Inserting Documents

4 files changed, 91 insertions, 16 deletions

diff --git a/archivist/cli.py b/archivist/cli.py
index 0b8abe1..8238a30 100644
--- a/archivist/cli.py
+++ b/archivist/cli.py
@@ -61,6 +61,12 @@ def list_prefixes():
     for p in Prefix.select():
         print("  * %s (builtin: %s; pseudo: %s)" % (p.name, p.builtin, p.pseudo))
 
+def prefix_tag_name(name, prefix=None):
+    if prefix:
+        return "%s:%s" % (prefix, name)
+    else:
+        return name
+
 class PrefixTag:
     def __init__(self, tag, prefix = None):
         self.tag = tag
@@ -72,6 +78,16 @@ class PrefixTag:
         else:
             return self.tag
 
+    def __iter__(self):
+        yield self.prefix
+        yield self.tag
+
+    def prefixed_name(self):
+        return prefix_tag_name(self.tag, self.prefix)
+
+    def __str__(self):
+        return self.prefixed_name()
+
 class PrefixTagType(click.ParamType):
     name = 'prefixed tag'
 
@@ -87,12 +103,8 @@ class PrefixTagType(click.ParamType):
 
 TAG = PrefixTagType()
 
-@tag.command('add')
-@click.argument('name', type = TAG)
-@click.argument('description', required = False)
-def add_tag(name, description):
+def create_tag(name, description=None):
     from .model import Tag, Prefix, db
-
     with db.atomic():
         if name.prefix:
             prefix, created = Prefix.get_or_create(name = name.prefix)
@@ -100,14 +112,20 @@ def add_tag(name, description):
         if not created and prefix.pseudo:
             raise click.UsageError("Prefix '%s' is not allowed to carry additional tags." % name.prefix)
 
-        tag, created = Tag.create_or_get(name = name.tag, prefix = name.prefix, description = description)
-        if not created:
-            print("Tag already existed:", tag)
+        return Tag.create_or_get(name = name.tag, prefix = name.prefix, description = description)
+
+@tag.command('add')
+@click.argument('name', type = TAG)
+@click.argument('description', required = False)
+def add_tag(name, description):
+    created, tag = create_tag(name, description)
+    if not created:
+        print("Tag already existed:", tag)
 
 @tag.command('edit')
-@click.option('--description')
 @click.argument('name', type = TAG)
-def edit_tag(description, name):
+@click.option('--description')
+def edit_tag(name, description):
     from .model import Tag, Prefix, db
 
     try:
@@ -118,3 +136,42 @@ def edit_tag(description, name):
     if description:
         tag.description = description
         tag.save()
+
+@cli.group()
+def doc():
+    """Document handling"""
+    pass
+
+@doc.command('add')
+@click.argument('file', type=click.File(mode = 'rb'))
+@click.argument('tags', type = TAG, nargs = -1)
+@click.option('--create-tags', '-c', is_flag = True)
+@click.option('--ignore-missing-tags', '-i', is_flag = True)
+def add_doc(file, tags, create_tags, ignore_missing_tags):
+    """Add a new document together with the given tags."""
+
+    from .model import Document, Tag, db, DocumentTag
+    from .peewee_ext import sqlite_tuple_in
+    import magic
+
+    with db.atomic():
+        if tags:
+            if create_tags:
+                tags = [create_tag(tag)[0] for tag in tags]
+            else:
+                fetched_tags = Tag.select().where(sqlite_tuple_in((Tag.prefix, Tag.name), tags)).execute()
+                if len(fetched_tags) < len(tags):
+                    print("Some tags were not present:", ', '.join(set(map(str, tags)).difference(map(str, fetched_tags))))
+                    if not ignore_missing_tags:
+                        raise click.ClickException("Not all tags present")
+                tags = fetched_tags
+
+        mimetype = magic.from_file(file.name, mime=True)
+
+        doc = Document.create(content = file.read(),
+                file_type = mimetype,
+                original_path = file.name,
+                direction = Document.Direction.IN)
+
+        for t in tags:
+            DocumentTag.create(document = doc, tag = t)
diff --git a/archivist/model.py b/archivist/model.py
index fd1bec3..bee6200 100644
--- a/archivist/model.py
+++ b/archivist/model.py
@@ -1,5 +1,5 @@
 from peewee import *
-from playhouse.fields import CompressedField
+from playhouse.fields import CompressedField as _CompressedField
 from playhouse.hybrid import *
 from playhouse.sqlite_ext import SqliteExtDatabase
 
@@ -31,6 +31,13 @@ class BaseModel(Model):
     class Meta:
         database = db
 
+class CompressedField(_CompressedField):
+    def db_value(self, value):
+        return value if value is None else self.compress(value)
+
+    def python_value(self, value):
+        return value if value is None else self.decompress(value)
+
 @table
 class Document(BaseModel):
     @unique
@@ -39,10 +46,11 @@ class Document(BaseModel):
         OUT = 1
 
     content = CompressedField()
-    created = DateField(default=datetime.datetime.now)
+    created = DateField(default=datetime.date.today)
     inserted = DateTimeField(default=datetime.datetime.now)
     description = CharField(null=True)
     original_path = CharField(null=True)
+    file_type = CharField(null=True)
     direction = EnumField(Direction, null=True)
 
     @classmethod
@@ -87,6 +95,7 @@ class Tag(BaseModel):
             (('name', 'prefix'), True)
         ]
 
+    @property
     def prefixed_name(self):
         if self.prefix_id:
             return "%s:%s" % (self.prefix_id, self.name)
@@ -94,12 +103,10 @@ class Tag(BaseModel):
             return self.name
 
     def __str__(self):
-        description = ' -- ' + self.description if self.description else ''
-
-        return self.prefixed_name() + description
+        return self.prefixed_name
 
     def __repr__(self):
-        return "<%s (#%d) %s>" % (self.__class__.__name__, self.id, self.prefixed_name())
+        return "<%s (#%d) %s>" % (self.__class__.__name__, self.id, self.prefixed_name)
 
 @table
 class DocumentTag(BaseModel):
diff --git a/archivist/peewee_ext.py b/archivist/peewee_ext.py
index 9fda66e..80bb6f4 100644
--- a/archivist/peewee_ext.py
+++ b/archivist/peewee_ext.py
@@ -1,6 +1,16 @@
 from playhouse.sqlite_ext import VirtualModel, VirtualIntegerField, VirtualCharField
 from peewee import Field
 
+from itertools import starmap
+from functools import reduce
+import operator as op
+
+def sqlite_tuple_in(fields, values):
+    """SQLite does not support (foo, bar) IN ((1,2),(3,4)).
+    So we construct a '(foo = 1 AND bar = 2) OR (foo = 3 AND bar = 4)' monstrum."""
+    subqueries = (reduce(op.and_, starmap(op.eq, zip(fields, value_tuple))) for value_tuple in values)
+    return reduce(op.or_, subqueries)
+
 class EnumField(Field):
     db_field = 'enum'
 
diff --git a/archivist/prefixes.py b/archivist/prefixes.py
index 4e73ca7..a039485 100644
--- a/archivist/prefixes.py
+++ b/archivist/prefixes.py
@@ -20,6 +20,7 @@ def _create_prefixes():
     
     FieldPseudoPrefix('direction', 'In- or outgoing', Document.direction)
     FieldPseudoPrefix('date', 'Date of creation', Document.created)
+    FieldPseudoPrefix('mime', 'Mime Type', Document.file_type)
     
     for part in ('day', 'month', 'year'):
         FieldPartPseudoPrefix(part, part.capitalize() + ' of creation', Document.created, part)