Allow configuring html_dir via the command line and fix run_example.

Support a directory argument of "." to allow outputting in the current
directory with an automatic suffix: ./_autoarchaologist

Author: alexjeffburke

.gitignore

@@ -8,3 +8,4 @@ docs/_build/
 
 # Temporary files
 _.*
+_autoarchaologist/

autoarchaeologist/ddhf/bitstore.py

@@ -8,8 +8,6 @@
 import mmap
 import urllib.request
 
-import ddhf_bitstore_metadata
-
 from ..base import artifact
 from ..base import excavation
 from ..container import simh_tap_file
@@ -165,10 +163,6 @@ def fetch_single(self, arg):
             print("Could not fetch", arg)
             return
 
-        meta = ddhf_bitstore_metadata.internals.metadata.MetadataBase(metatxt)
-        if meta is None:
-            return
-
         if self.media_types and not self.check_media_type(meta):
             return
 

autoarchaeologist/ddhf/decorated_context.py

@@ -6,6 +6,7 @@
 
 '''
 
+import argparse
 import os
 
 from ..base import excavation
@@ -84,13 +85,29 @@ def from_bitstore(self, *args, **kwargs):
     "AUTOARCHAEOLOGIST_BITSTORE_CACHE": "ddhf_bitstore_cache",
 }
 
-def main(job, html_subdir="tmp", **kwargs):
+def parse_arguments(argv=None):
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-o', '--out', default='/tmp/_autoarchaologist')
+
+    args = parser.parse_args(args=argv)
+    if args.out == '.':
+        args.out = os.path.join(os.getcwd(), "_autoarchaologist")
+    return args
+
+def main(job, html_subdir, **kwargs):
+    args = parse_arguments()
+    kwargs["html_dir"] = args.out
+
     ''' A standard main routine to reduce boiler-plate '''
     for key in os.environ:
         i = OK_ENVS.get(key)
         if i:
             kwargs[i] = os.environ[key]
 
+    if 'html_dir' not in kwargs:
+        raise AttributeError("missing: html_dir")
+
+
     kwargs['html_dir'] = os.path.join(kwargs['html_dir'], html_subdir)
     kwargs.setdefault('download_links', True)
     kwargs.setdefault('download_limit', 1 << 20)

run_example.py

@@ -1,5 +1,7 @@
 
+import argparse
 import os
+import sys
 
 import autoarchaeologist
 
@@ -8,10 +10,24 @@
 from autoarchaeologist.data_general.absbin import AbsBin
 from autoarchaeologist.data_general.papertapechecksum import DGC_PaperTapeCheckSum
 
+def parse_arguments(argv=None):
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-d", "--dir", default="/tmp/_autoarchaologist")
+
+    args = parser.parse_args(args=argv)
+    if args.dir == ".":
+        args.dir = os.path.join(os.getcwd(), "_autoarchaologist")
+    return args
 
 if __name__ == "__main__":
+    args = parse_arguments()
+
+    try:
+        os.mkdir(args.dir)
+    except FileExistsError:
+        pass
 
-    ctx = autoarchaeologist.Excavation()
+    ctx = autoarchaeologist.Excavation(html_dir=args.dir)
 
     ctx.add_examiner(BigDigits)
     ctx.add_examiner(AbsBin)
@@ -22,11 +38,6 @@
 
     ctx.start_examination()
 
-    try:
-        os.mkdir("/tmp/_autoarchaologist")
-    except FileExistsError:
-        pass
-
-    ctx.produce_html(html_dir="/tmp/_autoarchaologist")
+    ctx.produce_html()
 
     print("Now point your browser at", ctx.filename_for(ctx).link)