Implement --reject-log and --reject-data options.

pgloader.1.txt

@@ -104,6 +104,18 @@ refers to a PostgreSQL table into which some data is to be loaded.
 
     file where to log messages, see +log_min_messages+.
 
+-r, --reject-log::
+
+    Filename, with a single "%s" placeholder, where to store the bad data
+    logs (that's the error messages given by PostgreSQL). If you want to put
+    a percent in the file name, write it '%%'.
+
+-j, --reject-data::
+
+    Filename, with a single "%s" placeholder, where to store the bad data
+    (the exact lines that didn't make it from your input file).  If you want
+    to put a percent in the file name, write it '%%'.
+
 -s, --summary::
 
     makes pgloader print a 'nice' summary at the end of operations.

pgloader.py

@@ -58,6 +58,14 @@ def parse_options():
                       default = "/tmp/pgloader.log",
                       help    = "log file, defauts to /tmp/pgloader.log")
 
+    parser.add_option("-r", "--reject-log", dest = "reject_log",
+                      default = None,
+                      help    = "log file for rejected data error messages")
+
+    parser.add_option("-j", "--reject-data", dest = "reject_data",
+                      default = None,
+                      help    = "log file for rejected data, bad input data")
+
     parser.add_option("-s", "--summary", action = "store_true",
                       dest    = "summary",
                       default = False,
@@ -210,6 +218,30 @@ def parse_options():
 
     pgloader.options.LOG_FILE = opts.logfile
 
+    # reject file names must contain one %s which will get replaced by the
+    # section name
+    if opts.reject_log:
+        try:
+            unused = opts.reject_log % "that would be a section name"
+        except TypeError, e:
+            # TypeError: not all arguments converted during string formatting
+            # TypeError: not enough arguments for format string
+            print >>sys.stderr, \
+                "Error: reject log must contain a '%s' place holder for section name"
+            sys.exit(1)
+        pgloader.options.REJECT_LOG_FILE = opts.reject_log
+
+    if opts.reject_data:
+        try:
+            unused = opts.reject_data % "that would be a section name"
+        except TypeError, e:
+            # TypeError: not all arguments converted during string formatting
+            # TypeError: not enough arguments for format string
+            print >>sys.stderr, \
+                "Error: reject data must contain a '%s' place holder for section name"
+            sys.exit(1)
+        pgloader.options.REJECT_DATA_FILE = opts.reject_data
+
     if opts.loglevel:
         loglevel = pgloader.logger.level(opts.loglevel)
         pgloader.options.CLIENT_MIN_MESSAGES = loglevel

pgloader/options.py

@@ -50,3 +50,5 @@ LOG_MIN_MESSAGES    = DEBUG
 DEFAULT_LOG_FILE    = "/tmp/pgloader.log"
 LOG_FILE            = None
 
+REJECT_LOG_FILE  = '%s.rej.log'
+REJECT_DATA_FILE = '%s.rej'

pgloader/pgloader.py

@@ -21,6 +21,7 @@ from options import COPY_SEP, FIELD_SEP, CLOB_SEP, NULL, EMPTY_STRING
 from options import NEWLINE_ESCAPES
 from options import UDC_PREFIX
 from options import REFORMAT_PATH
+from options import REJECT_LOG_FILE, REJECT_DATA_FILE
 from options import MAX_PARALLEL_SECTIONS
 from options import DEFAULT_SECTION_THREADS, SECTION_THREADS, SPLIT_FILE_READING
 from options import RRQUEUE_SIZE
@@ -242,10 +243,10 @@ class PGLoader(threading.Thread):
                 self.reject_data = config.get(name, 'reject_data')
 
             if not self.template and 'reject_log' not in self.__dict__:
-                self.reject_log = os.path.join('/tmp', '%s.rej.log' % name)
+                self.reject_log = os.path.join('/tmp', REJECT_LOG_FILE % name)
 
             if not self.template and 'reject_data' not in self.__dict__:
-                self.reject_data = os.path.join('/tmp', '%s.rej' % name)
+                self.reject_data = os.path.join('/tmp', REJECT_DATA_FILE % name)
 
             # reject logging
             if not self.template: