Merge pull request #27 from wrznr/bibl_short

Compile the magic element "bibl"

mets_mods2teiHeader/api/mets.py

@@ -70,6 +70,7 @@ def __init__(self):
         self.collections = None
         self.languages = None
         self.extents = None
+        self.series = None
 
     @classmethod
     def read(cls, source):
@@ -108,8 +109,11 @@ def __spur(self):
         # alternatively identify the corresponding dmdSec via <structMap type="Logical" />
 
         #
-        # main title
-        self.title = self.tree.xpath("//mets:dmdSec[1]//mods:mods/mods:titleInfo/mods:title", namespaces=ns)[0].text
+        # main title and manuscript type
+        title = self.tree.xpath('//mets:structMap[@TYPE="LOGICAL"]/mets:div', namespaces=ns)
+        if title:
+            self.title = title[0].get("LABEL")
+            self.type = title[0].get("TYPE")
 
         #
         # sub titles
@@ -240,7 +244,7 @@ def __spur(self):
         collections = self.tree.xpath("//mets:dmdSec[1]//mods:mods/mods:relatedItem[@type='series']", namespaces=ns)
         self.collections = []
         for collection in collections:
-            title = collection.xpath("//mods:titleinfo/mods:title", namespaces=ns)
+            title = collection.xpath("./mods:titleInfo/mods:title", namespaces=ns)
             if title:
                 self.collections.append(title[0].text)
 

mets_mods2teiHeader/api/tei.py

@@ -35,6 +35,13 @@ def main_title(self):
         by the TEI Header.
         """
         return self.tree.xpath('//tei:titleStmt/tei:title[@type="main"]', namespaces=ns)[0].text
+
+    @property
+    def publication_level(self):
+        """
+        Returns the level of publication ('monographic' vs. 'analytic')
+        """
+        return self.tree.xpath('//tei:sourceDesc/tei:biblFull/tei:titleStmt/tei:title[@type="main"]', namespaces=ns)[0].get("level")
 
     @property
     def subtitles(self):
@@ -52,7 +59,7 @@ def authors(self):
         """
         authors = []
         for author in self.tree.xpath('//tei:fileDesc/tei:titleStmt/tei:author', namespaces=ns):
-            authors.append(" ".join(author.xpath('descendant-or-self::*/text()')))
+            authors.append(", ".join(author.xpath('descendant-or-self::*/text()')))
         return authors
 
     @property
@@ -119,13 +126,35 @@ def extents(self):
         """
         return [extent.text for extent in self.tree.xpath('//tei:msDesc/tei:physDesc/tei:objectDesc/tei:supportDesc/tei:extent', namespaces=ns)]
 
+    @property
+    def collections(self):
+        """
+        Returns information on the collections of the work represented
+        by the TEI Header.
+        """
+        return [collection.text for collection in self.tree.xpath('//tei:profileDesc/tei:creation', namespaces=ns)]
+
+    @property
+    def bibl(self):
+        """
+        Returns the short citation of the work represented
+        by the TEI Header.
+        """
+        return self.tree.xpath("//tei:fileDesc/tei:sourceDesc/tei:bibl", namespaces=ns)[0]
+
     def set_main_title(self, string):
         """
         Sets the main title of the title statements.
         """
         for main_title in self.tree.xpath('//tei:titleStmt/tei:title[@type="main"]', namespaces=ns):
             main_title.text = string
 
+    def set_publication_level(self, level):
+        """
+        Sets the level of publication ('monographic' vs. 'analytic')
+        """
+        self.tree.xpath('//tei:sourceDesc/tei:biblFull/tei:titleStmt/tei:title[@type="main"]', namespaces=ns)[0].set("level", level)
+
     def add_sub_title(self, string):
         """
         Adds a sub title to the title statements.
@@ -322,3 +351,33 @@ def add_extent(self, extent):
             support_desc = phys_desc.xpath('/tei:objectDesc/tei:supportDesc', namespaces=ns)[0]
         extent_elem = etree.SubElement(support_desc, "%sextent" % TEI)
         extent_elem.text = extent
+
+    def add_collection(self, collection):
+        """
+        Adds a (free-text) collection of the digital document
+        """
+        profile_desc = self.tree.xpath('//tei:profileDesc', namespaces=ns)[0]
+        creation = etree.SubElement(profile_desc, "%screation" % TEI)
+        creation.text = collection
+
+    def compile_bibl(self):
+        """
+        Compile the content of the short citation element 'bibl' based on the current state
+        """
+        if self.publication_level:
+            self.bibl.set("type", self.publication_level)
+        bibl_text = ""
+        if self.authors:
+            bibl_text += "; ".join(self.authors) + ": "
+        elif self.publication_level == "monograph":
+            bibl_text = "[N. N.], "
+        bibl_text += self.main_title + "."
+        if self.places:
+            bibl_text += " " + self.places[0].split(":")[1]
+            if len(self.places) > 1:
+                bibl_text += "u. a."
+        if self.dates:
+            if self.places:
+                bibl_text += ","
+            bibl_text += " " + self.dates[0] + "."
+        self.bibl.text = bibl_text

mets_mods2teiHeader/data/tei_skeleton.xml

@@ -33,9 +33,6 @@
     <profileDesc>
       <langUsage>
       </langUsage>
-      <textClass>
-        <classCode scheme="http://www.deutschestextarchiv.de/doku/klassifikation#DTACorpus">dtae</classCode>
-      </textClass>
     </profileDesc>
   </teiHeader>
   <text>

mets_mods2teiHeader/scripts/mets_mods2teiHeader.py

@@ -36,6 +36,9 @@ def cli(mets):
     # main title
     tei.set_main_title(mets.get_main_title())
 
+    # publication level
+    tei.set_publication_level(mets.type)
+
     # sub titles
     for sub_title in mets.get_sub_titles():
         tei.add_sub_title(sub_title)
@@ -101,6 +104,13 @@ def cli(mets):
     for extent in mets.extents:
         tei.add_extent(extent)
 
+    # collection
+    for collection in mets.collections:
+        tei.add_collection(collection)
+
+    # citation
+    tei.compile_bibl()
+
     click.echo(tei.tostring())
 
 

tests/test_mets.py

@@ -74,3 +74,6 @@ def test_data_assignment(subtests, datadir):
 
     with subtests.test("Check manuscript extent"):
         assert(mets.extents == ['[8] Bl., 783 S., [1] Bl.'])
+
+    with subtests.test("Check collections"):
+        assert(mets.collections == ['Drucke des 18. Jahrhunderts', 'Saxonica'])

tests/test_tei.py

@@ -23,6 +23,10 @@ def test_data_assignment(subtests):
         tei.set_main_title("Testbuch")
         assert(tei.main_title == "Testbuch")
 
+    with subtests.test("Check publication level"):
+        tei.set_publication_level("m")
+        assert(tei.publication_level == "m")
+
     with subtests.test("Check first subtitle"):
         tei.add_sub_title("Untertitel 1")
         assert(tei.subtitles == ["Untertitel 1"])
@@ -33,11 +37,11 @@ def test_data_assignment(subtests):
 
     with subtests.test("Check first author"):
         tei.add_author({'family': 'Mustermann', 'given': 'Max', 'date': '12.10.1956', 'title': 'Dr.'}, "personal")
-        assert(tei.authors == ["Mustermann Max Dr."])
+        assert(tei.authors == ["Mustermann, Max, Dr."])
 
     with subtests.test("Check further author (organisation)"):
         tei.add_author({'family': 'Mustermann', 'given': 'Max', 'date': '12.10.1956', 'title': 'Dr.'}, "corporate")
-        assert(tei.authors == ["Mustermann Max Dr.", "Mustermann Max 12.10.1956 Dr."])
+        assert(tei.authors == ["Mustermann, Max, Dr.", "Mustermann Max 12.10.1956 Dr."])
 
     with subtests.test("Check date(s)"):
         tei.add_date({"from": "01.01.1823", "to": "25.01.1823"})
@@ -74,3 +78,11 @@ def test_data_assignment(subtests):
     with subtests.test("Check further extent"):
         tei.add_extent("5 Abb.")
         assert(tei.extents == ["32 S.", "5 Abb."])
+
+    with subtests.test("Check collections"):
+        tei.add_collection("LDP")
+        assert(tei.collections == ["LDP"])
+
+    with subtests.test("Check bibl"):
+        tei.compile_bibl()
+        assert(tei.bibl.text == "Mustermann, Max, Dr.; Mustermann Max 12.10.1956 Dr.: Testbuch. Dresden, 01.01.1823.")