Added option to parse scan info from filename

git-svn-id: http://comictagger.googlecode.com/svn/trunk@592 6c5673fe-1810-88d6-992b-cd32ca31540c

comictaggerlib/comicarchive.py

@@ -1027,6 +1027,9 @@ class ComicArchive:
 			metadata.year = fnp.year
 		if fnp.issue_count != "":
 			metadata.issueCount = fnp.issue_count
+		if self.settings.parse_scan_info:
+			if fnp.remainder != "":
+				metadata.scanInfo = fnp.remainder
 
 		metadata.isEmpty = False
 

comictaggerlib/filenameparser.py

@@ -81,7 +81,7 @@ class FileNameParser:
 		found = False
 		issue = ''
 		
-		# first, look for multiple "--", this mean's it's formatted differently from most:
+		# first, look for multiple "--", this means it's formatted differently from most:
 		if "--" in filename:
 			# the pattern seems to be that anything to left of the first "--" is the series name followed by issue
 			filename = filename.split("--")[0]
@@ -191,6 +191,28 @@ class FileNameParser:
 			year = re.sub("[^0-9]", "", year)
 		return year
 
+	def getRemainder( self, filename, year, count ):
+		#make a guess at where the the non-interesting stuff begins
+		
+		remainder = ""
+		
+		if "--" in filename:
+			remainder = filename.split("--",1)[1]
+		elif "__" in filename:
+			remainder = filename.split("__",1)[1]
+		elif "(" in filename:
+			remainder = "(" + filename.split("(",1)[1]
+
+		remainder = self.fixSpaces(remainder, remove_dashes=False)
+		if year != "":
+			remainder = remainder.replace(year,"",1)
+		if count != "":
+			remainder = remainder.replace("of "+count,"",1)
+			
+		remainder = remainder.replace("()","")
+		
+		return remainder.strip()
+		
 	def parseFilename( self, filename ):
 
 		# remove the path
@@ -208,20 +230,12 @@ class FileNameParser:
 		if filename.count("_28") > 1 and filename.count("_29") > 1:
 			filename = filename.replace("_28", "(")
 			filename = filename.replace("_29", ")")
-		
-		# ----HACK  
-		# remove the first word that word is a 3 digit number.
-		# some story arcs collection packs do this, but it's ugly
-		# this will probably break something, i.e. "100 bullets"
-		#word = filename.split(' ')[0]
-		#if len(word) == 3 and word[0] =='0' and word.isdigit():
-		#	filename = filename[4:]
-		# ----HACK  -
 					
 		self.issue = self.getIssueNumber(filename)
 		self.series, self.volume = self.getSeriesName(filename, self.issue)
 		self.year = self.getYear(filename)
 		self.issue_count = self.getIssueCount(filename)
+		self.remainder = self.getRemainder( filename, self.year, self.issue_count )
 	
 		if self.issue != "":
 			# strip off leading zeros

comictaggerlib/settings.py

@@ -93,6 +93,9 @@ class ComicTaggerSettings:
 		self.dont_notify_about_this_version = ""
 		self.ask_about_usage_stats = True
 		
+		#filename parsing settings
+		self.parse_scan_info = False
+		
 		# Comic Vine settings
 		self.use_series_start_as_volume = False
 		
@@ -208,6 +211,9 @@ class ComicTaggerSettings:
 		if self.config.has_option('identifier', 'id_publisher_blacklist'):
 			self.id_publisher_blacklist =   self.config.get( 'identifier', 'id_publisher_blacklist' )
 
+		if self.config.has_option('filenameparser', 'parse_scan_info'):
+			self.parse_scan_info =   self.config.getboolean( 'filenameparser', 'parse_scan_info' )
+
 		if self.config.has_option('dialogflags', 'ask_about_cbi_in_rar'):
 			self.ask_about_cbi_in_rar =        self.config.getboolean( 'dialogflags', 'ask_about_cbi_in_rar' )		
 		if self.config.has_option('dialogflags', 'show_disclaimer'):
@@ -283,6 +289,11 @@ class ComicTaggerSettings:
 		self.config.set( 'dialogflags', 'show_disclaimer', self.show_disclaimer )
 		self.config.set( 'dialogflags', 'dont_notify_about_this_version', self.dont_notify_about_this_version )
 		self.config.set( 'dialogflags', 'ask_about_usage_stats', self.ask_about_usage_stats )
+
+		if not self.config.has_section( 'filenameparser' ):
+			self.config.add_section( 'filenameparser' )
+			
+		self.config.set( 'filenameparser', 'parse_scan_info', self.parse_scan_info )
 	
 		if not self.config.has_section( 'comicvine' ):
 			self.config.add_section( 'comicvine' )

comictaggerlib/settingswindow.py

@@ -122,6 +122,9 @@ class SettingsWindow(QtGui.QDialog):
 		if self.settings.check_for_new_version:
 			self.cbxCheckForNewVersion.setCheckState( QtCore.Qt.Checked)
 		
+		if self.settings.parse_scan_info:
+			self.cbxParseScanInfo.setCheckState( QtCore.Qt.Checked)
+			
 		if self.settings.use_series_start_as_volume:
 			self.cbxUseSeriesStartAsVolume.setCheckState( QtCore.Qt.Checked)
 	
@@ -170,7 +173,9 @@ class SettingsWindow(QtGui.QDialog):
 		
 		self.settings.id_length_delta_thresh = int(self.leNameLengthDeltaThresh.text())
 		self.settings.id_publisher_blacklist = str(self.tePublisherBlacklist.toPlainText())
-		
+			
+		self.settings.parse_scan_info = self.cbxParseScanInfo.isChecked()
+
 		self.settings.use_series_start_as_volume = self.cbxUseSeriesStartAsVolume.isChecked()
 				
 		self.settings.assume_lone_credit_is_primary = self.cbxAssumeLoneCreditIsPrimary.isChecked()
@@ -230,5 +235,5 @@ class SettingsWindow(QtGui.QDialog):
 			control.setText( str(fileList[0]) )
 
 	def showRenameTab( self ):
-		self.tabWidget.setCurrentIndex(4)	
+		self.tabWidget.setCurrentIndex(5)	
 		

comictaggerlib/ui/settingswindow.ui

@@ -314,6 +314,24 @@
          </item>
         </layout>
        </widget>
+       <widget class="QWidget" name="tab_6">
+        <attribute name="title">
+         <string>Filename Parser</string>
+        </attribute>
+        <widget class="QCheckBox" name="cbxParseScanInfo">
+         <property name="geometry">
+          <rect>
+           <x>30</x>
+           <y>30</y>
+           <width>421</width>
+           <height>25</height>
+          </rect>
+         </property>
+         <property name="text">
+          <string>Parse Scan Info From Filename (Experimental)</string>
+         </property>
+        </widget>
+       </widget>
        <widget class="QWidget" name="tab_3">
         <attribute name="title">
          <string>Comic Vine</string>