Changeset 221

Timestamp:

19-05-07 00:04:20 (6 years ago)

Author:

thomas

Message:

• moap/command/doap.py (Search, Search.addOptions, Search.handleOptions, Search.do, Search.foundURL, Doap): Implement moap doap search using either google or yahoo to do a search for your project's page rank based on your keyword query.
• moap/util/deps.py (genshi, genshi.fedora_install, google, google.fedora_install, yahoo, yahoo.fedora_install, handleImportError): Add deps for google and yahoo. Fix genshi dep.

Location:

trunk

Files:

• ChangeLog (modified) (1 diff)
• moap/command/doap.py (modified) (2 diffs)
• moap/util/deps.py (modified) (4 diffs)

trunk/ChangeLog

@@ +1 @@
+2007-05-19  Thomas Vander Stichele  <thomas at apestaart dot org>
+
+        * moap/command/doap.py (Search, Search.addOptions,
+          Search.handleOptions, Search.do, Search.foundURL, Doap):
+          Implement moap doap search using either google or yahoo
+          to do a search for your project's page rank based on your
+          keyword query.
+        * moap/util/deps.py (genshi, genshi.fedora_install, google,
+          google.fedora_install, yahoo, yahoo.fedora_install,
+          handleImportError):
+          Add deps for google and yahoo.  Fix genshi dep.
+
 2007-05-16  Thomas Vander Stichele  <thomas at apestaart dot org>
 

trunk/moap/command/doap.py

@@ -125 +125 @@
                 self.stderr.write("ERROR: %r\n" % e)
 
+class Search(util.LogCommand):
+    description = "look up rank of project's home page based on keywords"
+
+    _engines = ["google", "yahoo"]
+    _default = "yahoo"
+
+    def addOptions(self):
+        self.parser.add_option('-e', '--engine',
+            action="store", dest="engine", default=self._default,
+            help="search engine to use (out of %s; defaults to %s)" % (
+                ", ".join(self._engines), self._default))
+        self.parser.add_option('-l', '--limit',
+            action="store", dest="limit", default="100",
+            help="maximum number of results to look at")
+
+    def handleOptions(self, options):
+        self._limit = int(options.limit)
+        self._engine = options.engine
+
+    def do(self, args):
+        if not args:
+            self.stderr.write('Please provide a search query.\n')
+            return 3
+
+        d = self.parentCommand.doap
+        project = d.getProject()
+
+        rank = 0
+        found = False
+        query = " ".join(args)
+
+        def foundURL(target, url):
+            # returns true if the url is close enough to the target
+            if target.endswith('/'):
+                target = target[:-1]
+            if url.endswith('/'):
+                url = url[:-1]
+            return url == target
+
+        if self._engine == 'google':
+            from pygoogle import google
+
+            while not found:
+                self.debug('Doing Google search for %s starting from %d' % (
+                    " ".join(args), rank))
+                value = google.doGoogleSearch(" ".join(args), start=rank)
+                for result in value.results:
+                    rank += 1
+                    self.debug('Hit %d: URL %s' % (rank, result.URL))
+                    if foundURL(project.homepage, result.URL):
+                        found = True
+                        break
+
+                if rank >= self._limit:
+                    break
+
+        elif self._engine == 'yahoo':
+            from yahoo.search import web
+
+            # yahoo's start is 1-based
+            while not found:
+                search = web.WebSearch('moapmoap', query=query, start=rank+1)
+                info = search.parse_results()
+                for result in info.results:
+                    rank += 1
+                    self.debug('Hit %d: URL %s' % (rank, result['Url']))
+                    if foundURL(project.homepage, result['Url']):
+                        found = True
+                        break
+
+                if rank >= self._limit:
+                    break
+
+        else:
+            self.stderr.write("Unknown search engine '%s'.\n" % self._engine)
+            self.stderr.write("Please choose from %s.\n" %
+                ", ".join(self._engines))
+            return 3
+
+        if found:
+            self.stdout.write("Found homepage as hit %d\n." % rank)
+        else:
+            self.stdout.write("Did not find homepage in first %d hits.\n" %
+                self._limit)
+
 class Ical(util.LogCommand):
     description = "Output iCal stream from project releases"
@@ -366 +451 @@
     usage = "doap [doap-options] %command"
     description = "read and act on DOAP file"
-    subCommandClasses = [Freshmeat, Ical, Mail, Rss, Show, bug.Bug]
+    subCommandClasses = [Freshmeat, Ical, Mail, Rss, Search, Show, bug.Bug]
 
     doap = None

trunk/moap/util/deps.py

@@ -40 +40 @@
         return "Your version of Fedora does not have python-redland available."
 
-
 class genshi(Dependency):
-    module = 'genshii'
+    module = 'genshi'
     name = "Genshi templating language"
     homepage = "http://genshi.edgewall.com/"
@@ -48 +47 @@
     def fedora_install(self, distro):
         return "genshi is not yet available in Fedora Extras.\n"
+
+class google(Dependency):
+    module = 'google'
+    name = "A Python Interface to the Google API"
+    homepage = "http://pygoogle.sourceforge.net/"
+
+    def fedora_install(self, distro):
+        return "pygoogle is not yet available in Fedora Extras.\n"
+
+class yahoo(Dependency):
+    module = 'yahoo'
+    name = "A Python Interface to the Yahoo Web API"
+    homepage = "http://developer.yahoo.com/python/"
+
+    def fedora_install(self, distro):
+        return "python-yahoo is not yet available in Fedora Extras.\n"
 
 def handleImportError(exception):
@@ -60 +75 @@
     module = module.split('.')[0]
     deps = {}
-    for dep in [RDF(), genshi()]:
+    for dep in [RDF(), genshi(), google()]:
         deps[dep.module] = dep
 
@@ -92 +107 @@
         sys.stderr.write('import %s\n' % module)
 
-
         return
 
+    # re-raise if we didn't have it
     raise