Improve error reporting from uscan

by parsing out the warnings and error fields from the dehs output.

1 files changed, 112 insertions, 52 deletions

diff --git a/gbp/deb/uscan.py b/gbp/deb/uscan.py
index 76bde653..731ea78b 100644
--- a/gbp/deb/uscan.py
+++ b/gbp/deb/uscan.py
@@ -45,10 +45,6 @@ class Uscan(object):
         @type out: string
 
         >>> u = Uscan('http://example.com/')
-        >>> u._parse('<status>up to date</status>')
-        >>> u.tarball
-        >>> u.uptodate
-        True
         >>> u._parse('<target>virt-viewer_0.4.0.orig.tar.gz</target>')
         >>> u.tarball
         '../virt-viewer_0.4.0.orig.tar.gz'
@@ -60,59 +56,116 @@ class Uscan(object):
         UscanError: Couldn't find 'upstream-url' in uscan output
         """
         source = None
+        self._uptodate = False
+
+        # Check if uscan downloaded something
+        for row in out.split("\n"):
+            # uscan >= 2.10.70 has a target element:
+            m = re.match(r"<target>(.*)</target>", row)
+            if m:
+                source = '../%s' % m.group(1)
+                break
+            elif row.startswith('<messages>'):
+                m = re.match(r".*symlinked ([^\s]+) to it", row)
+                if m:
+                    source = "../%s" % m.group(1)
+                    break
+                m = re.match(r"Successfully downloaded updated package "
+                             "([^<]+)", row)
+                if m:
+                    source = "../%s" % m.group(1)
+                    break
+        # Try to determine the already downloaded sources name
+        else:
+            d = {}
+
+            try:
+                for row in out.split("\n"):
+                    for n in ('package',
+                              'upstream-version',
+                              'upstream-url'):
+                        m = re.match("<%s>(.*)</%s>" % (n,n), row)
+                        if m:
+                            d[n] = m.group(1)
+                d["ext"] = os.path.splitext(d['upstream-url'])[1]
+                # We want the name of the orig tarball if possible
+                source = ("../%(package)s_%(upstream-version)s."
+                          "orig.tar%(ext)s" % d)
+
+                # Fall back to the upstream source name otherwise
+                if not os.path.exists(source):
+                    source = "../%s" % d['upstream-url'].rsplit('/',1)[1]
+                    if not os.path.exists(source):
+                        raise UscanError("Couldn't find tarball at '%s'" %
+                                     source)
+            except KeyError as e:
+                raise UscanError("Couldn't find '%s' in uscan output" %
+                                 e.args[0])
+        self._tarball = source
+
+    def _parse_uptodate(self, out):
+        """
+        Check if the uscan reports that we're up to date.
+
+        @param out: uscan output
+        @type out: string
 
+        >>> u = Uscan('http://example.com/')
+        >>> u._parse_uptodate('<status>up to date</status>')
+        >>> u.tarball
+        >>> u.uptodate
+        True
+        >>> u._parse_uptodate('')
+        >>> u.tarball
+        >>> u.uptodate
+        False
+        """
         if "<status>up to date</status>" in out:
             self._uptodate = True
-            self._tarball = None
-            return
         else:
             self._uptodate = False
 
-            # Check if uscan downloaded something
-            for row in out.split("\n"):
-                # uscan >= 2.10.70 has a target element:
-                m = re.match(r"<target>(.*)</target>", row)
-                if m:
-                    source = '../%s' % m.group(1)
-                    break
-                elif row.startswith('<messages>'):
-                    m = re.match(r".*symlinked ([^\s]+) to it", row)
-                    if m:
-                        source = "../%s" % m.group(1)
-                        break
-                    m = re.match(r"Successfully downloaded updated package "
-                                 "([^<]+)", row)
-                    if m:
-                        source = "../%s" % m.group(1)
-                        break
-
-            # Try to determine the already downloaded sources name
-            else:
-                d = {}
-
-                try:
-                    for row in out.split("\n"):
-                        for n in ('package',
-                                  'upstream-version',
-                                  'upstream-url'):
-                            m = re.match("<%s>(.*)</%s>" % (n,n), row)
-                            if m:
-                                d[n] = m.group(1)
-                    d["ext"] = os.path.splitext(d['upstream-url'])[1]
-                    # We want the name of the orig tarball if possible
-                    source = ("../%(package)s_%(upstream-version)s."
-                              "orig.tar%(ext)s" % d)
-
-                    # Fall back to the upstream source name otherwise
-                    if not os.path.exists(source):
-                        source = "../%s" % d['upstream-url'].rsplit('/',1)[1]
-                        if not os.path.exists(source):
-                            raise UscanError("Couldn't find tarball at '%s'" %
-                                         source)
-                except KeyError as e:
-                    raise UscanError("Couldn't find '%s' in uscan output" %
-                                     e.args[0])
-            self._tarball = source
+    def _raise_error(self, out):
+        r"""
+        Parse the uscan output for errors and warnings and raise
+        a L{UscanError} exception based on this. If no error detail
+        is found a generic error message is used.
+
+        @param out: uscan output
+        @type out: string
+        @raises UscanError: exception raised
+
+        >>> u = Uscan('http://example.com/')
+        >>> u._raise_error("<warnings>uscan warning: "
+        ... "In watchfile debian/watch, reading webpage\n"
+        ... "http://a.b/ failed: 500 Cant connect "
+        ... "to example.com:80 (Bad hostname)</warnings>")
+        Traceback (most recent call last):
+        ...
+        UscanError: Uscan failed: uscan warning: In watchfile debian/watch, reading webpage
+        http://a.b/ failed: 500 Cant connect to example.com:80 (Bad hostname)
+        >>> u._raise_error("<errors>uscan: Can't use --verbose if "
+        ... "you're using --dehs!</errors>")
+        Traceback (most recent call last):
+        ...
+        UscanError: Uscan failed: uscan: Can't use --verbose if you're using --dehs!
+        >>> u = u._raise_error('')
+        Traceback (most recent call last):
+        ...
+        UscanError: Uscan failed - debug by running 'uscan --verbose'
+        """
+        msg = None
+
+        for n in ('errors', 'warnings'):
+            m = re.search("<%s>(.*)</%s>" % (n,n), out, re.DOTALL)
+            if m:
+                msg = "Uscan failed: %s" % m.group(1)
+                break
+
+        if not msg:
+            msg = "Uscan failed - debug by running 'uscan --verbose'"
+        raise UscanError(msg)
+
 
     def scan(self, destdir='..'):
         """Invoke uscan to fetch a new upstream version"""
@@ -121,6 +174,13 @@ class Uscan(object):
                              cwd=self._dir,
                              stdout=subprocess.PIPE)
         out = p.communicate()[0]
-        return self._parse(out)
+        # uscan exits with 1 in case of uptodate and when an error occured.
+        # Don't fail in the uptodate case:
+        self._parse_uptodate(out)
+        if not self.uptodate:
+            if p.returncode:
+                self._raise_error(out)
+            else:
+                self._parse(out)
 
 # vim:et:ts=4:sw=4:et:sts=4:ai:set list listchars=tab\:»·,trail\:·: