Parse Sources file to extract information about the source tarball url

2 files changed, 48 insertions, 0 deletions

diff --git a/export.py b/export.py
index 8bd72bd..3a93602 100644
--- a/export.py
+++ b/export.py
@@ -52,6 +52,7 @@ class PkgData(object):
         pkg_store = pd.HDFStore('pkg.h5')
         self.pkgs = pkg_store['packages']
         self.descs = pkg_store['descriptions']
+        self.srcs = pkg_store['sources']
         pkg_store.close()
 
         cp_store = pd.HDFStore('cp.h5')
@@ -224,6 +225,15 @@ def export_srcpkgs(data, name, srcpkg_names):
         for srcpkg in srcpkg_names])
     versions = {}
 
+    srcfiles=""
+    for srcpkg in srcpkg_names:
+        srcfile = data.srcs[data.srcs['_srcpkg'] == srcpkg]['srcfile'].values[0]
+        letter = srcfile[0]
+        if srcfile[:3] == 'lib':
+            letter = srcfile[:4]
+        srcfile = 'http://ftp.debian.org/debian/pool/main/%s/%s/%s' % (letter,srcpkg,srcfile)
+        srcfiles = srcfiles + srcfile + " "
+
     for (_i, pkg) in binpkgs.iterrows():
         versions[pkg['_srcpkg']] = pkg['Version']
 
@@ -267,6 +277,8 @@ def export_srcpkgs(data, name, srcpkg_names):
         ('Computer languages', ', '.join(langs)),
         ('Status', ''),
         ('Is GNU', 'No'),
+        ('Version identifier', pkg['Version']),
+        ('Version download', srcfiles),
         ('Submitted by', 'Debian import'),
         ('Submitted date', today())])
 
diff --git a/load_sources.py b/load_sources.py
new file mode 100644
index 0000000..f420796
--- /dev/null
+++ b/load_sources.py
@@ -0,0 +1,36 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import sys
+
+import debian.deb822
+import pandas as pd
+
+def pkg(para):
+    d = dict(para)
+    a = dict()
+
+    if 'Files' in d:
+        a['srcfile']=d['Files'].split('\n')[2].split(' ')[3]
+
+    if 'Source' in d:
+        # Source fields sometimes have the source version number; strip it.
+        a['_srcpkg'] = d['Source'].split(' ')[0]
+    else:
+        # No 'Source' field means that it has the same value as the 'Package'
+        # field.
+        a['_srcpkg'] = d['Package']
+
+    return a
+
+if __name__ == '__main__':
+    packages = debian.deb822.Packages.iter_paragraphs(sys.stdin)
+    df = pd.DataFrame([pkg(p) for p in packages])
+    store = pd.HDFStore('pkg.h5')
+
+    print df
+
+    store = pd.HDFStore('pkg.h5')
+    store['sources'] = df
+    store.close()
+