[WP] (csbusortdirlist): scan lines in file

util/csbusortdirlist.py

@@ -16,7 +16,7 @@ This program takes a directory list like created by csbulistcsfiles.sh and
 creates a sorted table which should support a completeness check.
 """
 # ----------------------------------------------------------------------------
-__version__ = "csbusortdirlist 2016-06-10"
+__version__ = "2016-06-10"
 __author__ = "Thomas Forbriger"
 # ----------------------------------------------------------------------------
 import sys
@@ -24,7 +24,24 @@ import os
 import string
 import getopt
 
+# ----------------------------------------------------------------------------
+
+# exit status for any specific error
+ERR_ERROR=100
+
+# DEBUG output - global variable
+DEBUG=False
+
+# verbose output - global variable
+VERBOSE=False
+
+# name of checksum files
+CHECKSUMFILE='checksumfile.cs'
+
 # ============================================================================
+# elementary functions
+# --------------------
+
 def usage():
     """
     print usage information
@@ -42,7 +59,8 @@ def help():
     print(
 """
 This program takes a directory list like created by csbulistcsfiles.sh and
-creates a sorted table which should support a completeness check.
+creates a sorted table which should support a completeness check. Pathnames
+are expected to contain the string 'checksumfile.cs'.
 
   listfile  list of subdirectory path names
   tablefile name of file to write table to
@@ -52,19 +70,122 @@ creates a sorted table which should support a completeness check.
 
 """)
     print('Call "pydoc csbusortdirlist" to learn more')
+
+# ----------------------------------------------------------------------------
+def debugout(m):
+    """
+    print debugging message
+    """
+    if DEBUG:
+        print("DEBUG: "+ m)
+
+# ----------------------------------------------------------------------------
+def verbose(m):
+    """
+    print verbose message
+    """
+    if VERBOSE:
+        print(m)
+
+# ============================================================================
+class Basetable():
+    """
+    Base class for table objects.
+    """
+
+# ============================================================================
+class SEEDlinktable(Basetable):
+    """
+    Class to hold a seedlink data structure
+    """
+
+# ============================================================================
+class UIPCtable(Basetable):
+    """
+    Class to hold a UIPC data structure
+    """
+
 # ============================================================================
+class DL1table(Basetable):
+    """
+    Class to hold a DL1 data structure
+    """
+
+# ============================================================================
+class HPMOtable(Basetable):
+    """
+    Class to hold a HPMO data structure
+    """
+
+# ============================================================================
+class PathLine():
+    """
+    Class to hold and parse a sinlge input line.
+    """
+
+    def __init__(self, line):
+        """
+        Constructor takes an input line string
+        """
+        self.datatype='Unknown'
+        for substring in line.split():
+            if substring.find(CHECKSUMFILE) > 0:
+                self.pathname=substring
+                if substring.find('seedlink') > 0:
+                    self.datatype='seedlink'
+                elif substring.find('DL1') > 0:
+                    self.datatype='DL1'
+                elif substring.find('HPMO') > 0:
+                    self.datatype='HPMO'
+                elif substring.find('UIPC') > 0:
+                    self.datatype='UIPC'
+
+# ============================================================================
+# processing functions
+# --------------------
+def readlist(filename):
+    """
+    Read a list of subdirectory path names and return a table object
+    """
+    debugout("entered function readlist")
+    verbose('read file %s' % filename)
+
+    for line in open(filename, 'r'):
+        inputline=line.rstrip('\n')
+        pathline=PathLine(inputline)
+        if pathline.datatype == 'seedlink':
+            debugout("seedlink")
+            debugout(pathline.pathname)
+        elif pathline.datatype == 'DL1':
+            debugout("DL1")
+            debugout(pathline.pathname)
+        elif pathline.datatype == 'HPMO':
+            debugout("HPMO")
+            debugout(pathline.pathname)
+        elif pathline.datatype == 'UIPC':
+            debugout("UIPC")
+            debugout(pathline.pathname)
+        else:
+            verbose("input line of unknown data type:")
+            verbose(line)
+
+    debugout("finished function readlist")
+
 # ============================================================================
 def main(argv=None):
+    """
+    Then main function
+    """
     argv = sys.argv
 
     (opts, args)=getopt.gnu_getopt(sys.argv[1:], 'voDh', ('help'))
    
-    verbose=False
     overwrite=False
    
     for opt in opts:
       if (opt[0]=='-v'):
-        verbose=True
+        global VERBOSE
+        VERBOSE=True
       elif (opt[0]=='-D'):
         global DEBUG
         DEBUG=True
@@ -77,13 +198,15 @@ def main(argv=None):
         usage()
         exit(0)
    
-    if len(args) < 1:
+    if len(args) < 2:
       usage()
       exit(0)
    
     inputfilename=args[0]
     outputfilename=args[1]
 
+    tabledata=readlist(inputfilename)
+   
     # check for existing file
     # if an exception is raised here, this is just fine: we do not want to
     # overwrite an existing file
@@ -91,8 +214,7 @@ def main(argv=None):
       try:
         output=open(outputfilename, 'r')
       except IOError:
-        if verbose:
-          print('create new file %s' % outputfilename)
+        verbose('create new file %s' % outputfilename)
       else:
         print('ERROR: output file %s exists' % outputfilename)
         exit(ERR_ERROR)