Start adding support for reprint

Xennis · Apr 24, 2024 · a8fd46e · a8fd46e
1 parent 0f77489
commit a8fd46e
Show file tree

Hide file tree

Showing 4 changed files with 98 additions and 1 deletion.
diff --git a/epidoc/body.py b/epidoc/body.py
@@ -26,3 +26,24 @@ def foreign_languages(body: Tag) -> dict[str, int]:
                 continue
             result[lang] = result.get(lang, 0) + 1
         return result
+
+
+class _Head:
+
+    @staticmethod
+    def reprint_from(body: Tag) -> list[str]:
+        result: list[str] = []
+        for elem in body.findAll("ref", type="reprint-from"):
+            n = _normalize(elem.attrs.get("n"))
+            if n:
+                result.append(n)
+        return result
+
+    @staticmethod
+    def reprint_in(body: Tag) -> list[str]:
+        result: list[str] = []
+        for elem in body.findAll("ref", type="reprint-in"):
+            n = _normalize(elem.attrs.get("n"))
+            if n:
+                result.append(n)
+        return result
diff --git a/tests/test_body.py b/tests/test_body.py
@@ -3,7 +3,7 @@
 
 from bs4 import BeautifulSoup
 
-from epidoc.body import _Edition
+from epidoc.body import _Edition, _Head
 
 TESTDATA_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "testdata")
 
@@ -30,3 +30,27 @@ def test_all(self):
                 self.assertEqual(want_language, actual_language, msg="language")
                 actual_foreign_languages = _Edition.foreign_languages(elem)
                 self.assertEqual(want_foreign_languages, actual_foreign_languages, msg="foreign languages")
+
+
+class HeadTest(unittest.TestCase):
+    def test_all(self):
+        tests = [
+            (
+                "3.xml",
+                ["p.oxy;10;1271", "chla;4;266"],
+                [],
+            ),
+            (
+                "4.xml",
+                [],
+                ["c.ep.lat;;213"],
+            ),
+        ]
+        for filename, want_reprint_from, want_reprint_in in tests:
+            with self.subTest(filename):
+                with open(os.path.join(TESTDATA_DIR, "body", filename)) as f:
+                    elem = BeautifulSoup(f.read(), features="lxml")
+                actual_reprint_from = _Head.reprint_from(elem)
+                self.assertEqual(want_reprint_from, actual_reprint_from, msg="reprint from")
+                actual_reprint_in = _Head.reprint_in(elem)
+                self.assertEqual(want_reprint_in, actual_reprint_in, msg="reprint in")
diff --git a/tests/testdata/body/3.xml b/tests/testdata/body/3.xml
@@ -0,0 +1,39 @@
+<!-- DDB_EpiDoc_XML/c.ep.lat/c.ep.lat.213.xml
+            <availability>
+               <p>© Duke Databank of Documentary Papyri. This work is licensed under a <ref type="license" target="http://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution 3.0 License</ref>.</p>
+            </availability>
+-->
+      <body>
+         <head n="21785" xml:lang="en">
+            <date>AD 246</date>
+            <placeName>Oxyrhynchus</placeName>
+            <ref n="p.oxy;10;1271" type="reprint-from">P.Oxy. 10 1271; CPL 179</ref>
+            <ref n="chla;4;266" type="reprint-from">ChLA 4 266</ref>
+         </head>
+         <div xml:lang="la" type="edition" xml:space="preserve"><ab>
+    <lb n="1"/><foreign xml:lang="grc">Οὐαλερίῳ Φίρμῳ ἐπάρχῳ Αἰγύπτου
+
+    <lb n="2"/>παρὰ Αὐρηλίας <app type="editorial"><lem resp="BL cf. 8.243">Μαρκιανῆς</lem><rdg>Μαικιανῆσ</rdg></app> <expan>Σιδήτ<ex>ιδος</ex></expan>
+
+    <lb n="3"/>βούλομαι κύριε ἐκπλεῦσαι διὰ Φάρου.
+
+    <lb n="4"/>ἀξιῶ γράψαι σε τῷ ἐπιτρόπῳ τῆς Φά
+
+    <lb n="5" break="no"/>ρου ἀπολῦσαί με κατὰ τὸ ἔθος
+
+    <lb n="6"/><app type="editorial"><lem resp="BL cf. 8.243"><num value="3">γ</num> <expan><ex cert="low">ἔτους</ex></expan> <expan>ἐ<ex>παγομένων</ex></expan></lem><rdg><unclear>Π</unclear><supplied reason="lost">α</supplied><unclear>χὼν</unclear></rdg></app> <num value="1">α</num> διευτύχει </foreign>
+
+    <lb n="7"/><handShift new="m2"/>Valerius Firmus
+
+    <lb n="8"/>Asclepiade salu<unclear>t</unclear>em
+
+    <lb n="9"/>dimitti <choice><reg>iussi</reg><orig>iusi</orig></choice> de <unclear>P</unclear><supplied reason="lost">haro</supplied>
+
+    <lb n="10"/><choice><reg>commendo</reg><orig>comendo</orig></choice> <unclear>t</unclear><supplied reason="lost">ibi</supplied>
+
+    <lb n="11"/><handShift new="m3"/>vale iu<supplied reason="lost">ssi</supplied>
+
+    <lb n="12"/>datum <num>XVII</num> <expan><unclear>K</unclear><supplied reason="lost">al<ex>endas</ex></supplied></expan><supplied reason="lost"> <app type="editorial"><lem resp="BL cf. 8.243"><expan>Oct<ex>obres</ex></expan></lem><rdg><expan>Iun<ex>ias</ex></expan></rdg></app></supplied>
+
+    <lb n="13"/>Presenti <unclear>A</unclear><supplied reason="lost">lbino <expan>co<ex>n</ex>s<ex>ulibus</ex></expan></supplied> </ab></div>
+      </body>
diff --git a/tests/testdata/body/4.xml b/tests/testdata/body/4.xml
@@ -0,0 +1,13 @@
+<!-- DDB_EpiDoc_XML/p.oxy/p.oxy.10/p.oxy.10.1271.xml
+            <availability>
+               <p>© Duke Databank of Documentary Papyri. This work is licensed under a <ref type="license" target="http://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution 3.0 License</ref>.</p>
+            </availability>
+-->
+      <body>
+         <head xml:lang="en">
+            <date>AD 159</date>
+            <placeName>Oxyrhynchus</placeName>
+            <ref n="c.ep.lat;;213" type="reprint-in">C.Epist.Lat. 213</ref>
+         </head>
+         <div xml:lang="grc" type="edition" xml:space="preserve"><ab/></div>
+      </body>