn=mfn,
+            count=count,
+            proc=proc
+        )
+    return cmd
+
+
+def execute(cmd):
+    # TODO usar subprocess
+    # TODO remover esta funcao deste modulo
+    os.system(cmd)
+
+
+def proc_update_doi(file_path, doi, doi_items):
+    """
+    Retorna a "proc" que atualiza
+    os campos v237 com DOI original e 337 todos os DOI com idioma
+    """
+    if not file_path:
+        raise ValueError("Invalid value for file_path")
+    if not doi:
+        raise ValueError("Invalid value for DOI")
+    proc_d = "'d337d237a237~{}~'".format(doi)
+    proc_a337 = [
+        ",'a337~^l{}^d{}~'".format(lang, _doi)
+        for lang, _doi in doi_items
+    ]
+    return "if 'hf':v706 and v702='{}' then {} fi".format(
+            file_path,
+            proc_d + ''.join(proc_a337)
+        )
+
+
+def proc_update_datetime(file_path, update_date, update_time):
+    """
+    Retorna a "proc" que atualiza
+    os campos v91 (data) e v92 (hora) da atualização
+    """
+    proc = "'d91d92a91~{}~a92~{}~'".format(update_date, update_time)
+    return "if v706='o' and v702='{}' then {} fi".format(
+            file_path,
+            proc
+        )
+
+
+def get_commands_to_update_ohf_records_of_a_doc(
+        db, file_path, h_mfn, doi, doi_items, update_date, update_time):
+    """
+    Gera dois comandos para atualizar 1 documento os registros:
+    - h e f: os campos v237 com DOI original e 337 todos os DOI com idioma
+    - o: os campos v91 (data) e v92 (hora) da atualização
+    """
+    o_mfn = str(int(h_mfn) - 1)
+    o_proc = proc_update_datetime(file_path, update_date, update_time)
+    o_cmd = update_command(db, o_mfn, 1, o_proc)
+    hf_proc = proc_update_doi(file_path, doi, doi_items)
+    hf_cmd = update_command(db, h_mfn, 2, hf_proc)
+    return [o_cmd, hf_cmd]

proc/scielo_doi/setup.py

@@ -0,0 +1,16 @@
+import setuptools
+
+
+setuptools.setup(
+    name="scielo_doi",
+    version="0.1",
+    author="SciELO",
+    author_email="scielo-dev@googlegroups.com",
+    description="",
+    license="2-clause BSD",
+    packages=setuptools.find_packages(
+        exclude=["*.tests", "*.tests.*", "tests.*", "tests"]
+    ),
+    include_package_data=True,
+    test_suite="tests",
+)

proc/scielo_doi/tests/test_cisis.py

@@ -0,0 +1,78 @@
+from unittest import TestCase
+
+
+from app import cisis
+
+
+class TestCisis(TestCase):
+
+    def test_update_command(self):
+        db = "base"
+        mfn = 100
+        count = 5
+        proc = '@a.proc'
+        result = cisis.update_command(db, mfn, count, proc)
+        self.assertEqual(
+            result,
+            "cisis/mx base from=100 count=5 \"proc=@a.proc\" copy=base now -all"
+        )
+
+    def test_proc_update_doi(self):
+        file_path = "xml/abc/v1n1/a.xml"
+        doi = "10.1590/1809-4392201902421"
+        doi_items = [
+            ("pt", "10.1590/1809-4392201902421.1"),
+            ("en", "10.1590/1809-4392201902421.2"),
+            ("es", "10.1590/1809-4392201902421.3"),
+        ]
+        result = cisis.proc_update_doi(file_path, doi, doi_items)
+        expected = ("if 'hf':v706 and v702='xml/abc/v1n1/a.xml' then "
+                    "'d337d237a237~10.1590/1809-4392201902421~',"
+                    "'a337~^lpt^d10.1590/1809-4392201902421.1~',"
+                    "'a337~^len^d10.1590/1809-4392201902421.2~',"
+                    "'a337~^les^d10.1590/1809-4392201902421.3~' fi")
+        self.assertEqual(result, expected)
+
+    def test_proc_update_datetime(self):
+        file_path = "xml/abc/v1n1/a.xml"
+        result = cisis.proc_update_datetime(file_path, '20100501', '1903')
+        expected = ("if v706='o' and v702='xml/abc/v1n1/a.xml' then"
+                    " 'd91d92a91~20100501~a92~1903~' fi")
+        self.assertEqual(result, expected)
+
+    def test_get_commands_to_update_ohf_records_of_a_doc(self):
+        db = "base"
+        file_path = "xml/a.xml"
+        h_mfn = "4"
+        doi = "10.1590/1809-4392201902421.1"
+        doi_items = [
+            ("pt", "10.1590/1809-4392201902421.1"),
+            ("en", "10.1590/1809-4392201902421.2"),
+            ("es", "10.1590/1809-4392201902421.3"),
+        ]
+        upd_date = '20100501'
+        upd_time = '1903'
+        result = cisis.get_commands_to_update_ohf_records_of_a_doc(
+            db, file_path, h_mfn, doi, doi_items, upd_date, upd_time)
+
+        self.assertEqual(
+            result[0],
+            (
+                "cisis/mx base from=3 count=1 \"proc="
+                "if v706='o' and v702='xml/a.xml' then"
+                " 'd91d92a91~20100501~a92~1903~' fi\" "
+                "copy=base now -all"
+            )
+        )
+        self.assertEqual(
+            result[1],
+            (
+                "cisis/mx base from=4 count=2 \"proc="
+                "if 'hf':v706 and v702='xml/a.xml' then "
+                "'d337d237a237~10.1590/1809-4392201902421.1~',"
+                "'a337~^lpt^d10.1590/1809-4392201902421.1~',"
+                "'a337~^len^d10.1590/1809-4392201902421.2~',"
+                "'a337~^les^d10.1590/1809-4392201902421.3~' fi\" "
+                "copy=base now -all"
+            )
+        )