Make akabei-db-import more testable: move main functionality in separate...

Make akabei-db-import more testable: move main functionality in separate function, have the package parser return errors and ignore malformed packages instead of just exiting on the spot. Runned again and it works.

scripts/akabei-db-import/akabei-db-import.py

@@ -91,6 +91,9 @@ def parseLocalDatabase(path):
             
             # Calls the package parser
             parser = PacmanPackageParser(desc, install, files)
+            # Ignore packages that we could not parse successfully.
+            if not parser.parsePackage():
+                continue
             packages.append(parser.getPackage())
     
     # show last progress in case the progress never became exactly 100 (almost always)
@@ -109,6 +112,27 @@ def usage():
     print "    -o, --db-output <file> Prints the Akabei database on <file> (default: local.db)"
     print "    -s, --db-scheme <file> Reads the database scheme from <file> (default: /usr/share/akabei/dbschema.sql)"
 
+def DoConversion(pacmanConf, outputFile, schemeFile):
+    # Gets the database path from the pacman configuration
+    print "[**] Reading Pacman database path from " + pacmanConf + "...",
+    dbpath = getDatabasePath(pacmanConf)
+    print "done."
+    if len(dbpath) == 0:
+        dbpath = "/var/lib/pacman" # if not explicitly set there, then go back to the default value
+    
+    localPath = os.path.join(dbpath, "local")
+    
+    # Get local packages from pacman, then write them into the akabei database in the right format
+    
+    print "[**] Parsing Pacman local database..."
+    localPackages = parseLocalDatabase(localPath)
+    print "done."
+    
+    dbwriter = AkabeiDbWriter(outputFile, schemeFile)
+    dbwriter.addPackages(localPackages)
+    
+    print "[**] Database converted successfully. Now enjoy akabei!"
+
 # Start point
 if __name__ == "__main__":
     
@@ -145,22 +169,4 @@ if __name__ == "__main__":
     print "This script may take some time to finish."
     print
    
-    # Gets the database path from the pacman configuration
-    print "[**] Reading Pacman database path from " + pacmanConf + "...",
-    dbpath = getDatabasePath(pacmanConf)
-    print "done."
-    if len(dbpath) == 0:
-        dbpath = "/var/lib/pacman" # if not explicitly set there, then go back to the default value
-    
-    localPath = os.path.join(dbpath, "local")
-    
-    # Get local packages from pacman, then write them into the akabei database in the right format
-    
-    print "[**] Parsing Pacman local database..."
-    localPackages = parseLocalDatabase(localPath)
-    print "done."
-    
-    dbwriter = AkabeiDbWriter(outputFile, schemeFile)
-    dbwriter.addPackages(localPackages)
-    
-    print "[**] Database converted successfully. Now enjoy akabei!"
+    DoConversion(pacmanConf, outputFile, schemeFile)

scripts/akabei-db-import/akabeipackageparser.py

@@ -16,34 +16,48 @@ import os
 class PacmanPackageParser:
     
     def __init__(self, d, i, f):
+        self.descFileName = d
+        self.installFileName = i
+        self.filesFileName = f
+
         self.desc = None
         self.install = None
         self.files = None
-        
-        if len(d) > 0:
-            self.desc = open(d, "r")
-        else:
-            print "[!!] Error: this package doesn't contain a desc file."
-            print "[!!] Your local database may be broken. Quitting now..."
-            sys.exit(-1)
-            
-        if len(i) > 0:
-            self.install = open(i, "r")
-            
-        if len(f) > 0:
-            self.files = open(f, "r")
+
+    def parsePackage(self):
+        """Main public API. Returns True/False for success/failure."""
+
+        if not self.descFileName:
+            print("[!!] Found malformed package: no desc file.")
+            return False
+
+        self.desc = open(self.descFileName, "r")
+
+        if self.installFileName:
+            self.install = open(self.installFileName, "r")
+        if self.filesFileName:
+            self.files = open(self.filesFileName, "r")
         
         self.package = AkabeiPackage()
         
         # Used by the parse* functions
         self.state = ""
         
-        self.parseDescription()
+        result = self.parseDescription()
+        # These are both optional files.
         self.parseInstall()
         self.parseFiles()
+        return result
         
     def parseDescription(self):
+        """Returns success/failure."""
+
         self.state = ""
+
+        # Two required parameters for a package: NAME and
+        # VERSION. If at least one is missing, returns False.
+        hasName = False
+        hasVersion = False
         
         for line in self.desc.readlines():
             line = line[:-1] # delete newline at the end
@@ -58,8 +72,10 @@ class PacmanPackageParser:
             else:
                 if self.state == "NAME":
                     self.package.name = line
+                    hasName = True
                 elif self.state == "VERSION":
                     self.package.version = line
+                    hasVersion = True
                 elif self.state == "DESC":
                     self.package.description = line
                 elif self.state == "GROUPS":
@@ -90,7 +106,14 @@ class PacmanPackageParser:
                 elif self.state == "REPLACES":
                     self.package.addReplace(line)
                 elif self.state == "CONFLICTS":
-                    self.package.addConflict(line)                        
+                    self.package.addConflict(line)
+
+        if not hasName:
+            print "[!!] Found malformed package with no name."
+        if not hasVersion:
+            print "[!!] Found malformed package with no version."
+
+        return hasName and hasVersion                  
 
     # Installation scriptlet
     def parseInstall(self):