add abstract xmlElement model

eoa/models.py

@@ -1,6 +1,7 @@
 from django.db import models
 from django.utils import timezone
 from django.core.validators import URLValidator, RegexValidator
+from django.contrib.postgres.fields import HStoreField
 
 #################################################
 #
@@ -48,13 +49,39 @@ def filepath(instance, filename):
     ('it', 'Italian'),
     )
 
+tagname_choice = (
+    ('hi', 'Highlight'),
+    ('foreign', 'Foreign'),
+    ('ref', 'Reference'),
+    ('note','Note'),
+    ('label','Label'),
+    ('language','Language'),
+    ('graphic','Graphic'),
+)
+
 default_char = {
     'max_length':200,
     'default':None,
     'blank':True,
     'null':True,
     }
 
+validkeys = [
+    'place',
+    'type',
+    'target',
+    'indexname',
+    'sortkey',
+    'xml:lang',
+    'pages',
+    'n',
+    'url',
+    'rendition',
+    'rend',
+    'height',
+    'notation'
+]
+
 ##########################################
 #
 # basic information on publications, citations and authors
@@ -99,26 +126,27 @@ class Meta:
 
 
 class Citation(models.Model):
-    pub_type = models.CharField(
+    pubtype = models.CharField(
         max_length=20,
         choices=citation_choice,
         default='book'
     )
-    title = models.CharField(max_length=200)
+    titlea = models.CharField(**default_char)
+    titlem = models.CharField(**default_char)
     date = models.IntegerField(
         blank=True,
         null=False,
         validators=[RegexValidator(r'^\d{4}$')])
-    place = models.CharField(max_length=200)
-    publisher = models.CharField(max_length=200)
+    pupPlace = models.CharField(**default_char)
+    publisher = models.CharField(**default_char)
     volume = models.CharField(**default_char)
     issue = models.CharField(**default_char)
-    pages = models.CharField(max_length=20, blank=True)
-    # No automatic idstring here, is there a way to achive this?
-    idstring = models.CharField(max_length=200)
+    page = models.CharField(**default_char)
+    idstring = models.CharField(max_length=200,blank=False)
+    note = models.TextField(blank=True)
 
     def __str__(self):
-        return self.title
+        return self.idstring + ': ' + self.title
 
     class Meta:
         ordering = ('title',)
@@ -143,6 +171,10 @@ class Author(models.Model):
         Citation,
         blank=True,
         related_name='authors')
+    editions = models.ManyToManyField(
+        Citation,
+        blank=True,
+        related_name='editors')
     born = models.DateField(
         blank=True, null=True)
     foto = models.ImageField(upload_to=filepath, blank=True)
@@ -218,6 +250,27 @@ class Subsubsection(CommonTextPart):
     section = models.ForeignKey(Section, blank=False)
     subsection = models.ForeignKey(Subsection, blank=False)
 
+class MixedContent(models.Model):
+    parent = models.ForeignKey(MixedContent, blank=True)
+    position = model.PositiveIntegerField(blank=False)
+    tagname = models.CharField(**default_char,
+                               choices=tagname_choice,
+                               default='de')
+    attributes = HStoreField(
+        validators=[
+            KeysValidator(
+                validkeys,
+                strict=True,
+                messages = {
+                    'missing_keys': _('Some keys were missing: %(keys)s'),
+                    'extra_keys': _('Some unknown keys were provided: %(keys)s'),
+                    }
+                )
+            ]
+        )
+
+    class Meta:
+        abstract = True
 
 class Paragraph(models.Model):
     publication = models.ForeignKey(Publication)
@@ -227,10 +280,22 @@ class Paragraph(models.Model):
     subsection = models.ForeignKey(Subsection, blank=True, null=True)
     subsubsection = models.ForeignKey(Subsubsection, blank=True, null=True)
     idstring = models.CharField(max_length=100,blank=True)
-    text = models.TextField(blank=False)
+    text = models.ForeignKey(MixedContent,blank=False)
 
     def __str__(self):
-       return (self.publication.__str__() + ' ' + str(self.text)[:30])
+       return (self.publication.__str__() + ' ' + str(self.idstring))
 
     class Meta:
         ordering = ('idstring',)
+
+class Text(models.Model):
+    parent = models.ForeignKey(MixedContent)
+    position = models.PositiveIntegerField(blank=False)
+    text = models.TextField(blank=True)
+    pass
+
+class Table(models.Model):
+    pass
+
+class Figure(models.Model):
+    pass

publications/management/commands/_importer_class.py

@@ -215,6 +215,64 @@ def paragraphs(self):
                 pass
         return ret
 
+    def biblio(self):
+        ret = []
+        for path in self.returnPath(['div','body'],self.dataDict):
+            for result in self.getFromDict(self.dataDict,path):
+                try:
+                    if result['head']['#text'] == 'Bibliography':
+                        for bib in result['listBibl']['biblStruct']:
+                            #print(bib.keys())
+                            retBib = {}
+                            retBib['id'] = bib['@id']
+                            retBib['pubtype'] = bib['@type']
+                            try:
+                                retBib['author'] = bib['analytic']['author']['surname'] + ',' + bib['analytic']['author']['forename']
+                                retBib['titlea'] = bib['analytic']['title']['#text']
+                            except:
+                                pass
+                            try:
+                                retBib['author'] = bib['monogr']['author']['surname'] + ',' + bib['monogr']['author']['forename']
+                            except:
+                                pass
+                            try:
+                                retBib['editor'] = bib['monogr']['editor']['surname'] + ',' + bib['monogr']['editor']['forename']
+                            except:
+                                pass
+                            try:
+                                retBib['titlem'] = bib['monogr']['title']['#text']
+                            except:
+                                pass
+                            for key in ['publisher','pubPlace','date']:
+                                try:
+                                    retBib[key] = (bib['monogr']['imprint'][key])
+                                except:
+                                    pass
+                            try:
+                                retBib['note'] = bib['monogr']['imprint']['note']['p']['#text']
+                            except:
+                                pass
+                            try:
+                                resTemp = bib['monogr']['imprint']['biblScope']
+                                if type(resTemp)==list:
+                                    for dic in resTemp:
+                                        retBib[dic['@unit']] = dic['#text']
+                                elif isinstance(resTemp,dict):
+                                    retBib[resTemp['@unit']] = resTemp['#text']
+                            except:
+                                pass
+                            ret.append(retBib)
+                        return ret
+                except:
+                    pass
+
+########################
+#
+# Helper functions to write to db
+#
+########################
+
+
 def createAuthors(text,publication):
     authorObjects = []
     authors = text.authors()
@@ -339,3 +397,12 @@ def createParagraphs(text,publication):
             )
         parList.append(parTemp)
     return parList
+
+def createCitations(text,publication):
+    citList = []
+    for cit in text.biblio():
+        citTemp, created = Citation.objects.update_or_create(
+            **cit
+            )
+        citList.append(citTemp)
+    return citList

publications/management/commands/tei_import.py

@@ -57,6 +57,8 @@ def handle(self,*args, **options):
 
         paragraphs = createParagraphs(text,pub)
 
+        citations = createCitations(text,pub)
+
         self.stdout.write(
             'Successfully read TEI file, by {0}'.format(text.authors())
                 )